EPA
United States
Environmental Protection
Agency
Office of
Research and
Development
Office of Environmental
Processes and Effects Research
Washington. D.C. 20460
June 1980
RESEARCH GUIDELINES FOR
REGIONAL MODELING
OFFINEPARTICULATES,
ACID DEPOSITION AND VISIBILITY
Report of a Workshop
Held at Port Deposit,
Maryland
-------
U.S. Environment.il P--:* Action Agency
l\-~ -.,.-v ROOT; ?'^"t r>V-:?ll-A
40'. *'. Street. S.W.
»m l"'-n«ton. DC 20460
-------
RESEARCH GUIDELINES FOR REGIONAL MODELING OF FINE PARTICULATES,
ACID DEPOSITION AND VISIBILITY
(Report of a Workshop Held at Port Deposit, Maryland,
October 29 - November 1, 1979)
R. G. Henderson
R. Fitter
J. Wisniewski
June 1980
MTR-80W00148
Sponsor: The United States Environmental Protection Agency
Office of Research and Development
Contract No.: 68-01-5-51
The MITRE Corporation
1820 Dolley Madison Boulevard
McLean, Virginia 22102
-------
-------
PREFACE
In recent years three air pollution phenomena became etched into
the public and political conscience. These are:
1. Acid Rain
2. Visibility Impairment by Anthropogenic Haze
3. Increasing Concentrations of Ambient Fine Particulate Matter
The three phenomena have several common physico/chemical features:
(a) they share the same types of chemical precursors, namely SOX,
NOX, and carbonaceous matter; (b) they require relatively long
periods (life-times) in the atmosphere for the final product to
develop from the original emissions; (c) they seem to prevail to a
greater extent in the Eastern U.S., where present emission rates of
the precursors, and/or the relative humidities are higher. They also
share a common legislative impediment for their control and mitiga-
tion none are at present categorized as "criteria pollutants."
Also, the three phenomena transcend state and national boundaries,
therefore the normal regulatory and enforcement mandates of the Clean
Air Act which are based on State Implementation Plans (SIPs), are not
applicable for controlling the above air pollutants. There are
several sections of the Clean Air Act Amendments of 1977 which could
be invoked to regulate the regional scale air pollution, viz., Sec-
tion 115 - International Air Pollution, Section 126 - Interstate Pol-
lution Abatement, and Section 169A - Visibility Protection for
Federal Class I Areas. However, none of them have yet been tested in
iii
-------
practice, and all of chem assume that there are available validated
air quality simulation models which can, with reasonable accuracy,
relate the effects to the distant source(s).
This Workshop was convened upon the recommendation of the Assis-
tant Administrator, Office of Research and Development, Dr. Stephen
Gage, with the objectives of assessing the state-of-the-art of air
quality simulation models for the above three pollution phenomena and
of identifying the research needs toward developing such models on a
timely and cost-effective basis.
These objectives the Workshop addressed satisfactorily. There
is nothing that scientists can agree on more heartily and existen-
tially than that more research is necessary before decisions can be
made. It is gratifying to conclude though, that the longest (and
perhaps crucial) research effort was estimated to take 10 years and
$40M. This effort is the continuous updating and refining of long-
range transport and transformation models through field measurements;
the average research gaps will be filled in 3-5 years. Perhaps a
conservative assessment is that after 5 years of research, at an
annual cost of about $6-10M (excluding the cost borne by EPA and
other agencies for ongoing programs), EPA's Office of Air Quality
Planning and Standards will have a user-oriented model for estimating
the contribution of individual and multiple sources to (1) the con-
centration and quality of inhalable fine particulates at ground
level; (2) the impairment of visibility in Federal Class I Areas; and
(3) the geographic and quantitative extent of acid deposition.
iv
-------
I would like to express my appreciation to my fellow members of
the task group that organized this Workshop, Drs. K. Demerjian,
R. Papetti, and L. Smith and to the MITRE Corporation, notably,
Drs. R. Henderson, R. Fitter and J. Wisniewski who provided the
executive secretariat and reporting.
Dan Golomb
Program Manager, Atmospheric Transport
and Transformation of Energy-Related
Pollutants
-------
-------
TABLE OF CONTENTS
Page
LIST OF TABLES ix
LIST OF FIGURES xili
1.0 INTRODUCTION AND SUMMARY 1
2.0 INTEGRATED RESEARCH GUIDELINES FOR REGIONAL MODELING 3
2.1 Emissions Data Bases 3
2.2 Monitoring and Data Analysis 4
2.3 Climatological Analysis 6
2.4 Laboratory Studies 6
2.5 Field Studies 7
2.6 Modeling 9
2.7 Equipment Development and Other Recommendations 13
2.8 Funding Guidelines 14
3.0 INVITED PAPERS 17
REGIONAL MODELS FOR FINE PARTICLES, VISIBILITY, AND
ACID PRECIPITATION: A REGULATORY PERSPECTIVE INTRODUCTION
John Bachman 19
STATE ENVIRONMENTAL OFFICE NEEDS FOR REGIONAL AIR
POLLUTION MODELING
Robert Hodanbosi 47
REGIONAL NEEDS FOR REGIONAL MODELSA SHORT GENERAL
COMMENT
William E. Belanger 53
INTERNATIONAL NEEDS FOR RAPM
G. A. McBean 57
REGIONAL EMISSIONS INVENTORIES
C. M. Benkoviiz 63
OBSERVED METEOROLOGICAL DATA BASES FOR POLLUTION
MODELING
J. L. Heffter 99
SULFATE, VISIBILITY, AND PRECIPITATION CHEMISTRY DATA
BASES AND RESULTS FOR REGIONAL MODELING
B. Nieman 115
vii
-------
TABLE OF CONTENTS (Concluded)
Page
HYBRID REGIONAL AIR POLLUTION MODELS
R. Drake 257
MODELING LONG RANGE TRANSPORT AND DIFFUSION
Arthur Bass 295
4.0 FINE PARTICULATES MODELING WORKING GROUP RECOMMENDATIONS 369
4.1 Background 369
4.2 Specific Recommendations 383
5.0 ACID-DEPOSITION MODELING WORKING GROUP RECOMMENDATIONS 391
5.1 Background 392
5.2 Specific Recommendations 399
6.0 VISIBILITY MODELING WORKING GROUP RECOMMENDATIONS 411
6.1 Background 411
6.2 Specific Recommendations 413
viii
-------
LIST OF TABLES
Table Number Page
Section 2.0: INTEGRATED RESEARCH GUIDELINES FOR REGIONAL
MODELING
1 Approximate Funding Levels 15
Section 3.0: INVITED PAPERS
REGIONAL MODELS FOR FINE PARTICLES, VISIBILITY, AND
ACID PRECIPITATION: A REGULATORY PERSPECTIVE
INTRODUCTION
1 Clean Air Act (CAA) Regulatory Options of
Interest 27
2 Major Uses of Regional Models by Regulatory
Program 30
3 Interdependence of Regional Pollutants 33
4 Regional Problems for Definition, Assessment,
and Regulatory Decisions Through 1985 35
5 Some Regional Strategy Issues 38
6 Tentative Timing of Size Specific Primary
Standard (Health) 41
7 Tentative Timing of FP Secondary Standard
(Visibility) - Acid PPT Decision 43
8 Tentative Timing of Class I Visibility
Regulations 45
SULFATE, VISIBILITY, AND PRECIPITATION CHEMISTRY
DATA BASES AND RESULTS FOR REGIONAL MODELING
1 Key to SURE II Station Number and Locations 126
2 Number of S0£ Sites per Day with Concentrations
> 25ug/m3, 1976 148
3 Number of Regional Elevated Sulfate Days
During 1960-1978 Over the Eastern and Western
U.S. 150
ix
-------
LIST OF TABLES (Continued)
Table Number
10
11
12
13
14
15
Page
Number of Days of Low Noontime Visibilities
with Relative Humidities Less than 70% at
Multiple Locations in the U.S. 180
Sulfate/Visibility Case Studies in the Four
Corners Area of the Western Pristine
Region 184
MAP3S Precipitation Chemistry Data for
July 18-22, 1977 207
Number of Sulfate Measurements by States in
the Eastern U.S. in the National Aerometric
Data Bank 239
Number of Total Suspended Particulate
Measurements by State in the Eastern U.S. in
the National Aerometric Data Bank 240
Number of Sulfate Measurements by State in the
Western U.S. in the National Aerometric
Data Bank 241
Number of Total Suspended Particulate
Measurements by State in the Western U.S. in
the National Aerometric Data Bank 242
Sulfate Concentration Trends by Eastern
Subregion 245
SO^ Air Quality Trends in the Six Ohio
River States 246
State Annual Average Sulfate Levels (ng/m3) 247
Northern Plains StatesState Annual Average
Sulfate Levels ((j,g/m3) 248
AQCR Annual Average Sulfate Levels (ug/nr*)
# of Monitors 249
-------
LIST OF TABLES (Concluded)
Table Number
MODELING LONG RANGE TRANSPORT AND DIFFUSION
1 Current Long Range Transport Models (USA)
Section 5.0: ACID-DEPOSITION MODELING WORKING GROUP
RECOMMENDATIONS
Page
351
1
Summary of Existing Types of Wet-Removal
Models
398
XI
-------
-------
LIST OF FIGURES
Figure Number Page
REGIONAL MODELS FOR FINE PARTICLES, VISIBILITY, AND
ACID PRECIPITATION: A REGULATORY PERSPECTIVE INTRODUCTION
1 Visual Range Isopleths, Summer 1975-77
(Trijonis and Shapland, 1979) 20
2 Regions Containing Lakes Sensitive to
Acidification 23
3 Projected Utility Sulfur Oxide Emissions
by Geographic Region (ICF, 1979) 24
REGIONAL EMISSIONS INVENTORIES
1 Inventory Update Distribution 73
2 Emissions Study 74
3 SIC Code Categories 75
4 TSP Emissions, Kilotons (M)/Yr 76
5 S02 Emissions, Kilotons (M)/Yr 77
6 NOY Emissions, Kilotons (M)/Yr 78
A
7 HC Emissions, Kilotons (M)/Yr 79
8 CO Emissions, Kilotons (M)/Yr 80
9 S02 Emissions (Tons (M)/Year) 81
10 Relative TSP Emission Levels (Tonnes/Year) 82
11 Relative S02 Emission Levels (Tonnes/Year) 83
12 Relative NOX Emission Levels (Tonnes/Year) 84
13 Relative HC Emission Levels (Tonnes/Year) 85
14 Relative CO Emission Levels (Tonnes/Year) 86
15 MAP3S Emission Inventory Point Source
Emissions Inventory July 1979 87
xiii
-------
LIST OF FIGURES (Continued)
Figure Number
16
17
18
19
20
MAP3S Emissions Inventory Area Source
Emissions Inventory July 1979
Emissions (Tonnes per Year)
S0? Emissions (Tonnes per Year)
NO Emissions (Tonnes per Year)
X
Comparison of Electric Power Plants in NEDS
and FPC Data Bases
Page
88
89
90
91
92
SULFATE, VISIBILITY, AND PRECIPITATION CHEMISTRY
DATA BASES AND RESULTS FOR REGIONAL MODELING
5
6
10
11
Flowchart of Master Data Base: Organization
and Episode Retrieval System 118
Sulfate Monitoring Locations 122
UTE Research Lab Hi-Volume Sampler
Network 123
Sulfate Monitoring Locations (1975-77) from
EPA National Aerometric Data Bank 124
SURE II Station Numbers and Locations 125
Sulfate Concentration Trends in Air Quality
Control Region 14 129
Sulfate Concentration Trends in Montana,
North Dakota and South Dakota 130
Boundaries of Subregions Considered in the
ORBES Regional Transport Model 131
Sulfate Concentration Trends in the Ohio
River Basin States (Illinois, Indiana,
Kentucky, Ohio, West Virginia, and Pennsylvania) 133
Sulfate Concentration Trends in Subregion III
(South of the Ohio River Basin States) 134
Sulfate Concentration Trends in Subregion V
(Northeast of the Ohio River Basin States) 135
xiv
-------
LIST OF FIGURES (Continued)
Figure Number
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Five-Year Average (1960-1964) of AQCR
Average Sulfate Concentrations (ug/m3)
Five-Year Average (1965-1969) of AQCR
Average Sulfate Concentrations
Five-Year Average (1970-1974) of AQCR
Average Sulfate Concentrations
Three-Year Average (1975-1977) of AQCR
Average Sulfate Concentrations
Isopleths of Summer Sulfate Concentrations
as a Percentage of Annual Average Concen-
trations, 1976
Isopleths of Winter Sulfate Concentrations
as a Percentage of Annual Average Concen-
trations, 1976
Annual Average Sulfate Concentrations in
M-g/m3 (SURE II Data)
Annual Average Nitrate Concentrations in
(j.g/m3 (SURE II Data)
AQCR Average SOT Concentrations ((Jg/m3)
on 19 July 1978
SURE II Sulfate Data for 19 July 1978
SURE II Sulfate Concentrations (fig/m-*) on
23 January 1978
AQCR Average SO? Concentrations (ug/m3) on
19 February 1978
Frequency Distribution of Sulfate Concentra-
tions in Ohio, Pennsylvania, and West
Virginia in 1976 and 1978
Frequency Distributions of NADB and SURE II
Sulfate Data
138
139
140
141
143
144
146
147
152
153
155
156
157
159
xv
-------
LIST OF FIGURES (Continued)
Figure Number_
26
27
28
29
30
31
32
33
34
35
36
37
Episodic ity of Daily Area Average Sulfate
Concentrations During 1974 Ohio, Pennsylvania,
and West Virginia Area
Episodicity of Daily Area Average Sulfate
Concentrations During 1976-1978 over Ohio,
Pennsylvania, and West Virginia Area
AQCR Average SO^ Concentrations ((jig/m) and
High Pressure System on September* 27, 1975
Sulfate Levels Along the RTI Aircraft
Flight Track and Path of High Pressure
System during September 27-30, 1975
304 (Rg/m3)
Aerosol Sulfur Concentrations at Brookhaven,
New York, and Sulfate Concentrations at
Duncan Falls, Ohio, and Indian River,
Delaware, on July 19-23, 1978
SURE II Aircraft Sulfate Data for 20 July
1978
Florida State University Streaker Sampling
Sites
Sulfur Concentrations (^g/m) @ 2 Hour
Intervals @ St. Louis, MO (top) Argonne,
IL (middle) , and Moadville, PA (bottom)
Inhalable Particulate Monitoring Sites
(Long Term)
Sources of Coarse and Fine Particulates
at Urban and Rural Sites
Shaded Isopleths of Yearly Average
Visibilities
Stations Used in 1948-72 Visibility
Analysis
161
163
164
165
167
168
169
171
172
173
176
177
xvi
-------
LIST OF FIGURES (Continued)
FjLgure Number
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
Episodicity: Fractional Contribution Made
fay Upper Percentile (20%) of the Extinction
Coefficient to the Total Dosage Integral
(Time of the Extinction Coefficient)
Contours of Low Noontime (EST) Visibilities
(in Miles) on 11 June 1976, Based on Data
from Selected Stations
Airborne MeasurementsRegional Survey
Visibility at 11:00 AM on December 13, 1974
Sulfate Episode in Arizona on December 13, 1974
Target Contrast at Canyonlands National Park,
Urah, for September, 1978
Visibilities (in Miles) at 1400 LST on
23 November 1978
Proposed Forty Station Fine Particle
Sampling Network for Western Energy Resources
Development Area
Current Precipitation Chemistry Monitoring
Networks
MAP3S Precipitation Chemistry Network
EPRI Precipitation Chemistry Network
Ontario Hydro Precipitation Chemistry Network
Wet Deposition of SO^ -S (gSm~2yr.1) in 1977
pH Distributions from Ontario Hydro Precipi-
tation Data (June 1976-December 1977)
Analysis of Satellite imagery on 20 July
1977 at 12:30 GMT
Cumulative Percent of Wet Sulfur Deposition
Events
179
182
185
186
187
189
190
191
195
197
198
200
202
204
206
209
xvii
-------
LIST OF FIGURES (Continued)
Figure Number
54
55
56
57
58
59
60
61
62
63
64
65
66
Page
Cumulative Percent of Precipitation
Events 210
Hourly Precipitation Chemistry at Brookhaven,
NY November 15-16, 1978 212
Distribution of pH from the BNL (Raynor)
Automatic Sequential Precipitation Sampler 214
Event Means of Precipitation Chemistry by
Length of Event from the BNL Sequential
Precipitation Data 216
Event Means of Precipitation Chemistry
by Precipitation Rate from the BNL Sequential
Precipitation Data 217
Spatial Variability 1977 219
Frequencies of Wet and Dry Periods
for 1977 221
Schematic Diagram of the Parameterization
of Wet Removal in the Episode Transport-
Removal Model 223
Number of Conventional Recording Rain
Gauges per State or Southern Portion of
Province (below 47° N) 224
Analysis of Satellite Imagery on 22 July 1978
at 11:30 GMT 226
Special Sources of Air Pollutant and
Meteorological Data
24-Hour Average 862 Concentrations (ppb)
at Ontario Hydro Monitors in Southern
Ontario on 19 January 1976
Average 24-hour S02 Concentrations ((Jig/m^)
AEP Monitors in the Upper Ohio River Area
5-6 January 1977
at
230
232
234
xviii
-------
LIST OF FIGURES (Concluded)
Figure Number
67 TVA Air Quality Monitoring Network
HYBRID REGIONAL AIR POLLUTION MODELS
1 Scale classification system for terrain
and meteorological phenomena, patterned
after that of Orlanski (3)
2 Examples of an Eulerian system XYZ and a
Lagrangian system X'Y'Z', where locations
a,b,c,d represent the pollutant cloud
at times ta
-------
-------
1.0 INTRODUCTION AND SUMMARY
A workshop on regional air pollution modeling was held on
October 29 through November 1, 1980 at Port Deposit, Maryland. The
workshop was sponsored by EPA's Office of Environmental Processes and
Effects Research (OEPER), Energy Effects Division. The MITRE Cor-
poration was contracted with to plan, hold and report on the results
of the workshop. The first day and a half of the workshop was de-
voted to a number of invited talks on the need for regional air
pollution models, their data requirements and the state-of-the-art of
regional models. Papers of some of the invited talks are included in
Section 3.0.
The last two and a half days of the workshop were devoted to
discussions of the research needs for regional models in three
working groups: Fine Particulates, Acid Deposition and Visibility.
In this report the recommendations of each of the working groups are
detailed and an integrated set of research study guidelines are de-
veloped. The research study guidelines were formulated by combining
the working group recommendations into a single set of research
requirements. This was necessary because of the large amount of
overlap between the working group recommendations. The overlaps,
given the nature of the problem, were inevitable. In all three
areas, fine particulates, acid deposition, and visibility, there are
many aspects of regional modeling which are common. Thus, with
perhaps some minor differences, each area requires the use of the
-------
same emissions input, the development of meteorological fields to
drive the model, the same dispersion and transport components and
much of the same atmospheric chemistry. The primary difference
between a fine particulate model and an acid deposition model is the
need for scavenging itechanisms, both in clouds and below clouds in
the acid deposition model. Similarly, the primary difference between
a visibility model and a fine particulate model is the need for
radiative transfer methods in the visibility model.
Because of the similarity of the three model types two general
research recommendations are immediately obvious:
Model development should be modular so that the basic
components can be used for all three types of model, and
* The research should be carefully coordinated to insure that
redundant studies are eliminated as much as possible - this
is particularly important in the area of field studies.
In section 2.0 the integrated research guidelines are presented.
Some of these guidelines fall into areas which may not be the respon-
sibility of the same funding unit which is responsible for the model-
ing research. An attempt has been made to include these guidelines
in section 2.7. The working group recommendations are given in sec-
tions 4.0, 5.0 and 6.0.
-------
2.0 INTEGRATED RESEARCH GUIDELINES FOR REGIONAL MODELING
The research recommendations that resulted from the working
groups on Acid Deposition, Fine Particulates and Visibility (see Sec-
tions 4.0, 5-0 and 6.0) were, in many instances, overlapping. In ad-
dition, some recommendations, while- not redundant as worded, could be
expanded to include other recommendations. In this section a general
set of recommended guidelines is put forth which cover the needs of
all three modeling areas. These guidelines are in the areas of:
Emission Data Bases, Monitoring and Data Analysis, Laboratory Stud-
ies, Field Studies and Model Development. In addition at the end of
this section a discussion of concurrent instrument development is
presented which would be necessary for the successful completion of
the studies recommended by the working groups. The funding levels
for the guidelines have been developed from the recommendations of
the working groups, however, because of the reorganization of the
recommendations, the funding levels of the guidelines do not neces-
sarily match the working groups funding estimates.
2.1 Emissions Data Bases
Accessible data bases containing emissions data and inventories
are required for the development, testing and implementation of
models. While considerable effort has been put into the development
of a comprehensive emissions data base, additional work is required
particularly if the emissions data base is to support visibility,
fine particulates and acid deposition modeling. In order that the
-------
emissions data base support all three areas it will be necessary that
it include various source types e.g., power plants, mining opera-
tions, synthetic fuel plants, smelting and urban areas and the fol-
lowing information:
* S02 emissions
NOX emissions
primary particulates, including size distribution and
chemical composition
soot emissions
hydrocarbons by species or class, and
source characteristics
Specifically, the following guidelines relating to emission data
bases are recommended:
GUIDELINE 1: Source Characterization^ Field studies should be
performed to determine size distributions and chemical composi-
tion of primary particulate emissions for each source type. In
addition the variability of these characteristics within each
source type should be determined. The possibility of relating
changes in the mix of emissions, size distributions and chemical
composition of primary particulate with more readily determined
factors such as operation mode or specified source characteris-
tics such as flow rate or temperature should also be considered.
Guideline 2: Emission InventorylJpdate. A continuous refine-
ment and updating of the emissions data base, with identifica-
tion of data gaps, should be undertaken.
2.2 Monitoring and Data Analysis
Data from a number of studies and monitoring networks are
available and could provide useful insight into the problems of the
sources and dynamics of aerosols. Full analysis of existing data
whenever possible is both cost effective and required if maximum
-------
progress is to be made in the regional modeling area. In this regard
a number of guidelines are recommended:
GUIDELINE 3_: Sources of Aerosols. Data sets archived by large
scale studies such as SURE II, the Florida State University
(FSU) Streaker Study, the Western Energy Environmental Moni-
toring Study (WEEMS) and aerosol sampling networks should be
screened for aerosol chemistry data at rural sampling sites.
These data, coupled with meteorological information for the
times of aerosol sample collection and basic emissions inven-
tories of the most proximate urban sources may be useful in
gaining insight into such questions as the contributions of
anthropogenic primary particulate emissions, natural emis-
sions, and secondary aerosol formation to total rural aerosol
concentrat ions.
GUIDELINE 4: Chemical Species Distribution. Aerosol chemistry
data sets archived by large scale studies should be analyzed to
study the spatial distribution of various chemical species in
aerosols as a function of particulate size. The spatial corre-
lations of various species will indicate the scales of transport
and hence the relative roles of removal mechanisms on various
chemical species components of aerosol.
GUIDELINE 5: AerosolParticle Dynamics. The rate of change of
an aerosol size distribution is governed by several mechanisms:
sources, coagulation, dry deposition, etc. Data analysis of
results from studies which have collected aerosol size distribu-
tions should be conducted. Data from the Visibility Impairment
Due to Sulfur Transport and Transformation in the Atmosphere
(VISTTA) study and others may be of sufficient quality to permit
rate of change analysis. Use of theoretical models should be
made to evaluate the roles of coagulation, gas-to-particle con-
version, diffusion and dry deposition on the size distribution
and mass concentration of the atmospheric aerosol.
In addition to these general guidelines some studies related
specifically to the individual modeling areas should also be pursued:
GUIDELINE 6: Visibility Monitoring Needs. The data from the
WEEMS program should be utilized to determine the need for ad-
ditional monitoring of visibility and aerosols in the western
U.S. to support model validation and initialization. Recom-
mendations for additional sites and instrumentation should be
developed.
-------
GUIDELINE 7: Analysis of Recent Precj-pitation Chemistry Network
Da t a Systematic analysis of available data should be pursued.
Specifically the data analysis should include:
1. Variable pair correlation
2. Ion balance.?
3. Factor analysis
4. Time-series analysis
5. Material budgets
6. Statistical modeling analysis
2.3 Climatological Analysis
When a sufficient data record exists, useful information can
often be obtained by doing a climatological analysis of various
parameters. This can. lead to a greater understanding of causes and
effect and also can provide data directly useful to model develop-
ment.
GUIDELINE 8: Climatology of Trajectories and Mixing Heights.
Use existing data to generate a climatology of various levels
of trajectories and mixing heights.
A study particular to visibility is the creation of a visibility-
meteorology climatology.
GUIDELINE 9: Relationship between Visibility Impairment and
Meteorology. Analyze WEEMS and other available data to de-
termine the relationship between visibility and meteorological
conditions. A climatology of visibility and related meteorology
should be developed, particularly in the western U.S.
2.4 Laboratory Studies
Aerosol particulates which contribute to the fine particulate
problem, visibility impairment and acid deposition arise predomi-
nately as secondary particulates which form as the results of chem-
ical reactions occurring in the atmosphere. The processes which
create these fine particulates, called gas-to-particle conversion,
-------
are not well understood. For this reason two laboratory study areas
have high priority:
GUIDELINE 10: Homogeneous Gas-to-Particle Conversion. Gas-
to-particle conversions may occur homogeneously(i.e. all reac-
tants existing in the same phase) at sufficient rates to provide
significant secondary sources of fine particulates that can con-
tribute to acid deposition and visibility impairment. The gen-
eralized reactions of most interest are:
S02
NOX
Volatile Hydrocarbons
Sulfates
Nitrates
Non-Volatile Organics
Studies of the kinetics of homogeneous reactions should be
conducted in smog chambers, investigating overall gas-to-
particle conversion under controlled conditions, and using
chemical kinetics to resolve the pathways and rates of the
various reactions.
GUIDELINE 11: Heterogeneous Gas-to-Particle Conversion. This
study involves the chemical kinetics of the generalized reac-
tions noted in study 10 as they take place among reactants in
different phases. Also important for consideration in the study
are the roles of nucleation and scavenging in particulate forma-
tion and fractionation.
In addition to the studies of the homogeneous and heterogeneous
chemistry of sulfur and nitrogen oxides in the atmosphere a better
understanding of the role of ammonia in these chemical processes is
required.
GUIDELINE 12: The Role of Ammonia in Atmospheric Chemistry.
Recent studies indicate that most previous work on the effects
of NH3 on the conversion of S02 to sulfate in aqueous solu-
tion may be improper. Further studies are needed to elucidate
the role of NH3/NH4+ in the conversion of S02 to sulfate,
and in the conversion of NOX to nitrate.
2.5 Field Studies
Field studies will be necessary to support all three areas of
modeling. Data from field studies will provide better understanding
7
-------
of the physical and chemical processes important to the long range
transport, transformation, and fate of pollutants. In addition, the
detailed data will allow more careful model initialization and vali-
dation than can be accomplished using only monitoring data. Field
studies are also the most cost-intensive component of the proposed
research studies and therefore careful planning is required to ensure
the optimal utilization of available resources. Because of the very
similar data requirements for support of visibility, fine particulate
and acid deposit ion modeling, all field studies should be designed to
support, to the extent possible, all three areas. Thus a generalized
set of studies is recommended, the details of which will change for
different areas of the country where one or two of the three modeling
areas may not be important.
GUIDELINE13: REGIONAL SCALE FIELD STUDIES. A series of three
to four week intensive field studies should be carried out. Use
of alternate years to perform the field studies in the east and
the west would allow for support by the field equipment to all
areas of concern. Four to five intensive studies could perhaps
be performed in one area during the year and analysis of the
data could be performed during the off year. In order to carry
out these field studies a national mobile rawinsonde network
should be implemented. This mobil network would consist of 50
mobil rawinsonde stations which could be deployed in the east
or the west as needed. During the field studies tetroons will
be employed to track air parcel movements for at least one and
a half diurnal cycles and meteorological data will be augmented
in both temporal and spatial coverage using National Weather
Service (NWS) type radiosondes. In addition to meteorological
data the field studies should collect, when appropriate, data on
visibility, dry deposition, transformation and removal processes
(using tracers), chemical conversion processes in clouds, etc.
The important point to make is that the resources required for
these field studies must be coordinated with the needs of the
three modeling groups as well as other EPA, TTF data require-
ments, including the study of the meteorology of complex ter-
rains .
8
-------
2.6 Modeling
Regional models for predicting source-receptor relationships in
the areas of acid deposition, fine particulates and visibility (where
the receptor is the "eye of the beholder") have a great deal in
common. The meteorological fields needed to drive the models are
essentially the same, as is a great deal of the transformation
chemistry, This is not surprising since, as pointed out previously,
all three areas are impacted by the same emissions and result, to a
large extent, from the same fine particulates. The major differences
in the models will result from different areas of application (i.e.
North East, South Western U.S., etc.), possibly different methods of
initialization and different end product modules. Thus a fine
particulate model will have as end product the atmospheric concentra-
tion at some level or levels of fine particulates while the acid
deposition model will have end product modules for dry and wet
deposition of important acidic and other species and the visibility
model will have end product modules which determine the optical
extinction coefficient and perhaps other parameters of importance.
Because of the similarity of the models in many areas it is
important that complex models be developed in a modular fashion with
separate modules for initialization (including meteorological fields
and emissions), transport, transformation, and end product. Within
each of these modules the use of submodules may also be useful. The
-------
transformation module may be divided into submodules which deal with
separate aspects of atmospheric chemistry; some of the transformation
submodules would then, perhaps, not be required for certain applica-
tions. The development of a fully modularized, complete set of
regional models will have to be evolutionary in nature with improve-
ments continuing to he made as the important physical and chemical
processes become better understood (the evolutionary nature of the
models is another argument for the use of a modular approach -each
module can be changed independently as new understanding becomes
available). The full development of a set of regional models will
require a large expense and a long time for completion (although in a
real sense, because of the evolutionary nature, the models may never
be "complete"). The expense is justified, however, because it will
be less, in the long run, than the expense of developing a number of
single unit models which cannot be easily improved and updated.
GUIDELINE 14: Evolutionary Model Development. In the short
term the basic structure of an evolutionary regional model
should be developed using the best existing techniques for each
of the modules and submodules. Where differing techniques are
available, and there is no clear preference for one over the
other, each technique should be implemented in modular form to
allow testing of the various modules in the overall model
program. In the long term, as insight and understanding is
gained from laboratory or field studies, the appropriate modules
should be altered or new modules should be written.
In addition to modeling the regional transport, transformation
and fate of the pollutants of interest, the meteorological fields
which will drive the regional model must also be derived. A number
of methods are currently employed to develop meteorological fields
10
-------
for regional models. These methods usually make use of single level
trajectories which are then interpolated and adjusted to obtain di-
vergence free wind fields. Dynamic models which employ thermodynamic
principles are an alternative method of generating the meteorological
fields required for driving regional models. Such models are being
used for weather forecasting and their use for regional air pollution
modeling should be investigated.
GUIDELINE 15: Dynamic Meteorological Model. Existing dynamic
meteorological models which include thermodynamic principles
should be developed for use to support regional air pollution
modeling. The dynamic meteorological model should produce spa-
tially variable, multi-layered wind fields, vertical temperature
structure and direct measures of atmospheric stability on a fine
spatial grid (20km x 20km). The model should include air flow
constraints imposed by actual terrain conditions. The dynamic
meteorological model should be tested against data derived from
various field experiments including tetroon tracking and refine-
ments to the model should be made where indicated.
The evolutionary modeling approach, along with the development
of dynamic meteorological models, will lead eventually to an accurate
capability to predict the effect of new emission sources and control
technologies. However this capability for accurate prediction is
probably at least five to ten years away. In the interim there is
a very real need to be able to make the best educated estimate of
the effects of new power plant placement, control technologies, etc.
Unfortunately, regulation wil1 not wait for the completion of an ac-
curate complex evolutionary model. For this reason simpler modeling
approaches which have the potential for providing "reasonable"
11
-------
answers to the regulatory questions arising over the next five years
should be pursued.
GUIDELINE Jj: Near Terra Regional Models. Modeling approaches
to the regional transport transformation and fate problem which
can provide reasonable estimates of source receptor relation-
ships in the near term should be pursued. This includes the
continued testing and development of current modeling approaches
such as LRTAP, EURMAP, ERT/ACHEX etc. Also simpler, statistical
modeling approaches such as statistical trajectory analysis of
source receptor relationships, should be supported.
In addition to these general study recommendations, a number of
modeling studies are required specifically for visibility and acid
deposition. For visibility, radiative transfer models and psycho-
visual models must be further developed. The models should be in-
corporated as modules to the evolutionary regional model.
GUIDELINE 17: Radiative Transfer Models. Radiative transfer
methods should be developed for computing important optical
parameters from reduced inputs as may be available from regional
TTF models (e.g. using particle concentrations in only two size
ranges rather than complete size distributions). This should
include an anlaysis of the sensitivity of the resulting optical
parameters to the resolution and accuracy of regional models
under a. number of field situations using data from the field
experiments.
GUIDELINE18: Psychovisual Models. Radiative transfer models
should be used co further investigate the relationship between
pollutant concentrations and subjective determination of visi-
bility impairment under a number of illumination conditions for
representative scenes.
In the area of acid deposition a detailed storm model, which
could also be used as a module for the evolutionary regional model,
should be developed.
12
-------
GUIDELINE 19: DetailedStorm Model. Existing detailed numer-
ical models of storm dynamics and physics can be modified to
incorporate scavenging characteristics. Existing storm models
can also be used in conjunction with field studies for experi-
mental diagnosis.
2.7 Equipment Development and Other Recommendations
A number of recommendations were made during the workshop for
the development or improvement of equipment particularly in regard to
field studies. While some of these developments will be necessary to
successfully carry out some aspects of the field studies the funding
for these development may more appropriately come from outside the
OEPER Energy Effects Division. The equipment recommendations in-
clude :
Cloud Water Sampler: This must be able to separate aerosols
from cloud droplets while collecting enough cloud water
within a short time to allow chemical analysis. Its goal
is to elucidate cloud chemistry: how quickly are freshly-
entrained aerosol particles scavenged; how does cloud water
solute vary from aerosols chemically, and is this due to
variable nucleation, scavenging or aqueous conversion of
gases?
Exotic Species Measurements: Methods for measuring the
concentration of exotic species in clouds such as H202
and HN03 should be developed.
Size Resolved Aerosol Chemistry Measurement: A better size-
resolved aerosol sampler, which is compatible with chemical
analyses, is needed in order to characterize the size dis-
tributions of individual chemical species, and conversely
to determine the chemical make-up of various size fractions
of the atmospheric aerosol.
Dual-Doppler Radar Facility: A dedicated dual doppler radar
facility should be developed for use with field studies in-
volving clouds.
13
-------
Tracers: There is a need to continue development of suitable
tracers for use in field studies.
Tetroons: Tetroons which can be used to track plume movement
over a period of one to two days should be developed.
In the visibility area a recommendation was made for the devel-
opment of a visibility perception criteria document:
A visibility criteria document should be produced. This
document would contain photographs of selected Class I scenes
under various atmospheric conditions and illumination condi-
tions along with associated physical parameters such as op-
tical extinction coefficient.
2.8 Funding Guidelines
It is difficult to assess accurately the funding required for
the total research program for regional modeling. In part this is
do to the uncertain nature of research and development. During the
workshop the participants were asked to estimate the funding and time
required for each of their recommendations. The figures in Table 1
represent an attempt to integrate these funding estimates into esti-
mates for the whole program. Included in Table 1 are all of the re-
search guidelines from sections 2.1 - 2.6 of this report. It must be
kept in mind that the funding levels and timing shown in Table 1 are
approximate and should be used only as general guidelines.
14
-------
= £ S = 5 s £
c c c c c:
^ tr- C C ^ W
_i
e
c cr c c - c:
C L" C = 1.- U^.
r- r^ f^
OS. >-,
RI ; c t, a. ',
< c. *- E
~ I 2
S - =
-- *- tf V-
c, C c v. '.".
isij;
E £.£
4,j ^ )- *j O
15
-------
-------
3.0 INVITED PAPERS
The papers in this section represent some of the invited talks
given during the workshop. The views expressed in these papers are
those of the respective authors.
17
-------
-------
REGIONAL MODELS FOR FINE PARTICLES, VISIBILITY, AND ACID
PRECIPITATION: A REGULATORY PERSPECTIVE INTRODUCTION
John Bachman; EPA, OAQPS
I would like to begin by identifying our component of the Envi-
ronmental Protection Agency. I am a member of the Strategies and Air
Standards Division of the Office of Air Quality Planning and Stan-
dards in Research Triangle Park, N.C. Our office identifies and
evaluates known or potential air pollutants, determines whether con-
trol may be needed, recommends appropriate regulatory options, and
develops control strategies. In particular, we are responsible for
reexamining existing and establishing new National Ambient Air Qual-
ity Standards (NAAQS). Over the past several years, we have been
involved in a number of evaluations of fine particles, acid precipi-
tation, and visibility. This paper will present some current
thoughts on the prospects for regulation in these areas and discuss
the importance of regional models to the regulatory process.
I would like to review a few of the reasons we are concerned
about these phenomena/problems very briefly, since I am sure the audi-
ence is familiar with them. Figure 1 is a map of summertime airport
visibility for 1974 through 1976 prepared by Trijonis and Shapland.
You can see that the area of reduced visual range extends over the
entire eastern United States, and, conversely, there is excellent
visibility in areas of the Southwest and the West in general. The
work of Trijonis, Husar, and others, suggests that visibility used to
19
-------
20
-------
be markedly better in the East than it has been over the past several
years. In the Southwest, there is no clear indication of strong
trends one way or the other, with some exceptions. This map suggests
a significant regional gradiant in visibility between the East and
West. I would argue that the map is not an unrealistic representa-
tion, at least in the summer, of areas with higher densities of fine
particulate matter. Certainly, the higher humidities and other
factors in the East are largely responsible for reduced summertime
visibility, but it is also probable that the heaviest regional load-
ing of fine particules in the U.S. occurs in the area of the Ohio
River Valley in the eastern United States. Thus, Figure 1 is indi-
cative not only of regions of reduced visibility, but also of regions
of concern for other potential effects of fine particles. Available
data indicates that a significant fraction of eastern fine particles
consists of secondary particles, particularly various sulfates.
Although several years of additional research will be needed to
determine the extent of any health risks posed by eastern fine par-
ticles, present evidence is at least indicative that the problem is
regional in scale and not solely derived from "traditional" local
sources.
The acid precipitation phenomenon has been widely discussed.
Analysis in recent years suggests that regional acid precipitation in
the East has expanded and the acidity has increased throughout this
region through the early 70"s. There are various known and potential
21
-------
problems associated with acid precipitation. While acid precipita-
tion may produce signficant effects on terrestrial ecosystems, crops,
and materials, the most convincing evidence of damage from acid
precipitation relates to aquatic ecosystems, in particular, poorly
buffered lakes of the Northeast. Figure 2, prepared by Galloway and
Cowling, indicates geographical areas with surface waters which may
be sensitive to acid-induced changes. This figure indicates one of
the reasons why we may have an international problem, since a large
percentage of sensitive surface waters are in Canada, in addition to
the rather large areas in the northern U.S.
All three problem areas, fine particles, visibility, and acid
precipitation, are in some measure related on a regional scale to
SOx emissions. Figure 3 indicates reasons for continued concern
over the next several decades. The map shows projected utility SOx
emissions over the next 30 years and is based on the analysis done
for the recently promulgated new source performance standard for
power plants. I must: point out a number of caveats regarding these
projections. The timing and extent of any possible emissions de-
creases depends strongly on assumptions concerning utility retirement
schedules, energy growth projections, and extent of nuclear growth.
Given these assumptions, the figure shows the power of one of the
regulatory mechanisms of the Clean Air Act, the new source perfor-
mance standard. The objective of this provision of the Act is to
ensure that the best control, considering economics, energy, and
22
-------
FIGURE 2
REGIONS CONTAINING LAKES SENSITIVE
TOACIDIFICATION25
23
-------
NORTH DAKOTA
SOUTH DAKOTA
NEBRASKA
KANSAS
MINNESOTA
IOWA
VISSOrRI
MAINE
NKW HAMPSHIRE
VERMONT
NKW YORK
MASSACHUSETTS
CONNECT!CUT
RHODE ISLAND
NKW JERSEY
PENNSYLVANIA
DEI .AWARE
L MARYLAND
WISCONSIN
II.MNOIS
INDIANA
MICHIGAN
;iHIO
IS 75 1990,19''5 Ml (I
TIME
197> 1990 1995 '010
TIME
WASHINGTON
ORKCON
CAI.IFORKIA
1971) 19901995200
TIME
1 1 1
MONTANA
IDAHO
WYOMING
NEVADA
UTAH
COLORADO
ARIZONA
SEW MEXICO
' r-
_
~^
1975 1990 1995 201
TIME
OKLAHOMA
ARKANSAS
TEXAS
I.OU I SI ANA
KENTUCKY
WEST VIRGINIA '
VIRGINIA
TENNESSEE
|- N. CAROLINA
S. CAROLINA
[_ MISSISSIPPI
ALABAMA
GEORGIA
FLOKIUA
I I I
1975 1990 1995 :OiO
TTMK
1975 199019952010
TIME
FIGURES
PROJECTED UTILITY SULFUR OXIDE EMISSIONS
BY GEOGRAPHIC REGION
(ICF, 1979)
24
-------
other impacts, is put in place on new units of various major source
categories.
Figure 3 indicates that for the next 20 years or so, sulfur
oxide emissions, already high in the east, will continue at roughly
current levels. In the long-term, we may have a significant strategy
for attacking problems associated with high regional SOX emissions.
After the year 2010 or so, assuming older plants retire and newer
plants are constructed meeting the new standard, regional emissions
of sulfur oxide should decline. There are three key points to be
made: (1) Assertions regarding massive increases in national SOX
emissions with increased coal use are probably in error. If any-
thing, we project essentially a leveling off of sulfur oxide emis-
sions and eventually a possible decrease in the East; current
regional emissions are however fairly large and it will be some time
before they decrease; (2) With respect to total acid deposition,
projections for NOX emissions indicate that, without additional
regulatory controls on new or existing sources, NOX emissions in
the regions of best visibility (Southwest) are going to double from
currently low levels. These increases may however be balanced by an
expected decrease in emissions from smelters suggesting that, on a
regional basis, things may balance off. Nevertheless, because these
areas are so sensitive to the effects of fine particles, we have a
special concern about trying to address the regional visibility
problem in the western United States from new sources, as well as a
need to worry about the continuing high loading in the eastern U.S.
25
-------
In summary, we have relatively high fine particle levels
throughout the eastern U.S. The prospects are that these levels are
going to be about the same for the next 20 or 30 years with a possi-
ble increase in emissions of acid precursors, mainly nitrogen oxides,
and perhaps some potential for degredation of regional visibility in
the western U.S. Regional models will be essential in evaluating all
of these problems.
CLEAN AIR ACT REGULATORY OPTIONS
Table 1 shows the Clean Air Act regulatory options of interest
to which we may address regional problems associated with fine par-
ticles, acid precipitation and visibility. Under Sections 108 and
109, primary (health) and secondary (welfare) national ambient air
quality standards (NAAQS) can be set for pollutants which are preva-
lent in ambient air and result from numerous and/or diverse station-
ary or mobile sources. Control is effected by state action under the
state implementation plans (SIP's). EPA is responsible for: the
development of criteria and control techniques documents; evaluating
alternatives and promulgating NAAQS; and providing guidance for the
development of SIP's. The most relevant standards to the problems
under discussion are those for particulate matter, sulfur oxides and
nitrogen oxides. Proposal of new or revised air quality standards
must be accompanied by a regulatory impact analysis. After the
standard is promulgated, states submit SIP's which must demonstrate
attainment of the standards in 3 - 5 years for the primary standard
26
-------
NAAQS
TABLE 1
CLEAN AIR ACT (CAA) REGULATORY OPTIONS OF INTEREST
Primary/Secondary
Criteria Document
Standards Analysis, including Impacts
Promulgation/SIP Guidance on Strategies
SIP Development by States 1 yr
110/126 Interstate Pollution
Implementation 3-5 years/"Reasonable Time"
NSPS
Utilities, Industrial Boilers, Other
Sources
BACT
Long-Term Strategy for Regional Emissions
PSD/Visibility
Class I Areas
National Goal
BART - Major Existing Sources
Long-Term Strategies (10-15 yr)
New Source Review
CAA Revisions
27
-------
and "reasonable time" for a secondary standard. Implementation con-
flicts can and do arise from interstate and international transport
of pollutants such as ozone and sulfur oxides. We are currently in
the process of revising the criteria document and reassessing the
standards for nitrogen oxides, particulate matter and sulfur oxides.
The New Source Performance Standard (NSPS) mechanism under Sec-
tion 111 of the Clean Air Act was outlined above. Protection of
visibility in Class I areas, through Section 169A and 165 (PSD) was
added to the Clean Air Act in 1977. Section 169A establishes a
national goal for enhancing and preserving visibility in Class I
areas where visibility is an important resource. PSD provides a
mechanism for implementing the national goal for significant new
source categories. There are 156 Class I areas where visibility is
important, mostly located in the western U.S. They include certain
national parks, monuments, and wilderness areas. Section 169A calls
for long-terra strategies for visibility protection, over a 10-15 year
period, designed for making reasonable further progress towards the
national visibility goal. Although initial regulations for visibil-
ity protection will likely be limited to single source impacts,
eventually visibility protection regulations may require regional
visibility protection strategies; hence regional modeling for visi-
bility impacts will ultimately be important for both existing and new
sources.
28
-------
In addition to regulatory options available under the Clean Air
Act, there is much discussion on the need to evaluate other possible
regulatory mechanisms for dealing with acid precipitation or other
related regional air problems. It is possible that the available
options do not effectively match the scale of regional air pollution
problems or provide for the most cost-effective controls. Regional
models may play an important role in evaluating alternative regula-
tory mechanisms for consideration in future deliberations on the
Clean Air Act.
MAJOR USES OF REGIONAL MODELS FOR REGULATION
The potential uses for regional models by regulatory programs
range from problem definition to implementation strategies, as out-
lined in Table 2. Regional air pollution models will have an impor-
tant role to play in defining the scope of the acid precipitation,
fine particle, and visibility problems and in providing an improved
basis for deciding whether and what kind of regulatory remedies may
be necessary for dealing with them. Regulators are increasingly
being asked to defend environmental programs through the use of
quantitative analyses of costs and benefits. Such analyses will be
particularly important in providing support for programs dealing with
the "welfare" effects of air pollution. Although our ability to
quantify the benefits associated with reduced pollutant loadings on
aquatic systems, soils, biota, visibility, and the like are para-
mount, regional models may provide the key link between the benefits
29
-------
TABLE 2
MAJOR USES OF REGIONAL MODELS BY REGULATORY PROGRAM
Benefit Analysis - Establishing Need for Regulation
Possible Basis for Visibility NAAQS
Acid PPT Program
Regulatory Analysis (Impacts)
E.G. NSPS
S02/PM NAAQS
Control Strategy Analysis
Least Cost/Most Equitable Strategies
SIP Guidance
Mid-Course Corrections
Interstate Disputes
Recommendation of Alternate Regulatory Mechanisms
30
-------
associated with reduced regional loadings and costs associated with
achieving such reductions.
The second major regulatory use for regional models is for the
regulatory analysis: shorthand for environmental, economic, energy,
and other impacts analyses. Regulatory impact analyses must accom-
pany proposals of all major air pollution regulations. In the
future, regional models might be used to predict the air quality/
deposition impact of changes in new source performance standards or
air quality standards. For example, we are currently conducting a
regulatory analysis associated with the review and possible revision
of the air quality standards for sulfur oxide and particulate matter.
To the extent that revisions under consideration affect national and
regional emissions, it would be desirable to evaluate possible
regional effects. Because regulatory analyses must of necessity deal
with broad national perspectives, the kinds of models needed and use-
ful for this purpose are of less detail and rigor than those ulti-
mately required for control strategy implications.
The most obvious use for regional models is in control strategy
analysis and development. Models used for this purpose should under-
go some validation and be scientifically credible, since significant
resources may be allocated in response to the model predictions.
Having defined some problem such as acid precipitation which is to be
regulated, it would be highly desirable to develop least cost, equit-
able control strategies for addressing the problem. In the case of
31
-------
possible air quality standards for fine particles or sulfates,
regional models will be necessary for developing guidance for state
implementation plans (SIP's).
We also may need regional models to make mid-course corrections
in existing air pollution programs. We are facing this problem right
now with ozone, since it too is a multi-state problem, necessitating
controls in one state to ensure the air quality standard is met in
another state. There have already been disputes between Ohio,
Pennsylvania, and West Virginia over the amount of particulate matter
that is transported across state boundaries, and the effects of these
particulates on meeting the TSP air quality standard. I am not sure
that the regional models currently in existence are sufficient to
convince the governor of one state of the need for additional con-
trols on SC>2 emissions to markedly improve TSP (or particulate
sulfates) levels in other states. If, in our revisions of the air
quality standard for particulate matter, we move to a size-specific
standard, then the problem of transported particles on a regional
scale might assume significantly increased importance. This, of
course, depends strongly upon the level of any such standard. In any
case, the need for development of regional models which can speak to
these issues over the next several years is clear.
EFFECTS AND AVERAGING TIMES OF CONCERN
Table 3 shows the interdependence of regional emissions of some
important pollutants and potential regional effects of concern.
32
-------
^ I
< n w 4_i
z s
.J w *- .J
-- I -i a 3 u
I 3
jj -^
J ! ^ 4J -H
I -J a
J I " - = 41
.- ! 13
I ^
33
-------
Thus, the acid precipitation/fine particle/visibility problems which
are the subject of this conference are all related. Priorities are,
in rough order - SOX emissions, NOX emissions, and primary par-
ticles, with organics and hydrocarbons of lesser concern. Besides
showing the interdependence of these problems, Table 3 also makes a
second important point: we have essentially two kinds of regional
air pollution problems - air quality and deposition. Regional air
quality tends to relate to effects caused by ground level concentra-
tions of pollutants, effects such as those on health and visibility
degredation. Deposition effects are certainly coupled in some ways
with regional air quality, but there may not be a one-to-one rela-
tionship between the two. This is important because the air quality
standard mechanism of the Clean Air Act is directed at solving air
quality related problems but is not as well suited for dealing with
deposition-related problems.
Table 4 outlines the three problem areas, potential effects, and
averaging times that I think will be of some concern in developing
possible regulations. In this case, I am attempting to project pos-
sible programs through 1985.
The key question for particulate matter is whether we will have
a TSP standard, a size selective (IP, FP or other) standard, or a
sulfates standard. In any case, it currently seems likely that we
will be concerned with both annual and 24-hour average particulate
levels. Although a health-based sulfate standard does not appear
34
-------
C/3 U1
crt ao
W a^
< PC
a
- p>
z q
o PC:
M pg
H H
W M
Q to
M
OS CJ
o w
BQ
§
s
PS
a)
CO £
)-< -H
CU H
CO
CO
O1
c -*
4-l
^-- r-l
cn 3
Q) W
,-H
U 00
H C
4-1 -H
(U C
C M
CO
CO
I
cn Q
vo cn
-H PL,
n) o
3 M
>, c
u )-i
H (U
H 4-t
H cn
^3 ca
H W
CO
H C
n a) CN
< 10 i-H 1I
O
H 14-1
cn tu
H (-1
oo r-
C cn
§
W
-------
likely at this time, new data over the next several months could
change this assessment. In that event, information from clinical
studies may suggest a concern over averaging times as short as 1
hour. Moreover, epidemiological and other studies of particulate
matter conducted over the next several years with improved monitoring
instrumentation may also suggest a need for a short-term (e.g. 1
hour) particulate standard. In dealing with regional visibility and
climatic impacts of fine particles in the eastern U.S., we will most
likely be concerned with annual and seasonal averaging times.
Although benefits analysis and standards development may require
information on episodic impacts in worst-day effects, in my judgment
we are more likely to try to implement regulatory programs for these
welfare effects using longer-term (at least seasonal) averaging
times. In this case, the regulatory mechanism might include improv-
ing new source performance standards, and fine particle or sulfate
secondary ambient air quality standards.
Visibility protection programs for Class I areas, and particu-
larly in the Southwest, will certainly be concerned about annual,
seasonal and daily visibility impacts. Although the effects of
plumes for averaging times as short as one hour may assume signifi-
cance, it is not clear that we will need regional models that predict
one hour visibility effects in Class I areas. Nevertheless, we are
likely to require regional models capable of predicting daily worst-
case visibilities in areas with complex terrain.
36
-------
Our current state of knowledge on acid and other deposition
problems suggests that effects on aquatic ecosystems are demonstrated
and potential effects may ensue on terrestrial ecosystems, crops, and
materials. We are currently concerned about the rate of titration of
lakes and streams and thus with annual deposition. Because of sensi-
tivity of biological systems and known seasonality in acid, nitrogen,
and sulfur deposition, we must also examine seasonal or monthly depo-
sition patterns. To the extent that biological and other effects
work demonstrates significant impacts associated with single storm
events, we may eventually be concerned with minimizing peak impacts.
Thus, regional acid deposition models must, at minimum, address
annual and seasonal deposition and possibly storm events. Because
the mechanism for regulation of acid precipitation, if needed, is not
clear, it is more difficult to predict the averaging times and spa-
tial extent to be addressed by regional models in implementing
control strategies.
KEY STRATEGY ISSUES
Table 5 lists some of the key control strategy issues which we
are likely to face in attempting to deal with these problems. The
list is not intended to be inclusive and there may be some disagree-
ment with respective specific issues. Clearly, the most important
issue in any control strategy is the effect of different emission
source regions on important receptor regions, whether that be sensi-
tive populations, important vistas, or sensitive lakes. The effect
37
-------
crt
on
O
w
w
M
&. Oi
S-i
QJ O
l-< ft,
3D
O CJ
H 0)
t-i en
CO
U-i 3
pa 4J
60
-------
of Call stacks versus shorter stacks on long distance transport is of
obvious importance. In a related question, what is the relative con-
tribution from stationary sources versus mobile sources, particularly
with respect to nitrogen oxides and acid precipitation. The relative
importance of SOX versus NOX emissions is more significant with
respect to acid precipitation because we are less concerned about
NOX from the standpoint of regional visibility or fine particles.
However, we need to investigate the possible impacts of regional
loadings of nitric acid vapor with respect to public health. Animal
and human testing will be conducted over the next several years to
further determine whether there is any significant basis for concern.
Clearly, the contribution of primary versus secondary particles to
regional fine particle loadings is most important with respect to
visibility degredation and health related particulate control
strategies.
In order to minimize control costs, it is important to identify
any significant variations in transport and/or effects with respect
to summer, winter, or other seasonal emissions. You might find, for
example, that seasonal, e.g. summertime, reductions in sulfur oxides
emissions might be" more effective for improving regional visibility;
at the same time, such a strategy might be ineffective at ameliorat-
ing acid surge problems derived from wintertime emissions that build
up in snow packs. The effects of single sources on a regional scale
can be important in the review procedure for permitting new sources
39
-------
under prevention of significant deterioration programs. This will
likely be more important for sources proposing to locate near Class I
areas in the Southwest where good visibility prevails or in the
Northern-midwest where geologically sensitive lake systems exist.
The need for improved understanding about the relative roles of wet
and dry deposition, single storm events, long-term depositions, and
snow melt is self-evident and will require additional information
from biological effects research. However, we also need models which
can tell us the relative amounts of hydrogen and various ions depo-
sited during these kinds of events. The effect of complex terrain is
clearly important for all three problems, as is the relative impact
of anthropogenic versus natural sources.
TENTATIVE TIMING OF POTENTIAL REGULATORY PROGRAMS
I would now like to present some thoughts on the possible timing
of any regulatory action on particulate matter, visibility, and acid
precipitation. It is important to note that these are just prelimi-
nary guesses as to the possible timing. New information or a number
of other factors could dramatically alter the timing of actions out-
lined here.
Table 6 shows possible timing of health-related size specific
particle air quality standards. The criteria document air quality
statement review/revision process for particulate matter and sulfur
oxides is already underway. We are considering the possibility of
40
-------
TABLE 6
TENTATIVE TIMING OF SIZE SPECIFIC PRIMARY STANDARD (HEALTH)
Regulatory Analysis for PM/SOX
(Impacts of Resultant S02, Primary FP
Emissions)
Potential Revised PM/SOX NAAQS, SIP Guidance
(TSP/IP/FP/S04/Other)
Implementation of Revised NAAQS
(Estimate Secondary, Other Fine Contribution
to PM, Develop Multi-State Strategies)
Continue Health Studies of FP/IP/S04
(Regional Model Support of Epidemiology
Episode Studies)
Initiate Criteria Document Revision Process,
Regulatory Impact Analysis for PM/SOX
(FP, S04?)
80
80
80-85+
80-84
85+
-------
TSP, inhalable particles (<15 microns), Fine Particles «2.5 mi-
crons), sulfates, or other size-specific particulate standards during
the course of this review process. A decision on the nature of the
new stardard and an accompanying regulatory analysis should be avail-
able during 1980. Implementation of any revised air quality standard
for particulate matter would occur in the 1980-85 timeframe and be-
yond. Depending on the nature of the standards, it may be important
to have regional models for assisting in the development of multi-
state control strategies. Studies of the health effects of par-
ticulates and sulfates will continue in the 1980s. Already under
consideration is the possibility of studies of regional fine particle
episodes on public health. Such studies would be assisted through
forecasts using regional air quality models. The results of con-
tinued research will be used in the next round of criteria documents/
standard reviews which, under the Clean Air Act Amendments must occur
within five years after the most recent revision, or approximately
1985. Newly developed information may permit or require consider-
ation of fine particles and or sulfate standards at that time.
Table 7 shows the tentative time frame for a possible secondary
fine particle standard related to visibility and climatic effects and
a decision on the appropriate approach for dealing with acid precipi-
tation in the near-term. Again, the starting point is the revision
process for the standards for particulate matter and sulfur oxides.
At this point in time, it does not appear likely that we will set a
42
-------
TABLE 7
TENTATIVE TIMING OF FP SECONDARY STANDARD
(VISIBILITY) - ACID PPT DECISION
Promulgate Revised SOX/PM NAAQS
Develop Basis for Standards - Benefits
Analyses, Preliminary Models
Develop Fine Particle/Sulfate - Visibility
Acid PPT Relationships
PEPE, MAP3S, SURE Results
Refine Regional Models
Regulatory Analysis, Preliminary Strategies
Analysis
Decision on NAAQS, Examination of
Alternative Regulatory Mechanisms
Implementation of Standards, Full-Scale
Models
12/80
80-82
80-82
80-82
81-82
82
82-83
84-90
-------
secondary air quality standard for visibility or acid precipitation
in the 1980 review/revision process due to inadequate information.
However, subsequent to the air quality standard process, and in par-
allel, we will be developing the basis for possible future standards
or regulatory actions for visibility/acid precipitation. This will
include several approaches to conducting benefits analysis and al-
ternative control strategy analysis. At the same time, the research
programs will be working to develop fine particle/sulfate visibility
relationships and presumably emission/air quality/deposition relation-
ships in field studies such as the STATE, MAP3S, and SURE program.
Hopefully, this additional work will result in a refinement of re-
gional models that will help to improve benefits analysis and permit
regulatory and preliminary strategy analyses in the 1981 time frame.
We hope that regional modelers can target some significant improve-
ments in existing models and some validation for some of these models
for use in these analyses. Such models will improve the decision-
making process for air quality standards and acid precipitation con-
trol programs in the 1983 time period.
Table 8 shows the tentative time frame for visibility protection
regulations in Class I areas under section 169 (a) and 165. The Ad-
vance Notice of Proposed Rulemaking has already been published in the
Federal Register and proposal of regulations is expected in May of
1980. Currently, we are suggesting that Phase 1 of the regulatory
program ignore to a Large extent the question of regional transport,
44
-------
TABLE 8
TENTATIVE TIMING OF CLASS I VISIBILITY REGULATIONS
Advance Notice
Evaluation of Single New Source
PSD Permits
Phase I Regulatory Analysis
Phase I Regulation Promulgation
VISTTA, Other Field Studies
Refine Regional Models for Western
Application
Develop Additional Guidance on
Regional Visibility Problem
Phase II Regulatory Analysis
Decision on Regulations
Fall/79
Ongo ing
Spring/80
Fall/80
79-82
80-82
83-85
83-85
85
45
-------
principally because available models are inadequate and because emis-
sions increases over the next several years do not appear significant
in western areas. Th<> area in which regional emissions seem to pro-
duce the greatest impact on visibility is in the East. However, the
question of what to do about protection of Class I areas from re-
gional visibility degredation in the East is intimately involved with
what to do with respect to visibility throughout the East and with
acid precipitation. Thus the focus of the Class I area program must
be on regional visibility in the West. We feel that it is important
within the next five or six years to markedly improve our ability to
predict the impacts of single and multiple sources associated with
energy in urban and general development in the West to enable us to
make improved decisions on potential controls in location of new
sources in the West. Thus improved regional models may well form
the most important component of any future "Phase 2" visibility
regulations.
-------
STATE ENVIRONMENTAL OFFICE NEEDS FOR
REGIONAL AIR POLLUTION MODELING
Robert Hodanbosi
Ohio Environmental Protection Agency
Office of Air Pollution Control
My name is Rober Hodanbosi, I am employed by the Ohio Environ-
mental Protection Agency. My position is Chief of the Division of
Air Quality Modeling and Planning within the Office of Air Pollution
Control. This division has responsibility for development of the
technical support for any regulations that are promulgated by our
agency. This would include atmospheric dispersion modeling.
I would like to give you some background specific to the situa-
tion in Ohio. Our state is the largest coal consuming state in the
nation. 95% of the electricity generated in Ohio is produced from
the combustion of coal. Industrial sources are also significant in
Ohio with a total annual consumption of more than five million tons
per year. There is also a large domestic coal mining industry in
Ohio with the bulk of the coal supplies being composed of medium to
high sulfur content. Requirements designed to control emissions that
affect visibility, fine particulates, and in particular acid precipi-
tation could have a substantial economic impact on our state.
Over the past few years the application of atmospheric disper-
sion models has progressed from a planning tool into an integral
element for the development of regulations, new source review proce-
dures, and enforcement cases. Whether the model be for visibility
47
-------
or for sulfur dioxide, some of the basic needs for a state agency
remain the same.
First, there needs to be a formalized mechanism for the release
of new models. At the present time our agency learns of new models
through the "grapevine," Let me site a recent example. A new source
applicant met with me to discuss what type of model would be required
in a specific application. After outlining the "approved model" the
applicant went to the Regional Office to discuss a Prevention of
Significant Deterioration permit. The applicant later called to
notify me that he will be required to use the MPTER model and asked
if our agency would approve of the use of this model. Since I had
never even heard of this model it was a question I could not fairly
answer. I later found out that this model is still being developed
at Research Triangle Park. To remedy this problem I suggest that all
models be announced in the Federal Register.
Once the models are developed, there should be formal training
sessions on the application of these models. The present gaussian
models are frequently misapplied. With the newer, more complex mod-
els being developed the training of personnel is a necessity.
Any models that are developed should be thoroughly field tested
before release. A few years ago the U.S. EPA utilized the RAM urban
model for the development of sulfur dioxide regulations. This was
done before there were any validation studies performed on the accur-
acy of this model.
48
-------
There must be consistency in the application of these models
from state to state. Due to the nature of these regional pollutants,
it is likely that the major impact from large sources is going to be
in a different state. Unless there is consistent application of mod-
els there will be different strategies developed for similar sources.
If time permits later in this workshop, I want to discuss the prob-
lems of implementating control strategies in one state due to pre-
dicted contributions to another states air quality problems.
In the area of long-range transport there must be greater coop-
eration between states. At the present time the State Implementation
Plan requires a demonstration that the ambient air quality standards
will be attained and maintained within the boundaries of the state.
The long-range transport is not being considered. Because of the
chemical transformation of NOX and SC>2 to nitrates and sulfates,
states developing revised Implementation Plans for total suspended
particulates are finding it difficult if not impossible to show
attainment of the standard. A significant portion of the particu-
lates may be due to long-range transport, not under control of the
state developing the plan. Our border state of Pennsylvania esti-
mates that in the western half of that state, sulfates amount to 20%
or more of the particulate matter. Our agency is having hi-vol
filter analysis performed throughout the state. The results are
similar for the areas tested so far.
49
-------
Here are some interim measures that EPA should consider until a
better perception of long-range transport is realized. First, EPA
should appraise the thought of allowing states to remove from the
particulate SIP revisions analyses, sulfates that can be attributable
to sources outside the state. This may remove the continued need for
emission offsets in order to allow new source growth.
EPA should support states in the development of a coal washing
or coal preparation regulation. Our agency is developing require-
ments that will require the use of washed coal at all utility plants
in the state. We estimate a reduction of up to 700,000 tons per year
of sulfur dioxide emissions statewide. Up to this point we have not
received any encouragement in this effort from EPA.
Although specifically prohibited in the Clean Air Act Amend-
ments, EPA should study the effectiveness of supplementary control
strategies for these regional pollutants. Under certain meteorologi-
cal conditions that accelerate sulfate formation low sulfur fuels
could be utilized by the large coal burning facilities. This may
provide some relief to the states downwind. I want to emphasize that
these are interim measures until the proper analytical techniques are
available.
The primary sources being studied as causes of acid precipita-
tion are coal combustion sources. Because both sulfates and nitrates
contribute to this problem, other sources of sulfur dioxide and
nitrogen oxides should also be examined. As with rural ozone being
50
-------
caused by long range transported pollutants from urban areas, EPA
should investigate what is the possible impact of urban transporta-
tion sources on visibility and acid precipitation.
I was recently made aware of some of the present research ef-
forts by the Environmental Protection Agency in this field. Results
from studies such as Project Midwest Interstate Sulfur Transformation
and Transport which examined the kinetics of the conversion of sulfur
dioxide to sulfates should be incorporated in the regional models.
EPA should not preclude results from the private sector such as the
Electric Power Research Institute, provided the validity of the re-
search can be documented. Further studies that are underway should
lead to a better understanding of the reaction mechanisms occurring
in the atmosphere. In turn this should lead to improvements in re-
gional air pollution modeling.
Presently, there are many consulting firms developing regional
models. Each of these models are likely to be unique in some par-
ticular fashion. In order for a state agency to review the results
of these models a set of criteria needs to be developed to measure
the regional models against. EPA should outline the basic elements
of the regional models that should be included in any effort.
Regional air pollution models that are developed should be
programmed to run efficiently on a computer. As a state agency we
run the present gaussian models a multitude of times. With our
regional U.S. EPA office insisting on five years of meteorological
51
-------
data, it is not difficult to run up a high computer bill. Presently
the air modeling being performed by our office accrued costs of over
$40,000 in computer time in a one month period. I believe many of
the present models could be written more efficiently, and I would
like any new models to be efficient.
Our agency, which I believe has similar modeling capabilities to
most states, has had no experience in the application of regional air
pollution models. I look forward to participating in the following
sessions and, speaking on behalf of the State of Ohio, we appreciate
the opportunity to be involved from the start on this important new
field of atmospheric dispersion analysis.
52
-------
REGIONAL NEEDS FOR REGIONAL MODELS
A SHORT GENERAL COMMENT
William E. Belanger, P.E. - EPA Region III
In spite of the similarity in name, regional Models are not
necessarily well-matched to the needs of EPA regional offices.
Instead, they are regarded as somewhat of a solution in search of
a problem. This does not imply that the regional models are not
needed, but only that EPA regional offices are unprepared to deal
with them.
By way of explanation, EPA has ten regional offices, each of
which consists of a staff of several hundred people and handles on
the order of five states (some more, some less). The central func-
tion of a regional office is to carry on the day-to-day operational
relations with state and local governments. Their primary function
is to act as the political and technical interface between EPA and
the states. In general there is a small staff which handles air pol-
lution models at each regional office. Regional offices are often
organized on a state-by-state basis so that each individual will
deal with only one or two states. There is therefore no mechanism
in place to deal with interstate problems.
In the few instances when a pollution problem has been inter-
state in nature, there have been severe political problems in ar-
riving at a joint solution. Even when short distances are involved
and no particularly sophisticated modeling is applied, there is a
great reluctance for one jurisdiction to control its emissions (and
53
-------
therefore incur an economic penalty) for the benefit of another
jurisdiction downwind. This is well illustrated in the Philadelphia
AQCR where there are four jurisdictions involved. It took over a
year of negotiations to come up with a control strategy for S02« A
part of the problem is that pollution control is not subject to the
political horse-trading that usually goes on because the upwind
jurisdiction usually will experience a pure economic loss for someone
else's benefit. There is nothing to "trade" for this economic loss,
and therefore a strong incentive for the upwind jurisdiction to con-
trol its emissions exists only so much as is needed to deal with its
own problems.
Regional models add a whole new complication to this procedure.
They would eventually ask a jurisdiction many miles removed from the
problem area for control not needed for local problems. There will
therefore be an even stronger resistance to the additional controls.
When this is coupled with the politician's mistrust of anything more
complicated than linear rollback, a very difficult situation would be
expected to develop. It is easy to imagine a situation where the
"upwind" jurisdictions mount an attack on a regional model to avoid
additional control of pollution emissions, and to visualize endless
haggling over the allocation of the control burden among many states.
Thus, regional models add a level of complication not previously
experienced by regional offices. Their use for regulatory purposes
will be quite difficult to implement. It may be necessary eventually
-------
to use the models to define a uniform emissions limit which would
apply over large geographic areas. Direct application of regional
models to define individual emission limits may not be possible.
One final note, though: The "solution in search of a problem"
phrase has been used beforeto describe the laser. Regional models
could enjoy a similar level of usage for applications where today's
gaussian models will not work, especially long-range transport and
rough-terrain problems. Thus the negative tone of this paper is not
a condemnation of regional models. It is only a warning that we will
have to tread softly in their application to interstate problems.
55
-------
-------
International Needs for RAPM
G.A. McBean
Atmospheric Environment Service
Downsview, Ontario, Canada
Introduction
Air pollution problems occur on many scales, ranging from those
in the immediate vicinity of a source to those of a global nature.
All except the very local problems can have international implica-
tions depending on the proximity of the source to an international
border. North America, Canada and the United States have, over the
past fifty years or more, dealt with air pollution problems that were
international but local in scale. The modeling and analysis needs
for such problems were appropriate to the scale. More recently there
has been a growing awareness that important air pollution problems
arise due to long-range transport; transport on the scale of hundreds
to thousands of kilometres. The LRTAP has added another important
(in some cases dominant) question to the international consideration
of air pollution. Canada and the United States have now begun
discussions which may lead to an air agreement. The form that the
agreement will take cannot as yet be stated but it is anticipated the
air pollution modeling will play a role in determining the control
strategies arising out of it.
Requirements for Models
In many ways, the international needs for RAPM will not be
different from the needs of individual states (or provinces) in their
discussions with other states (or provinces). As each jurisdiction
57
-------
considers its approach to air pollution problems, it will need to
know how much pollution is in its airshed and how much is being de-
posited on its ecosystem. It will further want to know the contri-
bution to these totals from within and from outside its area. Al-
though there are potentially several ways of determining this infor-
mation, atmospheric transport and deposition models are the most
promising way. The first requirement for RAPM'E is the computation
of transboundary fluxes on time and space scales appropriate to areas
under a single jurisdiction for establishing control policies or
protecting ecosystems. Hence, it will likely be necessary to refine
RAPM's to give transboundary fluxes into areas the size of typical
states and on monthly (or a least seasonal) time scales.
The second requirement will be to go beyond the transboundary
flux to the computation of depositions and the relative contributions
to these depositions from each source region. A typical question
that must be answered is what are the relative contributions to the
depositions on the Mjskoka region of Ontario from emissions at
Sudbury and Pittsburgh. Any other choice of receptor and sources
could be substituted.
Although acidic precipitation is the major problem of concern at
the present, a third requirement for models will be to be able to
deal with a wider range of pollutants. Some of the heavy metals and
persistent organic contaminants appear as probable priority candi-
dates.
58
-------
Another requirement for models is that they must include in a
reasonable way all the appropriate physical and chemical processes
that transport, transform and deposit the pollutants in both wet and
dry forms. At the same time the models must be suitable for running
on a variety of emission scenarios so that various control strategies
can be tested.
jProblemiS_ Facing Modellers
Most of the problems facing modellers with respect to RAPM for
international needs are again the same as for all sorts of RAPM.
However, because of their international implications there is a
special pressure put on them. One aspect is the time requirements.
Depending on the speed of the bilateral negotiations, there may be
very limited time available during which to develop models. At the
same time the models will be expected to give accurate results. The
results of models may be used to determine an international control
strategy which, once set, may be very difficult to change.
Another important aspect is that of model verification. Because
the costs of control are large and certain sectors of society need to
be convinced that modeling results are correct it is necessary to be
able to demonstrate their ability. Tests will have to be conducted
to compare modeling predictions and observed values, like of pollu-
tant concentrations or wet deposition. When the results of models
are presented, the verification statistics or appropriate qualifiers
must also be presented.
-------
Presently, models are still in considerable need of treatment
of the meteorological and chemical processes. For example, it is
inappropriate to couple a complex meteorological flow model with
first order chemistry or a simple advective box model with a multi-
equation, higher-order chemistry model. In dealing with the meteo-
rology one can stress the examination of episodes or try to produce
a good climatological estimate. Both have their advantages and each
type of modeller can learn from the other.
United States-Canada Research Consultatiot^ Group on LRTAP
Early in the discussions between the United States State Depart-
ment and Canadian Ministry of External Affairs regarding mutual air
pollution problems it was agreed to establish a United States-Canada
Research Consultation Group on the LRTAP. This Group was to aid in
the exchange of information and in the coordination of research be-
tween the two countries. It was also requested to provide reports
to the two governments on the status of the LRTAP situation and
research. The first report was completed earlier this month.
A major question put to the Group was to estimate the trans-
boundary fluxes and put into perspective local versus long-range,
cross-boundary transport. Only limited results were available to
the Group to use in preparation of the Report. A second report will
probably be produced by next summer and it is hoped to have more
extensive modeling results to which to refer. To provide a better
basis for reviewing and reporting on these results, the Research
60
-------
Consultation Group is establishing a Sub-group on Modeling. It is
hoped that this Workshop will provide the Sub-group with a good start
on its work.
61
-------
-------
REGIONAL EMISSIONS INVENTORIES
C. M. Benkovitz
Atmospheric Sciences Division
Brookhaven National Laboratory, Upton, N. Y.
The goal of the Multistate Atmospheric Power Production Pollu-
tion Study (MAP3S) is to improve understanding about transport,
transformation and fate of pollutants released by energy-related
activities. Tasks 1 and 2 of the MAP3S Program Plan (1) are
described briefly as follows:
Task 1 - Power Production Emissions
To specify and quantify the emissions of pollutants from present
power production plants, and to consider pollutants that may be emit-
ted as a result of an increased usage of coal and the introduction of
new power production processes.
Task 2 - Nonpower Production Emissions
To identify and quantify sources of pollutants that do not stem
directly from power production but that may affect the concentration,
distribution, transformation, and fate of pollutants.
The primary region of interest for the MAP3S program was defined
as the high pollution, energy intensive Northeastern quadrant of the
United States. Secondary regions of interest includes those areas
directly influencing air quality of the primary region.
63
-------
The MAP3S emissions inventory project is a part of the total
MAP3S Data Management effort. Given the available resources and
the lack of mandate to gather emissions data directly the most cost
effective approach was found to be the compilation of the inventory
from data supplied by other agencies.
The starting point in our data search was the Environmental Pro-
tection Agency (EPA). Under the Office of Air and Waste Management
the National Air Data Branch (NADB) has the responsibility to amass
source emissions, air quality and related data from the whole U.S.
into a single central repository under one centralized administrative
body. The National Emissions Data System (NEDS) was developed to
standardize storage formats and to collect and report on information
relating to sources of any of the five criteria pollutants. Respon-
sibility for NEDS data collection and error correction rests with
state agencies and the EPA regional offices; authority to insert data
into NEDS rests with NADB.
The type of data stored in NEDS is as follows:
Point Source^ Data
General source information - Name, address, types of source,
year of record, comments, etc.
Operating or production rate, fuel type, yearly estimated emis-
sions, control device type and efficiency on each criteria pollutant,
etc.
64
-------
Geographic location of source (using UTM coordinates), stack
height, and diameter, exhaust gas temperature, and flow rate.
Allowable emissions, applicable control regulations, compliance
status, and schedules, etc.
Area Source Data
General source information - Name and location of area (county)
source, population, year of record.
Activity levels - Countywide fuel activity level of each type
of area source. Sources are classified residential, commercial and
institutional, industrial, incineration; fuels are classified as an-
thracite and bituminous coal, distillate and residual oil, natural
gas and wood. Some data on transportation levels and its use of
fuels is also included.
Emissions data - Yearly emission estimates for the entire county
(for each criteria pollutant).
Spatial resolution is on a point-by-point basis for point
sources, county level for area sources. Temporal resolution is
yearly for both, with some scheduling information for point sources
ava ilab le.
Since the state agencies have the responsibility for the collec-
tion of the emissions data for NEDS most states have emissions inven-
tories in house. In general, these inventories contain the same data
that has been submitted to NEDS; however, a considerable time lag can
develop between state inventory compilation and NEDS submission.
Recent NADB activity has attempted to close down such gaps.
65
-------
The EnergyData System (EDS) was developed by the Strategies and
Air Standards Divison of the Office of Air Quality Planning and
Standards as a tool for assessing air quality and energy impacts of
environmental legislation. The system stores nationwide summaries of
air quality data and data related to the fuel consumption and atmos-
pheric emissions from power plants and large industrial fuel burners.
The primary sources of data for EDS are Federal Power Commission
(FPC) data obtained via questionnaires known as Form 67 and Form 423,
NEDS, SAROAD*, and the Compliance Data System. Thus, the EDS system
combines data from diverse sources into a single data base.
Plant data included in the EDS system can be classified in the
following general categories:
Regulations data
Compliance data
Projection data
Fuel data - Includes year, fuel type, fuel use and fuel
characteristics. Boiler operation data - Includes
monthly fuel consumption, fuel characteristics,
emission rates, output characteristics.
Plant emissions data
Plant waste product data
Boiler design data
Stack data - Height, diameter, exit flow rates, exit temperature
Plant diffusion modeling reports
Prohibition order data
*SAROAD = Storage and Retrieval of Aerometric Data, EPA
66
-------
Starting in 1968, a series of studies were conducted, some under
the auspices of the Air Pollution Control Office of EPA, some under
the auspices of the National Air Pollution Control Administration,
Division of Air Quality and Emissions Data of the Department of
Health, Education and Welfare. These studies were conducted to try
to estimate levels of air pollutant emissions and status of their
control. In general, the earlier surveys covered a limited geo-
graphic area (metropolitan areas, AQCR's, etc.) and were conducted
using a general survey procedure based on the document by Ozolins,
Guntis and Raymond Smith entitled, "Rapid Survey Techniques for Es-
timating Community Air Pollution Emissions" (Public Health Service
publication 999-AP-29, Oct. 1966). Numbers resulting from these sur-
veys usually covered source type, seasonal and geographical distribu-
tion within the study area and were included in the final report to
the sponsoring organization.
Other studies conducted during this time period included surveys
based on the use of questionnaires on a statewide basis. NEDS data
was sometimes used as a starting point; some results were scheduled
to be included in the NEDS system as well as summarized in the final
project report. A partial list of resulting reports is included as
an Appendix.
The purpose of the Regional Air Pollution Study (RAPS) was to
provide a data base and test bed for the development and verification
of air quality simulation models. The St. Louis metropolitan area
67
-------
was chosen "because it represented a fairly typical metropolitan
area, included light and heavy industry, was based on a coal-burning
economy and was situated in a self-contained air basin, separated
from the nearest metropolitan area by 350 miles" (see Ref. 2). Thus,
geographic coverage could be limited to the immediate counties in
Missouri and in Illinois. Time resolution requirements of short-term
diffusion models is hourly datathis time resolution was carried
over to the emissions inventory. The data cover the two years of
matching ambient and meteorological data gathering--!975 and 1976.
The final RAPS emissions inventory includes the following data:
Point sources - point identification, stack parameters, fuel
characteristics, hourly or annual process data, point scheduling,
emission factors. Hourly emission rates are obtained by calculation
applying emission factors to the hourly process data or to the annual
process data and scheduling parameters. Hourly data are obtainable
for the five criteria pollutants, non-criteria pollutants, heat,
sulfur trioxide, hydrocarbon breakdown and particle size distribu-
tion.
Area sources - area source data are stored in a variable size
grid. Grid parameters include size, geographic location, state,
county, population, annual emissions of criteria pollutants and heat
for highways, railroads, river vessels, stationary industrial,
fugitive dust, off-highway mobil and residential/commercial. Hourly
emissions and hydrocarbon breakdown are generated by retrieval
software containing source dependent apportioning schemes.
68
-------
Line sources - line location, length, traffic volume, average
daily emission rates, etc. Hourly emission rates are calculated by a
distribution scheme that includes the month of the year, day of the
week, and hour of the day.
The primary purpose of the Sulfate Regional Experiment (SURE)
program is to define the physical and chemical mechanisms that link
emissions of SOX, total emitted particulates (TEP), NOX and
hydrocarbons to ambient concentrations of SC>2 and 804. The major
goal of the SURE program is to develop a useful air quality model to
relate emissions to ambient SOX and 804 levels over distances of
1000 km.
The development of a detailed emissions inventory for the SURE
region was contracted to GCA/Technology Division, Bedford, Mass.
Regional coverage includes three areasprimary, secondary and
tertiaryencompassing the continental United States east of the
Rocky Mountains and southeast Canada. Spatial resolution is down to
the individual point for "major" sources (sources emitting >^ 7000
tons/year TEP or >_ 10,000 tons/year SOX or >^ 3,000 tons/year NOX
or _> 5,000 tons/year HC), aggregated to an 80 x 80 km grid for
smaller point and for area sources. Temporal resolution is seasonal
emissions, with hourly emissions data for large utility sources
during five SURE intensive periods. Pollutants included are S02,
504, NO, N02, low, moderate and highly reactive HC and TEP.
Base data for this inventory were obtained from NEDS. Auxiliary
data were obtained from state agencies. Seasonal variations were
69
-------
calculated. Hourly data for intensive periods were obtained through
questionnaires to the individual utilities.
The initial implementation of the MAP3S emissions inventory has
been designed to include data for the continental U.S. east of the
Mississippi River. The starting data set was extracted from the NEDS
system and because of the diverse end uses of such data at BNL, the
decision was made to use a generalized data base management system
(DBMS) to manipulate this inventory. Work accomplished to data
includes:
1. Design and loading of the data bases for point source and
area source data.
2. Missing or erroneous point source locations were found to be
the most numerous errors of critical importance to the MAP3S modeling
community. To attempt to correct these errors the following tasks
were undertaken:
a. Master Enumeration District List - expanded (MED-X) data
were obtained from the Bureau of the Census.
b. Correlation of geographic codes between NEDS and MED-X.
c. Development of computerized methodologies to determine
whether given point source locations are within county
of residence, or missing and if so, to attempt to
correct using census data.
3. In cooperation with the SURE program emissions inventory
project (described above), corrections to NEDS data developed by both
projects were interchanged and SURE data added to MAP3S inventory.
70
-------
4. Design and loading of auxiliary data bases containing data
from the FPC form "Steam Electric Plant Air and Water Quality Control
Data," (Form 67). After correlation of FPC/NEDS plant identification
codes, emissions and fuel summaries by plant were extracted from the
FPC data and added to the point source inventory as separate data
items.
5. Design and implementation of "checkpoint" data files.
Checkpoint files are sequential data files containing the subset of
inventory data needed for input to atmospheric models. No updating
of data will be done for these files; they will provide the modeling
community with an unchanging test base for model development.
The following figures present the results of some of our initial
studies of the inventory data. Methodologies used to produce these
and other displays and/or tabulations are available to MAP3S program
participants. The main emphasis for FY 80 of the MAP3S Data Man-
agement Project will be in the further development of these user-
oriented packages.
FIGURE 1 - Inventory Update Distribution. Count of number of points
updated by year.
FIGURE 2 - Emissions Study. Cumulative % of emission totals vs
number of point sources for all five criteria pollutants.
Data used included FPC emissions data for plants when NEDS
emissions data were missing.
71
-------
FIGURE 3 - Standard Industrial Classification (SIC) code (see Ref. 6)
groupings.
FIGURE 4-8 - Distribution of emission totals for 5 criteria
pollutants by SIC categories as described in Figure 3.
FIGURE 9 - Map of largest 200 point sources in inventory area.
FIGURE 10-14 - Relative emissions by state for 5 criteria pollutants.
FIGURE 15 - Point source emissions totals using NEDS emissions data.
FIGURE 16 - Area source emissions totals using NEDS emissions data.
FIGURE 17-19 - Scattergrams of plant emissions data obtained by using
NEDS and FPC as sources for these data.
Current work on :he inventories includes:
1. Preparation of Progress Report for the initial stage of the
inventory project.
2. Further develDpment of user access and data manipulation
methodology. To be included in this area among others are
computerized procedures needed for data transformation.
3. Annual update using NEDS data. Pertinent data are extracted
from NEDS, computerized at BNL and all previously developed
corrections found to be still pertinent are added.
4. Additional merging of FPC data into the inventory. Figure
20 presents some preliminary counts done in this area.
5. Additional quality checks of inventory data will be
performed on a time available basis.
6. Expansion of geographic coverage to include all of the
continental U.S. and Canada.
The major effort of this project to date has been in the data
acquisition and computerization areas. Both ad hoc data retrieval
and the systematic production of data subsets have been greatly
72
-------
O
D
m
I ->
n:
r i
n.
rt:
UJ
D3GOJ 00091 00091 OOOH Qfj'JZ\ 00001 0008
cmuodn siNiOcTON
tr.
K
Q
I-
oc
O
0009 OOC^ fJOOZ
73
-------
NO. POINTS
2
4
7
10
13
16
20
24
29
34
40
46
43
60
68
69
78
89
101
1 15
131
140
173
191
109
("' '.?
r ',9
307
748
395
.,.,9
-BO
590
'./J7
684
791
812
035
1083
1398
I 131
2 1 72
281 1
3715
4812
4B91
7 165
"125
10212
I IJ'f-35
i p : S2
1 3449
|4()b8
1 bO 1 0
: 7519
23128
2 '989
2»f,95
e i.931
e '3J9
55294
T1.,!' LL1 t
. 12
45
2 25
i - 02
4 .45
4.09
|).U6
7.55
>3 . 45
' i . 92
i ri . 97
t 2 . 29
1 i. 1 6
14. 1?
1 4 . 28
; ' .> . '6
1 H . 1 8
! " . 30
21 . 30
2 3 . 25
2'-* . J>6
<"';.? - 2-
r ' . 76
<0 . 137
^ 1 . 39
<3. :?
<'; 21
3 7 . «7
4! .118
43.10
'«4.£(l
47.13
..'i 'iS
40 - 1 1
42. 19
5'+ -3
V- . C6
b7 . £3
sa . B4
61 .73
r, ? . 45
_.4.3i
' -;: . H9
lift. 35
71 49
72 . 1 3
7 5 . (* -1
75.95
7H.b7
'.-'9 . ^6
i-^ . i6
B4 . 60
H5 . 4 8
04 . H9
H9 . 1 3
90 4!
93. 14
1 -r> . t">4
''8. 75
100.00
ISP
7B.21
S02 ~.OM
4. 12
a. ,?s
2 . 37
5.H2
:9 02
22 . 06
') . (jO
; a . i-;B
:2 . 47
.-.5 . B3
-.9.26
42.45
'5. '.'9
43.86
' 1 . 'J9
';=. '-6
45.43
~, 9 . £ -i
61 .64
64.72
67.77
73 . '.'9
5.99
7' 3. 90
715. 61
;"3 . h3
E.? . 2!
B4.37
8tS . 26
8H . 0 3
t:» . ;58
9'., .47
9c; . 67
Ji- . IB
rt": 99
9=:. 10
95 . 28
9F, . 50
96 . 93
97 . 96
95.61
9= . 1 1
9^ . b 3
99.81
99.94
fj'j - 96
100.00
100.00
',00.00
100.00
IPO. 00
100.00
100.00
100.00
too. oo
: :T oo
100.00
100.00
100 oo
iOO 00
see
1941 1228
t >,M CUM :
3.40
4 83
6.76
8.24
10.28
12.68
14.50
16.00
1 8 . ?:B
20.70
22 . 55
2^97
26 ^2
27^91
.'.8 - B8
58.98
40.51
43. 77
46.69
49.43
51 .65
54 . 90
5 7 . 79
HO . 88
61 .32
64. 14
b?. 17
L'3 58
71 .92
73.42
?5 . 09
'5 82
78. 1 1
78. 16
79.27
80.26
no . 45
81 .S3
*2 . 59
H3 '. 76
rj'-* . F52
H-i . 68
fir, . 50
H7 - 86
f!8 - 82
f'9 . 39
M . B-t
i! . 19
91 .26
,9.47
0 . 05
43. 10
56 . 1 Q
59 . 1 6
61 .77
54 . H2
b7 . BS
70 68
73.r,8
75 . 08
76.98
bD.09
H3. 17
06 . 17
IJ8.81
91 .84
<-N . b7
«6 . 75
99 . 80
1 00 . OD
t M 1 SS I ONS 1 N
NOX
709789?. 31
CO CUM t
.06
. 10
. 17
.22
.26
.32
. 38
.40
.46
.52
48
.67
.74
.81
.HQ
10.13
10. 18
1 C . 26
1 1 .98
12.05
12.1!
ft . 08
i6. 15
37 . 23
40 30
41 .67
42.21
45. 18
45.51
47. 18
50. IB
51 .66
54 . 75
54.92
57 . 95
59. 16
66.70
67 . 82
85.84
B7.57
eg. 20
H9 . 'J6
G9 . 76
91 .70
92.55
93. II
94. 15
95. 14
95.31
95. 77
95.94
96.08
96.70
97.51
97.73
9B.01
93. b7
98 . 95
99.55
100.00
K: TH1C ICA-j/YEAR
HC
24302534.0!
CO
9124534.74
FIGURE2
EMISSIONS STUDY
-------
CT>
T3 -*
C ^H
0) CO
OS
0\
OS
fl
U
3 a»
-a 43
a tu
o
o en
CO (0
O tfl
H iH
LI a.
xi
nj T3
-H oo a
4-1 C cr
X -H OJ
a a) u
00
CM
en ui
T3 0}
C (U
tfl T)
o
to o
CJ
U O
o
co nj to
00
a)
00
a) a>
00
LU
»s
c^
= 0
2 iu
o
o
75
-------
o
CD
re:
i -
en
f 3
CO
^
CD
* «
in
in
U"
n:
ct
r..
CO
z
o
I-
o
_j
* s<:
UJ . I
DC w ;
UJ
a.
or o c
LJO-
t .
3O ID O O O
^ O O O ('3 O
^ (:> GD cn o o
M M- V T" U"l 1C
I
oooc
OOOZ
OOSI
0001
005
- - o
D
76
-------
C3
t _,
'.3
CO
ft:
t
en
a
CO
^
O
in
en
IT
a'i
>
Cf.
1^
(.1
f;
CJ
fj
(f
U)
z
o
OO
u. w
CO
UJ
LJ
<7>CJ Ci CT) CT1 CT1 01 C
cno --r^-«r-(r'oooooooo
ooooooo
o o co en o o
S
i
OMU 000£1 OOOZ1 00011 0000! 0006
0009 OOO/ 0009 0005 000V 0001
'SNOISS1U3
-------
o
I -
D
(O
CO
r D
cn
^,
c ;>
CO
CO
ct
( 0
cj
i^
-
§
en
z
O
(0
Ul
" :'^ ci Ol
t.'^ hTi cn cn
c. cn ci cn
v * in cn
i - I t I I I I
r 1 i - T i ------ T ------ i ---- i -- r --
nrjSS CiOQS i.OSk TOOk 03SC OC10C C'35Z [10.1? COSI
'SN01SSIU3 XON
0001
DOS
78
-------
o
CO
re:
CO
11
en
xt
CD
LT)
CO
en en CJ> en Cf<
crv"cj>c7i!^C
LTVO ---- rn-«-s
I I I I I
P09i oo H ooei coo i 009 009
*U/(U)SN01G1]>I 'SN01SSIW1 OH
-- - o
0
Z
O
g
U| DC
los
5
UJ
o
79
-------
CD
I -
~J
rn
en
< i
in
s*.
o
tn
LO
c. I:JE ooo 3
r'.rg EOCS i.cn
ai/;w)iNuio"ii> 't'-.
oo
z
o
uj cc
cc w >
3 Z ii
o g s
U. (O
(A
i
UJ
o
o
80
-------
(A GC
Z <
gffij
00
CO (-
81
-------
10
J
U
flCE w
3 UJ UJ
= £§
UJ fc.
UJ
82
-------
( j
a)
\J
a)
r,J
o
LU HI
CO O
83
-------
y
N f>.,(K
jq,S2
CO
LU
CC 2 CO
3 Ul UJ
O *Z
rr O z
zo
UJ t
UJ
-------
CO
: uj
UJ I
UJ
85
-------
UJ
UJ
.O<
?o
UJ
££
^ t
Ul
cc
86
-------
** i£! - -1 j di nj o in r- us ru T j- 01 ru - - r- _r L j^ j"- LJ ro en o - UD CD
--tor~ in ru fxj - VJD ru .. . ru fo
-, nj oj
jr^)Mroj-inir)L03-^a>j-ir)iOj"iOj mLnmj-roLninruintoj-
O O O O O O O O O C.J O O CJ O CD CD O O O O O O O O U CD O O
UJ UJ UJ UHJ UJ UJ LJ Uf U (J tit UJ IJ LJ LJ U 1J IJ UJ UJ liJ UJ IJ U IJ UJ UJ
o o ru T r- CD -- j _j ro en 01 -- CD CD -1 j o ru ru - j 1/1 ai KI en u>
of^iDuiffl r-r-cnmj in m ru j- UTJ o o .T ru aj en CD - r- jrina
r'J^LonjCT'<\^ir)ix>G)j'<3rucio-^oicDfuj"iocDinrur**jCD'j"* en
tnr~'U}Mr---:r-^^in--Mj-i%jfu-i£""-iojfxj:fru-"- ru ru to
M 01 io r- o ID j co co en (.2 m M (- a> in if) f> :r un r r~ r- ai j- J en
04 - to m ir> ru - - ^o co o fL; ru r~ J -^ nj ^ ID
JlDTJ-lflJ-J-lOin^TT^-KllDj-lO
( J UJ lil lil IJ t iJ LJ \i) liJ IJ l«l lillil >«l M U !;J liJ UJ K] Ijj liJ l.J lit (.1 f.J |J |,j (jj
I. J; -j'- .'; r-'j j-iOTj; Hij [-(.:>' u n - - r~ ru o iy r ^ o iu OD o o
0 ft - - - t j as CD ro fu - -»j«.* fD '-* j- rj ii> iji ji HI rj en o CD m CD
- u*5 iD u") cr. j: o _r j- 'D to 01 - 'u -r> ix) o r- cj - - r- - - - r- *o(ijfo--cni;iinotojcn--u3f.iiiir"fXjLDMj[
<----
uj i0rucoi-.r ...-ip.- ruronjr-in
>- .
ui j 3- j ro in in o io m j in J in r j- .."* in m u"i in j ro in u1* ru kO in to
^ X fj C3 I J U O cy t-) O O O O O O O O , .' i 3 O O f J ! J ( J t 1 d ,'-} O O O
i;j0+44» + + + + + *4-t-t-t. + + + + + ++ + + + + + + f
- /! ;j :i,(;.. u M :j ' j id u u «.j i.j i,.' LJ u ' j u u u u NJ u ui l.J M L.J i j ul
s - -1 > j ro ji ru u~) IP if) i1^- (^ m u^ ILI '.3 nj - u) LD r^ j ro j (^ n_ ru CD
o o :o o M VD m en CD o o o m 13 : aj ^o ro en in j- r- ia r- 3- ui j- ro
a? t/i fv r~ r- r- nj - nj j) ro m i.n en ^) en nj CD o r- (.D L'I ru j- CD LO o ui
o:
»- IOMOJ T j--- -Kjftj r-*roioro--nj'-inmr^i/>.-njrxj--j'
UJ
x
cc «--t£)Ljiorumi£)jcoir)Lnajr^r^[najfUfioruu3r---'j-i£)o---i£)---
o un *
r- ^r iri fu 01 p* txj r* nj nj 10 P- '-ID aimr*)
z o
> !/ ru ui m ul r*l a"» un (D UD UD LTi ID in iD in T ifl in ID ID iD liT i u~ tO fU Jl LQ Lf>
S ..(^ '4 + + '+ + + + t4.4.+ 4.+ + + + + + 4 + +V + + + *+*
^ UJ IJ UJ UJ UJ UJ LJ U U LJ IjJ UJ UJ UJ UJ U U UJ UJ UJ U Ij.' UJ UJ LJ LJ UJ LJ
»-U^ ui --.*r(iuior-rLOicntDU3--rurocDcDoro jr-ccoMiDnjuio
^z j tnajf^r-njocDODcncjitOu^i^oin a.'iDo{cfDr~a3"LDCO
uO ----;ocncnrofi;a3j-cDU3roj-ru^- rocrj-jcjL^otDr^oco
>LO aj'-'-r-cnmru j-ru cnroj-j-rj ru tc j- in
- m a)
-- r*
^LJ HOJJ rocnjaiiviiDiD--ortiDruonjxtCPJ-LnLjCr)njomo
o
u>- ru ojrur^f*- ui ru 3- ^ aj a -tfi tornj-
ui tJ>C>CDOOOOC3aC3(O
ujcn cn-* + + * + -f + -*+-* + + + + + 'f + -*- + t+*->
h-UJLJUJLJWWLJtJM^l^LJLJLJUJLJtjJI^I^LJl^bJuJL^
uit njj-fu tDLncDrtjri-crij'inr^iinfXiMoi3-cDo--LDr'ruc3j'
m^ rxjinc)CDcr)LnGDcj)r^LnruciJinQj^r-coror*njj-roior'1Tr*-t
(I.-- njrvj'(^iPj
« o '
5 a. --rururu r-
M ---ixir^-ju!)in
m--roj-tfl----runjj- j-i^minin r- ru
u. u^ nj
r-r-nj cor- --inj-inr^C
tncDj-roiTJ
cnjipinocoL
. -ajfo JTCJ
fXJ
>« 3-ioin--ir>'-cDiniOJ«>tJCDf\i jjrL*iinr^cjitTcn--r*j-
' j--njiPtooj j-rou>--ru r^_rin -co ru
u, o> ru
^tn^r nj r^ cr: ru ru r- 1rcncOOT--roiniooLDrinjT'^Uij-pocn
-xtc?ij^."^f^'^ocnookn--tji : cjiotD^-iDTiPjioruczjcOLDo
O ^ ro -- nj r*^ iD \r* CO U) ru t j CJD cc no tc C" T CD ru if ru nj ip nj ^ ro ro
LJ
>
Z
LT)
r
LJ
Ln
ro
0.
*t
r
LJ
in
a ru
u ~
Ul
t/>
in
en t- £
en <
LJ (/)
>- Q O
-J LJ
3 i- in
g
§
a. o
en a
i- in
n
LU Z
> UJ
W Z >
"J z::
H
Z
O
a.
Z
<
Z
4
Z <
< O --
- o ?: _J z
_ ^u QiDz am^< >crLn-l/lIt/l I OXnrQ' j^J~OI~O
<^ jooo-roz Qri/iuLO'i iJtr-^i/.icjj>zcfn:cnin
. J O UJ J CJ _1 Z LJ < < < LJ LJ !*J O T 1J 3 T O UJ LJ LJ
15
87
-------
**
o
C ,)
in
ro
10
U
r-
r-
;n iT.
tf> i/")
< t O
.0
m
'->
r- ru - -
K; CD f J
iD lf> in
i i 1 1 i j
J
ru
in
t j
0'. ro to - (0 (OCO
ro j f-- ru (O
in in in io ID LH uo
(XXIJCDfDf-lO
in ID .3 r- r r- ru ru ji ro r~ to O)
a.' ru
^: ,; "
i
CD J 1-
\ 0
:» u"i in 01 - J PO
?o ID ro m j- en y
j-
ro
in
r j
(J
< i
r-
r-
*
LD
cn
LDID
[ J
^J
(1
T
o
w
lf> fJ3 J- tO
o ru j
in I.D in to
< i . ) n o
I J U Ld Ld
ro (D ID LD
jf - - Oj f\J
ro ro co ro
Ol
-
in
o
w
cn
ru *o - ru ro
m - - ru fu cn
- oj r,j in - in a* ru o ID nj --LDCD io iO ;r to CD
^ r~ f j ru ' nj j r- nj -
a:
o
\~
z
u
fr z
O--
^? y
cnr-
in-r cn
OLJ
CO U J
- cr ~>
S3
ro <
a u>
<: Cr
s: <
«
V lo
cr
>* »« ro
LO r*>
6
rr ro
t~ CD
UJ ro
5:
- ru
o ** in
in
3 ru .1
UJ <.) f3
m ^
uj
OJ
CD
T
T
), i..i i^J
^- 01 T
": -:TI o
fu ru i.n
u)
u ' i ^ rj
<-)
ro
ro
i j
PO
ru .1 1 in o T j
r- r r u. A. -.?<
in r i u) - * j ?*>
iL> 0) f\i n.] 7 'O
(jj . . ro - - ru ro
u 1 1: i in j u ' '.P
i.r ],1 i,J ',< i,< i,;
m 3 i > :n j: -
.- rl ^3 ' o>
ru
n cu
10 4
* - Id
<\l
(D
aj
"o r^i s
LJ CU C3
LJ [-1 L-!
-'D f r^J
tJ -- CD
b'l 'X
J-
CD
CD .1 fu - nj ai
.O U"- .1 T J .T
It 1 - I 1 J 1 3 C )
ro L'- CL _j ro ro
r- n... j j *) ro
rj [- j iD to
.j U! j ru c .1 (D
- j ru -
in if' j T- j- j
U,1 U (J (.; !j !.' f.j
U5 Hj f> ID r\j A_ -
r- j- f i -- r-o < 'i ro
If)
ro r- fu rO - -
i ) ro ro nj
cn r ' tn
ji o r- j
in .T UD r-
lOOj
in in
hi u
iDf-
cr- .t
J
LD T
(_> -"J
r^ jr
- -rj
- ru
ro
T .*
UJ IJ
j- r^
rjiM
LO
j- m
IJ !J
ID
UD r>
in in r
in --
3 m j-
0 0 t^
Id M Ld
if} ID ru
ru in rv
ro - rji
j io "
_T 0 ftj
J Id
(Dunru
ro ro ru
- tor-
in iDfu
- j- in
- ro
t LlUD
O « .1 O
UP UJ UJ Id l.J IJ
CD ro rx j 0) o
mr- r- in asarj
ro io o r* j o
::!
ru
3oi
52^
- - ro r-
j r- ru
ro ui LD
£
CJ> C 5 O
U UIU
to
i i CD a
.- ro --
LJ 1 J
ro -
ru io
ro r-
.- nj
ru f j
j- ro
in in1
-'- tD
ru -
o
i j id
om
Sm
in ro
ro
j m
m
ai
j ro
UJ
CD
in
r^
+
rj ro fu
r- ID ro in a rji
ru ro
y in
CZ) CD
id UJ
r- c^
*! (J}
~ O
m
~
o o
Ld Ld
m r-
10 ru
LD
00 JT
AJ ro
j- in
0 0
IxJUJ
o ro
UD CJ\
01
If)
OJ
or-
R
en
z
UJ
P
o) in
-J
.T o
a.
u ^~
O OJ
= 2
O (O
i
M
in
z
o
IU
CO
g
LU
08
ZCJ
|
01
CC
n r3 z -- a. -
t o o . j o z - - cr
.
. «r LO j: a ff
I U (3 t/1 4 U O
J< -- I T >-
< o
- Z _l
ir n o
< J ft L
> on <
.
Ln LJ I Lu O
< d
' Ifi
z > z
< ^ j in <_ . .. . _ _
j o u - j oj . j 7: i j <<<--- L.' _j u c
-> U O O O. C V 5 }' > L I' Z Z ^ .
< o r :?rno j-iT- ajo--r
1/1O
LJ~
T I
- .r tn if*
o tn r
oj r-
o: j-
LJ i
cr in
UJLJ
CD >-
88
-------
f- \
., I
ui O
m 52«
2 5 UJ
u-gz
(p
li'Jd
89
-------
cr
IvJ
n.
to
tf>
o
r O
\
:
.0_
S DC
"u, CO <
o Z UJ
«/ 00 Q >
1 ' HI 55 yj
-% ?:« Q.
8 K E oiZ
ii " O z
~ ^ CO O
DdJ - t ."JiSSIIIB
90
-------
LJ o
O 0,
«» O >
HI CO
UJ
!ii a.
HI
o
o o
-T> \*.
S'<
-ad - l .v. ; Sv .1
91
-------
NEDS Data
FPC Data
State
Alabama
Connecticut
Delaware
Florida
Georgia
Illinois
Indiana
Kentucky
Maine
Maryland
Massachusetts
Michigan
Mississippi
New Hampshire
New Jersey
New York
North Carolina
Ohio
Pennsylvania
Rhode Island
South Carolina
Tennessee
Vermont
Virginia
D.C.
West Virginia
Wisconsin
# plants with
some electric power
SIC Codes
3
27
5
46
14
50
48
20
11
16
28
62
12
7
26
59
14
30
65
4
14
9
5
14
2
12
27
// plants with
all electric power
SIC Codes
2
27
4
35
14
48
47
17
10
16
28
58
12
7
26
58
14
29
60
4
14
8
4
14
2
12
27
1974
# plants
15
11
4
44
15
43
29
21
4
12
18
31
13
3
20
34
17
39
42
2
14
8
2
14
2
13
25
FIGURE 20
COMPARISON OF ELECTRIC POWER PLANTS
IN NEDS AND FPC DATA BASES
92
-------
facilitated. Since the overall objective of the RAINE phase of
MAP3S is "To define the relationships between the emissions of air
pollutants, the deposition and the chemical quality of precipita-
tion," the maintenance, updating and upgrading of the inventory are
being planned as continuing efforts. Thus we are building a dynamic
information base from which researchers can draw to fulfill their
mandate.
93
-------
-------
APPENDIX I
Summary Reports on Air Pollution Emissions Inventories
Emissions Inventories for States
IDAHO, WASH
CALIFORNIA (EXCEPT LOS ANGELES COUNTY)
ONSIN, VIRGINIA
Aug 71
Jun 74
Feb 74
Nov 75
Jan 70
Sep 71
Nov 73
Oct 71
Aug 71
SEP 71
Dec 75
Jul 71
Jan 72
Dec 77
Aug 71
72
Sep 72
Aug 71
72
Apr 76
Apr 76
Nov 75
Aug 71
Oct 71
Nov 73
Oct 71
Sep 75
Aug 71
Dec 75
Sep 71
Aug 71
Emissions
Feb 74
Oct 75
Oct 75
Dec 75
Dec 75
Sep 75
Sep 75
Jul 76
Oct 75
Oct 75
Oct 69
PB
PB
PB
PB
PB
PB
PB
APTD
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
EPA
EPA
EPA
PB
PB
PB
PB
PB
PB
PB
PB
203 176
253 301
231 698
265 735
210 564
204 190
230 938
0732
204 383
265 750
202 255
212 606
275 388
203 083
260 120
274 454
204 384
222 835
904/9-76-005a
904/9-76-005b
908/1-76-002
220 211
203 812
230 930
204 382
210 787
265 743
203 503
210 430
ALASK,
ALASK,
CALIFi
COLOR
HAWAI
IDAHO
KANSA:
MINNE:
MISSOl
MONTA;
MONTAl
NEBRA:
NEW a
NEW HJ
NEW Jl
NEW Jj
NEW J:
NEW Ml
NEW Y(
NORTH
NORTH
NORTH
NORTH
OKLAHt
OREGOI
PUERTf
RHODE
SOUTH
SOUTH
VERM01
WYOMII
Inventories for Counties
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
231 696
250 386
250 387
256 004
256 005
250 382
250 383
258 139
258 137
258 138
205 276
CA ~ 1
GA - (
GA - (
GA - I
GA - I
GA - ]
GA -
TN - (
TN - 1
GA - I
GA -
VT - (
Los Angeles County
Chatham County Vol I
Chatham County Vol II
Dougherty County Vol I
Dougherty County Vol II
Fulton, DeKalb, Cobb, Clayton,
Gwinnett Counties Vol I
" Vol II
Cheatham, Davidson, Robertson, Ruthers-
ford, Sumner, Williamson & Wilson Co.
Hamilton County
Walker & CartoosaCounties Vol I
Vol II
Chittenden County
95
-------
Appendix I (continued)
C.
Emissions Inventories for Cities and Areas
Oct 69
Sep 75
Dec 77
Aug 68
Jul 70
Apr 70
Oct 69
Jun 70
Apr 68
Jan 70
Oct 69
Apr 70
Sep 75
Apr 70
Dec 68
Mar 70
Aug 68
Jan 71
Feb 70
May 70
Aug 70
Jun 70
Dec 68
May 77
Feb 71
May 70
Aug 70
Dec 70
Nov 68
Aug 69
Jul 69
Feb 70
Mar 70
Apr 67
Jan 71
Feb 69
Jan 71
Sep 69
Mar 70
May 70
Jul 69
Jun 69
Feb 70
Dec 68
Feb 69
Jul 70
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
PB
207
269
278
206
207
205
207
207
227
205
207
205
252
205
206
207
220
207
207
207
207
205
206
268
207
207
205
207
206
205
206
207
207
207
207
207
207
206
207
207
207
206
207
206
220
207
696
907
178
757
691
246
686
699
354
255
693
278
698
245
819
690
847
684
695
748
697
244
243
242
749
747
270
688
242
268
115
694
752
648
750
267
751
469
685
689
691
244
698
245
481
687
Phoenix - Tucson Metropolitan Area
Phoenix - Tucson " "
Phoenix Area
Denver
Mid-Connecticut, Lower Pioneer Valley
Miami - Fort Lauderdale - West Palm Beach
Atlanta Metropolitan Area
Treasure Valley Area, Idaho
Northwest Indiana
Baton Rouge Metropolitan Area
New Orleans Metropolitan Area
Portland Metropolitan Area
Pioneer Valley, Mass.
Merrimack Valley Metropolitan Area
Metropolitan Indianapolis
Fargo - Moorehead
St Louis
Billings, Montana
Omaha Metropolitan Area
Las Vegas Metropolitan Area
Reno Area
Albuquerque Metropolitan Area
Buffalo, NY
Capital District, NY
NY State Southern Tier West
Charlotte Metropolitan Area
Triangle Metropolitan Area
(Raleigh, Durham, Chapel Hill)
Columbus Metropolitan Area
Dayton
Springfield
Toledo
Oklahoma City Metropolitan Area
Willamette Valley Metropolitan Area
Pittsburgh Metropolitan Area
San Juan Puerto Rico
Providence - Pawtucket, New Bedford,
Fall River, R.I.
Sioux Falls Metropolitan Area
Memphis Metropolitan Area
Beaumont - Port Arthur, Texas
El Paso Metropolitan Area Texas
Houston
San Antonio Metropolitan Area
Salt Lake City - Provo - Ogden
Seattle - Tacoraa
Milwaukee Metropolitan Area
Cheyenne, Wyoming
96
-------
REFERENCES
1. MacCracken, M. C. (Coordinator), "The Multistate Atmospheric
Power Production Pollution Study-MAP3S Progress Report for FY
1977 and FY 1978," DOE/EV-0040 (July 1979).
2. Littman, F. E., "RAPS Emissions Inventory-A Retrospective
Evaluation," Proceedings of the Emissions Factors and Inventory
Specialty Conference, Air Pollution Control Association, Anaheim,
CA. (Nov. 1978).
3. Holland, T. C., Crenshaw, J. D., Wehe, A. H. and Potter, J. D.,
"Energy Data System Terminal Users Manual," EPA (Nov. 1976).
4. "Aeros Manual Series Volume I-Aeros Overview," EPA-
450/2-76-001 (OAQPS No. 1.2-038) (Feb. 1976).
5. Benkovitz, C. M., "Compiling a Multistate Emissions Inventory,"
submitted to the Journal of the Air Pollution Control Association
(Oct. 1979).
6. Executive Office of the President, Office of Management and
Budget, U.S. Government, "Standard Industrial Classification
Manual," (1972).
7. Klemm, H. A., and Brannan, R. J., "Emissions Inventory in the
SURE Region," GCA-TR-79-49-G Draft Final Report, (Sept. 1979).
97
-------
-------
OBSERVED METEOROLOGICAL DATA BASES FOR POLLUTION MODELING
J. L. Heffter - NOAA, Air Resources Laboratories
Introduction
When a pollutant is released into the atmosphere, it is trans-
ported by the winds and dispersed by diffusion and deposition proces-
ses. For releases near the ground, the portion of the pollutant that
remains within the lower atmosphere most affects surface air concen-
trations and deposition amounts. Boundary layer meteorological data,
therefore, are an essential input to every air pollution model in
order to determine pollutant transport and dispersion. This report
outlines various data groupings and discusses data bases within the
groups most applicable and accessible for pollution modeling.
Data Grouping
Observed meteorological data used in studies of the transport
and dispersion of pollutants can be grouped in several ways of which
two are considered here.
1. Synoptic grouping - observations at all locations for suc-
cessive time periods.
a) Station format
b) Gridded format
2. Station grouping - observations at one location for suc-
cessive time periods.
Synoptic grouping is considered the most applicable for air pollution
model input since calculations must take into account spacial varia-
tions with time. Emphasis here will be on station formatted data
99
-------
which include upper air and surface data bases. Station grouped
tower data will also be briefly discussed.
SynopticGrouping: UpperAir Data Bases
Global upper air observed meteorological data are collected by
the U.S. Air Force, sorted into a synoptic (time) grouping, and
stored on magnetic tape (one month of data on 2 tapes). The Air
Resources Laboratories (ARL) extracts observations in the lower
atmosphere from these tapes for specific geographic areas of inter-
est. A data base called NAMER-WINDTEMP has been created that con-
tains upper air winds and temperatures from rawinsonde and pibal
stations for North America (excluding Alaska) from the surface to 6
km (or 500 mb). Station identification information, including an
average terrain height at each station and observed winds, tempera-
tures and heights are recorded for four observation times per day
(00, 06, 12 and 182). The 00 and 12Z time sequences contain approxi-
mately 130 reports; 18Z contains about 20 reports, and 06Z about 10
reports. One year of data is stored on 2 to 3 magnetic tapes. Four
years of data (1975 to 1978) are presently archived at the National
Climatic Center (NCC), Asheville, NC. Detailed information on the
NAMER-WINDTEMP data tapes is given in Appendix A.
A second upper air data base is also available. Global data are
collected by the National Weather Service, NOAA, subjected to quality
control procedures, and archived at NCC as the TD-5681 synoptic
grouped data base. Data tapes contain rawinsonde observations only
100
-------
(no pibals) over the entire globe for 00 and 12Z (no 06Z or 182
reports). Significant level wind information is identified only by
pressure level (no heights are given). One month of data is stored
on 2 to 3 magnetic tapes. Eight years of data (1971 to 1978) are
presently available at NCC. Detailed data information is given in
Appendix B.
Surface j)ata Base
Hourly surface meteorological observations in synoptic grouping
for stations between 100°W and 60°W, 50°N and 20°N have been put on
magnetic tape by NCC. About 600 surface stations in this area report
each hour. As many as 35 identification and meteorological parame-
ters are recorded in each report at a station. One month of data is
stored on a magnetic tape. Four years of data (1975 to 1978) are
presently archived at NCC as the TD-9687 surface data base. Detailed
data information is given in Appendix C.
Station Grouping; Tower Data Base
Meteorological tower data are collected at approximately 90
power plant sites throughout the U.S. These data, contained in
Nuclear Regulatory Commission Reports for individual sites, are
available through public documentation. They are, by nature, station
grouped, but may be useful to augment the other data bases discussed
here.
101
-------
-------
APPENDIX A
NAMER-WINDTEMP DATA
MAGNETIC TAPE FORMAT
NAMER-WINDTEMP data tapes contain rawinsonde and pibal observations
for North America (excluding Alaska) from the surface to 6 km (or 500
mb) TAPE CHARACTERISTICS
TAPE - 9 track, 1600 bpi, EBCDIC
LABEL - None
RECORD FORMAT - FB
RECORD LENGTH - 30
BLOCK SIZE - 12000
TAPE ORGANIZATION
All reporting stations, in block-station sequence, are compiled for
each sequential observation time.
4 observation times per day (0,6,12,18 GMT)
2 files per month (day 01 to 15; day 16 to last)
12 files per tape (6 months)
DATA ORGANIZATION FOR EACH OBSERVATION
TIME TIME REC (FOR WINDS)
STA REC (STATION 1)
WIND REC (HEIGHT 1)
WIND REC (HEIGHT 2)
ETC.
103
-------
STA REC (STATION 2)
WIND REC (HEIGHT 1)
WIND REC (HEIGHT 2)
ETC.
ETC.
TIME REC (FOR TEMPERATURES)
STA REC (STATION 1)
TEMP REC (HEIGHT 1)
TEMP REC (HEIGHT 2)
ETC.
STA REC (STATION 2)
TEMP REC (HEIGHT 1)
TEMP REC (HEIGHT 2)
ETC.
ETC.
DATA FORMAT
TIME REC: MONTH (1st HOUR NUMBER OF NUMBER OF
3 LETTERS) YEAR DAY (GMT) REPORTS RECORDS MET FIELD
A3
14 12
12
15
Al
W=WINDS
T=TEMPS
STA REC: AVG
BLOCK LATITUDE LONGITUDE STATION HGT TERRAIN NUMBER OF
STATION (DEC*100) (DEC*100) (M, MSL) (HGT(M.MSL) LEVELS
15
15
17
15
15
12
WIND REC: WIND HGT WIND DIRECTION WIND SPEED
(M, MSL) (DEC) (M/S*10)
15
13
14
104
-------
TEMP REC: TEMPERATURE PRESSURE TEMPERATURE
HGT (M,MSL) (Mb*10) (DEC K*10)
14 15 14
NAMER-WINDTEMP data tapes starting for the year 1975 (refer to tape
deck #9743) are available at:
National Climatic Center, NOAA
Federal Building
Asheville, NC 28801
Attn: Steve Doty
(Tel: 704-258-2850, Ext. 203 or FTS: 672-0203)
105
-------
-------
APPENDIX B
GLOBAL RAWINSONDE DATA
MAGNETIC TAPE FORMAT
TAPE
POSITIONS
01-04
05-08
09-12
13-17
ELEMENT
BLOCK LENGTH
OBSERVATION LENGTH
DECK NUMBER
STATION NUMBER OR
LATITUDE
Number of bytes in this physi-
cal record - in binary. This
occurs once each block.
Number of bytes in this logical
record - in binary. This field
occurs at the beginning of each
observation.
Unique for each type or source
of data.
WMO block-index number or
latitude in degrees and tenths.
18-19 YEAR
20-21 MONTH
22-23 DAY
24-25 HOUR
26-27 NUMBER OF LEVELS
28-33 BLANK OR LONGITUDE
99LaLaLa = Ships
+
OOLaLaLa = Sirs
Signed Position:
+ = Northern Hemisphere
- = Southern Hemisphere
78 = 1978 etc.
01 - 12 = Jan.-Dec.
01 - 31 = Day of month
00 - 23 = GMT
Number of 25 character levels
contained in this observation.
Blank for land stations - west
longitude in degrees and tenths
for other observations.
107
-------
TAPE
POSITIONS
ELEMENT
34-38
39-43
PRESSURE
HEIGHT
44-46
TEMPERATURE
47-49
RELATIVE HUMIDITY
50-52
52-55
WIND DIRECTION
WIND SPEED
99LoLoLoLo = Ships
OOLoLoLoLo = Sirs
LoLoLoLo = 000.0-360.0
starting at
zero meridian
and increas-
ing in
westerly
direction.
090.0 = 90.0°W
270.0 = 90.0°E
Pressure of the level in mil-
libars and tenths.
Height of the level, above sea
level, in geopotential meters.
Signed plus = HGT above
sea level
Signed minus = HGT below
sea level
Temperature of the level in de-
grees Celsius and tenths.
Signed plus = Positive
temp
Signed minus = Negative
temp
Relative humidity of the level
in whole percent.
Signed plus = Actual RH
Signed minus - Statistical
RH
Wind direction of the level in
whole degrees.
Wind speed of the level in
meters per second.
108
-------
TAPE
POSITIONS
56
ELEMENT
HEIGHT INDICATOR
57
TEMPERATURE INDICATOR
Height indicator for the level.
Blank = HGT as reported
1 = HGT computed
2 = HGT recomputed
during QC
Temperature indicator for the
level.
58
BLANK
Blank = Temp as reported
1 = Temp computed
2 = Temp recomputed
during QC
This could be used for other QC
information.
Each data level is 25 bytes. Missing data fields are coded as all
9's with signed fields being signed minus. The first level is always
the surface level. All other levels then follow in decreasing pres~
sure or ascending height order. For those observations where no sur-
face data are available the first level is 9 filled.
Observations are packed as many as possible into variable length
blocks that do not exceed 6000 bytes.
109
-------
-------
APPENDIX C
SURFACE WEATHER OBSERVATIONS
MAGNETIC TAPE FORMAT
TAPE CHARACTERISTICS
TYPE - 9 track, 1600 bpi, BINARY (8 bits/byte)
LABEL - BLP
RECFM - FB
LRECL - 100
BLKSIZE - 1000
TAPE ORGANIZATION
One file per month per tape
DATA ORGANIZATION
All records are the same. Each record contains one observation
at one time for one station.
DATA FORMAT
Parameter, units or WMO code, and length in bytes are as fol-
lows for each record:
block station number, code 4
year, GMT, 2
month, GMT, 2
day, GMT, 2
hour, GMT, 2
minute, GMT, 2
report type, code, 2
observation type, code, 2
111
-------
latitude, degrees/minutes, 2
longitude, degrees/minutes, 2
station elevation, meters, 2
special ship, code, 2
quadrant, code, 2
wind indicator, code, 2
call letters, EBCDIC, 4
station control, code, 2
wind direction, degrees, 2
wind speed, tenths m/s, 2
wind gusts, tenths m/s, 2
sea level pressure, tenths millibars, 2
barometric tendency, code, 2
dry bulb temperature, tenths degrees Kelvin, 2
dew point depression, tenths degrees Kelvin, 2
altimeter setting, hundredths inches, 2
6 hr precipitation amount, code, 2
sky cover, code, 2
past weather, code, 2
visibility, meters, 4
visibility characteristic, code, 2
present weather, code, 2
present weather, code, 2
present weather, code, 2
present weather, code, 2
112
-------
station pressure, tenths millibars, 2
cloud cover, code, 2
type of low cloud, code, 2
height of low cloud, code, 2
type of middle cloud, code, 2
type of high cloud, code, 2
amount of cloud, code, 2
cloud classification, code, 2
cloud type, code, 2
height of cloud base, code, 2
cloud height device, code, 2
cloud layer characteristic, EBCDIC, 2
ceiling, code, 2
SPECIAL CODES
Missing data are given by -1 in the field.
113
-------
-------
SULFATE, VISIBILITY, AND PRECIPITATION CHEMISTRY
DATA BASES AND RESULTS FOR REGIONAL MODELING
B. Nieman - Teknekron Research, Inc.
1.0 INTRODUCTION
Regional air pollution models require considerable quantities of
input data to run and observed data to evaluate and refine their sim-
ulations. The emissions and meteorological data bases for regional
models are discussed in other invited papers. The purpose of this
paper is to provide an overview of the air quality, visibility and
precipitation chemistry data bases that are available and being used
in current regional modeling efforts. The principal focus on air
quality data bases in this paper is on sulfates since they are
thought to provide the best surrogate for fine particle (less than 5
micron diameter) concentrations until data from the EPA inhalable
particulate monitoring sites become available. The principal focus
on visibility data bases in this paper is on airport visual range and
turbidity measurements and satellite photographs since these are all
that is available until networks of more sophisticated measurements
have been fully implemented. The principal focus on precipitation
chemistry data bases in this paper is on that produced by so-called
event and subevent monitoring and the precipitation and nephanalysis
(cloud) data bases required for regional modeling of atmospheric de-
position (acid rain).
115
-------
Another purpose of this paper is to suggest that an integration
of air quality, precipitation, precipitation chemistry, and cloud and
satellite data bases is necessary for the successful modeling of wet
sulfur deposition on a regional scale. The latter type of modeling
is currently thought to be more difficult than regional modeling of
visibility degradation and both types of modeling require the basic
capability of modeling fine particulate concentrations on a regional
scale.
The concept of integrating data bases to support regional wet
sulfur deposition (acid rain) modeling can be illustrated with satel-
lite photographs during major sulfate episodes. One of the best ex-
amples is the major sulfate episode during July 18-23, 1978, over the
northeastern United States. The normal satellite photograph at mid-
day on July 22, 1978, at the height of the episode, shows a rather
faint milky colored air mass over most of the northeastern United
States and the distinct white clouds associated with a frontal system
moving southeastward across the Great Lakes. A digital enhancement
of the same satellite photograph on July 22, 1978, shows a bright
white in place of the faint milky colored air mass and a speckled
texture in place of the clouds. More importantly, this enhanced
photograph shows the polluted air mass being drawn in among the
clouds, thereby setting the stage for entrainment into the clouds and
wet removal by precipitation processes. The ambient sulfate and wet
sulfur deposition data on this day show elevated concentrations and
116
-------
depositions in the hazy and cloudy air masses, respectively. This
interesting case study will be discussed more throughout the paper.
There would seem to be four principal answers to the question
"why look at data?" in the context of regional modeling as follows:
A considerable quantity of data is available to everyone now
while participation in model development and applications
will probably be limited to certain groups
Data analysis provides basic insights into atmospheric
processes and source-receptor relationships
Results may allow for considerable simplification and
increased accuracy in regional modeling
Results will suggest improvements in future measurement
programs
The question of model accuracy is one that is frequently raised
in regulatory situations, but is one that should also be raised in
the sense of what the data bases will allow especially on the region-
al scale. There seems to be a growing consensus that the government
and industry have acquired a lot of data in recent years and now is
the time to provide funding and priority to the analysis of those
data.
The data base acquisition and results presented in this paper
are the result of work on the Ohio River Basin Energy Study (ORBES)
for the University of Illinois, Regional Air Quality Studies for the
EPA Office of Energy, Minerals, and Industry, and an Integrated Mon-
itoring Network for Acid Deposition for the EPA Office of Anticipa-
tory Research. Figure 1 is a flowchart of the master data base
organization and episode retrieval system developed for ORBES. This
117
-------
118
-------
figure shows how EPA SAROAD (Storage and Retrieval of Air Quality
Data), SCMD (Special Continuous Monitoring Data), NCC (National
Climatic Center), NESS (National Environmental Satellite Service),
and precipitation chemistry data bases are analyzed and integrated to
support subregional and regional modeling. The method of selecting
sulfate, visibility, and wet sulfur deposition episodes will be de-
scribed later in the paper.
The overview of sulfate, visibility, and precipitation chemistry
data bases and results for regional modeling is given in the next
three sections of the paper. The principal conclusions and recom-
mendations are given in the last section of the paper.
119
-------
-------
2.0 SULFATES
In this section, the monitoring locations, trends, and spatial
distribution of sulfates are discussed along with the selection of
episodes and examples of higher resolution data.
2.1 Monitoring Locations
The NADB (National Aeroinetric Data Bank) sulfate monitoring loca-
tions in the western states (see Figure 2) are mainly clustered around
population centers with large remote areas of interest to regional
modeling like the Four Corners area with very few monitors. Fortu-
nately, the monitoring gap in the Four Corners region has been par-
tially filled in recent years by the Ute Sulfate Monitoring Network
(see Figure 3). Unfortunately, the rather dense Ute Network has been
discontinued recently and is being replaced by a less dense fine par-
ticle sampling and visibility monitoring network to be discussed in
the next section (see Figure 45).
The NADB sulfate monitoring locations in the eastern states (see
Figure 4) show a somewhat more uniform spatial distribution than in
the west with the greatest density of monitors in Ohio, Pennsylvania
and Maryland. Since most of the NADB sulfate monitors are in urban or
industrialized areas, there is a need for monitors in rural and even
remote areas, a need which has been filled recently for all but the
extreme southern states by the SURE II (Sulfate Regional Experiment
Phase II) (see Figure 5 and Table 1). Additionally, EPA/OAQPS and
EPA/ESRL are establishing national ambient background sites that will
121
-------
F1GURE2
SULFATE MONITORING LOCATIONS
122
-------
LU
LU
V)
in
UJ
o
us
V)
UJ
UJ
123
-------
FIGURE4
SULFATE MONITORING LOCATIONS (1975-77) FROM EPA NATIONAL
AEROMETRIC DATA BANK
124
-------
FIGURES
SURE II STATION NUMBERS AND LOCATIONS
125
-------
TABLE 1
KEY TO SURE II STATION NUMBER AND LOCATIONS
Number
Station Name
City
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Montague
Scranton
Indian River
Philo
Rockport
Giles City
Ft. Wayne
Chapel Hill
Lewisburg
Beverly
Fall River
Albany
Oswego
Dunkirk
Roseton
Warren
Lewisville
Brush Valley
Wilmington
Parkersburg
Madison
Louisa
Sullivan
Detroit
Fort Huron
Springfield
Braidwood
Columbus
Conneaut
Toronto
Huntington
Loves Mill
Hytop
L-B-L
Paradise
Memphis
Hanover
St. Louis
Nekoosa
Ithaca
Amherst, MA
Scranton, PA
Millsboro, DE
Zanesville, OH
Owensboro, KY
Huntsville, AL
Ft. Wayne, IN
Raleigh, NC
Lewisburg, WV
Boston, MA
Boston, MA
Albany, NY
Oswego, NY
Buffalo, NY
Poughkeepsie, NY
Erie, PA
Johnstown, PA
Johnstown, PA
Wilmington, DE
Parkersburg, WV
Louisville, KY
Ashland, KY
Terre Haute, IN
Detroit, MI
Detroit, MI
Springfield, IL
Joliet, IL
Columbus, OH
Erie, PA
Toronto, Ont.
New York, NY
Bristol, TN
Chattanooga, TN
Nashville, TN
Madisonville, KY
Memphis, TN
Hanover, NH
St. Louis, MO
Madison, WI
Ithaca, NY
126
-------
TABLE 1 (Continued)
Number
Station Name
City
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Lafayette
Dayton
Mt. Storm
Chesterfield
Yorktown
Marshall
Weatherspoon
Atlanta
Columbia
Chester
Whiteface Mtn.
York
Rents Hill
Meredosia
Lafayette, IN
Dayton, OH
Morgantown, WV
Richmond, VA
Newport News, VA
Charlotte, NC
Lumberton, NC
Atlanta, GA
Columbia, SC
New York, NY
Plattsburg, NY
York, PA
Augusta, ME
Jacksonville, IL
127
-------
provide sulfate as well as other parameters like ozone and precipi-
tation chemistry.
2.2 Trends
Although there are serious limitations in the historical sulfate
data (1960-1978) which preclude a rigorous trends analysis, it is of
interest to compute the yearly and seasonal average values over all
the measurements and monitoring sites in the NADB for various AQCRs
(Air Quality Control Regions), states, and subregions to see what the
variations are. Later in this paper, the results of a more rigorous
trends analysis for a 3 state region (Ohio, Pennsylvania, and West
Virginia) with consistently elevated sulfate levels and a period
(1974-1978) with a large number of measurements from a large number
of stable monitors in urban or industrial areas is presented.
The suifate concentration "trends" in the Four Corners Air Qual-
ity Control Region (AQCR Number 14; see Figure 6) on a yearly basis
and in the summer and winter seasons show considerable fluctuations
before and after reaching a peak value in 1973-1974. In contrast,
the sulfate concentration "trends" in the Northern Great Plains States
(Montana, North Dakota, and South Dakota; see Figure 7) are generally
based on more measurements and have higher average concentrations than
for AQCR 14, but show no discernible trends during 1960-1978.
The sulfate trends analysis for the eastern United States was
made for the same subregion used in the ORBES regional transport
model (see Figure 8) except subregions 1 and 2 were combined into one
128
-------
s
O
O
O
( m/Srf)
129
-------
t
o
NTER
he
above
average
The numbe
<
<
*2
i°
* *
2 <
S°
II
K «
ZS
8<
5<
a: »-
^°
z ^:
UJ 4
»1
8t
gs
LL
_l
3
VI
( m/Srf)
130
-------
Lower ORBES
Upper ORBES
South
12345678 9101112131415161718192021222324252627282930
Note: Dashed lines show extrapolated state
boundaries outside the study region.
FIGURES
BOUNDARIES OF SUBREGIONS CONSIDERED IN THE ORBES REGIONAL
TRANSPORT MODEL
131
-------
subregion and extended eastward to include the entire state of Pennsyl-
vania. The sulfate "trends" in the 6 ORBES states (see Figure 9) have
generally declined on an annual basis, except in 1978 when it rose,
fluctuated about a nearly constant value in the summer season, and
declined in the winter season. The fact that the winter (December-
February) season sulfate concentration was greater than that for the
summer (June-August) season in 1978 makes that an interesting year for
analysis and regional modeling. The general trends in annual and
seasonal average sulfates may be due primarily to changes in SOX
emissions from various source categories. During the 1960's, there
was a larger contribution to sulfates from urban sources whose plumes
were trapped near the ground especially in the winter season while in
the 1970's there was a larger contribution from elevated sources in
rural areas whose emissions were transported over longer distances and
subject to chemical conversions that were more active in the summer
season.
The sulfate "trends" in the subregions to the south and north-
east of ORBES (Figures 10 and 11) show no discernible trends and a
general decline, respectively, on an annual basis. Interestingly, the
summer season sulfate "trend" in the subregion south of ORBES shows a
general increase while the winter season shows a general decrease.
These "trends" results suggest some interesting features for
multi-year regional models to try to simulate.
132
-------
( ui/£rf) suof j
<2z
- (C E
5
o 2 <
ujxi
« 2= MOW
u.
UJ Q.
?
-------
O
J3
0V
0
r-
m
u-i
f~
r~4
rH
i 1
< I
CO
-H
,-4
CO
<
c
C
^
CO
in
l-x.
S
R
S
0
CM
-
N
O
>
3
OO
r-.
r--
1C
t-l
rt
QJ
(SJ *J
^ G>
O
^-i XI
o 5
^ j:
ad M
^
O\ H M
i£> S -^
M
s ?
H
£0
>
r*- a!
a*
* 5
m in
C
* B
s 2
01
CO 01
sD 1)
tM
C^ O
!j
-------
C3£
Uj <*
EC 1~
fflW
=> ?
w w
Z <
cc
Il
LU O
O i-
ZW
ni
135
-------
It should be emphasized that any firm conclusions regarding
trends in the various AQCRs, states, and subregions based on the NADB
data base must take into account the changes in monitor locations,
ratio of urban to rural monitors, and number of measurements from year
to year.
2.3 Spat i a 1 Pi s t r i but i on
The "trends" in the spatial distribution of sulfates over the
eastern United States during 1960-1978 are of interest to the analysis
and modeling of source-receptor relationships. Again it should be
mentioned there are serious limitations in historical sulfate data
that precludes their use in a rigorous analysis of trends in spatial
distributions; however, it is still of interest to see what the vari-
ations are and what: insight they provide to regional modeling.
Because the number of sulfate measurements in any one AQCR in any
one year, especially during 1960-1969, is not sufficient to provide a
representative average value, the yearly values for multi-year periods
were computed and then isopleths were drawn by the author. The
periods 1960-1964, 1965-1969, 1970-1974, and 1975-1977 were selected
for computing the AQCR averages and these periods were found to
generally have the same number of stagnating anticyclones according to
criteria. Their result suggests that the affect of year-to-year
variations in meteorology that affects regional sulfate concentrations
was normalized to a first approximation by the averaging periods
selected.
136
-------
The five-year average (1960-1964) of AQCR average sulfate concen-
trations (see Figure 12) shows a region of elevated (greater than 12
(ig/m3) concentrations extending eastward from the upper Ohio River
Basin to the Atlantic Coast with a maximum of about 22 (jLg/m3 cen-
tered over the Huntington, West Virginia area. By contrast, the five-
year average (1965-1969) of AQCR average sulfate concentrations (see
Figure 12) shows a smaller region of elevated concentrations than dur-
ing 1960-1964 and a lower maximum concentration (about 16 ^g/m3). A
subregion of elevated sulfate concentrations along the extreme south-
eastern coast of Georgia and Florida is also apparent in Figure 13.
It should be noted that the region of maximum sulfate concentrations
during 1960-1969 generally coincides with the AQCR with the largest
number of measurements. This means that data base generally allows
one to resolve the region of maximum sulfate concentrations better
than the regions of low sulfate concentrations.
The five-year average (1970-1975) of AQCR average sulfate con-
centrations (see Figure 14) shows the region of elevated sulfate con-
centrations has extended west along the Ohio River Basin similar to
its position for the 1960-1964 period, while the maximum concentra-
tion has increased to about 20 |ag/m3 centered over the West Virginia
panhandle compared to the 1965-1969 period. The development of a sub-
region of elevated sulfate concentrations in northern Illinois is also
apparent in Figure 14. The three-year average (1975-1977) of AQCR
average sulfate concentrations (see Figure 15) shows the region of
137
-------
FIGURE 12
FIVE-YEAR AVERAGE (1960-1964) OF AQCR AVERAGE
SULFATE CONCENTRATIONS (M9/m3)
138
-------
FIGURE 13
FIVE-YEAR AVERAGE (1965-1969) OF AQCR AVERAGE
SULFATE CONCENTRATIONS
139
-------
FIGURE 14
FIVE-YEAR AVERAGE (1970-1974) OF AQCR AVERAGE
SULFATE CONCENTRATIONS (Mg/m3)
140
-------
10
10
FIGURE15
THREE-YEAR AVERAGE (1975-1977) OF AQCR AVERAGE
SULFATE CONCENTRATIONS
141
-------
elevated sulfate concentrations has decreased in size and shifted
more westward into Illinois.
In all 4 periods analyzed, the region covered by the sulfate con-
centrations isopleth of 8 fig/ra3 was generally the same while the
region covered by 10 fig/ra3 was generally the same in all the periods
but 1965-1969, when the broad regional pattern seemed to break down
into pockets of subregional size. Thus, the main changes in spatial
distribution that occurred during 1960-1978 appear to be in the size
and general location of the region of elevated concentrations (greater
than 12 |j.g/ra3) and in the maximum concentration. It is of interest
to see if these features can be simulated by a multi-year regional
model using the changes in emissions from various source categories
that have occurred.
The strong seasonality of sulfate concentrations in various sub-
regions of the eastern United States is apparent in the isopleths of
summer and winter season sulfate concentrations to annual average con-
centrations (see Figures 16 and 17). The year 1976 was selected as an
example because it had the largest number of sulfate measurements of
any year during 1974-1978. The isopleths of seasonal to annual sul-
fate concentrations show values of 150% over the upper Ohio River
Basin, and central Georgia and North Carolina in the summer season and
values greater than 100% over Illinois and eastern South Carolina in
the winter season.
While the NADB sulfate data used in the analysis of trends in
spatial distributions represents a mixture of urban and rural
142
-------
100
FIGURE 16
ISOPLETHS OF SUMMER SULFATE CONCENTRATIONS AS A
PERCENTAGE OF ANNUAL AVERAGE CONCENTRATIONS, 1976
143
-------
100
FIGURE 17
ISOPLOTHS OF WINTER SULFATE CONCENTRATIONS AS A PERCENTAGE
OF ANNUAL AVERAGE CONCENTRATIONS, 1976
144
-------
measurement sites, the SURE II sites were purposely selected to
represent rural locations entirely. The "annual average" sulfate and
nitrate concentrations (see Figures 18 and 19) show very different
magnitude and patterns from one another. The maximum average sulfate
concentrations in the SURE II were about 8 (j.g/ra3 centered over the
Ohio River Basin while the maximum average nitrate concentrations were
only about 1 jig/mB, displaced well north of the Basin and located
over or downwind of major urban areas. As with the NADB data, SURE II
data isopleth analysis is based on averages containing from 4-11
months of data at the 54 sites. The general conclusions above are the
same if only the same four months of data at all 54 sites are used
instead of all the data at all the sites.
2.4 Episodes
The identification, frequency, and spatial distribution of epi-
sodes are discussed in this subsection along with the frequency dis-
tribution and "episodicity" of annual sulfate concentrations in the
3-state region (Ohio, Pennsylvania, and West Virginia) with a high
density of measurements.
The number of sulfate sites per day with concentrations greater
than 25 HLg/ra3 in 1976 (see Table 2) shows 3 days with a large number
of sites exceeding that value (June 11, August 22, and September 9).
Unfortunately there are a large number of blanks in this sulfate
episode matrix due to the 6-12 day sampling interval for sites
reporting data to the NADB. This rather infrequent sampling schedule
145
-------
Note: Contours are based on 4-11 months of data
collected during August 1977 to June 1978
at SURE II stations.
FIGURE 18
ANNUAL AVERAGE SULFATE CONCENTRATIONS IN
(SURE II DATA)
g/m3
146
-------
Note: Contours are based on 4-11 months of data
collected during August 1977 to June 1978 at
SURE II stations.
FIGURE 19
ANNUAL AVERAGE NITRATE CONCENTRATIONS IN
(SURE II DATA)
147
9/m3
-------
o
W
a
o
z
00
n
1
og
u
hJ
M
f>
B
CN
Al
o>
C
o
H
OJ
O
C
O
CJ
t-i
-H
3
a
O
ex,
w
en
-3-
vi
ro
p-
1
i
so
vO
I
vc
I
vo
1
esi
^D * i
t 1
BS
Key
>C
i£!
I
40
C
'H
4-1
u
o
Q.
V
b
a:
a>
Percentage
N)
j.
in
n sr
O
sO
r^
I
1
c-j
in
r»
1
w
oo
1^-
1
ber
1
(S
z.
<
-
oo
*
I
OJ f>
-3- I
O
i
ojcsjOvic-jCNJcNogrgojojr^m
148
-------
means the frequency of regional sulfate episodes is probably
underestimated using the NADB data. Using sulfate episode matrices
for each year of the 1960-1978 period, the number of regional elevated
sulfate days over the eastern and western United States have been
determined (see Table 3). In addition, the number of regional
elevated TSP days over the eastern United States have been determined
for 1975-1977. Table 3 shows that 1966 and 1978 were exceptional
regional sulfate years in the eastern U.S. and 1973 and 1976 were
exceptional years in the west. The year 1977 had the most regional
TSP episodes in the eastern U.S. during 1975-1977.
One of the most prominent regional sulfate episodes during the
summer of 1978 occurred during mid-July. Unfortunately, the 6-12 day
sampling interval of NADB stations did not generally coincide with the
peak days of the episode so that sulfate data is available for only
part of the eastern U.S. (see Figure 20). In this figure, the sulfate
data from the Ontario Hydro monitoring sites has been averaged for
Ontario province to supplement the NADB data. The SURE II sulfate
concentrations on the same day as the NADB (July 19) are generally the
same as the AQCR average sulfate concentrations (see Figure 21) in the
regions of overlapping data. Both sources of data show the highest
concentrations over the western Pennsylvania-New York border region.
The occurrence of regional sulfate episodes during the tradition-
al nonsummer months has recently been discovered in the NADB and SURE
II data bases. Two of the most prominent examples of these so-called
149
-------
TABLE 3: Number of Regional Elevated Sulfate Days During 1960-1978
Over the Eastern and Western U.S.
Year
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
EJML.O 1
Number of
Locations
5d>
10<2>
6
11
7
13
7
13
8
15
12
24
11
22
11
21
13
25
17
34
25
51
27
54
37
74
Number of
Days
4<3>
2
0
1
1
11
24
8
22
6
21
91
184
5
13
6
21
7
21
5
17
1
9
4
16
Year
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
WtST
Number of
Locations
2
1
2
1
3
2
3
1
3
2
5
2
6
3
7
3
9
5
10
5
12
6
19
10
23
12
Number of
Days
12
12
13
15
1
6
10
4
12
1
5
6
0
2
0
3
4
1
11
2
6
2
10
3
8
5
150
-------
TABLE 3 (Continued)
Year
1973
1974
1975
1976
1977
1978
CADi -
Number of
Locations
38
75
58
116
132(158) (6)
57(33)
114(164)
40(31)
80(156)
36
72
Number of
Days
4
9
0
2
3(2) <7)
13(4)(8)
10(2)
24(5)
5(7)
30(9)
10
35
Year
1973
1974
1975
1976
1977
1978
WILJl
Number of
Locations
18
9
27
14
25
13
31
16
33
17
28
14
Number of
Days
30
11
8
3
0
0
12
8
4
0
8
2
(1) 10% of the SO^ monitoring sites
(2) 20% of the S0| monitoring sites
T
(3) when > 10% of monitoring sites had > 20p.g/mJ
(4) when > 20% of monitoring sites had >10|dg/nr
(5) 1% of the TSP monitoring sites
(6) 5% of the TSP monitoring sites
(7) when 2 1% of monitoring sites had TSP > 260p.g/m
(8) when > 5% of monitoring sites had TSP >
151
-------
FIGURE 20
AQCR AVERAGE SO-= CONCENTRATIONS (Mg/m3) ON 19 JULY 1978
152
-------
FIGURE 21
SURE II SULFATE DATA FOR 19 JULY 1978
153
-------
"cold sulfate episodes" in the SURE II and NADB data occurred under
very cold, moist conditions with considerable snow cover on the ground
in January and February 1978 (see Figure 22 and 23). It may be noted
that these regional sulfate episodes have elevated sulfate concen-
trations of such magnitude and spatial extent as to be similar to
conventional warm weather episodes, but are probably produced by
heterogeneous rather than homogeneous conversion processes. These
cold sulfate episodes as well as nonepisodes in the summer season
would seem to present a considerable challenge to the treatment of
S02~sulfate chemistry in regional models.
One would like to see a more rigorous analysis of the frequency
of elevated sulfate concentrations and the "episodiality" of annual
sulfate concentrations, using the best data bases available. The
"episodiality" is a measure of how few episode days it takes to make
up 30% or more of the annual average concentrations. The best data
base of NADB sulfate data is that for the 3-state region (Ohio, Penn-
sylvania, and West Virginia) with the biggest density of measurements
for the 1974-1978 period. The frequency distributions for all the
sulfate measurements in the 3-state region for 1976 and 1979 (see
Figure 24) are different except for the sulfate concentrations greater
than about 24 jig/m^. The most frequent sulfate concentration in
1976 was more frequent but lower in concentration than the correspond-
ing concentration in 1978. In general, sulfate concentrations in the
range from 8-24 |u.g/m3 were more frequent in 1978 than in 1976.
154
-------
FIGURE 22
SURE II SULFATE CONCENTRATIONS
155
ON 23 JANUARY 1978
-------
FIGURE 23
AQCR AVERAGE SO 4 = CONCENTRATIONS (ng/m3) ON 19 FEBRUARY 1978
156
-------
11-
10-
02 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36
SO, Concentration (jj.g/m )
1976
1978
FIGURE24
FREQUENCY DISTRIBUTION OF SULFATE CONCENTRATIONS IN OHIO,
PENNSYLVANIA, AND WEST VIRGINIA IN 1976 AND 1978
157
-------
Since the sulfate concentrations in the 3-state region are obvi-
ously biased toward urban and industrial influences, it is of inter-
est to see how different the frequency distributions for the 3-state
region and the composite SURE II data are (see Figure 25). As ex-
pected, the 3-state NADB and composite SURE II sulfate data frequency
distribution are quite different with the former showing a pronounced
peak frequency at about 8 ^g/m3 and the latter showing a peak fre-
quency for the 0-4 p g/m3 band.
In order to minimize local influences on sulfate concentrations,
all the concentrations on individual days with 10 or more stations
reporting in the 3-state region were averaged before standard sta-
tistical analysis, episode identification and episodicity evaluations
were performed. The statistical values of daily area average sulfate
concentrations during 1974-1978 show the annual averages in 1975-1977
were essentially the same while those in 1974 and 1978 were signifi-
cantly higher. The statistical values for the 3-state region also
show the standard deviations and coefficients of spatial variability
were about the same during 1974-1978, but the number of observations
in 1974 was very low while the number in 1976 was very high compared
to the average number in the other years (about 6000).
The dates when the daily area average sulfate concentrations
were 2.20 jag/m3 during 1974-1978 are generally in the late spring to
early fall period except for November 10, 1978, which was an excep-
tional episode for that time of year. The 14 dates of 3-state area
158
-------
11-
10- -
9- -
8- -
o
e 7 ,
0)
cr
4)
4-
3--
2--
Nattonal Aerometric
Data Bank 1977-1978
SURE II 8/77-7/78
Source: Mueller,
et al. 1979
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46
_ 3
SO, Concentration <>g/m )
FIGURE 25
FREQUENCY DISTRIBUTIONS OF NADB AND SURE 11 SULFATE DATA
159
-------
sulfate episodes are distributed nearly equally, but only among 4 of
the 5 years, with no episodes satisfying the stated criteria in 1977.
The sulfate episodes of August 24, 1978, and June 11, 1976, had the
first and second highest area average sulfate concentrations.
The episodicity of annual sulfate concentrations is computed by
first putting the concentrations in rank order from the largest to
the smallest and then computing a conventional cumulative frequency
distribution. The episodicity analysis of daily area average sulfate
concentrations during 1974 in the 3-state region (see Figure 26) shows
that it took only about 4 episode days to produce about 30% of the
annual average concentration. In contrast, it took about 9 and 11
episode days in 1976 and 1978, respectively, to produce about 30% of
the annual average concentrations in those two years over the 3-state
region. The episodicity result for 1978 is probably more reliable
than for 1974 and 1976 because that year had a larger number of daily
average observations (61) than in 1974 and a higher annual average
(13.2 (ig/m3) than in 1976.
If the annual average sulfate concentrations display a high epi-
sodicity for an area the implication would seem to be that a regional
model must be able to simulate the episode concentrations in order to
simulate the annual average concentrations as well.
2.5 Higher Resolution Data
The term higher resolution data is used here to refer to aircraft
measurements, ground based measurements at 2-3 hour intervals, and
160
-------
M
c
o
n)
M
01
en
<§
'o*
u
M
0)
P*
(U
H
Si
o o
z cc
o>
OH
UJ (0
IJQ
(/> <
< z
UJ
cc
do
I
Q.
UJ
'os
jo
161
-------
size fractionation and chemical compositions measurements of fine
particulates in general.
Prior to the SURE II, there were very few cases when airborne
sulfate measurements coincided with the NADB sampling schedule so
that regional sulfate levels could be compared with long path sulfate
measurements above ground. One of the best examples of a pre-SURE II
match-up period is that in late September 1975 when the Research Tri-
angle Institute aircraft was making a broad area survey flight over
the eastern U.S. The AQCR average sulfate concentrations on September
27, 1975 (see Figure 28), were elevated but generally below regional
episode values (20-25 |o.g/m3) under the influence of a very large
high pressure system centered well to the west of its normal position
over the extreme southeastern U.S. The aircraft sulfate concentra-
tions on September 27 were higher than the AQCR average values at
ground level. There are a number of possible explanations for the
difference between ground based and airborne sulfate measurements.
If the differences are real, which there is certainly reason to be-
lieve in the case of the SURE II data, then this presents an addi-
tional challenge to regional modeling.
There are a very limited number of sulfate or sulfur measure-
ments for less than 24-hour intervals with the major data bases being
the 9 Class I stations during the 6 intensive months of the SURE II,
the Brookhaven filter pack data from the MAP3S program, and the
Florida State University Streaker Sampler data. The aerosol sulfur
162
-------
o
O
O
00
CO
c
-3
4J
rt
t
01
CO
§
b*
in
g
u
M
ill
Q)
H
T 1
I
3
CO
z
Z
O co
«UJ
o> _
DO
u. z
OE
O
O
o
CO
OL
UJ
o
o
o
ON
o
co
o
CNI
'os
jo
163
-------
FIGURE 28
AQCR AVERAGE SO4= CONCENTRATIONS kg/m3) AND HIGH PRESSURE
SYSTEM ON SEPTEMBER 27,1975
164
-------
ittsburgtv
29_ A^ 9/30/75
(3500')
Dayton
9/27/75 H
(3500') 27
= - A
21.2
Bedford
RTI Aircraft Flight Track
FIGURE29
SULFATE LEVELS ALONG THE RTI AIRCRAFT FLIGHT TRACK AND PATH
OF HIGH PRESSURE SYSTEM DURING SEPTEMBER 27-30,1975
165
-------
concentrations (generally every 2 hours) at the Brookhaven, New York,
MAP3S site and the sulfate concentrations (every 3 hours) at the
Duncan Falls, Ohio, and Indian River, Delaware, SURE II sites during
July 19-23, 1978 (see Figure 30), show generally consistent results
(i.e., sulfate concentrations are about 3 times the sulfur concen-
trations). Both types of data at the 3 sites show that the diurnal
variations in amplitude concentration are rather limited in amplitude
until the peak of the episode. The peak of the episode on July 22,
1978, at the Delaware site coincides well with the region of dense
haze seen in satellite imagery. In addition, the SURE II aircraft
sulfate data on July 20, 1978 (see Figure 31), is also consistent
with the data at Duncan Falls (Figure 30) and ground based SURE II
data (not shown) except in some areas the airborne sulfate concen-
trations are significantly higher in either the a.m. or p.m. flights
or both than the ground values. If the average of the a.m. and p.m.
flight sulfate values is comparable to the 24-hour average ground
based values and the former are larger, then there may be a signifi-
cant flux of elevated sulfate concentrations downwind which must be
dealt with in regional models.
There has been only very limited reduction and analysis of the
FSU streaker data, but this data base appears to have real value in
regional analysis and modeling. The stations are located generally
in the midwest (see Figure 32) and the 2 hourly values at 3 of the 14
stations during July 1976 show rather diurnal fluctuations and large
166
-------
Aerosol Sulfur Concentrations (ppb)
( ui/S'H)
167
-------
FIGURE31
SURE II AIRCRAFT SULFATE DATA FOR 20 JULY 1978
168
-------
KEY
1 Manhattan, KS 8
2 St. Louis, MO 9
3 St. Louis, MO (S) 10
4 Argonne, IL 11
5 Remington, IN 12
6 Forest, IN 13
7 Angola, IN 14
Frankfort, KY
Delaware, OH
Meadville, PA
University Park, PA
Annapolis, MD (N)
Annapolis, MD (S)
West Thornton, NH
FIGURE32
FLORIDA STATE UNIVERSITY STREAKER SAMPLING SITES
169
-------
multi-day variations associated with frontal passages and stagnation
(see Figure 33).
The SURE II and EPA Inhalable Particulate Matter (IPM) networks
have and will provide the main data bases on size fractionation and
chemical composition for regional analysis and modeling. The EPA IPM
network and other networks provide about 50 sites now and plans call
for as many as 200 sites over the next 3-5 years (see Figure 34).
The majority of the current IPM sites are in urban areas while most of
the future sites will be in rural areas. In fact, there is a critical
need to use data analysis and regional models to aid the design of a
rural IPM network.
Dichotomous samplers in St. Louis Regional Air Pollution Study
(RAPS) and at other locations have been used to identify the sources
of both coarse and fine particulates at urban and rural sites (see
Figure 35). In this example, the St. Louis rural and Smoky Mountain
sites had similar sulfate levels, but lower primary motor vehicle
impacts, than did the St. Louis urban sites. Furthermore, about 60
percent of the pine moss at the Smoky Mountain site is from sulfur
oxides sources suggesting long range transport from source regions
like the Tennessee Valley Authority.
170
-------
FIGURE 33
SULFUR CONCENTRATIONS (..g/FTT') 2 HOUR INTERVALS ; ST. LOUIS.
MO. (TOP). ARGONNE IL IMIDDLE). AND MEADVILLE. PA (BOTTOM)
171
-------
s
o
o
^
,0
nt
C
0
4J
rt
u
u
o.
o
^
§
W
Z
C-r
t-4
O
^
ON
ON
^
f.
^>
f^
cc
3
eo
3
c
o
a
c
o
H
4J
rt
Oi
Q.
O
^
§
4-i
01
z
IX,
/
ON
r-.
i i
«.
^
u
(li
fN
O
4J
y
O
j>^
_fN
n
c
0
jj
«
u
01
a
o
'fl
y;
I>4
§
c
a)
^j
O
C
Ov
r-
en
K
o
=3
O
cc
a.
UJ
_i
CQ
X
z
172
-------
Fine Fraction
45-
40-
e
~~- 35-
a.
5
-H
rt 25-
vJ
g 20-
CJ
3 15'
n 10-
9
5-
irni
1
m
LLLU
11
rrrrr
s
I
"*"*
mj
in
m
-
n
liu^
Unknown
[ Other
: Shale
| Motor Vehic]
- (NnVso=
,
BBMBI
BWW
103 105 106 108 112 115 118 120 122 124 Smoky
Sampling Site Mts
Coarse Fraction
Unknown
Other
Limestone
Shale
Motor Vehicle
103 105 106 108 112 115 118 120 122 124 Smoky
Sampling Site Mts
Source: EPA Report to Congress on Protection of Visibility (1979)
Note: Compared with the urban sites in St. Louis, the rural sites near St. Louis (122, 124)
and the Smoky Mountain site have similar sulfate levels but significantly lower primary motor
vehicle impacts ( 10 percent). Significantly, about 60 percent of the fine mass in the Smokies
is from sulfur oxides sources. The unknown fraction probably contains water, organics, and
nitrates. Almost all the coarse particle mass at all sites is accounted for by dust from the
earth's crust (Dzubay 1979).
FIGURE 35
SOURCES OF COARSE AND FINE PARTICULATES AT URBAN AND
RURAL SITES
173
-------
-------
3.0 VISIBILITIES
In this section, the national distribution, trends and episodes
are discussed along with the analyses of case studies of regional
visibility degradation and satellite data.
3.1 National Distribution
The general visibility conditions across the United States based
on routine airport visual range measurements as presented in the EPA
Report to Congress on Protecting Visibility (see Figure 36) show the
best visibility exists in the mountainous southwest while the poorest
visibility exists along the Ohio River Basin, the central Atlantic
coast, and along the southern Gulf Coast. The airport stations used
in this and other analyses are shown in Figure 37. Two of the ob-
vious regional modeling problems are (1) to predict the visibility
degradation in the mountainous southwest as a function of emissions
from energy facilities within the region and from urban areas within
and outside the region and (2) the reasons for the "better visibility
island" in the Smoky Mountains and along the eastern slope of the
Appalachian Mountains and the effect of regional emission increases
on visibilities in these areas. Several less obvious regional model-
ing problems are the effect of blowing dust on the regional visibili-
ties especially in the eastern Colorado-Oklahoma panhandle region and
the effect of controlled burning in the Pacific Northwest.
175
-------
CO
CO
UJ
H
_J
ffi
CO
>
UJ
a
<
:
a. u.
O
(O
z
h-
U
Q.
O
CO
o
UJ
o
<
X
CO
176
-------
(0
H
a
o
4-1
00
(0
rt
M
n)
(0
O
-
ffl
CO >
LU CM
Q
tu
(O
Z)
CO
z.
o
U)
177
-------
3.2 Trends
An extensive analysis of trends in airport visual ranges during
1948-1974 over the eastern United States has also been presented in
the EPA Report to Congress on Protecting Visibility. The airport
visual ranges showed a marked decline in the summertime throughout
the east with the largest decrease in the southeastern U.S. These
trends in visual range are generally consistent with the trends in
regional coal consumption, but a multi-year regional model is re-
quired to quantify the source-receptor relationships.
3.3 Episodes
The episodicity of the extinction coefficient (inverse of the
visual range)- is a measure of how important a relatively few cases of
high extinction coefficient (very low visual range) are to the annual
average extinction coefficient. The extreme northeastern U.S. is
highly episodic while the Ohio River Basin is unepisodic (see Figure
38) based on the airport stations shown previously (see Figure 37).
Visual range episode matrices similar to those for sulfates (see
Table 2) have been generated using the data from selected stations
across the U.S. (see Figure 37). The number of days of low noontime
visibilities with relative humidities less than 70% (see Table 4)
shows that 1970 had the largest number of regional episodes. When
the more recent period of data (1973-1978) is available it should be
analyzed in the same way to provide the identification and frequency
of regional episodes for regional modeling. It should be noted that
178
-------
[ | Insufficient Information
Kilometers
0 200 400
I'.' ' '
I I I
0 100 200
Miles
Source: Husar, et al. 1979
FIGURE 38
EPISODICITY: FRACTIONAL CONTRIBUTION MADE BY UPPER
PERCENTILE (20%) OF THE EXTINCTION COEFFICIENT TO THE TOTAL
DOSAGE INTEGRAL (TIME OF THE EXTINCTION COEFFICIENT)
179
-------
TABLE 4: Number of Days of Low Noontime Visibilities with Relative
Humidities Less Than 70% at Multiple Locations in the U.S.*
Year
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
Number of
Locations
12(D
6^
13
7
13
7
13
7
13
7
13
7
13
7
13
7
13
7
13
7
14
7
14
7
14
7
14
7
14
7
14
7
14
7
14
7
14
7
14
7
Number of
Days
12(3)
O^
1
0
13
0
6
0
21
11
12
3
4
1
11
0
10
1
3
0
6
0
3
0
10
0
8
1
9
1
21
5
11
1
7
0
24
4
23
2
Year
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
(1)
(2)
(3)
Number of
Locations
14
7
14
7
14
7
14
7
14
7
10% of sites
5% of sites
number of days
Number of
Days
26
2
27
3
36
3
29
1
28
4
when visibility
<6 miles at >10% of sites
(4)
number of days
<3 miles at >5%
when visibility
of sites
180
-------
the regional episode frequencies for low visibilities are based on
daily data rather than the 6-12 day sampling for sulfate data.
An excellent example of a regional visibility episode in the
eastern U.S. identified from the sulfate episode matrix is that on
June 11, 1976 (see Table 2). The contours of low noontime (EST)
visibilities based on data from stations shown by the asterisks
(see Figure 39) show most of the eastern U.S. had visual ranges of
6 miles or less. These contours are based on an objective analysis
computer routine. This episode "exploded" over most of the eastern
U.S. within a few days due to not only a recirculatory flow over high
emission density areas, but also due to pronounced warm air advection
at the 500 mb level which probably accelerated the conversion of S02
to sulfates and the associated degradation in regional visibilities.
3 .4 Case
The problem of determining the existence and frequency of region-
al low visibility episodes in the mountainous southwest is more diffi-
cult than in the east because the historical airport visual range data
has only been digitized for a relatively few stations and, as dis-
cussed earlier, the historical sulfate data base for the region (AQCR
14) is very sparse in stations and numbers of measurements. The his-
torical sulfate data base, the meteorological conditions thought to
produce regional low visibility episodes and recent visibility mea-
surements, were used to select 15 case studies for which the visual
range and other meteorological parameters for about 50 support
181
-------
FIGURE 39
CONTOURS OF LOW NOONTIME (EST) VISIBILITIES (IN MILES) ON
11 JUNE 1976, BASED ON DATA FROM SELECTED STATIONS
182
-------
stations were digitized by hand (see Table 5). The meteorological
conditions thought to be most responsible for regional low visibil-
ity episodes is a period of stagnation over the high emission areas
of southeastern Nevada or southern California followed by strong
persistent winds produced by an approaching cold front. These mete-
orological periods were identified from either synoptic weather maps
or periods of extremely persistent winds from the south as measured
on a meteorological tower in central Utah. The recent visibility
measurements were from the EPA VIEW and VISTTA programs. The VISTTA
program includes both power plant and smelter plume tracking flights
and regional survey flights (see Figure 40).
The December 11-14, 1974 sulfate visibility case study was
selected on the basis of the pronounced sulfate episode over south-
eastern Arizona. The airport visual ranges at 11:00 a.m. on December
13, 1974 showed generally reduced values (60 miles or less) at most
locations (see Figure 41) while the sulfate data showed elevated
concentrations at most of the locations (greater than 5 ^g/rn^) and
very high values for the southwest in south central Arizona (see Fig-
ure 42). It should be noted that the extremely low visual ranges at
Eagle and Leadville, Colorado, are thought to be due to local effects
of terrain and/or clouds.
The September 18-23, 1978 sulfate/visibility case study was
selected on the basis of the regional survey flight in VISTTA and the
special visibility measurements (target contrast) at the Canyonlands
183
-------
43
to
a
c
tfl
nj w
C C
O O
f, N N
TO -H -H
J-' !-i ^
P < <
C C C
O
H
,n
tfl
CD to
)-ip,p.ai
ooooi-'oca) H
NNNN.IN *Q 60 t>0 f
H -H -H ~H cfl -H CU 01 C C -M
Vi M IH C
cx ft a 4-1
05 M U) tt
a. a.
H -H fd n) C
CD TO CO -H
C C C O CO H -H
tti cfl cfl CU4-l
C ^ rH
ca 14-1
l-li-H
cn-n-i i
(J O O
00
vo CT% i r*. CM m
GO
g
l^i
T3
O
I oo r- (1 1
CT\ CN
^H « |
O 00 fl
o
H
M
a.
ISA
-------
Four Corners
Power Plant
FIGURE 40
AIRBORNE MEASUREMENTSREGIONAL SURVEY
185
-------
Salt Lake City 4 10«
Vernal
5-Eagle 60»Denver
. ^ , 2»Leadville 75«Limon
40.Grand
Junction
Colorado
100 springs
40, 10* 90»Pueblo
Montrose Salida
La Junta
Cediir City
Trinidad 65
40 Bryce Canyon
35 Farmington
30 Grand Canyon
* Natl. Park
40«Santa Fe
30 Tucumcari*
60«Albuquerque
20»Roswell
50«Truth or Consequences
40»Holloman AFB
40»Silver City
60»Tucson
Fort 6Q
Huachuca*
Las Vegas
65
= Phoenix
40 = Visibility (Miles)
FIGURE41
VISIBILITY AT 11:00 A.M. ON DECEMBER 13,1974
186
-------
14.2* 18.7
11.2. 17.5*.
FIGURE 42
SULFATE EPISODE IN ARIZONA ON DECEMBER 13, 1974
137
-------
National Park, Utah, which showed a relative minimum in target con-
trast on September 24 (see Figure 43). The visibilities at 1400 LSI
on September 23 were generally reduced (60 miles or less) (see Figure
44) and the sulfate levels were somewhat elevated (not shown).
These case studies suggest the existence of more regional scale
lower visibility episodes in the southwest; however, the most convinc-
ing case studies will result from a combination of the airport visual
range and VIEW program data consisting of special measurements at 14
sites in the region. Eventually the entire Southwestern Energy Re-
source Development area may be covered by a combination fine particle
sampling and visibility monitoring network laid out in a more or less
regular array (see Figure 45) to provide high resolution data for re-
gional and even interregional modeling.
3.5 Satellite Data
Examples of satellite data showing regional sulfate/low visibil-
ity episodes in the eastern U.S. have already been mentioned in the
introduction and will be discussed further in the next section on
precipitation chemistry data bases. Apparently no one has been able
to find a regional sulfate/low visibility episode in the southwest in
satellite data either because the sulfate concentrations are not high
enough to produce a visible haze or the satellite data coverage is not
complete enough. However, a very distinct example of the accumulation
and transport of a subregional scale haze has been captured in a high
quality astronaut photo during the Apollo 7 flight on October 12,
188
-------
X
<
cc
o.
_l
<
z
O
W
Q
IT Z
to
OJ
s
41
u
O.
01
O
w
Z
O
o
CD
oc
I
in
1
o
LTl
CM
I
O
189
-------
6 Salt Lake City
20 Vernal
. Grand 30 Leadville
Junction 50 Colorado
Springs
Montrose 70 Pueblo
Salida 20
La Junta
Green Rivert
Cedar City
o Bryce Canyon
Trinidad * 50
75 Farraington
e Taos
Cuba
40 *Santa Fe
50 Grand Canyon
Natl ?ark
40 « Albuquerque
15 Corona
Socorro
Roswell
f>5 m
Prescott Show Low
Silver City
Las Vegas
Needles
Blythe
Fort » Tucson
_Huachucaa
-------
z
UJ tu
0°
H EC
< UJ
« S
u-
Q
ai
o
cc
Q.
19Z
-------
1968, over southern California. In this photograph we can actually
see the urban plume going in three directions: one part blowing
northward to Ventura County, another part blowing eastward through
the Cajon Pass and into the high desert, and the third part blowing
westward out over the Pacific Ocean. The Los Angeles urban plume has
been tracked to the California-Arizona border. This situation calls
attention to the need to monitor and model the impact of urban plumes
on remote pristine areas in the southwestern U.S.
192
-------
4.0 PRECIPITATION CHEMISTRY
In this section, the monitoring locations, the spatial distribu-
tion of pH and wet deposition, and wet episode characteristics are
discussed along with examples of higher resolution data, and the use
of precipitation and nephanalysis (cloud) data bases in regional mod-
eling of wet sulfur deposition.
4.1 Monitoring Locations
The daily or longer period average precipitation amounts have
been measured at over 12,000 locations in the United States for many
years using standard gauges, and these measurements have been summa-
rized into climatological normals and extremes. Hourly precipitation
rates have been measured at only about 3,000 of the 12,000 locations
in more recent years using several types of recording gauges (see Fig-
ure 62, discussed later). By stark contrast, the chemical composition
of precipitation and its changes over time have been measured at only
a very few stations continuously during the 1970 decade. Unfortunate-
ly, even some of these "trends" stations have changed location during
their existence: some sites have been found to be subject to more
local than regional influences, and the trends from the five years of
available data for the stations that have not been moved are
inconclusive due to the short period of record among other factors.
The number of precipitation chemistry monitors has increased
rapidly during the past 3 years to the point where there are about 30
networks with about 250 monitors in the United States and about 100
193
-------
monitors in Canada (see Figure 46). Unfortunately, the data from the
current "composite network" cannot be readily integrated to support
research on ecological effects and atmospheric processes over the
entire United States because the networks use different brands and
types of monitors, collect with different sampling schedules, and
analyze different parameters in different ways. In addition, the
density of monitors varies widely over the United States from very
dense in the northeast to very low in the Great Basin of the west.
One of the major problems in integrating the individual networks
is the different sampling schedules of which there are four basic
types, namely, hourly, event (storm), weekly, and monthly. There ap-
pears to be only two locations with ongoing hourly sampling of precipi-
tation chemistry in the United States: the city of Philadelphia and
the Brookhaven National Laboratory in New York (see Figures 55-58,
discussed later). Hourly precipitation chemistry data generally shows
significant variations in pH and sulfate ion concentrations during the
course of an event (see Figure 55, discussed later). In fact, hourly
precipitation chemistry data may be the only way to really understand
atmospheric transformations and processes. Even event sampling,
usually defined as the precipitation during a single storm of less
than 24 hours, shows considerable spatial variability in the pH
frequency distributions and wet sulfur deposition over distances of
only several hundred kilometers (see Figures 50-51, discussed later).
Current thinking is that event sampling may be the only way to
194
-------
V)
X.
cc
o
5
LU
z
o
z
cc
o
O
s
%£
sS
CD UJ
z
o
CC
CL
t-
z
111
cc
DC
3
O
195
-------
really understand short-term ecological effects. The majority of the
current monitors are operating on a weekly or a monthly schedule
principally because the costs of event or hourly monitoring in these
networks would be prohibitive. However, there are those who argue
that reduced network size with event monitoring would be preferable to
larger network size with only weekly or monthly sampling intervals.
There are three event sampling networks in the eastern U.S. and
one in the west. The MAP3S precipitation chemistry network (see
Figure 47) consists of 8 stations laid out along roughly north-south
and east-west lines with the basic purpose of providing a data base
for analyzing and modeling atmospheric processes. Unfortunately, the
MAP3S network probably needs a greater density of stations sampling at
subevent frequency with concurrent meteorological and air quality
measurements to meet its basic purpose of providing wet removal rates}
source-receptor relationships, and well documented episodes for the
evaluation of regional models.
The EPRI precipitation chemistry network (see Figure 48) consists
of 9 stations located at or near the nine Class I stations in the SURE
II program with the basic purpose of providing a data base for trends,
analysis, and modeling. The EPRI event network has the virtue of
concurrent meteorological and air quality measurements and repli-
cate samples at all sites, but has the drawback of low sample den-
sity and a limited lifetime in that it may be disbanded in June 1980.
196
-------
t Brookhaven
Perm State
t Oxford r> f Vireinia
Precipitaion Chemistry Monitors
FIGURE47
MAP3S PRECIPITATION CHEMISTRY NETWORK
197
-------
FallsVl
id " V»T«J4«« D-;
Eastern Regional Monitors
ASeven Adirondack Monitors
FIGURE 48
EPRI PRECIPITATION CHEMISTRY NETWORK
198
-------
The Ontario Hydro precipitation chemistry network (see Figure 49)
consists of 7 stations, 6 of which Lie along a southwest-northeast
line, with the basic purpose of providing a data base for assessment
of energy use impacts. The Ontario Hydro event network also has the
virtue of some concurrent meteorological and air quality parameters,
although not as complete as at the EPRI stations. Perhaps the com-
posite of the 3 event networks provides sufficient sampling density to
resolve the wet depositions from storm events as they move across the
northeastern U.S. and southeastern Canada. An analysis of the simul-
taneous events in all three networks is in process to determine if
this is the situation and to extract the wet episodes with the most
event measurements in all three networks for evaluation of regional
models.
The fourth event network in the U.S. is in northern California
(not shown) and is operated by the University of California at
Berkeley for the California Air Resources Board principally to study
ecological impacts from winter storms depositions (November-April).
Experience from this network and analyses of the frequency distribu-
tions of wet and dry periods and the spatial-variability of precipi-
tation in the western U.S. suggest that dense networks of wet and dry
event sampling are necessary for characterizing the infrequent wet
events with high spatial variability and the extended dry periods
that are prevalent in this region.
199
-------
cc
O
LU
Z
>-
DC
CO
s
UJ
I
O
gz
£2
O t
C9:
O
UJ
cc
Q.
O
CC
Q
>-
r
O
cc
z
O
200
-------
4.2 Spatial Distribu11on
The spatial distribution of wet sulfur deposition for the eastern
U.S. and southeastern Canada is difficult to ascertain because there
was no network covering the entire eastern U.S. operating for several
years prior to 1978. There are problems with intercoraparisons of
monthly, weekly and event data, and the data for 1978-1979 is only
now being published. A compilation of published and unpublished data
during 1977 has been made by Whelpdale and Galloway (1979) and iso-
plethed by hand by the author (see Figure 50). These authors point
out that monthly deposition values may be too high compared to event
or weekly values due to possible sample contamination and evaporation
from being in the field so long. This situation could give rise to
artificial gradients in wet sulfur deposition between areas like
Canada with primarily monthly sampling and areas in the eastern U.S.
with primarily weekly or event sampling. The wet sulfur deposition
in 1977 shows a clear area of elevated values (greater than 2 grams
of sulfur per square meter per year) in the extreme southern part of
Ontario Province. The need for wet sulfur deposition data in the Ohio
River Basin, which is a major sulfur oxide emission source region, is
also readily apparent (see Figure 50). This situation is currently
being rectified by the analysis of all the 1978 data which includes
that for the MAP3S and EPRI stations in or near the Ohio Basin.
Although the 1977 spatial distribution of wet sulfur deposition
may be misleading due to the problem discussed above, it does suggest
201
-------
0.22»
FIGURE 50
WET DEPOSITION OF SO4= - S (gSm
1977
202
-------
that the regions of elevated values are more subregional in scale
(i.e., southern Ontario Province) than regional in scale as was the
case with ambient sulfate concentrations.
The pH distributions from the Ontario Hydro event network
(see Figure 49) for June 1976 to December 1977 (see Figure 51)
differ significantly from site to site for the 5 stations along the
southwest-northeast line of about 200 km. In fact, the pH frequency
distributions at the two Toronto sites differ very dramatically with
one major peak between 3.8 and 4.2 at the Toronto-Neilson Drive site
and two smaller peaks, one between 4.5 and 4.8 and the other between
6.3 and 6.6, at the Toronto Research Building site. Further analyses
are required to explain these differences, but it would seem that the
two sites are affected by different local pollution sources and/or
microscale meteorology. It is clear from this analysis that even
small changes in location can result in significant differences in
monitoring results. These results dramatize the large spatial vari-
ability that can occur within short distances and demonstrates the
need for rather highly dense event monitoring to characterize fully
the precipitation chemistry variability within a subregion or region.
4.3 Episodes
Wet episodes at individual sites can be identified by locating
the 10 lowest pH events and the 10 highest wet sulfur depositions
(sulfate ion concentration times the precipitation amount) in each
203
-------
peacock Paint
36 events
2.7 3.0 3.3 3.6 3.9 4.2 4.5 i.3 5.1 5.4 5.7 6.0 6.3 6.6 6.9 7.2 7.5 7.8
Burlington
26 events
2.7 3.0 3.3 3.6 3.9 4.2 4.5 4.8 5.1 5.4 5.7 6.0 6.3 6.6 6.9 7.2 7.5 7
Toronto-Nellson Drive
2 3 even t s
7 3.0 3.3 3.6 3.9 4.2 4.5 4.8 5.1 5.4 5.7 6.0 6.3 6.6 6.9 7.2 7.5 7.8 pH
2.
Toronto-Research Builditij;
53 events
2.7 3.0 3.3 3.6 3.9 4.2 4.5 4.8 5.1 5.4 5.7 6.0 6.3 6.6 6.9 7.2 7.5 7.8
« 25-,
S 20-
o 15-
Fenelon Falls
40 events
2.7 3.0 3.3 3.6 3.9 4.2 4.5 4.3 5.1 5.4 5.7 6.0 6.3 6.6 6.9 7.2 7.5 7.8 pH
FIGURE51
pH DISTRIBUTIONS FROM ONTARIO HYDRO PRECIPITATION DATA
(JUNE 1976DECEMBER 1977)
204
-------
year of monitoring data. Subregional or even regional scale wet epi-
sodes can then be identified by locating the common dates, especially
for adjacent monitors, of the 10 lowest or 10 highest values at all
the individual sites. One of the most prominent wet episodes that
was identified in the MAP3S data on the basis of the 10 lowest pH
values in 1977 at several adjacent stations was that during July
18-22, 1977. It was initially thought that this wet episode would
have been associated with a major frontal system over the eastern
U.S., but an examination of the satellite imagery on July 20, 1977,
the peak of the wet episode according to the MAP3S data, showed just
a large isolated cloud system surrounded by dense haze. The implica-
tion of the analysis of the satellite imagery on July 20, 1977, is
that the dense haze was due to elevated sulfate concentations from a
regional sulfate episode and sulfates were entrained into the major
cloud system and that system just happened to rain over 3 of the MAP3S
stations. This mid-July 1977 period experienced a major sulfate-
oxidant episode over most of the eastern U.S. and this episode has
been analyzed in detail by Tong et al. (1979). The MAP3S precipita-
tion chemistry data for July 18-22, 1977 (see Table 6), shows 6 of
the 7 pH values were less than 4 and the sulfate ion concentations
were generally elevated (greater than 100 umol/1).
The episodicity of wet sulfur depositing events is of consider-
able importance to regional modeling of annual depositions just as
the episodicity of ambient sulfate concentrations is to the regional
205
-------
o
°
r-»
»^
o>
T
>
3
^
o
CM
So
a *"
2 411
O O
<
v>
LL
O
V)
206
-------
00
60 25
II -*
Q) II
-------
modeling of the annual sulfate concentrations. Smith and Hunt (1978)
appear to have made the first analysis of wet sulfur deposition acid-
ity in western Europe using the data base from OECD Long Range Trans-
port of Air Pollution program. The researchers defined "episode days"
as those days with the highest wet deposition which, when summed, make
up 30% of the annual wet deposition total. They then defined episodi-
city as the ratio of "episode days" to the annual number of wet days
expressed as a percentage. Finally, they defined, somewhat arbitrari-
ly, the "episodicity" to be high, moderate, or low if the percentage
ratios were less than 5%, 5-10%, or greater than 10%, respectively.
Following the convention of Smith and Hunt (1978) and the same
computational procedure outlined for the episodicity analysis of sul-
fates described earlier, an extensive analysis of wet sulfur deposi-
tion and precipitation episodicity was made for the available data
from all three event networks. Only an example of the results are
presented here, while the complete results are presented elsewhere.
The episodicity of wet sulfur deposition events at the MAP3S
Whiteface Mountain station in 1978 (see Figure 53) shows a moderate to
high value (5.5%) in that it takes only 3 of the 54 events in that
year to produce 30% of the total wet sulfur deposition. On the other
hand, the episodicity of precipitation events at the MAP3S Whiteface
Mountain station in 1978 (see Figure 54) shows a low value (12%) in
that it takes 6.5 of the 54 events in that year to produce 30% of the
total annual precipitation. Thus, it may be concluded that the wet
208
-------
%
0
,. *:0
. 55
cr, »
Pi U *
COO A
o w ^
*-> O Q>
o U "6 *
O 11 3 A
J d. Z
S1U3A3 C W _
<
1 111 11
-
-
-
-
1
1 1
O
r-f
O
o
OO
o
o
s
o
o
rsl
O
o
LLJ
O
Q.
til
Q
u.
3 =» UJ
*, «5
0 En.
UJ
o
cc
UJ
a.
15
r>
o
o o
mJInS
jo
209
-------
o
:^
*.
^
*
6*5
<*1
*
' * %
* 4)
*«* GO
US r-» 0
(y p_^ i/-j fP
u
M~) ^
ixi *o co V
i-t o a ^
.e o cu A
3 m >
Od b3 A
.. u^ iw 0
c o o _
0
H -0 t,
4J O <1J ^ A
a -H ja
o M B
O 41 3 w
iJ eu z
s qua A3 5
i i i > i i
-
"
-
* .
* . 9
1 1
rH
o
o
DO
o
r
0
o
in
o
o
o
r>J
o
^^
DOOO OOOOOO
DiTsoor*. ^o m ^r r^i c^i i 4
en
C
9)
£
O
^_)
iJ
CQ
a
^_i
o
01
u
a-
IM
O
iJ
s
u
0)
Cu
CU
5
a)
^-]
3
U
»-
SUJ
Z
O
^~
^*
(
E
O
9 UJ
in
P
*«c
_J
3
s
O
210
-------
sulfur deposition at the site in 1978 was moderate to high in and of
itself and not just due to the inherent episodicity of the precipita-
t ion.
4.4 Higher Resolution Data
The Brookhaven National Laboratories have an automatic sequen-
tial precipitation sampler (Raynor and McNeil, 1979) and operated it
for more than 4 years. The BNL sampler generally provides hourly sam-
ples of precipitation for chemical analysis. The hourly samples have
been analyzed for pH and a large number of anions, cations, and ele-
ments. In addition, the BNL precipitation chemistry data bases have
been supplemented with a large number of meteorological parameters
that are very useful in correlative analysis.
Several ambient and wet sulfate episodes occurred during the
first half of November 1978. It may be recalled that November 10,
1978, was the first major sulfate episode over the 3-state region
(Ohio, Pennsylvania, and West Virginia) in a winter season during
1974-1978 (recall Table 4A). Interestingly, the BNL data base con-
tains several longer period events during the first half of November
and both show significant variations in pH or other chemical charac-
teristics during the events. The hourly precipitation chemistry at
Brookhaven, New York, during November 15-16, 1978 (see Figure 55),
showed a strong drop in pH in the first three hours of the event with
an associated rise in sulfate ion concentration. By contrast, the
nitrogen concentration and the precipitation amounts were about the
211
-------
(ran)
t
o
W
L
(H3in/SlN31VAin030HDIW)
o o o
$
o
LLl
OC
O.
O
I
212
-------
same for each hour of the event. It is generally thought that mate-
rials incorporated during cloud formation processes (rainout) should
have a relatively uniform concentration throughout the event while
materials removed from the atmosphere below cloud level (washout)
would be found predominantly in the early samples of the event. It
would seem that the materials removed from the atmosphere below cloud
level on November 15 at Brookhaven were nearly alkaline and the mate-
rials incorporated during cloud formation were acidic since the pH
dropped from about 7.0 to 4.5 in the first 3 hours and then apparently
remained nearly constant for the rest of the event.
The distributions of pH from the BNL automatic sequential pre-
cipitation sampler for mid-1976 to mid-1979 (see Figure 56) show very
significant differences in shape and location of peak values. The
distributions for the last 6 months of 1976 and the first 5 months
of 1979 are quite flat while those for the complete years of 1977
and 1978 show pronounced peaks at pH's of about 3.9 and 4.8, respec-
tively. The year to year variability is undoubtedly related to the
variability in air mass trajectories and chemistry affecting the
precipitation at BNL.
Raynor (1979) has performed extensive statistical analyses of
this unique data base to determine the mean and variability of pre-
cipitation characteristics by event, month, season, and year and to
relate concentrations of other constituents. Two examples of these
analyses are the event means of precipitation chemistry by length of
213
-------
10n
.0
2.7 3.0 3.3 3.6 3.9 4.2 4.5 4.3 5.1 5.4 5.7 6.0 6.3 6.6 6.9 7.2 7.5 7.8 pH
JUNE 15-DECEMBER 31, 1976
lO-i
I I ~I T 1 1 | ( I T ' T I ! I I I I
2.7 3.0 3.3 3.6 3.9 4.2 4.5 4.8 5.1 5.4 5.7 6.0 6.3 6.6 6.9 7.2 7.5 7.3 PH
JANUARY 1-DECEMBER 31, 1977
5-
x 7
2.7 3.0 3.3 3.6 3.9 4.2 4.5 4.8 5.1 5.4 5.7 6.0 6.3 6.6 6.9 7.2 7.5 7.9 pH
JANUARY 1-DECEMBER 31, 1978
ion
o
u
3
SS
5H
2.7 3.0 3.3 3.6 3.9 4.2 4.5 4.8 5.1 5.4 5.7 6.0 6.3 6.6 6.9 7.2 7.5 7.8
JANUARY 1-MAY 31, 1979
FIGURE 56
DISTRIBUTION OF pH FROM THE BNL (RAYNOR) AUTOMATIC
SEQUENTIAL PRECIPITATION SAMPLER
214
-------
event and by precipitation rates which show (see Figures 57 and 58)
that the weighted mean concentrations of all major constituents, ex-
cept the chlorine ion and the sodium anion, decrease with increasing
event length and precipitation rate. This behavior is as expected
due to the atmospheric cleansing and precipitation dilution effects
of longer and more intense wet events. The contrary behavior of the
chlorine ion and the sodium anion is probably due to local sources of
sea salt due to the near coastal location of the BNL monitor.
4.5 High Resolution Precipitation Data
High resolution precipitation data over a region is needed to
determine the spatial variability of wet sulphur deposition and the
frequency distribution of the durations of wet and dry periods, and
to parameterize wet removal in regional models. In addition, high
resolution precipitation data at specific locations is needed to de-
termine the number of wet events in weekly and monthly precipitation
chemistry samples and wet and dry event durations to specify realistic
exposures for simulated acid rainfalls.
High resolution precipitation data is available for recent years
at about 3000 recording rain gauge stations in the United States from
the National Climatic Center and at about 1000 stations in southern
Canada from the Canadian Atmospheric Environment Service. The number
of conventional recording rain gauges per state or southern portion
of the eastern Canadian Provinces are discussed later (see Figure 62).
215
-------
so
lOOr
90
80
70
01 60
50
T3
-------
100
90
80
70
H
0)
cr
a)
60
50
oo 40
H
OJ
3
30
20
10
Cond.
Precip. Amt.
(Units are mm)
0-
1.0
1.1
2.5
2.6-
6.0
Rate (rnm/hr)
* S04
A CI-
O H+
D N
* NA+
A Cond.
Precip. Amt.
Source: Raynor 1979.
FIGURE58
EVENT MEANS OF PRECIPITATION CHEMISTRY BY PRECIPITATION RATE
FROM THE BNL SEQUENTIAL PRECIPITATION DATA
217
-------
The spatial variability of precipitation over individual states
has been computed from the daily average precipitation values for
1974-1978. The spatial variability is defined as the standard de-
viation of the individual stations from the mean of all the stations
divided by the mean of all the stations expressed as a percentage.
The monthly, seasonal, and annual mean precipitation totals and spa-
tial variabilities for all the states and each of the 5 years are
summarized elsewhere. The total precipitation and the spatial vari-
ability for each station in 1977 (see Figure 59) shows states with
all four possible combinations of both, namely, (1) high total and
high variability (Washington), (2) high total and low variability
(Connecticut), (3) low total and low variability (North Dakota), and
(4) low total and high variability (California). In general, the
total precipitation and the spatial variability show less variation
from state to state for states east of the Mississippi River than for
states west of the Mississippi River. Ignoring differences in emis-
sions, chemistry, and transport for the time being, the implications
of these results are that point measurements of wet sulfur deposition
should be more representative of a larger area and wet removal in
longer period (seasonal and annual) regional models should be easier
to parameterize because of the more uniform distribution of precipi-
tation. Analyses of precipitation amounts and spatial variability
for hourly time periods and over stations within 80 x 80 km grid
squares over the eastern U.S. during selected wet episodes are
218
-------
219
-------
currently in process to see how complex the precipitation character-
istics are on these shorter time and smaller space scales needed for
regional modeling of episodes.
The inventory of current networks has shown that the majority of
the monitors in all the networks are not collocated at a site with an
hourly recording rain gauge. The collocation of monitors is important
not only so that the number of events which contribute to the weekly
or monthly total wet depositions can be determined, but so that
trends, or the lack thereof in precipitation chemistry data, can be
put in the overall perspective of the longer period of precipitation
amount at the site.
The frequency distributions of the durations of wet and dry
events are important to objectively specifying the optimum sampling
intervals in various regions of the U.S. for trends monitoring and
the parameterization of wet and dry removal in regional models. The
monthly, seasonal, and annual frequency distribution of the durations
of wet and dry events for all the states and each of the 5 years
(1974-1978) are summarized elsewhere. The frequencies of wet and dry
periods for 1977 {see Figure 60) show the most frequent durations of
wet events to be one hour in both the east and the west and 4-6 days
and 2-3 weeks for dry periods between wet events in the east and west,
respectively.
High resolution precipitation and precipitation chemistry data
are both needed on a grid base for regional modeling of wet episodes.
220
-------
1-1 y2 *D 1-* e*\ (M
tOe-*rj«-t oi*4tsi(*»
e irt v-' i 2 eo -^ i
-------
A schematic diagram of the parameterization of wet removal in an epi-
sode transport-removal model (see Figure 61) shows that mean and var-
iance of precipitation rate would be used to compute the mean removal
and a measure of the spatial variability of the wet removal due to the
spatial variability of the rainfall intensity over the grid square.
The percentage of stations reporting precipitation to the total number
of stations in the grid square would also be a useful way of parame-
terizing the amount of pollutant over the entire grid size that is
subject to wet removal. If there are no stations in a grid square or
too few to compute a meaningful spatial variability then an interpo-
lation based on the surrounding grid squares would have to be used.
The schematic diagram also is meant to suggest that the regional model
simulations would be compared to both the mean sulfate ion concentra-
tion or wet sulfur deposition over all the stations in the grid square
and the variance about the mean if there are at least several monitors
in the grid square. Ideally, some grid squares would have clusters of
event monitors during limited periods to provide a better measure of
the spatial variability in chemical constituents and their relation-
ships to the spatial variability in precipitation as well as factors
between the emissions and the measured concentrations. The role of
satellite and cloud data in parameterization of wet removal will be
discussed in the next section under the general heading of nephanal-
ysis data.
222
-------
Digital
Enhanced
Satellite Photo: Cloud
xBoundaries
--"Holes" in
Clouds
3DNEPH: Cloud Types
Cloud Amounts
niCloud Heights
Recording Rain Gauge Station (Hourly)
R Grid Average of Precipitation Rate
(ram per hour)
Vr Grid Average of Precipitation Rate
(mm per hour
0 Precipitation Chemistry Monitor (Event)
C Grid Average of Precipitation Chemistry
Concentration (yj.g per liter)
Vc Grid Variance of Precipitation Chemistry
Concentrations (^g per liter)
FIGURE 61
SCHEMATIC DIAGRAM OF THE PARAMETERIZATION OF WET REMOVAL
IN THE EPISODE TRANSPORT-REMOVAL MODEL
223
-------
DC
O
UJ
<
w
£2
a o
UJ
CO
~z>
(0 Z Q
tu Q oc
ac a Q-
si
li. M
O
£E
UJ
m
S
22A
-------
Finally, the number of conventional recording rain gauges per
state or southern portion of the eastern Canadian Provinces (see
Figure 62) range from a high of 282 in California to a low of 2 in
Delaware. On the other hand, the density of rain gauges per square
kilometer ranges from a high of about 10 in New Jersey to a low of
about 1 in Nevada.
4.6 Nephanalysis Data
A quantitative analysis of the satellite imagery on 22 July 1978
at 11:30 GMT for the case described in the introduction to this paper
shows dense haze on the southern boundary of a large cloud system over
the extreme northeastern United States. The quantitative analysis of
the digitally enhanced satellite imagery also shows several isolated
cloud areas surrounded by haze. The detailed analysis of the satel-
lite imagery and associated meteorological conditions for a number of
other regional ambient sulfate and wet episodes has been presented
elsewhere. Basically, these additional case studies show the value
of quantitative satellite imagery in determining whether the air pol-
lution has become entrained into the cloud systems or not and the
position of the haze gradients and cloud boundaries in relation to
the precipitation chemistry monitors.
One of the critical needs for the development of regional models
of wet removal is for a data base on cloud characteristics. A neph-
analysis (cloud) data base could be analyzed to provide a climatology
of cloud characteristics for use in specifying typical seasonal and
225
-------
41 3
X O
D
E
O
CO
f-
0>
CM
tM
oca
3 UJ
a o
UJ
(A
U.
O
CO
O)
226
-------
annual cloud types and amounts and their respective frequencies for
use in longer period models. The data base could also be used to
extract the specific cloud types, amounts, and heights for each grid
square during period models. The availability of high resolution
cloud type information would permit the more complete use of scaveng-
ing coefficients that are functions of cloud type as well as precipi-
tation intensity like those of Scott (1978).
Fortunately, the Air Force has assembled a computerized nephanal-
ysis data base (3DNEPH) which has real potential for use in regional
analysis and modeling of wet episodes. With this data base, one
should be able to prepare a cloud climatology and extract the detailed
cloud characteristics during the wet episodes. For example, the cloud
types and amounts in 3DNEPH should be related to the total precipita-
tion and its spatial variability over individual grid squares. The
use of the 3DNEPH and digitally enhanced satellite imagery data bases
for parameterizing wet removal in episode wet removal models is also
shown in the earlier schematic diagram (recall Figure 61).
227
-------
-------
5.0 CONCLUSIONS AND RECOMMENDATIONS
The principal conclusions of this survey of sulfate, visibility,
and precipitation chemistry data bases and results for regional
modeling are as follows:
An integration of precipitation chemistry, precipitation,
air quality, cloud and satellite data should be of use to
acid rain modeling.
The national and regional air quality data bases have been
organized and screened to the point where common data sets
can be used now in evaluations and intercomparisons of
regional models.
An extended period of at least sulfate and wet sulfur deposi-
tion data is required to not only determine trends, but also
obtain a representative cross-section of episodes.
The degree of episodicity of ambient sulfate concentrations
and wet sulfur depositions has an important bearing on the
time scales of modeling that are required (episode or annual
average or both).
The principal recommendations are as follows:
A workshop on data bases for the development and evaluation
of regional models should be held soon.
EPA, NASA and NOAA should coordinate air quality, meteoro-
logical, precipitation chemistry, aircraft, and satellite
measurement programs more closely in the U.S. and with
Canada to provide more integrated data bases.
Regional modeling should include S02 episodes and annual
averages.
With regard to the last recommendation, the use of special con-
tinuous monitoring data (recall Figure 1) is generally preferred over
the 24-hour average S02 measurements only every 6-12 days that are
part of the NADB. An inventory of special sources of air pollutant
and meteorological data (see Figure 64) has revealed a wealth of S02
229
-------
. In house
On order
ORequested by EPA
FIGURE 64
SPECIAL SOURCES OF AIR POLLUTANT AND METEOROLOGICAL DATA
230
-------
data especially in the Ohio River Basin that has and is currently
being acquired by utility and industrial companies. Even though it
has taken several years to identify, acquire and analyze the best
sources of special S02 and meteorological data in the eastern U.S.,
the interesting case studies and their indications of long range
transport of S02 have made the effort very worthwhile.
Generally elevated S02 levels on a subregional or regional
scale are limited to the winter season when low level air mass trap-
ping conditions and greatly reduced conversion to sulfate, except
under certain cold, moist conditions as discussed previously, favor
the accumulation of SC>2 during stagnation and its long range trans-
port when the stagnation is replaced by stronger winds. These exam-
ples of accumulation and/or transport of S02 on a subregional scale
have been selected from the monitoring data of the three largest
utility companies in North America, namely Ontario Hydro, American
Electric Power (AEP), and the Tennessee Valley Authority (TVA). The
first example is from the Ontario Hydro monitoring data for January
19, 1976, when the yearly highest 24-hour S02 concentration oc-
curred at a monitor 9 kilometers north-northeast (NNEA) of the
Nantuoke Generating Station. Interestingly, on that day the 24-hour
S02 concentrations at most of the monitors along Lake Erie upwind
of the plant were elevated and the winds were rather strong and per-
sistent from the southerly direction at both the Erie, Pennsylvania,
and the Nantuoke meteorological tower. The implication of these
231
-------
o
cc
a
z<
CD O>
o. '-
0. Z
ll
o
«
O
UJ
O
CM
232
-------
results is that transport of sulfur dioxide from sources in Ohio and
Pennsylvania was responsible for the elevated background concentra-
tions in the Nantuoke network. Episodes with even higher "back-
ground" 24-hour S(>2 concentrations in the Nantuoke network have been
subsequently recorded in January 1977 and January 1978.
The second example of at least a subregional scale S02 epi-
sode is from the AEP monitoring data in the upper Ohio River Basin.
During early January 1977, an extremely cold mass of air moved out of
Canada across the upper Ohio River Basin. A sequence of high pres-
sure systems and cold fronts produced stagnation, trapping, and flow
reversals so that elevated S02 concentrations were first transported
from northeast to southwest and then vice versa along the general
orientation of the river. The 24-hour S02 concentrations at the AEP
monitors in the upper Ohio River Basin on January 5 and 6, 1977, show
generally higher values on the 5th than the 6th to the southwest and
higher values on the 6th than the 5th to the northeast with some ex-
ceptions.
The third example of a subregional scale S02 episode is from
the TVA monitoring data in Tennessee and Kentucky. During January
23, 1975, a small subsynoptic high pressure cell was located over
southern Illinois, producing a trapping condition with light and
variable winds over 4 TVA power plants with very large emissions of
sulfur oxides. Trajectories calculations for this day and location
even suggest the air may have circulated at least halfway around this
233
-------
KEY
^ Existing power plants
Monitor
OHIO
I
220/215
Tidd
nd
,. l
Cardinal
212/157
/nd/
/ *
57*
d/175
Philo|
» nd
183/202*
/ »154/144
/ * '
319/nd»/^ 209/230
209/230
144/147-
KENTUCKY
* 144/97. *«« .144/1
149/110 ] 419/36
141/92 Mitchell
y
209/134*
196/115*
246/128/lgg/89
Gavin J^f 188/94
152/7S£^rn
' »nd/71
139/98
1191/nd*
H Amos
154/71 136/97
/ 136/63
f107/58
131/71. *
x ,128/73
318 ST,dey," \ WEST VIRGINIA
Note; "nd" indicates that no data were available.
FIGURE 66
AVERAGE 24-HOUR SO, CONCENTRATIONS (Mg/m3) AT AEP MONITORS
IN THE UPPER OHIO RIVER AREA 5/6 JANUARY 1977
234
-------
abnormally small high pressure system. The 24-hour SC^ concentra-
tions at a number of monitors around 3 of the 4 TVA plants showed
elevated concentrations (>50 H-g/m^) quite different from the im-
pacts associated with coning or limited mixing conditions. The
observed pattern suggests a more area-wide accumulation of S(>2 and
a subsequent mixing down to the ground where it affected a number of
monitors around each of the 3 plants at about the same time. The
reasons for these processes being absent at the Cumberland plant are
not readily apparent, but are probably related to mesoscale varia-
tions in the inversion heights and the vertical mixing processes.
Interestingly, the subsynoptic high cell moved rapidly eastward and
disappeared on January 24, 1975, and was replaced by strong persis-
tent winds in the Ohio River Basin. The AEP monitors on January 24
and 25, 1975, show elevated background SO-^ concentrations moving
from the lower to upper basin. The implication of this result is
that the S(>2 concentrations that accumulated over the TVA sources
on January 23 were then transported by persistent winds through the
Ohio River Basin on subsequent days.
These 3 case studies serve to illustrate why regional modeling
should include S02 episodes as well as sulfate episodes. In fact,
episodes of the two pollutants can occur at the same time with the
January 23-24, 1975, and January 23, 1978, episodes presented in this
paper being two of the best examples identified so far.
235
-------
cc
O
cc z
= O
OS
u. >
O
cc
236
-------
Appendix A
Sulfate and TSP Measurements by State 1974-1978
237
-------
-------
TABLE 7 : Number of Sulfate Measurements by States in the Eastern
U.S. in the National Aerometric Data Bank*
STATE
Alabama
Arkansas
Connecticut
Delaware
D.C.
Florida
Georgia
Illinois
Indiana
Iowa
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
New Hampshire
New Jersey
New York
North Carolina
Ohio
Pennsylvania
Rhode Island
South Carolina
Tennessee
Vermont
Virginia
West Virginia
Wisconsin
1974
1963
93
371
47
78
131
65
118
383
151
116
106
41
43
136
120
1686
59
76
44
209
2923
1209
245
343
107
80
3106
56
205
56
101
1975
1107
91
348
49
200
250
77
217
376
141
104
88
32
46
155
158
523
51
156
39
216
2451
428
1418
4157
181
75
1754
48
192
63
131
IliAK
1976
129
117
292
72
271
751
79
189
466
124
96
107
24
38
118
129
78
58
192
28
194
2435
116
3558
5775
292
82
270
36
232
75
123
1977
116
52
60
60
137
698
73
55
153
66
116
64
49
1378
36
94
53
48
49
42
155
2401
98
249
4341
259
44
244
33
241
54
127
1978
29
13
37
6
95
227
6
0
95
26
24
48
219
627
32
31
153
23
21
437
64
1445
30
2063
5296
263
0
94
27
82
16
48
*As of June 1979
239
-------
TABLE 8 : Number of Total Suspended Particulate Measurements by
State in the Eastern U.S. in the National Aerometric
Data Bank*
STATE
Alabama
Arkansas
Connecticut
Delaware
D.C.
Florida
Georgia
Illinois
Indiana
Iowa
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
New Hampshire
New Jersey
New York
North Carolina
Ohio
Pennsylvania
Rhode Island
South Carolina
Tennessee
Vermont
Virginia
West Virginia
Wisconsin
1974
8098
2691
3364
849
209
5705
3214
3333
5372
2620
6202
844
1019
4404
3501
6499
4984
1662
3367
1374
3675
17159
8665
17734
8251
914
4244
10540
342
9688
2096
5024
1975
6258
2561
3668
864
329
5518
3375
12423
5857
2627
6782
1390
1146
3742
4777
6225
5448
1594
3636
1657
3405
15783
7522
20943
7104
925
4460
9567
532
9958
2537
6103
ICAtt.
1976
7579
2708
2699
259
601
8152
3870
10650
8153
3393
6886
1974
1403
5429
3258
6965
4829
1641
3783
2036
3627
18350
6912
25747
9184
946
4994
9054
623
5420
2387
5571
1977 1978
5912
2410
2716
755
464
8624
2949
9755
5507
3593
6398
1475
2353
4709
2626
7123
4393
1310
3492
1871
3416
16171
5197
25409
7622
752
4352
5433
521
10007
1944
6806
*As of June 1979
240
-------
TABLE 9 : Number of Sulfate Measurements by State in the Western U.S,
in the National Aerometric Data Bank*
STATE
Arizona
California
Colorado
Idaho
Kansas
Montana
Nebraska
Nevada
New Mexico
North Dakota
Oklahoma
Oregon
South Dakota
Texas
Utah
Washington
Wyoming
1974
2076
2529
27
27
82
35
82
69
26
642
75
27
29
1183
1324
45
36
1975
2116
1613
63
51
92
67
64
74
12
804
84
36
36
578
738
100
57
IHAK.
1976
1421
460
53
50
82
1280
75
75
24
1268
89
17
53
2354
28
75
24
1977
1481
472
552
50
83
691
42
40
132
1732
39
56
24
2084
376
85
15
1978
1540
119
4010
41
43
523
0
0
60
1422
12
36
141
2023
152
65
0
*As of June 1979
241
-------
TABLE 10: Number of Total Suspended Particulate Measurements by
State in the Western U.S. in the National Aerometric
Data Bank*
STATE
Arizona
California
Colorado
Idaho
Kansas
Montana
Nebraska
Nevada
New Mexico
North Dakota
Oklahoma
Oregon
South Dakota
Texas
Utah
Washington
Wyoming
1974
2809
6995
5361
1775
2911
2359
1767
1676
2566
865
4538
2690
590
6550
5321
2465
937
1975
3127
5493
5179
2645
3025
3431
1793
2185
2590
1135
2472
2826
912
3657
5846
3724
2058
YEAR
1976
3809
5347
5683
1847
3321
2903
2009
2062
3256
1306
2788
2641
839
9766
6558
3767
2380
1977
3463
4726
5565
3099
3106
3448
2035
2371
3611
1841
2766
2323
905
10076
7296
3449
2057
1978
*As of June 1979
242
-------
Appendix B
Annual Sulfate Concentrations for Eastern Subregions
and States and Western AQCRs and States
243
-------
-------
g
§.
E
00
o
V.
c
:z
-3-
o
CO
m
Bi
H
§
s
E-
g
J
S
«
u
J
PC
B
"^
00
E
00
2A5
-------
ffl
H
i*
oo
H
4->
(0
bJ
H
<
H
O
M
EC
E
-^
DC
H
x:
O
pa 1-1
H c/j
o-
H-
-------
1)
60
0}
O
13
O
f-f
O
a
§
N
H
CN VD CM CM 00
i i O i
CM
VD l i l -vT
i iCMOOvO
247
-------
60 c^'-<»-*CJ\'I iI"li< CM <I iI i< CM O"> I"- "? ^ 1 . . . .
tfl /->/^/-v ^ GO CM O
PI ^ CMCMCMCMU1OOOCMCMCM'£>00'-l' -
"&0 ° r-^io ^c^rn O O O co r^ CM C3 ^ r-T ^ ^ ^ CM CM
CO
w >
w 5>
H hJ
< cfl
OT M-j « ^^v^^^^s^v^^^v^^v^s.,-,-/^^-.-'^
pi r {y
60
z as
I I
O ^H <^^v ^
Z o^-s/-s
d
O
0)
O^ O^ O^ O^
i It I. li t
248
-------
CO
6
m m
« t>o
>-) a
ca vj
<; a>
H >
Vj
o
«->
*rH
c
£
CN n co O tn co
i i r i i
249
-------
-------
REFERENCES
Report on Fine Particulates, Prepared by the Task Force on Fine
Participates, Working Party on Air Pollution Problems (Ninth
Session, 9-12 January 1979) and Senior Advisors to ECE
Governments on Environmental Problems, United Nations Economic
and Social Council, ENV/WP.1/R.44, 12 September 1978.
Kreitzberg, C.W., 1979: Observing, Analyzing and Modeling Mesoscale
Weather Phenomena, paper to be published, 44 pp.
Altshuller, A.P., 1973: Atmospheric Sulfur Dioxide and Sulfate Dis-
tribution of concentration at urban and non-urban sites in
United States, Environmental Science and Technology, 7,8, 709-
712.
Altshuller, A.P., 1979: Seasonal and Episodic Trends in Sulfate
Concentrations in the Eastern United States, paper presented at
the American Chemical Society Meetings, September 13, 1979,
Washington, D.C.
Monitoring and Data Analysis Division, 1978: Status Report on Long-
Range Transport of Fine Particulates Matter, Report from Office
of Air Quality Planning and Standards, U.S. Environmental
Protection Agency, December, 18 pp.
Hanna, S.R. , 1978: In Air Quality Meteorology and Atmospheric
Ozone, edited by A. L. Morris and R. C. Barras, American Society
for Testing and Materials, Philadelphia, PA.
Benkovitz, C.M., 1979: Status and Plans for the MAP3S Data Bank,
Presentation at the MAP3S Review Meeting, Lewes, Delaware, June
4.
Protecting Visibility - An EPA Report to Congress, Office of Air
Quality Planning and Standards, U.S. Environmental Protection
Agency, Research Triangle Park, N.C. , 1979.
Schedule of ORD Research Milestones for Development of Dispersion
Models in Complex Terrain, May 1979.
Rodes, C.E. , 1979: Protocol for Establishment of a Nationwide Inhal-
able Particulate Network, Paper from the Environmental Monitor-
ing and Support Laboratory, U.S. Environmental Protection
Agency, Research Triangle Park, NC, May 15.
Mueller, P.K, 1979: Personal Communication, December 10.
251
-------
EPA Development Plan for Review of the SC^ NAAQS to be published in
May 1980 (regulatory alternative of a change to a statistical
standard, rather than the present determination standard).
Pace, T.G., and Meyer, E.L., Jr., 1979: Preliminary Characterization
of Inhalable Particulates in Urban Areas, Paper 79-47.2 pre-
sented at the 72nd Annual Meeting of the Air Pollution Control
Association, June 24-29, 1979, Cincinnati, OH.
Dockery, D.W., and Spengler, J.D., 1979: Personal Exposure to
Respirable Particulates and Sulfates, paper submitted to the
journal of the Air Pollution Control Association.
Raynor, G.S., and McNeil, J.P., 1979: An Automatic Sequential
Precipitation Sampler, Atmospheric Environment, 13, 149-155.
Raynor, G.S., 1979: Private communications, July 18, and September
13.
Smith, F.B., and Hunt, R.D., 1979: The Dispersion of Sulfur Pollu-
tants over Western Europe, Phil., Trans. R. Soc. Land. A, 290,
523-542.
Reports of Working Groups Workshop on Regional Air Pollution Studies,
June 7-10, 1976, conducted by the Triangles Universities Consor-
tium on Air Pollution under contract with The Environmental Pro-
tection Agency, Environmental Sciences Research Laboratory,
Research Triangle Park, NC.
Mueller, P.K., et al., 1979: The Occurrences of Atmospheric Aerosols
in the Northeastern United States, Paper to be published in the
proceedings of the conference on aerosols: Anthropogenic and
Natural Sources and Transport, New York Academy of Science, New
York, New York.
Warren, K. and Mueller, P.K., 1979: Data Available from the EPRI/
SURE Data Bank, ERT Document P-5042DB2, October.
Workshop Draft Reports, Seminar/Workshop on Persistent Elevated Pol-
lution Episodes (PEPE), March 19-23, 1979, Ramada Inn Downtown,
Durham, NC.
Memorandum to MAP3S Modelers, Some Suggestions on Where MAP3S Model-
ing Should Be Going, May 4, 1978, Lawrence Livermore
Laboratories.
252
-------
MacCracken, M.C., editor, 1979: The Multi-State Atmospheric Power
Production Pollution Study-MAP3S, Progress Report for FY 1977
and FY 1978. DOE/EV-0040. Livermore, CA.: University of
California, Lawrence Livermore Laboratories, July.
Husar, R.B., et al. 1979: Trends of Eastern U.S. Hazeness Since
1948. In Proceedings of the Fourth AMS Symposium on Atmospheric
Turbulent Diffusion and Air Pollution. Reno, Nevada, January
15-18, 1979.
Anthes, R.G., and Warner, T.F., 1978: Applications of General
Meteorological Models to Air Quality Problems. Air Quality,
Meteorology, and Atmospheric Ozone, American Society for Testing
and Materials, Philadelphia, PA, 445-457.
Baker, M., et al., 1979: Evaluation of the Spatial and Temporal
Measurement Requirements of Remote Sensors for Monitoring
Regional Air Pollution Episodes, NASA contract or report
1590922, ERT, Concord, MA, July.
Niemann, B.L., 1979: Meteorological Analyses of Persistent Elevated
Pollution Episodes. Final Report. Prepared under Contract No.
68-02-3092 and Subcontract No. 3-340-1714 (EPA Prime Contract
No. 68-02-3000). Berkeley, CA: Teknekron, Inc.
Niemann, B.L., et al., 1979: An Integrated Monitoring Network for
Acid Deposition: A Proposed Strategy, Interim Report, R-
0230EPA-79, Prepared for the EPA Office of Anticipatory
Research, Office of Research and Development, Washington, D.C.,
November.
Scott, B.C., 1978: Parameterization of Sulfate Removal by Precipi-
tation, Journal of Applied Meteorology, 17, 1375-89.
Whelpdale, D.M., and Galloway, J.N., 1979: An Atmospheric Sulfur
Budget for Eastern North America. In Proceedings of the WMO
Symposium on Long Range Transport of Pollutants, Sofia,
Bulgaria, 1-5 October, 1979.
Freiberg, J. 1979: Effects of Humidity and Temperature on Warm and
Cold Sulfate Episodes, Paper in Proceedings of the Symposium on
Sulfur Emissions and the Environment, May 5-8, 1979, London,
U.K.
Smith, L.T., and Mells, M.J., 1979: Regional Air Quality Analyses
and Its Application in the Ohio River Basin, Paper Presented at
the Fourth US-USSR Symposium on Comprehensive Analysis of the
Environment, October 22-27, 1979, Jackson, Wyoming.
253
-------
Smith, L.T. , and Niemann, B.L. , 1979: The Ohio River Basin Energy
Study: The Future of Air Resources and Other Factors Affecting
Energy Development - An Update and Future Plans: Utility
Monitoring Data, P-003-EPA-79/RI, Teknekron, Inc. Berkeley, CA,
February.
Dzubay, T.G. , 1979: Chemical Element Balance Method Applied to
Dichotomous Sampler Data, Paper to be published in the Proceed-
ings of the conference on aerosols: Anthropogenic and Natural
Sources and Transport, New York Academy of Sciences, New York,
New York.
Fye, T.K., 1978: The AFGWC Automated Cloud Analysis Model, Air Force
Global Weather Central Technical Memorandum 78-002, June.
Jenne, R.L., 1979: Data Sets for Meteorological Research, NCAR
Technical Note TN/1A-111, July.
Baker, M.B., et al., 1979: Simple Stochastic Models for the Sources
and Sinks of Two Aerosol Types, Tellus, 31, 39-51.
Tryonis, J., and Shapland, D. , 1979: Existing Visibility Levels in
the U.S., Prepared by Technology Service Corporation for the
U.S. Environmental Protection Agency, Grant No. 802815, Research
Triangle Park, NC.
Niemann, B.L., 1980: An Integrated Monitoring Network for Acid
Deposition II: Analyses of Data Bases to Support Final Recom-
mendations, report in preparation.
Niemann, B.L. , 1979: Results of Multiscale Air Quality Impact
Assessment for the Southwest-Rocky Mountain - Northern Great
Plains Region, Part 1 - Slide Script R-005-EPA-79/R2, Berkeley,
CA, Teknekron, Inc., July.
Fox, J.D., 1979: Testing and Documentation of Programs Used to
Transform Climatological Precipitation Data to a Geographically
Gridded Format, Pacific Northwest Laboratory Annual Report for
1978 to the DOE Assistant Secretary for the Environment, Part 3,
Atmospheric Sciences, PNL-2850, UC-11, Richland, Washington.
Berg, W.W., et al., 1977: Time Dependent Sulfur and Trace Metal
Correlations in Non-Urban Aerosols from an Eastern U.S. Meso-
scale Network, Paper Presented at the Symposium on Atmospheric
Sulfur Compounds, Formation, and Removal Processes, AICHE 10th
Annual Meeting, New York, November 13-17, 1977.
254
-------
Nieraann, B.L., and Hall, B.R. , 1979: Data Compilation of Sulfate and
Oxidant Episodes, Data Set VC, RM-039-EPA-78, Prepared for the
Ohio River Basin Energy Study and the U.S. EPA Office of Energy,
Minerals, and Industry, Berkeley, CA, Teknekron, Inc., May.
Barnes, R. A., 1979: The Long Range Transport of Air Pollutants - A
Review of European Experience, of Air Pollution Control Assoc.,
29, 12, 1219-1235.
Malm, W.C., et al., 1979: Visibility in the southwest, Technical
Paper, Environmental Monitoring Systems Laboratory, U.S. Envi-
ronmental Protection Agency, Las Vegas, Nevada.
Tong, E.Y., et al., 1979: Characterization of Regional Sulfate/
Oxidant Episodes in the Eastern United States and Canada, Paper
presented at the 72nd Annual Meeting of the Air Pollution Con-
trol Association, Cincinnati, OH, June 25-28, 1979.
Niemann, B.L., and Apodoca, R. , 1980: The Ohio River Basin Energy
Study: The Future of Air Resources and Other Factors Affecting
Energy Development, Paper No. 5-Regional Haze Patterns from
Satellite and Air Quality Data, p-017-EPA-79, Berkeley, CA,
Teknekron, Inc., draft.
Niemann, B.L., et al., 1979: Regional Air Quality Assessment for the
Tennessee Valley Authority: Task 1-Initial Results and Interim
Recommendations on Regional Air Quality Issues, RM-084-TVA-79,
Berkeley, CA, Teknekron Research, Inc., October.
Mahan, A.L., et al., 1979: Characteristics and Origins of Sulfur
Dioxide, Total Suspended Particulates, and Sulfates in Western
Pennsylvania, R-019-EPA-79, Berkeley, CA, Teknekron, Inc., June.
Niemann, B.L,, 1980: Atmospheric Transboundary Flux of Sulfur
Compounds in the Great Lakes (St. Lawrence Region), draft report
in preparation.
255
-------
-------
HYBRID REGIONAL AIR POLLUTION MODELS
R. Drake - Pacific Northwest Lab.
INTRODUCTION
This discussion deals with a family of air quality models for
predicting and analyzing the fine particulate loading in the atmo-
sphere, for assessing the extent and degree of visibility impairment,
and for determining the potential of pollutants for increasing the
acidity of soils and water. The major horizontal scales of interest
are from 400 km to 2000 km; and the time scales may vary from several
hours, to days, weeks, and a few months or years, depending on the
EPA regulations being addressed.
To set the stage for this discussion, we first indicate the role
air quality models play in the general family of atmospheric simu-
lation models. Then, we discuss the characteristics of a well-
designed, comprehensive air quality model. Following this, the
specific objectives of this workshop are outlined and their modeling
implications are summarized.
There are significant modeling differences produced by the
choice of the coordinate system, whether it be the fixed Eulerian sys-
tem, the moving Lagrangian system, or some hybrid of the two. These
three systems are briefly discussed, and a list of hybrid models that
are currently in use are given. Finally, the PNL regional transport
model is outlined and a number of research needs are listed.
257
-------
AIR QUALITY MODELS
A mathematical simulation is a model, or working analogy, of a
physical phenomenon. It is used to analyze and study that phenomenon
and to communicate results about that phenomenon to others. The
major elements of a model are:
A temporal and spatial domain of computation that determines
the range of influence of the model.
A set of unknown quantities {the dependent variables) that is
specified by a set of conservation equations and constituent
relationships.
A set of system parameters that defines proportionality rela-
tionships in the governing equations.
Input data that determines the auxiliary conditions (boundary
and initial conditions) and the system parameters.
A solution process that is consistent with the nature of the
governing equations and the time and space resolution
required by the application of the model.
Output and data management schemes that process and exhibit
the results of a simulation.
A testing protocol for checking the output of the model
against independent data sets, and for examining the internal
consistency of the model.
A comprehensive model of the atmosphere and its constituents
consists of a set of nonlinear, coupled differential and integral
equations describing atmospheric processes and their interactions
between the underlying soil and water surfaces. The governing system
of equations consists of the following:
For the air - equation of state; equations of motion; ther-
modynaraic equation; conservation of mass equations for dry
air, water vapor and other trace constituents, including air-
borne pollutants; a solar radiation balance equation.
258
-------
For the soil - balance equations for water and heat.
Forthe water - equations of motion; thermal equation; con-
servation of mass equations for ice and salt.
The wind surface stress, radiant flux, evaporation and precipitation
couple the atmospheric equations to the water equations, while only
radiant flux, evaporation and precipitation couple the air and soil.
The family of air quality models (AQM's) is a subset of the mod-
els describing the atmosphere and its constituents since an AQM only
consists of the conservation of mass equations for the airborne pol-
lutants of interest, along with certain auxiliary modules. These
auxiliary modules describe the atmospheric forcing functions, such as
wind, diffusion and temperature.
Well-designed AQM's are playing an increasing role in air qual-
ity decision-making and management. Models organize and present data
logically and coherently so decision makers can understand key issues
and differences of opinion. They play an integral part in increasing
cost-effectiveness of air quality management, such as determining the
best balance between the costs of releasing emissions and the costs
of containment or control of these emissions. The AQM's are used for
simple screening and trend analysis of airborne pollutants, long-
range control plans and emission control strategies, short- and long-
term forecasting, facility and monitor network siting, land-use
planning, analysis of short-terra episodes, analysis of long-term
assessments, and research on the fate of airborne pollutants.
259
-------
CHARACTERISTICS OF AQM's
An AQM is used to predict the geographic and chemical fate of
airborne emissions. The model takes emission characteristics as
input and produces as output estimates of ambient air concentrations
and deposition rates of material deposited on surfaces. This output
can then be used as input to visibility models, soil and water pH
models, and various ecosystem pathway models. The air pathway pro-
cesses that control the fate of pollutants are transport, diffusion,
transformation and removal. The first of these processes determines
where pollutants from a given source will be found, and the remaining
processes determine their concentration and chemical form.
A generalized AQM should contain the following elements, or mod-
ules;
Terrain - accounts for topography, roughness elements, ground
cover, albedo, and surface fluxes of heat and moisture.
Source - accounts for natural and anthropogenic sources of
gases and particles, both local and background values, and
from urban, rural and industrial locations.
Water - accounts for the atmospheric distribution of humid-
ity, fog, clouds and liquid and solid precipitation.
Radiation - gives the solar radiation quantities needed to
evaluate photochemical reactions, impairment of visibility,
and surface temperatures.
Temperature - defines the regional, inesoscale and local
atmospheric temperature structures, including inversion sur-
faces and mixing depths.
State Variables - gives the atmospheric pressure and density
in a manner consistent with the temperature and humidity dis-
tributions.
260
-------
Transport - gives the mean winds that transport the "center
of mass" of the pollutants on a regional scale (400 to
2000km), the mesoscale (40 to 400km), and a local scale (1 to
40km).
Diffusion - accounts for the dilution and spread of the pol-
lutant about the "center of mass" as it moves with the mean
wind.
Chemistry of Gases - accounts for thermal and photochemical
reactions of gases that result in gaseous products.
Gas-to-Particles - accounts for the production of secondary
particles from gaseous reactions.
Particle Interactions - accounts for physical and chemical
interactions between gases and particles, and particles and
particles.
Dry Removal - considers the removal of airborne gases and
particles by impaction, absorption and chemical interaction
with various surfaces.
Wet Removal - considers the removal of airborne gases and
particles by rain, snow and fog.
Solution Algorithm - describes the method of solution for the
equations describing the inputs and processes listed above.
Output Algorithm - specifies the form of the output for air
concentrations of pollutants and deposited pollutants in a
manner consistent with post-processing modules (i.e. visibil-
ity and pH calculations) and requirements specified by EPA
regulations.
Visibility - accepts air concentrations of pollutants and
converts them into optical measures of the atmosphere, such
as visual range, contrast, blue-red luminance ratio, and
plume perceptibility.
Acidic Soil and Water - accepts the deposited pollutants and
converts them into the pH of the soil and water.
A SET OF SPECIFIC REQUIREMENTS FOR AN AQM
The main objective of this workshop is an assessment of the
state-of-the-art of regional air quality models (RAQM's) used for the
261
-------
prediction and analysis of the airborne concentrations of fine par-
ticles, the impairment of visibility, and the pH of soil and water.
Fine Particles
Airborne concentrations of fine particles (diameters less than
2 microns) originate from anthropogenic and natural sources as pri-
mary particles or as gaseous precursors that form particles along
an air pathway. The transition of atmospheric compounds from gas to
aerosol phase is difficult to quantify but rather easy to demonstrate
qualitatively. Gas-to-particle conversion occurs over all chemical
classes, with a spectrum of elements, vapor pressures, and functional
groups being represented [1]. Conversion takes place by either homo-
geneous nucleation or heterogeneous condensation, depending on the
degree of supersaturation of the pollutant vapors and the concen-
tration of small airborne particles that act as condensation nuclei.
The important processes producing particles by these two modes are
(23:
Homogeneous Nucleation
Physical Processes Producing Supersaturation
1. Adiabatic expansion
2. Mixing
3. Conductive cooling
4. Radiative cooling
- Gas Phase Chemical Reaction
1. Single condensable species (classical theory)
Heterogeneous Condensation
- Transport Limited
262
-------
1. Diffusion, if particle diameter is less than the mean-
free-path in air.
2. Molecular bombardment, if particle diameter is greater
than the mean-free-path in air.
- Surface Controlled Chemical Reaction.
- Particulate Phase Controlled Chemical Reaction.
Investigators have shown that from 1/3 to 1/2 of the aerosol
mass in the Los Angeles Basin is due to gas-to-particle conversion
[2]. The concentrations of pollutants in these aerosols were found
in the order: organics, nitrates, sulfates. However, based on the
statistical evaluation of the data, the nitrates were more efficient
than organics in visibility degradation. Typical submicron particles
found in the Los Angeles aerosols are ammonium sulfate, ammonium ni-
trate, oxygenated organic species, water and primary materials such
as lead and salts.
VisibilityImpairment
The impairment of visibility and increased haziness in the
atmosphere are due to the scattering and absorption of light by fine
particles and gases. The greatest effects occur for particles rang-
ing in size from 0.1 to 1.0 microns in radius and for the gas N02.
Nitrogen dioxide produces a yellow-brown cloud over an area because
it is strongly absorbent over the blue-green part of the visible
spectrum and thus produces an overbalance to the yellow-red part of
the spectrum. The presence of sufficient concentration of particles
will mask the N02 effects due to enhanced scattering, resulting in
263
-------
a whitish haze. The airborne particles that usually produce this
effect are sulfates, nitrates and natural and anthropogenic organics.
Visibility impairment: is also produced by natural causes, such as
fog, precipitation and windblown dust.
pH of Soil and Water
The pH of water and poorly buffered soils is affected by the wet
and dry deposition of pollutants on these media. If the hydrogen ion
content in these media increases, the pH decreases and the soil and
water become more acidic. Proton donors in the atmosphere (wet re-
moval) or at the surface (wet and dry removal) that are major sources
of acidity are the inorganic acids HNOj and I^SO^; other sources are
HC1 and organic acids.
EPA Regulations
The Clean Air Act and Amendments set primary and secondary
Ambient Air Quality Standards (AAQS) for particulate matter and the
precursor gases that are important in fine particulate formation,
impaired visibility and the decrease of pH in soil and water. Reg-
ulations will be promulgated in 1980 for control of the optical ef-
fects of plumes (plume blight). These visibility regulations will
probably be concerned with one or more of the following: visual
range, contrast, blue-red luminance ratio and plume perceptibility.
Future regulations will probably be concerned with veiled haze, re-
gional visibility and acid precipitation.
264
-------
The EPA regulations for PSD and criteria pollutants are in terms
of 1, 3, 8 or 24 hour averages, or in terms of annual averages, for
ground-level air pollutant concentrations. Determining these hourly
averages and their frequency of occurrence, along with the annual
averages, are important considerations in constructing a Regional Air
Quality Model (RAQM).
MODELING IMPLICATIONS OF THESE REQUIREMENTS
The application of RAQM's for the prediction and analysis of
fine particulate loading in the atmosphere, visibility impairment and
acid precipitation is taking and will take place over all areas of
the country. For example, RAQM's are being and will be used over the
industrial Northeast, the oil and gas producing areas bordering the
Gulf of Mexico, the tar sand and oil shale areas of the Rocky Moun-
tains and the coal and oil producing areas of Alaska.
Th_e_ Regional Domain
A domain of computation that has a horizontal scale of 400km to
2000km will contain a variety of natural and anthropogenic sources
of primary particles and precursor gases, precursors of the fine par-
ticles and acidic rain and snow. In addition, the domain will span
over a variety of terrain features, such as forests, grasslands,
urban areas, mountains, hills and valleys, bays, gulfs and oceans.
The transport and diffusion of the pollutants within the three-
dimensional domain of computation wil1 be governed by the gradient
winds of the atmosphere, atmospheric stability, surface roughness,
265
-------
thermal forcing due to variable heating in the mountains and at land/
water interfaces, and by the dynamic effects produced by mountainous
terrain.
Secondary Fine Part ic1e s
The formation of secondary particles in the atmosphere requires
knowledge of many precursor species, their concentrations, and their
thermal and photochemical reaction chains. Once fine particulate
matter is formed in the atmosphere, the particle spectra evolve by
condensational growth of nuclei and by the coagulation mechanisms of
Brownian motion, sedimentation and turbulent and laminar shear.
Atmpspheric Optic s
Particulate matter and gases in air affect atmospheric optics
through scattering and absorption; the main culprits being N02,
nitrates, sulfates and organics. For particles, the time-evolving
size spectra are very important. For example, if the median particle
size changes from 0.4 microns to 2.0 microns there is a factor of
three reduction in visibility, assuming certain quantities are held
constant. In addition, the visual effects of aerosols are greatly
dependent on chemical composition. Chemical composition affects
visibility through the extinction properties of the chemicals. Many
atmospheric compounds; including sulfates, nitrates and soil par-
ticles, are transparent to light and will act as scatterers only.
Other particles, such as those containing carbon, are completely
opaque as their size increases; these particles will both scatter
266
-------
and absorb radiation. For a mixture of pure tranparent particles and
pure opaque particles the extinction effects are additive. However,
for an aerosol whose particles contain both transparent and opaque
substances, such as an absorbing core surrounded by a water soluble,
transparent layer, the extinction effects are not additive and must
be calculated from the Mie theory. Other factors that influence vis-
ibility are the magnitude of the solar flux, background intensities
(i.e., blue sky, white clouds, mountains), and the geometry of the
observer with respect to the plume.
Removal^ Processes
Wet and dry removal of the airborne gases and particles con-
tribute to the variations in soil and water pH. Dry removal is the
direct transfer of a material from the atmosphere to the earth's
surface by adsorption, absorption and chemical fixation. Dry removal
mechanisms include gravitational settling, transport by atmospheric
turbulence, impaction, chemical reactions, and concurrent surface
fluxes such as the flux of water vapor. Wet removal is the scaveng-
ing of pollutants by precipitation. Below-cloud scavenging is the
collection of aerosols and gases beneath the visible cloud by preci-
pitation. In-cloud scavenging occurs during the formation and growth
of cloud particles. The scavenging of gases and particles depends on
the following parameters and quantities:
The solubility of gases in water and the amount of gas
absorbed by a water drop.
The chemical species of the gas and aerosol.
267
-------
The aerosol number distribution.
The dominant collision mechanism, such as Brownian motion,
geometric sweepout and turbulence agitation.
The type and intensity of precipitation, length of the pre-
cipitation event, and the size of raindrops, snowflakes or
fog particles.
The turbulence and electrical characteristics of the atmo-
sphere.
EPA Regulations
The implications of the EPA regulations on modeling are due to
the limits placed on various hourly averaged concentrations of pollu-
tants, the limits on annual averages, the upcoming limits on various
measures of atmospheric optics, and future regulations on acid precip-
itation and regional visibility. In addition, the specific use of a
model will determine its form and complexity. Models may be used for
reviewing the effects of new sources, for determining PSD impacts,
for assessing visibility impairment, and for analyzing the impacts of
new industries in nonattainment areas.
Summary
The form, complexity and resolution of a RAQM can be determined
once the regulatory use, required outputs and region of application
are set. For the particular set of requirements specified in this
workshop, candidate RAQM's should contain all of the elements listed
in the section "Characteristics of AQM's." The complexity of these
elements are dependent upon the factors given above.
268
-------
SCALE CONSIDERATIONS OF RAQM's
The resolvable scales (the model scales) for a RAQM are usually
in the following ranges:
Horizontal - minimum scales from 10 to 30km, maximum scales
from 400 to 2000km.
Vertical - minimum scales from 10's of meters to 100's of
meters, maximum scales of a few 1000's of meters.
Time - minimum scales of 10's of minutes to one or two hours,
maximum scales of several hours to three days.
Climatological - minimum scales of a few weeks to a month or
season, maximum scales of a year or more.
The "time" scale refers to the RAQM's used for EPA hourly-type
regulatory problems, while the "climatological" scale refers to the
RAQM's used for annual-average regulatory problems and for long-term
assessments.
Subscale phenomena are meteorological, terrain, and chemical
phenomena that occur on time and space scales smaller than the mini-
mum model scales, while smperscales refer to phenomena occurring on
scales greater than the maximum model scales. The subscale phenomena
in a RAQM consist of mixing processes and small-scale turbulence,
many of the removal mechanisms, and many of the transformation pro-
cesses. The superscale phenomena are used to set the boundary and
initial conditions of a RAQM. In a RAQM, the model scales are great-
ly influenced by the underlying topography, the release of latent
heat, and nonlinear interaction between subscale and superscale
phenomena. A major problem in regional air quality modeling is the
269
-------
lack of routine measurements of meteorological and chemical phenomena
on the scales required by a RAQM.
Figure 1 depicts examples of terrain and meteorological phe-
nomena in the model scale, subscale and superscale categories. For
example, the meteorological phenomena falling in the model scale cat-
egory are air masses, fronts, thermo-tidal waves, lee waves, valley-
plain winds and certain channeling effects. Subscale meteorological
phenomena are single thunderstorms, airflow separations and wake
effects, and small-scale turbulence. Superscale terrain phenomena
are effects produced by global and continental mountain ranges, and
oceans and large gulfs. Superscale meteorological phenomena are
global wind patterns, long wave ridges and troughs, storm tracks,
and patterns of cyclones and anticyclones.
270
-------
m
Hs
1000
300
700
JO
Km*
As
1O1
10*
10'
10'
1
10 '
10"
\-
\
MACROSCAlf.
a
MACRO SCALE
/?
(SYNOPTIC!
MESOSCALE
a
MESOSCALE:
<*
MESOSCALE
V
MtCROSCALE
a
MICROSCALE
0
MICROSCALE
y
ORLANSKI
CLASSIFICATION
GLOBAL
MOUNTAIN
AREA
CONTINENTAL
MOUNTAIN
AREA
REGIONAL
MOUNTAIN
AREA
MOUNTAIN
VALLEY (PLAIN!
BASIN
ISLAND
HILLS
RIDGES
GORGE
CANYON
CLIFFS
MESAS
TERRACES
GAP
CLIFFS
LARGE ROUGHNE
ROUGHNESS
TREES
TREES
VEGETATION
SMALL
ROUGHNESS
GENERAL
LANDFORM
OR
ROUGHNESS
MONTH DAY HOUR MINUTE SECOND
SUBTROPICA
GLOBAL WINC
LONG WAVE 1
TROUGHS
MONSOONS
SS
. JET STREAMS
> PATTERNS
IIOGESAND
STORM TRACKS
CYCLONES AND
ANTICYCLONES
AIR MA
FRONTS
CYCLOC
iSES
ENESIS
THERMO TIDAL
WINDS
LEE WAVES
SLOPE VALLEY WIND
VALLEY PLAIN WIND
CHANNELING
WAKE EFFECTS
BLOCKING
AIRFLOW SPEED I
WAKE EFFECTS
CHANNELING
CANYON
WIND
AIRFLOW
SEPARATIONS
AND WAKES
CHANNELING
p
AIRFLOW
SEPARATION
VERTICAL
WIND
PROFILES
TURBULENCE
VERTK
1 1
CLIMATOLOGICAL SYNOPTIC AND MESO
SCALE PLANETARY SCALE SCALE
1
I |
AL WIND
PROFILES
TUHIl
LENCE
MICROSCALE
FIGURE 1
SCALE CLASSIFICATION SYSTEM FOR TERRAIN
AND METEOROLOGICAL PHENOMENA,
PATTERNED AFTER THAT OF ORLANSKI [3]
271
-------
THE HYBRID APPROACH
The choice of a coordinate system fundamentally affects the formulation
of a RAQM. Two systems normally used in RAQM's are: The Eulerian system
which is fixed at some point on the earth's surface, and the Lagrangian
system which is fixed to a moving pollutant cloud {Figure 2). An Eulerian
quantity Q is expressed as a function of the spatial coordinates x and time
t,
Q = f(x,t\ where f(x,tQ) = Q0 = given,
and a Lagrangian quantity P is a function of the initial location of the
quantity x0 and time t,
P = g(x0,t), where g{x0,tQ) = P0 = given.
The relationship between the two systems is given by the "point transforma-
tion"
;< = x(xQ.t) or ^ = x^x.t) .
z1
+-Y'
FIGURE 2
EXAMPLES OF AN EULERIAN SYSTEM XYZ
AND A LAGRANGIAN SYSTEM X' Y' Z1,
WHERE LOCATIONS a, b, c, d, REPRESENT
THE POLLUTANT CLOUD AT TIMES ta
-------
This transformation represents a certain fluid "particle" that is at
position x0 at time t0 and at position x at time t _> to.
If Q is a random variable, as is the case with atmospheric
phenomena, we usually are seeking some statistical measure of this
variable, such as its mean, standard deviation, or some higher order
moment. These measures can be expressed in terms of either Eulerian
or Lagrangian statistics. In atmospheric sciences the turbulent
diffusion of pollutants is most easily formulated in the Lagrangian
sense, while most pollutant concentration data have been obtained in
the Eulerian sense. Hence, it is convenient to have a relationship
between Eulerian and Lagrangian statistics. A relationship has been
theoretically derived for atmospheric diffusion, but the formula is
extremely difficult to apply to real flows. However, some expirical
relationships have been obtained for diffusion and these are used to
relate Lagrangian formulations to Eulerian data.
Figure 3 summarizes some of the known results for Eulerian and
Lagrangian atmospheric diffusion, as well as some relationships be-
tween the two systems. In this figure the quantity c represents the
mean concentration of pollutant "particles" at location x and time t.
The quantity S(x',t') in the Lagrangian system is a known function
describing the distribution of sources of the diffusing particles.
The quantity P(x,t; x'.t') is the conditional probability density
for a diffusing particle released at (x1, t1) to be found at (x,t),
where t _>. t' . The quantity v'c' in the Eulerian system is an unknown
273
-------
second moment v'c1 can be approximated by the well-known K-theory
method, or by a variety of higher-order approximations [7,8,9].
Eulerian RAQM1s are usually K-theory models, while Lagrangian
RAQM1s are usually based on the Gaussian puff concept. For the pre-
sent applications the RAQM1s must be far more comprehensive than
those indicated in Figure 3. A hybrid RAQM combines features from
both the Sulerian and Lagrangian types.
Hybrid RAQM's may be of the following types:
Lagrangian models where complex processes, such as precipita-
tion events, surface radiation fluxes, visibility calcula-
tions, and complex terrain effects, are accounted for by
Eulerian submodels or modules.
Trajectory models where pollutant parcels follow along curved
trajectories while diffusion in perpendicular planes is han-
dled by K-theory methods.
Grid models where the governing equations are reduced in com-
plexity through the use of fractional steps [10J; some of the
processes in the reduced system are treated by Lagrangian
methods, while others are treated by Eulerian methods.
EXAMPLES OF HYBRID MODELS
Examples of hybrid models are identified in this section. The
scales considered in the models are local, mesoscale or regional.
Relationships Between Eulerian and Lagrangian Statistics
Several investigators have conducted computer experiments to
simulate Lagrangian statistics of particles moving through Eulerian
velocity fields. One of the earliest experiments was performed by
Patterson and Corrsin [11]. In this experiment, the velocity field
was one-dimensional, random and binary. Because of the crudeness of
276
-------
the model, some of the generated particle statistics were contradic-
tory. Kraichnan [12] and Riley and Corrsin [13] improved the model
in [11] by considering a three-dimensional, random Eulerian velocity
field. However, this velocity field developed differently in time
than a Navier-Stokes fluid because of the lack of interaction among
the wave numbers synthesizing the turbulent field. Hence, the newer
models still produced some unrealistic results.
Thompson [14] used a model similar to that in [13] to analyze
the dispersion of smoke from a stack in complex terrain. The simu-
lation accounted for horizontal and vertical wind shear, buoyancy
and anisotropic turbulence. The resulting smoke patterns obtained
from the ensemble of particles approximated reality.
Other papers of this general type are:
Justus and Hicks [15] studied the anisotropic diffusion of
particles in three-dimensional, stratified, shear flows.
Liu and Thompson [16] improved the one-dimensional model in
[11] by considering a stationary, homogeneous Eulerian ve-
locity field with correct first and second order statistics.
Watson and Barr [17] developed a Monte Carlo simulation of
regional diffusion of a bomb-produced cloud of debris using
a transport wind derived from 12 hour radiosonde measurements
enhanced by a random component for intermediate time scales
plus a random component for small time scales.
Particle Models
Particle models are hybrids which follow a pollutant passing
through an Eulerian grid. In this method the spatial distribution
of the pollutant is represented by a large number of Lagrangian par-
ticles of constant mass that are advected in a fictitious velocity
277
-------
field consisting of the true velocity field plus a turbulent flux
velocity field. The fixed Eulerian grid divides physical space into
cells and the particles carry pollution from cell to cell as they are
moved by the fictitious velocity field. This field causes the parti-
cles to move apart or together, resulting in uneven distributions.
In order to satisfactorily simulate spatial distribution of pollu-
tion, a large number of particles must be used in each grid cell.
The particles are initially placed at random within each cell. The
number of particles in a cell depends on the initial concentration
specified for that cell.
Some of the particle models that have been developed in the past
are:
The Marker-and-Cell (MAC) technique is one of the earliest
methods. This method uses massless particles to indicate the
presence of fluid and to define the positions of free liquid
surfaces [18;.
The Particle-in-Cell (PIC) is a modification of the MAC
method for the treatment of compressible flow problems. In
this method each marker has a mass associated with it [19]
The PICK code of Sklarew et al. [20] is a PIC code with K-
theory diffusion. The code relates the diffusion of pollu-
tants to the continuity equation for a compreisible fluid.
The PICK code can be used to study photochemistry as well as
nonreactive chemistry. Various modifications are given in
[21] to [23].
The ADPIC code is a newer version of the PICK code and the
salient features are a variable wind field, dry and wet
removal modules, an expanding grid version for release of
single puffs, and a fixed grid for instantaneous and continu-
ous releases near or at the ground [24,25].
278
-------
Kao et al. [26] developed a three-dimensional, large-scale
model for the movement of particles around the globe during
the winter months.
Sheih [27,28,29] developed a particle model that reduced the
time requirements of the previous versions such as ADPIC and
PICK. In this model, Lagrangian puffs are advected by the
mean wind through the use of a small number of tracer parti-
cles.
References [30,31,32] deal with the random walk of particles.
Hall [30] predicted particle concentrations over short and
medium ranges that were consistent with observation. Reid
[31] studied the effects of vertical dispersion in the neu-
tral surface layer for surface and elevated releases. Anbar
[32] used Brownian motion processes to design a practical
field monitoring system.
Langrangian Parcels and Lagrangian-Corrected Grid Models
There are several models that consist of a parcel being advected
along a trajectory by the wind, while at the same time undergoing
diffusion and transformation in a plane perpendicular to the trajec-
tory. Other models use this basic idea to correct Eulerian grid
models for pseudo-diffusive errors. Examples of these models are:
SAI developed two versions of a reactive plume model: RPM-I,
[33]; RPM-II [34]. RPM-I consists of an advecting, well-
mixed, expanding slab of reactive pollutants traveling along
a curved trajectory. RPM-II is an improved version of RPM-I,
with a better treatment of photochemistry and variable diffu-
sion in the transverse direction.
Hales [35] has developed a model called PROTEUS that is
similar in structure to RPM-II, except K-theory is used in
both the horizontal and vertical directions perpendicular to
the path of the parcel. The kinetics package of this model
also contains a gas-to-particle and aerosol package [36].
Gillani [37] developed a quasi-steady, Lagrangian model for
the assessment of power plant plumes for distances up to
250km. This model incorporates the diurnal variability of
the planetary boundary layer structure, chemical conversion
and dry removal.
279
-------
Tangermann [38] used a mixed Lagrangian-Eulerian finite dif-
ference scheme to solve the three-dimensional dispersion
equation in a stratified planetary boundary layer. The model
includes vertical wind shear, ground roughness and atmos-
pheric stability.
Several investigators have reduced the pseudo-diffusive
errors of the standard Eulerian grid models through the use
of a Lagrangian approach. Egan and Mahoney [39] developed a
moment-conserving numerical technique that reduced artificial
diffusion in their SC^- transport model; Rao et al. [40]
and Martinez et al. [41] extended this work to a regional
S02/sulfate model. Boris and Book [42] introduced a more
sophisticated scheme of correction, called the flux-corrected
transport algorithms. These authors feel their scheme is
competitive in computer time and accuracy.
Runca and Sardei [43] used a mixed Eulerian-Lagrangian finite
difference scheme to horizontally advect and vertically dif-
fuse pollutants from a line source. The scheme is mass con-
servative and avoids the artificial diffusion of Eulerian
grid systems.
Leakey [44,45] used a moving cell model to study the concen-
trations of S02, NOX and CO over New York City and
Edmonton, Alberta. Drivas et al. [46] developed a well-mixed
moving parcel model to study the evolution of photochemical
smog in urban areas. The EPA trajectory model, DIFKIN, is
similar to the Drivas model, except vertical diffusion is
considered and different chemical kinetics are used [47].
Sheih [48] developed a statistical trajectory model to study
the S02~sulfate concentrations over the northeastern United
States. This model simulates the release of pollutant puffs
from many sources and the vertical distribution of S02 and
sulfate concentrations are developed from a vertical,
advection-diffusion equation, with added terms for removal.
The Air Resources Laboratory of NOAA developed a regional-
continental scale transport, diffusion and deposition model
[49]. This model is the foundation of both the work at
Brookhaven National Laboratory (BNL) [50], and the work at
Colorado State University (CSU) [51]. The regional models at
CSU and BNL are both being used to study the S02~sulfate
problems in the northeastern United States. A regional model
developed at PNL is also being used to study these same
problems [52]. The PNL work is briefly discussed in the next
section.
280
-------
REGIONAL MODELING AT PNL
The PNL long-term regional model [52] is a versatile multi-
optioned code which simulates SC>2 plume element transport by tra-
jectories on a grid. It then calculates the plume contributions of
S02 and SO^ on each grid element based on the solution of coupled
mass conservation equations. The model simulates transport, diffu-
sion, wet and dry deposition, and SC>2 to 804 transformation over
one to twelve hour intervals. Results are averaged to provide
monthly values.
Input to the model includes gridded wind and precipitation
fields, SC>2 emissions and various operating or physical parameters.
Several model versions, options, and tests have been coded:
diurnal stability and mixing depth variability
horizontal diffusion by plume segment geometry for short-term
predictions
Gaussian to uniform vertical diffusion
time average or hourly precipitation
variations in emission source specifications
deposition by Horst's surface depletion approximations [53]
gridded deposition velocities based on ANL work
output modifications for validation against MAP3S/SURE data.
To improve the wind fields used in this regional model, we have
developed a generalized, direct method for adjusting windfields [54].
This mass consistent model computes windfields over complex terrain
with a terrain conformal coordinate system, using a "modified"
281
-------
Newton's method of solution. This method does not involve the itera-
tive solution of the Poisson equation. Thus, the model is more
flexible and efficient than the other balanced windfield models and
converges more rapidly to the final solution.
An Eight Layer Diabatic Regional Air Pollution model (ELDRAP)
has been used to simulate short-term regional patterns of SC>2 and
sulfate concentrations and cumulative wet and dry deposition of sul-
fur during precipitation events [55]. It is also being tested for
acid rain cases. Necessary data for input to the model are transport
winds, potential temperatures, mixing ratio and precipitation on a
regional scale, and emissions. A time series of eight layers of
gridded winds, potential temperature and mixing ratio as well as
gridded hourly precipitation are required for a simulation. Sources
can be point sources or area sources. ELDRAP calculates horizontal
dispersion during the Lagrangian transport by the variations in the
mean windfield. Vertical dispersion is estimated using Gaussian dif-
fusion. Dry deposition velocities, S02~sulfate transformation
rates, emission rates, and stability profiles are included. Wet
removal as a function of hourly precipitation is also included for
scavenging applications.
RESEARCH NEEDS
A mathematical model of a parcel of contaminated air moving over
complex terrain sprinkled with pollutant sources and under the influ-
ence of complicated meteorology is indeed a highly complex system.
282
-------
This system contains many elements that are often poorly formulated,
Imperfectly calibrated and incorrectly evaluated. Because of the
preciseness demanded by some of the EPA regulations, the complexity
of atmospheric chemistry, the importance of the meteorological driv-
ing mechanisms, and the economic impacts of improper and severe con-
trols predicted by poor modeling procedures, the existing simulation
models must be improved. In this section we list some research needs
that are required to improve the current models.
Background Pollutant Levels and Pollutant Sources
The following items are required for better background pollutant
level and pollutant source assessment:
Improved emission inventories of natural and anthropogenic
sources of primary and secondary particles, nitrates and
organics.
Better emission inventories and assessments of fugitive emis-
sions of particles and gases.
Improved and wider-spread monitoring of NMHC, NOX, O^ and
other oxidants.
Development of in-situ techniques for measuring NH^,
HNC>3, H2S, PAN, and free radicals in the ppb concentra-
tion range.
Use of tuned infrared laser spectroscopy with ranging capa-
bilities to extend the data bases of background pollutants.
Development of better remote sensing capabilities for measur-
ing and determining the causes of background Oj and visibil-
ity impairment.
Meteorological Measurements
More research is required to determine the properties of atmos-
pheric turbulence and wind fields over complex terrain. We need to
283
-------
know when the wind fields in deep valleys are coupled with and when
they are uncoupled from the upper-air winds, on a diurnal and sea-
sonal basis. The thermal structure of the atmosphere over complex
mountain-valley and land-water terrain must be better understood and
predicted. Storm precipitation statistics, i.e. cloud height and
areal extent, frequency of occurrence, storm duration, amount and
intensity of precipitation), solar radiation intensity and atmos-
pheric optics, must be better defined in terms of terrain conditions,
altitude, season of the year and airborne pollutant loading.
Atmospheric Chemistry
The most complex and least understood parts of a RAQM may well
be the kinetic and removal modules. These modules require informa-
tion on the presence of precursor, intermediate and product gases in
the atmosphere, data on the presence of primary and secondary aero-
sols, information on the important and rate-limiting reactions be-
tween the gaseous species, data on the gas-to-particle mechanisms and
other heterogeneous chemical reactions, and information on the inter-
actions between aerosol species and the removal processes operating
on gases and particles. Some items requiring new or further research
are:
Homogeneous and heterogeneous oxidation of S(>2 in pho-
tochemical smog.
The probable route of formation of nitrate and organic aer-
osols in urban photochemical smog and over rural areas.
A careful search for atmospheric precursors of aerosols.
284
-------
» Improved rate constants for reactions involving radicals such
as OH and H(>2
* Improved scavenging rates for gases and particles within com-
plex nonprecipitating and precipitating clouds and storms.
Improved dry deposition and resuspension rates for gases and
particles over various forest types, mountainous terrain,
water surfaces, ice and snow, and deserts.
Determination of removal rates as a function of the diurnal
cycle and season of the year.
Determination of the scattering and absorption properties of
cloud types, aerosol species and gaseous pollutants (notably
N(>2 and nitrous acid), and their resultant effects on
visibility.
Mathematics and Computer Algorithms
Proper use of solution and data management techniques, and com-
puter hardware and software will allow investigators to construct
more accurate, efficient and perceptive models. These techniques and
computer characteristics are concerned with addressing temporal and
spatial scale issues for meteorological and air quality input data
interpolation, resolving temporal and spatial scale issues related to
computational problems and minimization of numerical diffusion, and
treating issues of temporal and spatial averaging of model output.
At present, very few investigators have fully utilized the known or
potential capabilities of mathematics and computer algorithms for
RAQM's. Thus, the first order of business, computationally speaking,
is to make greater use of the techniques and algorithms currently
available. When these are exhausted or when the dynamical and air
285
-------
quality problems become too large and complex for current computa-
tional systems, improvements at all levels of numerical computation
will be required.
Some required improvements are:
The rapid growth in computational system performance will
probably begin to level out, thus the discovery and devel-
opment of better solution algorithms becomes crucial.
Better data archiving systems are required for large com-
prehensive RAQM'S.
Because of the large development costs, it is becoming
important to develop high quality, tested, documented,
standardized and portable software for all elements of a
RAQM.
Because of their expense, large-scale, computer systems may
become a rarity and more and more "distributed intelligence"
systems will be developed. Modelers of RAQM's must use these
distributed systems efficiently and economically.
Chemical and Aerosol Modeling
Some items that would improve the chemical, aerosol and overall
modeling capabilities of RAQM's are:
Use information from ongoing and future field programs to
improve input data modules and model validation procedures.
Determine the level of sophistication beyond which improve-
ments in deterministic models are unwarranted in light of
inadequate input information.
Determine to what extent deterministic models can treat the
inherently stochastic nature of the atmosphere and its con-
stituents .
Modify air quality modeling over complex terrain to account
for local flows (i.e., slope and valley winds) and complex
thermal and boundary layer structures.
Improve and simplify kinetic modules by identifying and
characterizing generic or surrogate rather than specific
reaction sequences.
286
-------
Extend the applicability of kinetic modules to diurnal and
seasonal cycles, cloudy conditions, and poorly exposed moun-
tain valleys.
Improve gas-to-particle, particle-particle, and atmospheric
optics modules to better account for these mechanisms over
rural and urban areas.
287
-------
-------
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
10.
Graedel, T.E. , Chemical Compounds in the Atmosphere,
Press, New York, NY, 440 pp. 1978.
Academic
Friedlander, S.K., "Fundamentals of Gas-to-Particle Conversion
-A Review." Atlantic City, NJ: AICHE Meeting on Aerosol
Science and Technology, August 31, 1976.
Orlanski, I., "A Rational Subdivision of Scales for Atmospheric
Processes." Bull. Arne r. Me t e o r. S o c . , 65: 527-530, 1975.
Lamb, R.G., H. Hogo and L.E. Reid, "A Lagrangian-Monte Carlo
Model of Air Pollutant Transport, Diffusion, and Removal Pro-
cesses." Reno, NV: AMS 4th. Symposium on Turbulence, Diffusion
and Air Pollution, January 15-18, 1979, pp. 381-388.
Lamb, R.G., W.H. Chen and J.H. Seinfeld, "Numerical-Empirical
Analysis of Atmospheric Diffusion Theories." J. Atmos. Sci.,
J32j 1794-1807, 1975.
Lamb, R.G., "A Numerical Simulation of Dispersion from an Eleva-
ted Point Source in the Convection Planetary Boundary Layer."
Atmos. Environ., 12: 1297-1304, 1978.
Donaldson, C. duP., "Construction of a Dynamic Model of the
Production of Atmospheric Turbulence and the Dispersion of
Atmospheric Pollutants." Workshop on Micrometeorology, Boston,
MA: Amer. Meteor. Soc., pp. 292-313, 1973.
Donaldosn, C. duP., Derivation of a Non-Boussinesq Set of Equa-
tions for an Atmospheric Shear Layer. Research Triangle Park,
NC: Environmental Protection AGency, Meteor. Lab., 1973.
EPA-R4073-016h.
Lewellen, W.S. and M.E. Tesks, "Second-Order Closure Modeling of
Diffusion in the Atmospheric Boundary Layer." Boundary-Layer
Meteor., 10-69-90. 1976.
Yanenko, N.N., The Method of Fractional Steps (The Solution oj^
Problems of Mathematical Physics in Several Variables). Trans-
lated by M. Holt, New York: Springer-Verlag, 160 pp., 1971.
11. Patterson, G.S. Jr. and S. Corrsin, "Computer Experiments on
Random Walks with Both Eulerian and Lagrangian Statistics."
Dynamics of Fluid and Plasmas, pp. 275-307, 1966.
289
-------
12. Kraichnan, R.H., "Diffusion by a Random Velocity Field." Phys.
Fluids, 13:22-31, 1970.
13. Riley, J. and S. Corrsin, "Simulation and Computation of Disper-
sion in Turbulent Shear Flow." Proc.of Conf. on Air Pollution
Meteorology, Raleigh, NC., pp. 16-21, April 5-9. 1971.
14. Thompson, R., "Numeric Calculation of Turbulent Diffusion."
Quart. J. Roy. Met. Soc.. 97^93-98, 1971.
15. Justus, C.G. and J.E. Hicks, "Modeling the Anisotropy in Strati-
fied Shear Flow." Proc. of the Symposium on Air Pollution,
Turbulence and Diffusion. Ed. H.W. Church and R.E. Luna,
Albuquerque, NM: Sandia Labs, pp. 202-207, 1972.
16. Liu, H.T. and R. Thompson, Lagrangian Statistics for Isotropic
Turbulence - One Dimension. Boulder, CO: NCAR Manuscript,
National Center for Atmospheric Research (Unpublished), November
1972.
17. Watson, C.W. and S. Barr, Monte Carlo Simulation of the Turbu-
lent Transport of Airborne Contaminantj. Los Alamos, NM: Los
Alamos Sci. Lab., 28 pp., LA-6013, 1976.
18. Harlow, F.H. and J.E. Welch, "Numerical Calculation of Time-
Dependent Viscous Incompressible Flow." Phys. of Fluids,
8^(12).-2182-2189, 1965.
19. Amsden, A.A. , Th«^ Particle-in-Ce 11 Method for Calculation of the
Dynamics of Compressible Fluids. Los Alamos Scientific Labora-
tory, Los Alamos, NM, LA-3566, 1966.
20. Sklarew, R.C., A.J. Fabrick and J.E. Prager, A Pajrticle-in-Cell
Method for Numerical Solution of the Atmospheric Diffusion
Equation, and Applications to Air Pollution Problems. Systems,
Science and Software, LaJolla, CA, SSR-844, Vol. 1, 1971.
21. Sklarew, R.C., A.J. Fabrick and J.E. Prager, "Mathematical
Modeling of Photochemical Smog Using the PICK Method." J. APCA,
2^:865-869, 1972.
22. Teuscher, L.H. and L.E. Hauser, Development of Modeling Tech-
nique for Photochemical Air Pollution. Research Triangle Park,
NC: Environmental Protection Agency, EPA-650/4-74-003, 1974.
23. Fabrick, A.J., P.I. Nakayame and E.J. Fredericksen, A Methodol-
ogy for Treating Large Localized Emissions of Reactive Pollu-
tants. Research Triangle Park, NC: Environmental Protection
Agency, EPA-650/4-74-006, 1974.
290
-------
24. Lange, R., ADPIC - A Three-Dimensional Computer Code forforthe
Study of Pollutant Dispersal and Deposition Under Complex Condi-
tions. Livermore, CA: Lawrence Livermore Laboratory, Report
UCRL-51462, 1973.
25. Lange, R., "ADPIC - A Three-Dimensional Particle-in-Cell Model
for the Dispersal of Atmospheric Pollutants and Its Comparison
to Regional Tracer Studies." J. Appl. Meteor^, 17;320-329,
1978.
26. Kao, S.K., C.N. Chi and W.M. Washington, "Statistical Character-
istics of Three-Dimensional Particle Movement in the NCAR
General Circulation Model." J. Atmos. Sci., 33:1042-1049. 1976.
27. Sheih, C.M., "A Puff-Grid Model for Predicting Pollutant Trans-
port Over an Urban Area." J. APCA, ^7_:784-785, 1977.
28. Sheih, C.M., "A Puff-Grid Model for Predicting Pollutant
Transport and Diffusion." J. Appl. Meteor. , _T7_: 140-147, 1978.
29. Sheih, C.M., "A Puff Pollutant Dispersion Model with Wind Shear
and Dynamic Plume Rise." Atmos. Environ., 12: 1933-1938, 1978.
30. Hall, C.D., "The Simulation of Particle Motion in the Atmosphere
by a Numerical Random-Walk Model." Quart. J.Roy. Meteor. Soc.,
^£1^:235-244, 1975.
31. Reid, J.D. , "Markov Chain Simulations of Vertical Dispersion in
the Neutral Surface Layer for Surface and Elevated Releases."
Boundary-Layer Meteor., J_6^:3-22, 1979.
32. Anbar, D., "A Diffusion Model for Use with Directional
Samplers." Atmos. Environ., 12:2131-2138, 1978.
33. Liu, M.K., A Simple Reactive Plume Model. San Rafael, CA:
Systems Applications, Inc., EF 75-12, 1975.
34. Liu, M.K., D.A. Stewart and P.M. Roth. "An Improved Version of
the Reactive Plume Model (RPM-II)." Toronto, Canada: Presented
at NATO/CCMS Air Pollution Pilot Study: Assessment, Methodology
and Modeling. August 28-31, 1978.
35. Hales, J.M., "A User-Oriented Computer Code for Solution of the
Genral Equations of Continuity for Trace Substances in the
Atmosphere." Richland, WA: Pacific Northwest Laboratory Annual
Report for 1974 to the USAEC Division of Biomedical and Environ-
mental Research, Part 3, Atmospheric Sciences, BNWL-1950,
PT3:77-78, 1975.
291
-------
47. Martinez, J.R. , R.A. Nordsieck and M.A. Hirschberg, User's Guide
of Diffusion/Kinetics (DIFKIN) Code. Research Triangle Park,
HC: Environmental Protection Agency, EPA-R4-73-012, Vol. b,
1973.
48. Sheih, C.M., "Application of a Statistical Trajectory Model
to the Simulation of Sulfur Pollution Over Northeastern United
States." Atmos. Environ. , JJ.: 173-178, 1977.
49. Heffter, J.L., A.D. Taylor and G.J. Ferber, A Regional-
50.
51.
Continental Scale Transport, Diffusion and Deposition Model.
Silver Spring, MD: NOAA, Air Resources Laboratories, Tech.
Memo ERL-ARL-50, 1975.
Meyers, R.E. , et al., "Modeling Sulfur Oxide Concentrations in
the Eastern United States: Model Sensibility, Verification and
Applications." Reno, NV: AMS 4th. Symposium on Turbulence,
Diffusion and Air Pollution, pp. 673-676, January 15-18, 1979.
Henmi, T. and E.R. Reiter, Long-Range Transport and Trans forma-
tion of S02 and Sulfate. Research Triangle Park, NC: Environ-
(CSU
mental Protect ion Agency, (CSU Envir.
Contract No. R805271-01, 1978.
Res. Paper No. 15) EPA
52. Powell, D.C., D.J. McNaughton, L.L. Wendell and R.L. Drake, _A
Variable Trajectory Model for Regional Assessments of Air Pol-
lution from Sulfur Compounds. Richland, WA: Pacific Northwest
Laboratory, PNL-2734, 1979.
53. Horst, T.W. , "A Review of Gaussian Diffusion-Deposition Models."
Gatlinburg, TN: Symposium on Environmental and Health Effects
of Atmospheric Sulfur Deposition. October 14-18, 1979.
54. Huang, C.H. and R.L. Drake, "A Direct Method of Adjusting
Windfield over Complex Terrain." Preprint Volume: Fourteenth
Conference on Agriculture and Forest Meteorology arid Fourth
Conference on Biometeorology, pp. 102-104, April 1979.
55. Davis, W.E., "Comparison of the Results of an Eight Layer
Regional Model Versus a Single Layer Regional Model for a
Short Term Assessment." Presented at the World Meteorological
Organization SyroppsJ-um on Long-Range Transport^ of Pollutants,
Sophia, Bulgaria, October 1-5, 1979.
292
-------
36. Drake, R.L., "A Comprehensive Aerosol Growth Model." Atmo-
spheric Aerosols: Their Optical Properties and Effects,
Williamsburg, VA: Optical Society of America, December 13-15,
1976.
37. Gillani, N.V. , "Project MISTT: Mesoscale Plume Modeling of the
Dispersion, Transformation and Ground Removal of S02-" Atmos.
Environ., 12:569-588, 1978.
38. Tangermann, G., "Numerical Simulations of Air Pollutant Dis-
persion in a Stratified Planetary Boundary Layer." Atmos.
Environ., 12:1365-1369, 1978.
39. Egan, B.A. and J.R. Mahoney, "Numerical Modeling of Advection
and Diffusion of Urban Area Source Pollutants." J. Appl.
Meteor., 11:312-322, 1972.
40. Rao, K.S., I. Thomson and B.A. Egan, "Regional Transport Model
of Atmospheric Sulfates." Portland, OR: Presented at the 69th.
Annual Meeting of the APCA, Paper No. 76-34.3, June 27 - July 1,
1976.
41. Martinez, J.R., et al. Development of an Atmospheric Model for
Sulfate Format^oru Washington, DC: National Science Founda-
tion,, Doc. No. P-1534, 1977.
42. Boris, J.P. and D.L. Book, "Flux-Corrected Transport.
I. SHASTA, A Fluid Transport Algorithm that Works." J. Comp.
Phys. , _U:38-69, 1973. J. Comp. Phys., 11:38-69, 1973.
43. Runca, E. and F. Sardei, "Numerical Treatment of Time Dependent
Advection and Diffusion of Air Pollutants." Atmos. Environ.,
9^:69-80, 1975.
44. Leahey, D.M., "An Advective Model for Predicting Air Pollution
Within an Urban Heat Island with Applications to New York City."
J. Air Poll. Contr. Assoc., 22(7):548-550, 1972.
45. Leahey, D.M., "An Application of a Simple Advective Pollution
Model to the City of Edmonton." Atmos. Environ. , 9^:871-823,
1975.
46. Drivas, P.J., M. Chan and L.G. Wayne, "Validation of an Improved
Photochemical Air Quality Simulation Model." Salt Lake City,
UT: AMS joint Conference on Application of Air Pollution Mete-
pro logy, pp. 255-260, November 29 - December 2, 1977.
293
-------
-------
MODELING LONG RANGE TRANSPORT AND DIFFUSION
Arthur Bass
Environmental Research & Technology, Inc.
Concord, Massachusetts 01742
1.0 INTRODUCTION
The growing concerns over the long range transport and trans-
formation of sulfur oxides and other industrial effluents, and their
subsequent consequences for human welfare and for the environment,
are truly international. Studies to characterize and to model the
regional and long range atmospheric transport, transformation, and
removal of S02 have grown to become inter-regional and trans-
frontier in scopeas evidenced by, for example, the OECD program on
the long Range Transport of Air Pollutants (LRTAP). Closer to home,
the recent establishment of a joint U.S.-Canada Research Consultation
Group on the LRTAP problem, and the ongoing major initiatives by
DOE/EPA in the Multistate Atmospheric Power Production Pollution
Study (MAP3S) and by EPRI in the Sultate Regional Experiment (SURE)
to characterize and model the regional formation and dispersion of
sulfates (two of the many important observation/modeling investiga-
tions of sulfur oxides transport at long ranges) clearly underscore
the critical requirement for accurate means to predict and to verify
the atmospheric dispersion of 862 over large spatial and temporal
scales.
The title of this review is something of a misnomer, for it is
hardly necessary to review for this audience the evolution of long
295
-------
range transport modeling. Indeed, several excellent, timely reviews
abound: notable, for example, are the recent papers by Pack et al.
(1978), Fisher (1978), and the earlier work of Eliassen and Saltbones
(1975), Fisher (1975), Scriven and Fisher (1975), among others, cap-
ture well the seminal results and the key conclusions obtained by the
community of long range transport modelers, at least to the near
present. It would be fatuous to attempt to distill here, in a few
pages, the wealth of detail to be found in the recent conference
literature on long range transport modeling studies: most notably,
the Dubrovnik Symposium on Sulfur in the Atmosphere (Husar et al.
(1978)), the Sofia WHO Symposium on the Long Range Transport of Pol-
lutants (WHO (1979)), and numerous workshops and specialty confer-
ences, as well as the sessions devoted to papers on this subject at
the several AMS Symposia on Atmospheric Turbulence, Diffusion and Air
Pollution (Santa Barabara 1974; Raleigh 1976; Reno 1979); and the
AMS/APCA Joint Conferences on Applications of Air Pollution Meteoro-
logy (Salt Lake City 1977; New Orleans 1980).
The perspective taken in this review of long range transport
models is much more limited in scope - for the most part, oriented to
issues of concern for practical model applications to long range
transport problems of pressing regulatory concern in the United
States - that is, the transport of sulfur oxides and nitrogen oxides
emitted principally from major point sources toward distant pristine
areas (the so-called "Class I" Clean Air regions of the country); as
296
-------
well as the associated problem of acid precipitation impacts on
regional and longer ranges, and that of preserving visual quality in
Class I regions.
We do not, therefore, emphasize the several research-grade mod-
els in active development; but rather, we stress only currently oper-
ational, easily-obtained long range transport models. Specifically:
models developed by U.S. research groups, that are already
(or are soon to be) in the public domain;
models that appear suitable to at least some present or anti-
cipatable near-term regulatory applications;
models that are practical, cost-effective, computationally
manageable and highly user-oriented; or models that could be
so made with minimal efforts.
This review tries to answer (if only in a preliminary way) the fol-
lowing questions, in keeping with the theme of this Workshop - mod-
eling in support of air quality regulations:
Is the j>r;_aepical, routinely achievable state-of-the-art in
long range transport modeling reflected in present regulatory
practices or model guideline requirements?
Are there impediments, in the present understanding of long
range transport and dispersion, that militate against devel-
opment of better verified, better justified long range
transport models for regulatory use? What can be done, in
the near term, to bring about useful improvements in long
range transport modeling tools for practical regulatory ap-
plications?
What prospects do the frontier areas of active research and
development in long range transport modeling offer for
improving predictive capabilities?
With this perspective, therefore, our review is "biased" towards
models for long range transport and dispersion of weakly reactive
297
-------
pollutantsin practice, S02 - because the present state-of-the-art
of photochemical dispersion modeling does not extend much beyond
local or subregional scales (characteristic transport ranges of, say,
50 km or less). For purposes of practical applications, rather than
for basic research or new modeling technique development, we will be
concerned with models that include highly parameterized (essentially
linear) plume chemistry. (Although important progress has been
achieved in developing and verifying multispecies nonlinear plume
chemistry mechanisms, these are not yet practicable for embedding
within a mesoscale or long range transport model for routine regula-
tory use.)
Another major "bias" throughout this review is its focus on
point-source-specific rather than area-source modelsbecause by and
large most regulatory discusions (and certainly the planning/site
evaluation studies chat precede a new source permit application) are
concerned with incremental ambient pollution burdens that might be
imposed by one or at most a small number of new sources. This is not
true, of course, for regional-scale or national-scale planning
studies (of the kind, say, that might be undertaken to support
national energy policy on coal conversion, on interregional shipment
of low sulfur coal, and the like); but the technical resources and
modeling tools needed for studies of this scope are for reasons of
economics and staffing limitations only available to the very largest
research groupsthe National Laboratories, EPA, EPRI, universities,
298
-------
and a few consulting organizations; the resources available in prac-
tice Co a new source applicant, or a beleaguered local or state regu-
latory group, will usually be considerably less powerful and less
sophisticatedand it is precisely these prospective users who have
the most need for practical, credible, and regulatorily acceptable
long range transport models.
A third major "bias", one that most underlies our later recom-
mendations for model use, is toward models that can in principle be
used to address short-term average ground-level concentrations (24-
hour averages, and perhaps, 3-hour averages) at regional and long
transport ranges. We emphasize here the short-term average models
because the short-term PSD Class I 502 concentration increments to
be protected under the 1977 Clean Air Act Amendments are usually more
constraining to a new source applicant than are the annual average
S(>2 increments. In view of the present and possible future regula-
tory needs for PSD and SIP review, we emphasize, and make recommenda-
tions concerning, models that are economical for evaluating the long
range impacts of one or at most several sources (on the order of ten,
say), rather than large grid-type multisource models best suited for
regional type inventories.
We will show a consistent bias toward models that are easy to
use and easy to explain to the nonspecialist:
models that are suitable for a broad range of meteorological
dispersion conditions - both worst-case (episodic) and aver-
age dispersion;
299
-------
* models that require only readily obtainable meteorogolical
and other input data;
models that are flexible, modular, readily transported and
modified for use on computer systems of conventional size;
models that are well documented for use outside the origi-
nating organization;
models designed from the beginning for efficient, economical
simulations, if possible, both in the near field of indivi-
dual sources, and in the far field where combined effects of
multiple sources may dominate.
Finally, we have tried to emphasize models that recognize expli-
citly the physical processes that govern turbulent plume transport on
regional and synoptic scalesbut not models that, in our subjective
opinion, overemphasize secondary parts of the modeling problem, while
ignoring or giving short shrift to the primary, crucial issue: real-
istic description of mesoscale and long range turbulent transport in
a spatially and temporally varying mixed layer driven by synoptic-
scale weather systems.
For purposes of discussion, we will sometimes use the term
"long-range" loosely, to mean transport distances beyond about 50 km,
notwithstanding the important physical distinctions between transport
on the (3-mesoscale range (on the order of 25-250 km) and transport
in the o-mesoscale range (on the order of 250-2500 km)see Gillani
(1978) for further discussion. Both pose important problems for
regulatory desicion making: the problem of significant near-field
ground-level concentrations of S02 falls within the short range end
of the p-mesoscale; whereas the problem of ambient sulfate impacts,
300
-------
including issues of acid precipitation and regional visibility degra-
dation, extends well into the o-mesoscale.
It is important to bear in mind, therefore, the current regula-
tory position on long range transport - as stated in the EPA Guide-
line on Air Quality Models (EPA 1978), estimation of source impacts
is highly uncertain beyond 50 km: the choices of dispersion coef-
ficients are "tenuous", and diurnal variations in meteorology over
such transport times "are more likely to alter plume trajectories and
dispersion characteristics." The Guideline asserts that no widely
applicable model is available for dealing with long range transport
and removal - and that source impacts at large transport distances
"should be considered on a case-by-case basis with available tech-
niques." This paper attempts to focus attention, therefore, on the
full range of techniques available now for practical use.
As basis for the suggestions we will subsequently make, we
illustrate several generic approaches to Lagrangian modeling on
regional and longer scales, describe some of the many possible
algorithm choices adopted by different research groups, briefly sur-
vey the available model sensitivity and model verification studies
for selected models, review the requirements for Lagrangian long
range transport models in light of current and possibly near term
ambient air quality and air quality-related regulations, and then
list some available long range transport models (principally
Lagrangian models but also some numerical grid models) that are
301
-------
currently operational and already (or soon to be) in the public
domain, so far as we know.
Finally we identify some of the current directions of active
research and development in long range transport modeling, and sug-
gest some of the future directions in which such modeling efforts
might be directed for achieving practical model development objec-
tives.
2.0 CURRENT REGULATORY NEEDS FOR LONG RANGE TRANSPORT MODELS
At present, reliable, practical long range transport modeling
tools are urgently required, both for regulatory applications and for
other planning needs for example:
for development and revision of SIP-related regional control
strategies;
» for New Source Review programs to attain and maintain
National Ambient Air Quality Standards (NAAQS), and for the
Prevention of Significant Deterioration (PSD) in Clean Air
regions; and
for a wide range of policy and planning studies to assess the
large-scale, long-range environmental, economic, and social
implications of new coal-based energy resource development,
fuel switching and coal conversion orders, delays and/or
cancellation of nuclear hydroelectric power generation capa-
city, and like issues.
The modeling techniques specifically referenced in the
GuidelineHeffter and Ferber et al. (1975); Rao et al. (1976);
Scriven and Fisher (1975); and Hales et al. (1976)have been sup-
planted by more powerful and more useful models, better able to
address directly the specific spatial and temporal resolution and
302
-------
averaging requirements imposed by the current ambient air quality
regulations.
To meet these ambient air quality source review requirements,
the more urgent need now is for credible, well-verified short-term
average (episode-type) models; long-terra average models are of lesser
priority at present. However, where the regulatory decision making
process must deal with secondary pollutant effects such as long-term
average acidification and biological impacts, the need for climato-
logical models on regional and longer spacial ranges will be much
greater. As driven by present regulatory practice, the near-term
requirement is for models of spatial resolution and domains of
validity appropriate to the (3-mesoscale (for a typical p-mesoscale
impact problemthe maximum ground-level concentration of SC>2 from
an elevated point source at a range on the order of 100 km); other
models, or perhaps the same models with resolution more suitable to
the a-mesoscale range, would be best for treating, say, the problem
of modeling maximum ground-level concentrations of sulfate formed by
conversion of S(>2 from elevated plumes ducted aloft for long dis-
tances.
As a rule, the computational approach and resolution of a model
can be economically matched only to problems of a certain spatial/
temporal range. For example, the "square puff" trajectory model
approach described later is extremely efficient for long-term average
dispersion over large distances; the same model, however, will
303
-------
underpredict drastically in the near field of a source. As another
example, a long-range transport model that requires immediate uniform
vertical mixing of an effluent plume, from the outset, may be
entirely suitable if the concern is only about impacts over travel
times comparable to the diurnal cycle of the mixed layer, but would
entirely misrepresent the ground-level impacts in the near field of
an elevated point source under stable nighttime flow.
To meet the short-term average requirements, the needed temporal
resolution of long-range models for regulatory application is on the
order of one to a few hours. Yet, because it is necessary to predict
maximum 3-hour average or maximum 24-hour average ground-level con-
centrations not to be exceeded more than once per year, models that
can economically simulate one year or more of 3-hourly average con-
centrations are also needed now.
Among other regulatory applications of long-range transport mod-
els, it is possible that modeling of long-range visibility impairment
will play an important future role in PSD source review, especially
in the western United States. Visibility degradation is both a
cc-mesoscale and an p-mesoscale problem.
Plume blight and plume discoloration are related chiefly to
ambient concentrations of suspended particulates and N(>2 within,
say, 100 km of a typical new major coal-burning source. To model
visibility impacts from individual sources economically on these
transport scales, Lagrangian transport models coupled to radiative
304
-------
transfer algorithms will be needed; several such models are already
available. On longer transport scales, say at a distance of several
hundred kilometers, the visibility impairment problem is one of
regional haze; and successful, credible simulations of multisource
impacts on regional visibility at such scales are not likely to be
achieved very soon.
We may also anticipate an evolving requirement for the use of
long-range transport models to address regulatory needs for modeling
acid precipitation impacts. In contrast to visibility impairment
problems, which are related to instantaneous ambient concentrations
of species causing light extinction, acid precipitation problems
include both episodic and long-term average concentrations.
Greater emphasis can be expected on policy/planning and regula-
tory applications of long-range transport models for inter-regional
transport and diffusion across state boundaries, as for example, the
growing controversy over SC>2 emission levels in the Ohio River
Valley, or the increasing concern over transnational boundary fluxes
of S02 and sulfates between the United States and Canada.
Sometime in the future, large numerical grid models for long-
range transport will play a major (perhaps dominant?) role in prac-
ticable regulatory decision-making; but at present, and probably in
the near term, for most applications Lagrangian long-range transport
models will continue to be more suitablefor reasons of near-field
resolution, data requirements, computer resources, costs, and ease of
305
-------
useat least for single sources, or for small ensembles of sources
for which the plume chemistry is sensibly linear.
3.0 LAGRANGIAN MODELS FOR LONG RANGE TRANSPORT
A Lagrangian variable trajectory plume model represents a con-
tinuous plume emitted by a point source by the transport and disper-
sion of a succession of discrete plume elements (air parcels or
massless trajectory points.) These plume elements or air parcels are
independently advected and diffused by a spatially and temporally
varying wind field. Each plume element carries an independent time
historyplume chemistry, dry deposition and scavenging. The time-
average ground-level impact of the (continuous) plume at a given
point is simulated by combining the contributions from all elements
that independently traverse that point during the specified averaging
time. Lagrangian models thus offer the opportunity to resolve the
near-field small-scale impacts of individual point sources with re-
solution that grid models cannot achieve, at least not economically.
(For this reason, some grid models (e.g., Liu and Wojcik 1979)
include subgrid scale plume or puff-type modules to carry the early
evolution of a plume element to the time when the element has grown
sufficiently large to be "handed over" to a grid cell, thus prevent-
ing unacceptably large initial dilution.)
Lagrangian models fall into two broad classes. The nonreactive
models employ simple linear or psuedo-first-order chemistry; handle
multiple sources by superposition; and usually have a detailed treat-
ment of horizontal plume transport and diffusion but generally little
306
-------
vertical resolution. Nonreactive Lagrangian models usually attempt
to recreate in detail the mesoscale meteorological variations that
dominate plume dispersion on these transport scales. These models
tend to be relatively inexpensive per source simulated.
Reactive Lagrangian models, by contrast, include complex multi-
species chemical mechanisms; use a single "reactor" volume advected
by the local wind; and, typically, have only very limited horizontal
coverage: essentially two-dimensional along the trajectory. The
vertical resolution is much more detailed, because the plume chemis-
try will typically vary strongly with altitude. Reactive Lagrangian
models use, as a rule, only simple treatment of plume diffusion.
Nonreactive Lagrangian models (the only ones to be discussed
here) can be further partitioned into short-term average and long-
term average models. Short-term average Lagrangian models attempt to
resolve individual plume trajectories using sequential meteorology.
Because the short-term models include so much detail about the
transport, diffusion, and transformation of many individual plume
elements, some may be computationally limited to handling a small
number of individual sources.
Some long-term average Lagrangian models include only a simple
treatment of horizontal diffusion or may ignore horizontal diffusion
altogether, as does the PNL regional model (Wendell et al. 1976).
Other long-term average Lagrangian models have described both hori-
zontal and vertical dispersion processes, each in great detail, but
307
-------
as completely decoupled problems, for example, Bolin and Persson1s
(1975) model or the ASTRAP model (Shannon 1979a). Still other long-
term average models may employ simple assumptions on the vertical
distribution of plume materialfor example, the uniform mixing
assumption of the EURMAP-1 model (Johnson et al. 1978). These mod-
els are designed, foremost, to be computationally economical for one
or more sources over Long periods of time; they cannot therefore also
provide the near-field detailed spatial resolution that may be needed
for regulatory evaluations.
Most Lagrangian plume models begin from the premise that long-
range transport and diffusion (strictly, for transport over times
longer than a few averaging times) is dominated by plume meander due
to mesoscale variations in the wind fieldthat is, by large eddies
of characteristic scale greater than the plume dimensionwhat Powell
(see Nappo 1978) has called "mesoscale turbulence"--rather than by
small scale turbulent eddies smaller than the plume size. Figure 1
illustrates schematically three different regional-scale Lagrangian
trajectory modeling approaches. (Other lagrangian models, such as
Sheih's (1977) statistical trajectory approach or his six-particle
puff model with wind shear (Sheih 1978a), are less readily
visualized.)
The puff superposition model represents a continuous plume as a
series of discrete puff elements of circular horizontal cross sec-
tion; the vertical mass concentration field may be treated as Gaus-
sian, as uniformly mixed, or as prescribed by a finite-difference
308
-------
representation of the vertical diffusion equation (see for example
Draxler 1979). The puff release rate and puff advection times must
be chosen in a consistent manner, so that consecutive puffs will
overlap sufficiently to ensure a reasonable representation to the
continuous plume. Ludwig (see Nappo 1978) provides guidance for the
appropriate puff spacing to achieve coverage comparable to the con-
tinuous plume.
The segmented plume approach shown in Figure 1 differs from the
conventional Gaussian plume model in that the segmented plume may be
continuously deformed by a temporally varying horizontal wind field.
The plume is treated as divided into contiguous segments. Each seg-
ment describes a portion of plume behavior between successive time
periods, and the end points of each segment are advected in a
Lagrangian sense.
Under conditions of uniform, homogenous flow, the segmented
plume and puff superposition models will give about the same results,
and the plume model is typically about 50% cheaper to run. Why then
are puff models favored by many workers? This issue was strongly de-
bated at the 1977 Trajectory Modeling Workshop (Nappo 1978)debated,
but not entirely resolved.
The preference for the puff model derives not so much from the-
oretical considerations as from the practical advantages of the puff
model: as expressed by Sheih (see Nappo 1978) puff models are gener-
ally more credible and less subjective to pathological behavior,
309
-------
(0
o
W
Ui
X
o
<
o
CC
CL
Q.
<
,-ts
iuz
CC Ij
=> LU
O O
CC
O
LU
5
CC
310
-------
especially under conditions of recirculating or strongly sheared wind
flows, conditions for which a plume model may offer poor coverages at
the interfaces where plume segments are "attached." And, especially
under stagnating flow conditions, the plume model has an explicit
inverse-wind speed singularity; when the winds approach calm, the
model can violate the basic assumption of the plume approachthat
mean advection is greater than turbulent diffusion in the along-wind
direction.
Although the conceptual basis for puff superposition and plume
segment models was not fully resolved at the Workshopand remains
unresolved at presentthe rejoinder by proponents of these modeling
approaches is that (a) they make better sense physically under irre-
gular flow situations than the conventional straight-line Gaussian
plume model they are meant to replace, and that (b) in practice they
work better for regional scale and long-range flows.
A central issue that must be confronted in using these variable
trajectory models is the problem of how to partition the turbulent
spreading into (a) large scale meandering of the plume centerline and
(b) small scale turbulent spreading relative to the centerline, that
is, how to partition the total crosswind spread. In practice, most
modeling groups make a somewhat arbitrary choice of averaging time to
describe the mean wind field (for example, one hour average winds).
This mean wind field is used to advect the centroid of the puff or
plume element. Lateral dispersion relative to the instantaneous
311
-------
position of the centroid is then prescribed as a function of travel
time or downwind distance. For example, Start and Wendell (1974) use
Yansky's sigma-curves; Heffter et al. (1975) permit the horizontal
puff radius to grow linearly with time; Johnson et al. (1978) in the
EURMAP-2 model, allow u^ to depend on the local deformation field;
and finally, the ERT group, to make their puff model more consistent
with routine regulatory usage, assume that u^ evolves as described
by the conventional PGT curves to a distance of 100 kilometers, and
thereafter assume (with Heffter) that puffs grow as simple linear
functions of time.
It is important to recognize that there is no rigorous theoreti-
cal justification for the artificial decomposition of turbulent dis-
persion into mean and small scale relative diffusion; and certainly
not for the use of PGT or other conventional turbulence typing
schemes based on the continuous straight line plume. But again, as
expressed by several participants in the Trajectory Modeling Work-
shop, these puff and segmented plume schemes, however artificial
their decomposition of mean and relative diffusion, are useful.
The third kind of Lagrangian trajectory model shown in Figure 1
does not appear to have a standard nameso, for want of one more de-
scriptive, we have called it the "square puff" model or the "massless
point in box" approach. This schemea computationally efficient,
long-term model developed by Powell and co-workers at PNLis con-
ceptually a descendent of the Start-Wendell MESODIF puff model (Start
312
-------
and Wendell 1974) except that only the puff trajectory remains; the
horizontal diffusion of puffs is always fixed at the size of one grid
cell. Powell et al. (1979) argue that for long-term averages at
large transport distances, horizontal diffusion can be ignored; for
if the trajectories are not systematically biased, then on the long-
term average adjacent cells will see comparable diffusive contribu-
tions, and the net effect on a regional average basis will be about
the same as if actual lateral dispersion were included. This is
indeed an attractive, efficient approach, if the concern is not with
near-field impacts (for which the model must necessarily underpredict
the maximum ground-level concentrations, and overpredict virtually
everywhere else), nor where the emphasis is on short-term predictions
(where the lateral diffusive contributions of individual puffs are
all-important).
Finally, Sheih (1977a) and Shannon (1979a) use a large ensemble
of individual trajectory calculations to develop empirical equivalent
puff dispersion statistics for their long-term long-range statistical
puff model.
Just as many approaches have been explored for characterization
of the horizontal transport and diffusion of plumes over long trans-
port ranges, so too have numerous schemes been suggested to charac-
terize vertical diffusion. Several Lagrangian models describe verti-
cal diffusion by a Gaussian profile - where the vertical plume spread
statistic,
-------
GZ curves, or else is specified as a function of vertical eddy
diffusivity Kz and of travel time. Lagrangian models that specify
az in a manner similar to the PGT curves (which corresponds most
closely to current regulatory useage of straight line model at
shorter ranges) include for example: STRAM (Hales et al. 1977),
EURMAP-2 (Mancuso et al. 1979), and MESOPUFF (Bass et al. 1979),
among others. By comparison, the ARL-ATAD model (Heffter 1980)
specifies crz by a Fickian diffusion law. Many of these models also
permit, as an option (if not as a fixed requirement) that the ver-
tical Gaussian profile be replaced by uniform vertical mixing, either
from the outset or when the puff/plume element has grown to appre-
ciable vertical extent. Finally, several more recent Lagrangian puff
type models use K-theoretic treatments of vertical diffusion by
numerically differencing and solving the vertical diffusion equation
as a multilevel problem including sources and sinks. Example models
include the NOAA/ARL Mesoscale Trajectory and Diffusion model
(Draxler 1979) and the BNL AIRSOX model (Meyers et al. 1979).
Shannon (1979b) has also used a Gaussian moment-conservation
method for vertical diffusion in the ASTRAP modelthis is a general-
ized Gaussian puff scheme, consisting of Lagrangian advection (coun-
tergradiently) by a fictitious diffusion velocity, followed by
Eulerian decomposition and mass and moment conserving interpolation
back to the Eulerian grid.
314
-------
4.0 NUMERICAL GRID MODELS FOR LONG RANGE TRANSPORT
A detailed discussion of grid modeling approaches to regional
scale transport is well beyond the scope of this review, but detailed
descriptions of these modeling approaches and their verification
histories are amply documented elsewhere.
We should note the ongoing development of the EPRI Sulfate
Regional Experiment and its associated model refinement/verification
activities (the SURAD model). Other important grid model development
initiatives under the MAP3S program are being actively pursued by the
BNL group (P. Michaels, personal communication) - notably, SCHEMATIC
(MAP3S Modeling Abstracts, in preparation). A regional scale grid
model using the pseudospectral method of Prahm has been tested
against the SURE intensive data base (Niemann et al. 1979) with some
success.
It should also be noted that the EPA Meteorology Laboratory has
undertaken an ambitious multiyear research and development program to
develop a "regional super model" including reactive chemistry, spa-
tially and temporally varying mesoscale meteorology, and a rigorous
treatment of turbulent dispersion on local and regional scales (K.
Demerjian, personal communication).
But, for practical applications to regulatory problems, in the
near term at least, we would judge that only the Northern Great
Plains Regional Model {Liu and Durran 1977) and the MESOGRID model
(derived from the EPRI/SURE regional model SULFA3D - Egan et al.
1976, Morris et al. 1979) are currently available.
315
-------
5.0 METEOROLOGICAL DATA REQUIREMENTS FOR LONG RANGE TRANSPORT
MODELING
5.1 Wind Field Synthesis
The experience of many groups suggests that the predictive
success of long range transport models is affected most by the repre-
sentativeness and adequacy of the available mesoscale-synoptic
meteorological data base. Judging by the many published model
sensitivity analyses, and the as yet fragmentary but rapidly improv-
ing efforts at verification of long range transport models, it
appears that the principal present limitation to greater accuracy of
these models is insufficient spatial and temporal resolution of the
wind fields.
The effective spatial resolution of the upper level wind field
observations is set by the characteristic distance between rawinsonde
stations. The spatial resolution of the upper air sounding program
in the U.S. is of order 300-500 kilometers, marginally adequate at
best even under ideal conditions of uniform topography and reasonably
homogenous flows. The temporal resolution of the upper air sounding
program is a yet more serious limitationsoundings are only made at
12 hour intervals, and important diurnal variations in wind field,
mixing height, and turbulent structure occur on much smaller time
scales.
One possible approach used to transcend the limited spatial
resolution of the upper level mesoscale wind data is that of Heffter
and co-workers (Heffter et al. 1975, Draxler 1977, Draxler 1979); in
316
-------
their approach, the twice-daily upper air data is supplemented by the
much denser set of hourly surface station data, interpolated upwards
to represent the layer-averaged wind fields. This approach is not
likely, of course, to be successful in regions of strong topographic
influences or under severe convective conditions that may dominate
mesoscale flow fieldsfor example, in the southwestern United
States.
As another possibility, a mesoscale Numerical Weather Prediction
(NWP) model of demonstrated prognostic skill in the geographic region
of interest (over a period of, say, 12 to 18 hours) could be used to
drive a trajectory model. This is the approach of the Drexel-NCR-BNL
Limited Area Mesoscale Prediction System (LAMPS)a system designed
to simulate the evolution of mesoscale weather systems for periods of
24 to 36 hours, with grid resolution of 35 to 140 kilometers (MAP3S
Modeling Abstracts, in preparation). (Preliminary forms of this
modeling system have been exercised at BNL and additional work is
planned to couple the mesoscale prediction system to a trajectory
model.) Indeed, looking to the future, such combined mesoscale
weather prediction-plume transport models may offer a practical way
in which to supplement sparse meteorological data on mesoscales. For
long range transport modeling applications in the near future, how-
ever, one is constrained to "make due" with the available observa-
tional network.
317
-------
We have stressed the controlling influence of mesoscale meteoro-
logical flow conditions over long range plume transport and disper-
sion; and a brief, graphic example is perhaps appropriate here. This
example is taken from a recent simulation study of regional scale air
pollution associated with coal-based energy resource development in
the Four Corners region of the southwestern United States (Bass et
al. 1979). One of the principal study conclusions was that plume
dispersion in the Four Corners region appears to be dominated by
large scale variations in mesoscale wind fields, and by the spatial
and temporal variations in mixing heights and stability classbut
not, as a rule, by small scale dispersion. (This point has been made
repeatedly, in the past, by the SRI, NOAA/ARL, and Battelle PNL
modeling groups, as well as in the earlier work of the OECD LRTAP
investigators, among others.)
Figure 2 and 3 illustrates the surface weather map and upper air
flow fields for a typical winter "best-day" regional dispersion situ-
ation in the southwestern United States. The 700 mb winds were
generated from the regional rawinsonde network data by the MESOPAC
mesoscale meteorological preprocessor (Benkley and Bass 1979c). The
westerly upper level flows are rapid and zonal (winds up to 25 m/s);
the surface (gradient) winds are also westerly. Computed mixing
depths (Benkley and Schulman 1979) were about 1,000 meters, caused by
strong mechanical forcing of the surface boundary layer with little
diurnal variation.
318
-------
CM
i-
cc
Q
3
W
319
-------
12/341/77
0/342/77
12/341/77
FIGURES
UPPER AIR METEROLOGY
320
-------
The 24-hour average ground level concentrations of S02 and
corresponding to this best-day winter dispersion case in the
Four Corners area are shown in Figure 4. The S02 plumes, clearly
disjoint, are advected strongly toward the eastern grid boundary,
with rapid downwind decrease of concentration. The regional-scale
ground-level 864 concentrations are negligiblethe maximum sulfate
concentrations, on the order of only 0.1-0.5 (j.g/iiH appear only in
the downwind portions of individual plumes.
This best-case situation may be contrasted with the next
example'worst-case1 dispersion. Figures 5 and 6 show the surface
and upper level meteorological flows for a typical "worst-case"
regional dispersion situation in the Four Corners areaa stagnant
summer day characterized by weak surface pressure patterns and a
surface heat flow over the southwestern portion of the region. The
upper level flows at 700 and 500 millibars are very weak and disor-
ganized (wind field less than 2 m/s), with no generally distinguish-
able transport direction. The computed mixing depths are low during
the night (on the order of 100-400 meters), and high during the day
(on the order of 2,000-3,000 meters). Plume elements more than one
day old are well mixed through a deep layer. Evidently, deep mixing
does not preclude large sulfate accumulations on regional scales.
The 24-hour average ground-level concentrations of S02 and
SO^ for this worst-case summer day are shown in Figure 7. They
differ strikingly from the previous example: the S02 field shows
321
-------
S04
DECEMBER 6, 1977
SO2
SO4
DECEMBER 7. 1977
DAILY AVERAGE GROUND LEVEL CONCENTRATIONS (M9/.m3l
1977 ERD INVENTORY (SOURCE: EPA)
MESOPUFF/MESOPAC MODEL
FIGURE 4
'BEST-CASE' REGIONAL DISPERSION
FOUR CORNERS AREA
(WINTER ZONAL REGIME)
322
-------
323
-------
12/206/77
0/207/77
12/206/77
FIGURE 6
UPPER AIR METEOROLOGY
324
-------
SO4
JULY 24. 1977
JULY 25. 1977
DAILY AVERAGE GROUND LEVEL CONCENTRATIONS
1977 ERD INVENTORY (SOURCE: EPA)
MESOPUFF/MESOPAC MODEL
FIGURE?
'WORST-CASE' REGIONAL DISPERSION
FOUR CORNERS AREA
(SUMMER STAGNATION REGIME)
325
-------
considerable merging of individual plumes, and accumulation in the
north central portion of the computational region. The sulfate field
shows dramatic accumulation in the north central portion of the grid,
with maximum concentrations greater than 10 ug/nH.
In all, 106 days during 1977 were modeled; these shown here were
the best-day and worst-day regional dispersion regimes. The success
of a long range transport model is, these results suggest, largely
dependent on the physical verisimilitude of the meteorological "pack-
age" used to drive the modelsamong the least well understood parts
of the long range transport modeling problem.
It is not surprising, therefore, that the available models dif-
fer appreciably in the way they define and use the meteorological
information. The models differ especially in the definition of the
advective wind field, that is, in the choice of representative level,
in the scheme used to analyze and interpolate wind data, in the time
averaging assumptions, and in the dynamical or kinematic constraints,
if any, imposed in constructing the mean flow field.
The central issue is: How is the mean wind field to be speci-
fied? Lagrangian trajectory models require a single transporting
wind field, assumed to be representative (in some sense) of mean
transport layer flow conditions over the scale of interest. (Some
vertically averaged grid modelslike the SAI modelalso use a
vertically averaged mean wind field.) The wind field may be that at
a single level (for example at a constant height above the surface),
326
-------
or on a constant pressure surface, or may be described as a layer
average in any of several ways.
For example, in one of the forerunner studies of long range
transport with a Lagrangian model, Eliassen and Saltbones (1975) used
the 850 mb wind field, linearly interpolated in time between six-hour
observations. Among more recent models, the SRI EURMAP-1 model
(Johnson et al. 1978) for long-term interregional transport of 502
and 804 in western Europe, also uses the 850 mb wind field, but
adjusted to give a "more representative layer average.
The SRI short-term model, EURMAP-2, (Mancuso et al. 1979)
requires finer resolution. It uses a weighted average of winds in
two layers, 0-300 meters and 300-1,000 meters; within each layer the
layer-average winds are obtained by integrating with a power law of
the form
P(stability)
In areas dominated by maritime meteorology, (e.g., Scandinavia,
the United Kingdom, and other parts of western Europe), the trajec-
tory errors made when 850 mb winds (or, for that matter, surface
geostrophic winds) are used may be tolerable (Smith and Hunt 1978).
327
-------
By contrast, under more continental mesoscale flow regimes (especial-
ly in the winter over the U.S. Northern Great Plains, for example),
the atmosphere is strongly stratified, and the surface and 850 mb
winds are uncoupled. The 850mb wind field (or in moderate to ele-
vated terrain, the 700 mb winds) might be a more appropriate choice
to describe mean transport conditions. But without a detailed veri-
fication study, like that made by Draxler (1979) on the sensitivity
of his model to different assumptions on wind speed and wind veering
with height, it is impossible to "know" which assumptions are best,
particularly for a new application in a region for which adequate
raesoscale meteorological data coverage does not exist.
The possible schemes for constructing mean wind fields are
endlesswhich is unfortunate, because there has been little syste-
matic attention given to model intercomparisons with consistently
defined wind fields (but cf. Bass et al. 1979).
Each modeling group adopts its own unique scheme (or schemes).
For example, in the northern Great Plains regional model (Liu and
Durran 1977; Liu and Wojcik 1979), the 850 rab geopotential height
field, spline interpolated, was used to generate a geostrophic wind
that was then iteratively relaxed.
In the PNL regional scale model Wendell et al. (1977) used a
layer average wind field averaged through a fixed layer (100-1,000
meters). They began from 12-hourly NWC rawinsonde station data,
interpolated linearly in time, and used an inverse square spatial
328
-------
weighting to interpolate the discrete data points to a uniform grid.
(The subject of interpolation schemes might well take an entire
review paper; Goodin et al. 1979 have made an excellent summary of
interpolation methods for sparse data applied to wind and concentra-
tion fieldsto which the reader is strongly recommended.)
Among the other alternative approaches to wind field definition,
Heffter and co-workers (Heffter 1980) define a spatially and tem-
porally variable layer for averaging the wind field, based on the
height of any nonsurface-based inversion as obtained from vertical
temperature profiles. By contrast, Heffter1s earlier long-term model
used a single fixed layer for averaging, 0-1,000 meters.
Draxler's mesoscale model (Draxler 1979) carries Heffter's
approach furtherhe averages the winds through a vertical layer of
variable thicknessdetermined as the vertical extent within which
90% of the columnar vertical mass distribution is found. This layer
is computed from numerical solutions to the vertical diffusion equa-
tion with specified vertical eddy diffusivity profiles.
In both the Heffter and the Draxler schemes, the winds are
averaged with a weighting that depends on the thickness between raid-
points of observation levels -the surface wind data when used is
adjusted upwards to represent the layer-averaged winds, and grid
point interpolations are made within a radius of influence on the
order of 300 km for upper air stations, and 150 km for surface
stations.
329
-------
Not surprisingly, each modeling group has its own ideas about
appropriate 'regions of influence" for station data interpolation,
and about relative weights to be assigned to observation stations in
performing the wind interpolations. Clark and Eskridge (1977) have
made a variation on the Liu-Goodin scheme (Liu and Goodin 1976) that
allows the user to specify a reliability or confidence factor for
each observation station and to construct the interpolated wind
fields by a composite of these weighted observations. The NOAA/ARL
group carry the 'region of influence' concept one step beyond: their
models (for example, ARL-ATAD) also give greater weight to observa-
tional data from stations that are aligned within a narrow angular
range of the direction of the local trajectory segment. Draxler
shows that such directional weighting can make a significant improve-
ment to the predictive success of the model.
Finally, ERT's multipurpose MESOPUFF model uses wind fields
developed from data at user-specified levels (nominally 850 mb, for
example)interpolated to a uniform gridand iteratively adjusted to
within a specified maximum value of absolute local divergence (e.g.,
lO-^ s-l).
In regions of widespread high terrain, the Rocky Mountain area
for example, and especially where terrain extends well up to about
the 700 mb level, the choice of a 'most representative1 mean advect-
ing wind field is extremely problematic at present (and likely to re-
main so). The conventional upper air rawinsonde network cannot begin
330
-------
to resolve terrain-induced spatial variations in the upper level wind
field, the surface station data is often useless, and local (e.g.,
source applicant sponsored) sounding programs can only indicate ini-
tial plume direction. In modeling the regional impacts of energy
resource development in the high terrain flow regimes of the Four
Corners region, Bass et al. (I979a) chose therefore to represent the
mean transport wind by the winds at 700 mbrecognizing fully that
the choice was entirely arbitrarybut accepting also the complete
impracticality of attempting a more rigorous definition of the wind
field for routine multiday modeling applications.
Where constraints intrinsic to practical modeling applications
are not overriding issues, the use of sophisticated dynamically-
constrained wind field models is being vigorously and successfully
pursuedfor example, the complex mesoscale modeling systems being
developed at the Lawrence Liver-more Laboratory. Sherman (1978) has
described MATHEW, a terrain-consistent variational wind field genera-
tion model, and Lange (1978) has used MATHEW output in driving the
regional version of the ADPIC model for verification against regional
tracer studies. Such wind field synthesis programs represent the
leading edge of the state of the artbut require computer resources
that fall well beyond those practical or attainable for routine
multiday impact assessments. For the present and in the near term,
then, we will often have to select, arbitrarily, the mean transport
wind levels and the wind field interpolation schemes.
331
-------
Over local transport distances, a small error made in predicting
or estimating initial plume trajectory direction can often mean a
completely missed receptor far down wind. Figure 8 illustrates just
how long the initial error can grow to be, assuming straight line
flow thereafter (hardly a realistic assumption, in general). The
figure shows the plume lateral (crosswind) trajectory error as a
function of initial directional error and downwind distance, and com-
pares the lateral error to the characteristic half-width of the plume
(taken as 2.15
-------
I
Q
a
cr
O
LU
5
a.
en
cr
LU
cr
O
cc
cc
LU
CO
UJ
CC
=>
C3
LU
cc _j
HI QL
< W
LU
% =! z -
i 5 < s
2pfcs
O
333
-------
have to judge that for many routine regulatory applications, espe-
cially to PSD Class 1. increment consumption, a long-range transport
model that does not permit the separation of plume vertical spread
from that of the mixed layer (or what might be even worse, assumes a
climatologically-fixed mixed layer height) is inadequateeven for
so-called "screening" applications.
Because of the sensitive relationship between plume height and
mixed layer height, it is not necessarily the case that increasing
the height of the mixed layer will result in a decrease in the
ground-level concentrationscertainly not, if an increased mixed
layer height causes an elevated plume to be fumigated to the surface.
In the far field of a source, plume concentrations are usually
greater if the mixing depth is lower and, conversely, because plume
elements far from a source will usually have experienced at least one
diurnal mixing depth cycle and will have been fully entrained. In
contrast, because plume entrainment is episode-specific, no clear
relationship can be assumed to exist in the near field of a source.
It is therefore important to consider how the various models
differ in their representations of mixing height variations and
interpolation schemes. Some models, the EURMAP-1 model, for example,
use spatially uniform climatological valuesbut only for long-term
averages. Other models, for example, the short-term ATAD model, use
spatially and temporally varying values. The MESOPUFF model updates
the mixing height field hourly, as the larger of a mechanical mixing
depth or a convective mixing depth (Benkley and Bass 1979b).
336
-------
In Heffter's ATAD model, the daytime transport layer depth is
calculated as the height in the critical inversion layer, where the
temperature is two degrees above the temperature at the inversion
base. At night, the transport layer depth is 22 and SO^ transformation and removal.
The ASTRAP model, for example, incorporates dry deposition rates
that depend on season and time of day. The 862 and 804 transfor-
mation is linear, but the rate is seasonally and diurnally dependent.
By contrast, the PNL regional model assumes constant rates for dry
deposition and for S02 to sulfate transformation; but the wet re-
moval rate, taken as linear with precipitation rate, is shown to
depend critically upon whether precipitation is represented as (1)
uniform and constant over the entire computational region; (b) con-
stant but spatially variable; or (c) varying hourly with space and
time (Wendell et al. 1977).
337
-------
The choices of plume removal mechanisms in the model can depend
on whether the model is intended for long-term or for short-term
impact assessments. For example, the EURMAP-1 long-term model uses a
constant dry deposition rate; but its short-term analogue, EURMAP-2,
uses dry deposition rates that depend on the height of the puff ele-
ment above the surface. In many models, the dry deposition rate
depends on the ground-level concentration, because the surface flux
is parameterized in terms of a deposition velocity. Here again, if
the ground-level concentrations are misrepresented by the model (be-
cause the plume is prematurely mixed to the ground when it shouldn1t
be, or conversely), the deposition rate will be significantly
different, and will therefore change the subsequent evolution of the
plume.
The dry deposition rates may be made to depend on the underlying
surface. Lavery et al. (1980) have found, using the advanced EPRI/
SURE regional grid model (SURAD), improved model performance using a
dry deposition mechanism that varies with the underlying surface
vegetative cover and with diurnal variations in turbulent eddy
stresses.
Virtually every modeling group has tested the sensitivity of
their models to variations in deposition and removal terms, but, to
our knowledge, the work of Lavery et al. and Johnson et al. (1979)
are among the few studies that seek to verify the effects of such
model variations with actual field data.
338
-------
The transformation rate of SC>2 to 804 in a plume is, it is
becoming increasingly clear, a sensitive, complex function of the air
mass characteristics, especially the relative humidity and tempera-
ture. Hidy et al. (1976) and more recently long (1979) have explored
this relationship using the SURE data base with models that assume
linear, first order conversion rates.
More elaborate, nonlinear S02~S04 transformation mechanisms
have been proposed, based on detailed in situ plume observationsof
the Labadie plume, for example. But for present, or for near-term
regulatory applications, a prospective user is not likely to have
sufficient information on which to base a source-specific, nonlinear
plume transportation rate.
Another of the outstanding problem areas in long-range transport
modeling, and especially in view of the increasing concern over acid
precipitation, is that of wet removal. Rainfall is extremely effi-
cient at removing ambient S02 and, during active precipitation,
washout easily dominates removal (Smith and Hunt 1978). Various
prescriptions are used for modeling washout in terms of rainfall rate
and characteristic scavenging efficiency; these offer at least
formally the capability to describe wet removal. But because pre-
cipitation rates can be highly irregular, spatially-limited, and
episodic, especially during active convective situations, it is
difficult if not impossible to categorize rainfall rate on a scale
adequate to describe the fate of a plume, especially in its early
339
-------
time history. Several studies have indicated the important episode-
like nature of the plume rainout problem, and point thereby to the
difficulty of the acid precipitation modeling problem: the signifi-
cant uncertainties in plume transport and mixing are compounded by
the uncertainties in the nature, extent, and duration of the pre-
cipitation pattern experienced by a plume in the course of long-range
transport. Because of the disproportionate importance of only a few
episodes to long-term acid precipitation burden, it will be important
to make modeling studies with spatially and temporally well-resolved
information on precipitation rates during characteristic episodes.
If wet deposition is to be represented systematically on a regional
basis, we will need objective schemes for rainfall rate analyses
schemes that have yet to be developed, at least for practical meso-
scale modeling. Here again, we may look to the LAMPS program for
significant new developments (but not, realistically, in the near
term).
6.0 MODEL SENSITIVITY AND MODEL VERIFICATION STUDIES
Until quite recently, attempts at verifying trajectory models at
long transport distances have been severely hampered for lack of
suitable inert tracers detectable at infinitesimal concentration
levels, yet easily distinguishable from natural or anthropogenic
background concentration levels. Of late, however, the NOAA/ARL
Idaho Falls laboratory, in particular, has pioneered in the use of
special tracer techniques for long-range transport measurement
programs. Techniques have also been developed to exploit a tracer
340
-------
of opportunity (Kyrpton-85)j emitted routinely from the Savannah
River Plant. Detailed descriptions of more recent experiments with
SF^, CD^, and Krypton-85 releases from the Savannah River Plant,
and release of SFfc, heavy methane, and fluorocarbons from the Idaho
National Engineering Laboratory are given by Draxler (1979).
Numerous trajectory modeling studies have been conducted to com-
pare predicted and observed long-term average regional-scale S02
and S04 concentration budgets in Scandinavia, over western
Europe, in the eastern United States, and elsewhere. The models
used simple, multisource trajectory models, or grid models of
modest spatial resolution have met with variable to good success
on these scales (see below).
But, in proposing to use regional-scale models for regions in
which tracer data or other adequate field monitoring concentration
does not exist for model verification, modelers have often been
forced to rely solely on model sensitivity studies. Not only for
lack of adequate data bases, but also to better understand the capa-
bilities and limitations of their techniques, most modeling groups
have made extensive model sensitivity tests. Unfortunately, only a
fragment of this important collective experience has been reported.
As study of the long-range transport modeling literature will
quickly reveal, it is frustrating to model sensitivity results of
different modeling groups, because no two models begin from substan-
tially identical physical assumptions, initial conditions, spatial
341
-------
and temporal resolution, meteorological, emissions, and so forth.
Different workers have tended to stress, perhaps as a matter of per-
sonal predilection or common interest, different areas of long-range
model sensitivity.
For example, Wendell and collegues (Wendell et al. 1977) have
looked in detail at the effects on their model of time-averaged
versus real-time precipitation in wet removal of SC>2; they found
that time-averaged precipitation causes significant overremoval of
S(>2, and very different spatial distributions of ambient sulfate
levels. And Powell et al. (1979) found significant differences
between model results under assumptions of variable, or constant,
stability class; these were related to differences in plume aging and
degree of prior vertical mixing.
Many long-range transport studies (e.g., Liu et al. 1977; Bass
et al. 1979a) have illustrated that model sensitivity to changes in
one parameter cannot be assumed independent of changes in another
parameter. For example, Powell et al. (1979) found that changing
from constant neutral stability to variable stability caused changes
in relative deposition of SC>2 and 304 because, in their model,
dry deposition is proportional to ground-level concentrations, but
wet deposition is proportional to the vertical layer-averaged
concentration field. Shannon has pointed out the importance of
including diurnal variations in vertical stability to describe the
time-dependent coupling of elevated plumes to surface boundary layer
processes. Rao et al. (1976), Johnson et al. (1978), Powell et al.
342
-------
(1979), and many others have described the relative sensitivity of
their modeling results to different choices of plume transformation
and removal, wind speed and mixing height assumptions, grid resolu-
tion, and other model parameters and input variables. But, to date,
it would seem no group has undertaken fully the task of systematic-
ally comparing the performance of the various suggested modeling
approaches with a consistent set of experiments beginning with the
same meteorology, the same emissions, the same observed ambient air
quality, geometry, resolution, etc., and as similar a choice of base
case model parameters as is possible. (It is currently planned,
however, to conduct a comparative analysis of the performance of
three trajectory-type models and one grid model developed at the
various national laboratories against the EPRI/SURE and MAP3S data
bases (personal communication, Paul Michael).)
It is not far amiss to describe the model sensitivity testing of
regional-scale and long-range transport models as anarchicmany
groups are repeating much the same kind of experiment, but no con-
sistent, uniform protocols exist by which to compare and evaluate the
respective performance of different models.
More generally, one of the crucial issues for verification of
regional-scale models is that of suitable performance evaluation
criteria. In recent EPA-sponsored studies, several such criteria
have been advanced (Hayes 1979; Hillyer et al. 1979; Gelinas and Vajk
1979). Other groups, in particular, the EPRI Plume Model Validation
343
-------
study currently underway, are also confronting squarely the issues of
appropriate model performance evaluation criteria. Until, however, a
reasonably broad consensus for acceptability criteria emerges among
the long-range transport modeling community, it will remain difficult
to prove, convincingly, that one or another model is clearly prefer-
able for a given regulatory application. Indeed, as perceived by
interested regulatory groups, the choice of long-range transport
model approaches may appear to be a matter of subjective taste; and
arguments about physical realism may not be especially persuasive to
those not familiar with the complex nature of the long-range trans-
port problem.
Under NOAA sponsorship, the ERT group has made an attempt at
systematizing such a set of comparison criteria for long-range trans-
port models, especially for model sensitivity studies. They compared
the respective sensitivity of the plume segment, puff, and grid
(moment-method) modeling approaches under identical meteorological
and resolution conditions. These are reported in considerable detail
in Bass et al. 1979a.
6.2 Verification of Long-Term Models
It is fair to say, we believe, that the verification history for
long-range transport models is, as yet, fragmentary, although efforts
have accelerated considerably within the last several years. Some
verification is available for long-term average regional-scale
models. For example, Mancuso et al. (1979) has reported on com-
parisons of modeled and observed monthly-averaged ground-level
344
-------
concentrations of S02 and SO^ for the western European LRTAP data
base using the EURMAP-1 model. They obtained correlation coeffi-
cients ranging from 0.7-0.8 for S02 and from 0.6-0.7 for SO^.
The regional distribution patterns seem qualitatively reasonable.
Heffter et al. (1979) examined a 2-1/2-year long set of weekly
average samples of Krypton-85 measured at 13 locations at distances
of 30-150 kilometers from the Savannah River Plant. He used both his
long-term model (version A) and his short-term (H) model (fore-
runners of the ARL-ATAD model). His results show considerable
scatter, but no systematic bias.
Pendergast (1977; 1979) also looked at this data base and com-
pared monthly and 10-hour averages using Kern's segmented plume
model. Pendergast's results suggest that hourly stability class,
rather than constant stability, made little difference; but, it will
be noted, these are near-surface releases (63 meters).
Meyers et al. (1979a), using essentially a version of Heffter's
puff model with more complex vertical diffusion (a forerunner of
AIRSOX), looked at monthly average concentrations of S02 and S04
using the NADP and EPRI/SURE data bases for the eastern United
States. They report correlations of 0.7+ for S02 and 0.6+ for
SO^, with the best correlations as high as 0.8 for SO^ during
some months.
Shannon (1979a) describes long-term average verification studies
with the ASTRAP statistical trajectory approach, with generally
favorable results.
345
-------
And finally, the extensive model verification history compiled
in the EPRI/SURE program has demonstrated crucial relationships
between characteristic meteorology and episodes of high sulfate
concentrations (Hidy et al. 1976; Hidy et al. 1979; Lavery et al.
1978).
Recently, McNaughton (1980) has provided another set of com-
parisons of the SURE and MAP3S data base observations with long-term
predictions of the PNL regional model; his work emphasizes the impor-
tance of studying plume transformation and precipitation scavenging
regimes.
Recognizing present and near-term limitations, it is important,
nevertheless, to assemble as thorough a verification history as pos-
sible for long-range transport models that may be prospectively
valuable for regulatory and other practical applications. It is
likely that the long-term average model approaches will be verified
sooner than the short-term modelsfor all the problems common to
verification of conventional short-range models, exacerbated by the
special considerations of long-range transport.
6.3 Verification of Short-Term Models
The verification history of available short-term models is
skimpy at best, but improving rapidly. Heffter (1977) has reported
on the use of his short-term model for verification of a few selected
raultiday episodes of Krypton-85 transport over scales of 1,000 km.
The agreement is encouraging, but much more work is needed. The "H"
346
-------
version of his model, with somewhat better physics than the "A" ver-
sion, found 50 percent of calculated weekly averages within a factor
of 2 of observations, and on the order of 90 percent within a factor
of 10. But, Heffter points out, the improvement can be attributed
essentially to use of hourly average, rather than monthly average,
release rates. He emphasizes that verification success is largely
a question of the availability of adequately resolved wind data.
Mancuso et al. (1979) has recently reported on some preliminary
results of their short-term EURMAP-1 model against the LRTAP data
base for one daily average comparison test. The qualitative com-
parison appears quite reasonable, but hardly conclusive.
The most persuasive verification experiment reported to date for
a short-term average mesoscale trajectory model, at transport dis-
tances on the order of 100 km, are the recent Draxler (1979) results.
Especially noteworthy are the differences observed in model calcula-
tions for different choices of wind data. Draxler found that surface
winds did poorest; wind obtained hourly from a 60-meter tower, al-
though not local, provided much better estimates of mean transport;
and hourly surface winds adjusted to describe a layer of average
wind fields provided the best correspondence between actual time of
arrival and duration of the tracer material.
Draxler's conclusion about the required sampling density for
wind station data (on the order of 25 km) points to the real diffi-
culty confronting the proposed use of the mesoscale transport model
347
-------
under considerably less well-resolved wind field conditions. He does
suggest, however, that for cases of relatively constant and uniform
flow fields, the wind field resolution available from the surface
wind data network (on average, on the order of 100 km) may be ade-
quate.
We may conclude, overall, that the verification of a short-term
trajectory model is a major undertaking, one that a regulatory group
contemplating the use of the model for a new region will probably
not have the resources to undertake. The burden will remain with
the principal model development groups. We understand that the SRP
data base will be made available to other interested users for model
verification purposes (personal communication, J. Heffter), and all
groups proposing to submit long-range transport models for review
as candidate guideline models should be strongly encouraged to avail
themselves of this opportunity.
7.0 PRACTICAL LONG-RANGE MODELS FOR REGULATORY APPLICATIONS
The picture painted above may appear bleak; it is not. Much
progress, more than was anticipated even a few years ago, is being
made to refine and validate long-range transport models, and more
will be forthcoming. Yet, the present regulatory requirements for
near-term decision-making about source impacts at long transport
ranges, and the increasing emphasis on secondary effects of major
pollution sources (acid precipitation, regional haze, and the like),
underscore the practical necessity for selecting long-range trans-
port models for use now. The available models, all of them, are
348
-------
imperfect: and their respective authors would be the first to under-
score their limitations. Yet, considerable thought and ingenuity has
gone into the elaboration and implementation of at least some of
these models, efforts to make the models more flexible, more respon-
sive to a range of possible uses, more efficient and modest in compu-
tational storage requirements, and more applicable directly to
regulatory-type questions.
If the models are to be generally useful for simulation of
regional-scale plume transport and dispersion under different meteor-
logical regimes, they should respond to mesoscale spatial and tem-
poral variations in wind field, mixing depth, and ambient turbulence
levels. The input meteorology fields necessary to drive the models
must be easily generated from readily available data if the models
are to be useful to a wide user community. The dispersion models
must have moderate execution time and computer storage requirements
if they are to be practical for multiple-source, multiple-day simu-
lation exercises. And the model design should also allow for easy
adaptation to future research needs and regulatory applications;
thus, the programs should be highly modular, having components that
are easily modified or readily substituted.
Other important features that contribute to the usefulness of
models for regulatory and other practical applications include:
(a) the ease with which model input data requirements, parameters,
and algorithms can be changed to facilitate use for different
349
-------
applications; (b) the ease with which the models can be used to
simulate short-term and long-average ambient ground-level concentra-
tions; and (c) the availability of preprocessing and postprocessing
systems to prepare input and analyze and display model geographic
extent and spatial resolution, and to modify (replace) the meteoro-
logical preprocessing system.
Table 1 lists a number of long-range transport models in the
public domain. The table stratifies the available models by three
classes:
short-term models that may be appropriate for practical
regulatory applications;
long-term models that may be appropriate for practical
regulatory applications; and
* a representative sampling (but far from a complete list) of
current research-grade long-range transport models under
active development at the national laboratories and else-
where.
The models identified in the first two classes are all generally
available (or will soon be) and documented for external users; some
of the models in the third class are also so documented. As empha-
sized in the Introduction, we believe the short-term models are more
urgently required at this time, but one may anticipate increasing
demand for long-term average models as well.
The models suggested as for regulatory applications are all, to
our knowledge, well-coded, readily transportable, easy to use and to
modify and, in general, appropriate tools to begin the analysis of
long-range transport and dispersion. But it should be emphasized
350
-------
til.-; lur
pli.Mliont
KL-pres..-:ltali
-mil DcVi'losMl
i:tiKMA >-:'
Mi.-..-j' :rrv
Sums-;-,;- l'i .i:is|i.,[ I
I :
'. i.i! Jl '-V^-; | ;
r-ru! j MKSoa* lli . I-:KT ; Mnrri-. i-t i
Si.it i*t u-.il ; AJ-.-.ju;-t-i.i Mati.-ilicai . AfKOr.iK- Natinit.ii l..i!i | Sit.innn;i 197
IlL-Lfu-r 19V)
: .in'i IAI^-. I
.: Dun.ir. :M." '
I'uff/Vcrl i.-iil
Finite
Oil IcreiK-i.-
I'ul'l'/Viirl leal
Finite
OH K-rs.-iK-.-
Piil'l
PI inn.-
N'l-j-tiiwi'.'. r L.-il>
Ml i.-\A A i f
| s,,ourc,.. Ul,
! SKI
,K>lmsim ill .il. lt;J
Me::u>?,i:.al.' '1 ra jvi-tory iNnAA Air Ru&t.ui t-.t^s Lab Draxivr 1979
and liiffunicm Modcl i
A1K.SDX
! llr.iokhavfn Niil ional Meyers ut al. !9;'l
- L.ili
i ;
Sfl|-.nu'titi!d tMunc Model I Savannah Rivt-r Lab ! I'.-ttdur^.lsc 1CH9
Purl ich'-in- I Al)l'li:-.VATliKW
r,- 1 i I
Grid i SL'LFAJC
CriJ : SfK,',!;
Criil
CriJ
frjic
KI'KI/KRT
Ei'Si/EST
.TL'kneknni
1 ttt}', ional SnpL-rrruidt;! ' , EVA Meli-oroln^y Liit:
il.i^ilfU A:L-,. Mk-:.u:^c.:(lll- DCL-XL; l-N.IAK-laM.
i'rt-0 ii-l inn SysU-in
(LAMi'S)
| H.-io el al. 11!f>
':.,n-i-ry .-t ol. 11-n
Ml Its .mil llir.-rt.-) IT,'a
351
-------
that skilled judgment is still critical in their use and interpreta-
tion. The different models have, of course, different advantages and
disadvantages; the prospective user is strongly urged to consult the
specific references in greater detail and, almost of necessity, to
speak to the authors as well. (We might note that the ERT models
MESOPUFF, MESOPLUME, and MESOGRIDhave been designed from the outset
specifically to be practical for regulatory purposes, and are under
consideration for wider distribution as EPA UNAMAP "INDIVIDUAL"
models.)
By and large, the short-term puff- and plume-segment models
are economical for simulating one or a small number of sources, but
would not be appropriate for simulating large multisource inventories
(e.g., heavily industrialized regions). For such problems, grid mod-
els are more practical.
The long-term trajectory models and, particularly, the statis-
tical trajectory and "square puff" approaches, are especially eco-
nomical. If the loss of near-field resolution is tolerable, they may
well offer the most efficient path to long-term average computations.
Of course, a short-term model can be used to develop long-term
averages by "brute force" sequential calculations. There is consid-
erable precedent for this approach in using conventional short-range
Gaussian plume models. It is indeed possible to do so for a raeso-
scale trajectory or grid model, but the costs may be prohibitive, es-
pecially for a large number of sources using a short-term puff model
352
-------
(MESOPUFF) 24-hour average concentrations were computed for 106 days
during calendar year 1977 for the Four Corners region, and these were
weighted to simulate the annual average concentrations of SC>2 and
304 in the region. Simulations of this magnitude may cost tens of
thousands of dollars; if the objective is only to develop long-term
(but not maximum short-term) average concentrations, the long-term
average modeling approaches are more attractive.
Finally, the usual spatial resolution of these suggested models
is typically on the order of 50 km (+_ a factor of 2). By and large,
their resolution is adequate forQmesoscale transport problems, and
adequate or marginally adequate on the pmesoscale.
8.0 FUTURE DIRECTIONS AND RESEARCH NEEDS FOR LONG-RANGE TRANSPORT
MODELS
The needs may be put simply:
standard model verification data sets;
* standard baseline cases;
standard protocols for model testing; and
uniform criteria for model performance evaluation and
ve rification.
Meeting these needs, unfortunately, is not simple. The evalua-
tion criteria should, practical experience suggests, reflect intended
regulatory usage, that is, the model's ability to predict highest or
second highest local values, not just grid-cell averages or area-wide
averages, or total pollutant fluxes and budgets. We alluded earlier
353
-------
to the problem of choosing suitable statistical measures. Conven-
tional measures, such as correlation coefficients, have inherent
problems for regional-scale plume model validation with limited sam-
pling station coverage, given the characteristic spikiness of the
primary SC>2 field; but are more successful with the smoother, more
regular behavior of the regional and long-range 864 fields.
Several workshops have underscored the requirement for addi-
tional tracer studies to investigate the chemistry and the physics of
plume transport on regional scales, and the need for further studies
of regional ambient air quality (not sensibly dominated by local
source contributions). To verify short averaging time plume models
on the p mesoscale range, ideally, mobile airborne sampling stations
are needed. Even then, the sampling problem is formidable. Sheih
et al. (1978b) have looked in detail at the mobile network sampling
problem; for short-term plume measurements 100 km downward from the
source, made to characterize the crosswind plume spread o"y, one would
need several hundred independent aircraft traverses under identical
atmospheric stability and mechanical turbulence conditions. Experi-
mental resources of this scale are virtually without precedent in the
long-range transport modeling field.
Better techniques, including satellite plume tracking and real-
time prognostic mesoscale wind field prediction modeling, could help
greatly in the forecasting of short-term plume movements and near-
real-time positioning of mobile samplers to promote greater capture
354
-------
statistics. But, except for short, intensive field programs like
those suggested by Sheih et al., for practical reasons an observa-
tional network for verification of a raesoscale trajectory model will
be fixed. Because the effects of large-scale plume meander will
probably dominate small-scale turbulent diffusion for many flow
situations, it may be hard to develop reliable estimates of relative
diffusion; so, for fixed station networks, the greatest emphasis
should probably be placed on verifying cross-wind-average or long-
term average predictions. Fixed-s tat ion networks may be practical
for long-term average trajectory models, or for models that essen-
tially ignore horizontal diffusion, or for models in which relative
small-scale diffusion and large-scale meander are not specifically
dist inguished.
9.0 FUTURE DIRECTIONS IN LAGRANGIAN LONG-RANGE TRANSPORT MODEL
DEVELOPMENT
Because the predictive success of regional and long-range
trajectory models is ultimately constrained most by the spatial and
temporal resolution of the available mesoscale meteorological data,
in the future, regional-scale diffusion models may be driven by the
output of prognostic fine-mesh numerical weather prediction models.
We may hope to understand better, on theoretical grounds, the key
problem of turbulent dispersion in the so-called "spectral gap"
(between the large, energetic baroclinic scales at which available
potential energy and kinetic energy are created, and the small,
355
-------
turbulent scales at which energy and entropy are dissipated). Per-
haps it will be clearer, not many years hence, whether the approxi-
mations made for plume growth with distance at long travel times are
of general or more limited validity (see Gifford 1976 for a review of
the basic data and theoretical issues).
As the cost of computation continues to fall, we can look for-
ward to models that make much greater demand on large-scale compu-
tational resources. Such models may become practical for routine
assessment use of techniques that are currently much too computer--
intensive to be practical (for example, the ADPIC model). We may
anticipate that trajectory modules may be embedded routinely within
larger grid models to provide subgrid-scale detailalready being
attempted in a rudimentary way by several modeling groups.
It will remain important to keep in perspective, however, the
real constraints on model improvement that arise from imperfect un-
derstanding of the physics and chemistry of long-range plume trans-
port. As such knowledge accumulates materially, it will support
modeling approaches of increasing sophisticationbut it is difficult
to imagine that more complex models will necessarily be improvements
over simpler models until the available mesoscale meteorological data
is greatly expanded--either by a denser network of direct observa-
tions, or by skillful interpolation with powerful objective analysis
or prognostic methods.
356
-------
For practical applications of long-range transport models, it
would be very helpful to see the development of interactive "problem
definition" programs (for example, as developed for the LIRAQ model-
ing system) to assist the nonspecialist in using sophisticated models
for routine applications.
We should encourage the use of these models to address regula-
tory questions couched in probabilistic termsthat is, not just
"What is the highest or second highest concentration observed in a
given time period?" but, perhaps, "What is the expected return period
for a given concentration value?" Impetus is developing for regula-
tory decisions based on probabilistic interpretations of familiar
short-range models; it is surely no less appropriate to use long-
range transport models in similar probabilistic ways.
Some of the research and development models identified in group
(c) of Table 1 may, in time, also become practical for regulatory
useespecially as the cost of large computations decreases, and
large-core machines are more generally available to user groups. Of
course, many models should properly remain research models only, as
test beds for new concepts and interpretations of turbulent plume
transport and transformation over long range.
10.0 RECAPITULATION
Overall, as we have seen, the available modeling resources for
practical regulatory applications of long-range plume transport and
impact are at an "adolescent" stage: promising but not yet mature.
357
-------
Although the verification history for these models is incomplete at
best, they are still better tools for regulatory decision making on
long transport scales than are the conventional short-range
straight-line Gaussian plume models. Important limitations will
continue, in the near term at least, to constrain the accuracy and
reliability of such models; but the alternative, to continue to use
straight-line techniques that are clearly inadequate to describe
mesoscale and long-range transport cannot be acceptable. We would
suggest that, in the present state of understanding, the use of these
models with reasonably conservative parameter choices could well
serve as "screening" tools for regulatory decision makingmore
detailed analyses, with models specifically verified, if possible,
for a given region, might be required at a later date.
The need for more frequent, denser meteorological data sets has
been emphasized, as has the difficulty of defining the mean wind
field (and other meteorological parameters) in regions of rough
terrain or under active weather conditions, and has underscored the
prospective importance, in the longer term, of coupled mesoscale
prediction/plume transport systems such as LAMPS.
Sensitivity of the available long-range transport models to the
meteorological, chemical transformation, and removal parameters (not
to speak of the intrinsic limitations of one or another numerical
scheme) requires much more study. Such further work would be more
purposeful, we would suggest, if coordinated within a commonly
358
-------
accepted framework of problem definition, meteorology, and physical
processes. The need for commonly accepted criteria for model perfor-
mance verification cannot be overemphasized, and the importance of
choosing evaluation criteria that relatedirectly to practical end
uses of these models must be addressed with real urgency.
Working with joint purpose, the developers of long-range trans-
port models, and those charged with making regulatory decisions, can
together advance both state-of-the-art of long-range transport model-
ing, and the national goals of energy development and industrial
growth consistent with environmental protection.
Ac know 1 e d gme n t
We acknowledge, with gratitude, the kind assistance of Dr. Paul
Michael, BNL, in providing a preliminary copy of the MAP3S Modeling
Abstracts.
359
-------
-------
REFERENCES
Bass, A., C.W. Benkley, J.S. Scire, and C.S. Morris 1979a. Devel-
opment of Mesoscale Air Quality Simulation Models. Volume I. Com-
parative Sensitivity Studies of Puff, Plume, and Grid Models for
Long-Distance Dispersion Modeling.EPA 600/7-79-XX, Environmental
Protection Agency, Research Triangle Park, NC, 238 pp.
Bass, A., C.W. Benkley, and J.S. Scire 1979b. Energy-Related
Regional Air Pollution in the Four Corners Area, 1977-1987: Simula-
tion Studies with the MESOPUFF Model. EPA-600/7-79-XX.
Benkley, C.W. and A. Bass 1979a. Development of Mesoscale Air Qual-
ity Simulation Models. Volume 2. User's Guide to MESOPLUME (Meso-
scale Plume Segment) Model. EPA 600/7-79-XXX, Environmental
Protection Agency, Research Triangle Park, NC, 141 pp.
Benkley, C.W. and A. Bass 1979b. Development of Mesoscale Air Qual-
ity Simulation Models. Volume 3. User's Guide to MESOPUFF (Meso-
scale Puff) Model. EPA 600/7-79-XXX, Environmental Protection
Agency, Research Triangle Park, NC, 124 pp.
Benkley, C.W. and A. Bass 1979c. Development of Mesoscale Air Qual-
ity Simulation Models. Volume6.User's Guide to MESOPAC (Mesoscale
Meteorology Package). EPA 660/7-79-XXX, Environmental Protection
Agency, Research Triangle Park, NC, 76 pp.
Benkley, C.W. and L.L. Schulman 1979. Estimating Hourly Mixing
Depths from Historical Meteorological Data. J. Appl. Meteor.
18:772-780.
Bolin, B. and C. Persson 1975. Regional Dispersion and Deposition of
Atmospheric Pollutants with Particular Application to Sulfur Pollu-
tion Over Western Europe. Tellus 2^:281-310.
Clark. T.L. and R.E. Eskridge 1977. Nondivergent Wind Analysis Algo-
rithm for the St. Louis RAPS Network. EPA 600/4-77-049. Environ-
mental Protection Agency, Research Triangle Park, NC, 63 pp.
Draxler, R.R. 1977. A Mesoscale Transport and Diffusion Model.
National Oceanic and Atmospheric Administration Tech. Memo.
ERL-ARL-64, Air Resources Laboratories, Silver Spring, MD.
Draxler, R.R. 1979. Modeling the Results of Two Recent Mesoscale
Dispersion Experiments. Atmospheric Environment H:1523-1533.
361
-------
REFERENCES (Continued)
Egan, B.A., S. Rao, and A. Bass 1976. A Three-Dimensional Advective-
Diffusive Model for Long-Range Sulfate Transport and Tranformation.
Seventh International Technical Meeting on Air Pollution and Its
Application, Airlie, VA, September 7-10, 1976, pp. 697-714.
Eliassen, A. 1978. The OECD Study of Long-Range Transport of Air
Pollutants: Long-Range Transport Modeling. Atmospheric Environment
_U:479-488.
Eliassen, A. and J. Saltbones 1975. Decay and Transformation Rates
of Sulphur Dioxide as Estimated from Emission Data, Trajectories, and
Measured Air Concentrations. Atmospheric Environment 9:425-429.
EPA 1978. Guideline on Air Quality Models. EPA-450/2-78-027. Envi-
ronmental Protection Agency, Office of Air Quality Planning and Stan-
dards, Research Triangle Park, NC.
Fisher, B.E.A. 1975. The Long-Range Transport of Sulfur Dioxide.
Atmos phe r ic Environment ^:1063-1070.
Fisher, B.E.A. 1978. The Calculation of Long-Term Sulfur Deposition
in Europe. Atmospheric Environment 12:489-501.
Gelinas, R.J. and J.P. Vajk 1979. Systema t i c Sensitivit y Analyses of
Air Quality Simulation Models. EPA-600/4-79-035. Environmental
Protection Agency.
Gifford, F.A. 1976. Tropospheric Relative Diffusion Observations.
J. Appl. Meteor. 16:311-313.
Gillani, N.V. 1978. MISTT: Mesoscale Plume Modeling of the Disper-
sion, Transportation, and Ground Removal of S02 Atmg_sp_heric Envi-
ronment 12:569-588.
Goodwin, W.R., G.J. McRae, and J.H. Seinfeld 1979. A Comparison of
Interpolation Methods for Sparse Data: Application to Wind and Con-
centration Fields. J. Appl. Meteor. 18:761-771.
Hales, J.M., D.C. Powell, and T.D. Fox 1977. _STRAM_-_An Air Pollu-
tion Model Incorporating Non-Linear Chemistry, Variable Trajectories,
and Plume Segment Diffusion. EPA 450/3-77-012. Environmental Pro-
tection Agency, Research Triangle Park, NC, 147 pp.
Hayes, S.R. 1979. Performance Measures and Standards for Air Quality
Simulation Models. Report EF78-93R2. Systems Applications, Inc.,
San Rafael, CA.
362
-------
REFERENCES (Continued)
Heffter, J.L. 1980. Air Resources Laboratories Atomospheric Trans-
port and Dispersion Model (ARL-ATAD). National Oceanic and Atmo-
spheric Administration, Tech. Memo. ERL-ARL-81. Air Resources
Laboratories, Silver Spring, MD.
Heffter, J.L. , G.J. Ferber, and A.D. Taylor 1975. A Regional-
Continental Scale Transport, Diffusion, and Deposition Model.
National Oceanic and Atmospheric Administration, Tech. Memo.
ERL-ARL-50. Air Resources Laboratories, Silver Spring, MD.
Heffter, J.L. and G.L. Gerber 1977. Development and Verification of
the ARL Regional-Continental Transport and Dispersion Model. Pro-
ceedings, Joint Conference on Applications of Air Pollution Meteorol-
ogy, November 29-December 2, 1977, Salt Lake City, UT, pp. 400-407.
Heffter, J.L., G.J. Ferber, and K. Telegadas 1979. Verification of
the ARL Transport and Dispersion Model at 30-150 km. Preprint Vol-
ume, Fourth Symposium on Turbulence, Diffusion, and Air Pollution.
January 15-19, 1979, Reno, NV, pp. 372-375.
Hidy, G.M. et al. 1976. Design of the Sulfate Regional Experiment
(SURE) Vol. I, Supporting Data and Analysis. Report No. EC-125,
Electric Power Research Institute.
Hidy, G.M., P.K. Mueller, T.F. Lavery, and K.K. Warren 1979. Assess-
ment of Regional Air Pollution Over the Eastern United States:
Results from the Sulfate Regional Experiment (SURE). WNO Symposium
on the Long-Range Transport of Pollutants. Sofia, Bulgaria, October
1-5, 1979, pp. 65-76.
Hillyer, M.J., S.D. Reynolds, and P.M. Toth 1979. Procedures for
Evaluating the Performance of Air Quality Simulation Models. Report
EF79-25R. Systems Applications, Inc., San Rafael, CA.
Husar, R.B., J.P. Lodge, and D.J. Moore (Eds) 1978. Sulfur in the
Atmosphere, Proceedings of the International Symposium, Dubrovnik,
Yugoslavia, September 7-14, 1977. Atmospheric Environment 12:1-796.
Johnson, W.B., D.E. Wolf, and R.L. Mancuso 1978. Long-Term Regional
Patterns and Transfrontier Exchanges of Airborne Sulfur Pollution in
Europe. Atmospheric Environment 12:511-527.
Kreitzber, C.W. and M.J. Leach 1978. Diagnosis and Prediction of
Tropospheric Trajectories and Cleansing. Proceedings, 85th National
Meeting. American Institute of Chemical Engineers. Philadephia, PA,
June 4-8.
363
-------
REFERENCES (Continued)
Lange, R. 1978. ADPIC - A Three-Dimensional Transport-Diffusion
Model for the Dispersal of Atmospheric Pollutants and Its Validation
Against Regional Tracer Studies. J. Appl. Meteor. 17:320-329.
Lavery, T.F., J.W. Thrasher, D.H. Gooden, A.C. Lloyd, and G.M. Hidy
1978. Regional Transport and Photochemical Model of Atmospheric
Sulfates. Proceedings of the Ninth International Technical Meeting
on Air Pollution Modeling and Its Application, CCMS/NATO,
Unweltbundesant, Berlin.
Lavery, T.F., R.L. Baskett, J.W. Thrasher, N.J. Lordi, A.C. Lloyd,
and G.M. Hidy 1980. Development and Validation of a Regional Model
to Simulate Atmospheric Concentrations of S02 and Sulfate. Pro-
ceedings of the AMS/APCA Second Joint Conference on Application of
Air Pollution Meteorology, American Meteorological Society, Boston,
MA.
Liu, C.Y. and W.R. Goodin 1976. An Interactive Algorithm for Objec-
tive Wind Field Analysis. Mon. Wea. Rev. 104:784-792.
Liu, M.K. and D. Durran 1977. The Development of a Regional Air Pol-
lution Model and Its Application to the Northern Great Plains.
EPA-908/1-77-001. Systems Applications, Inc., San Rafael, CA.
Mancuso, R.L., C.M. Bhuraralkar, D.E. Wolf, and W.B. Johnson 1979.
The Exchange of Sulfur Pollution Between the Various Countries of
Europe Based on the SURMAP Model. Preprint Volume, Third Symposium
onAtmospheric Turbulence, Diffusion, and AirQuality. Raleigh, NC.,
pp. 345-354.
McNaughton, D.J. 1980. Initial Comparison of SURE/MAP3S Sulfur Oxide
Observations with Long-Term Regional Model Predictions. Atmospheric
Environment 14:55-63.
Meyers, R.E., R.T. Cederwall, and W.D. Ohmstede 1979. Modeling
Regional Atmospheric Transport and Diffusion: Some Environmental
Applications. In: Advances in Environmental Science and Engineer-
_ing_. J.R. Pfafflin and E. Ziegler (Eds.) New York: Gordon and
Breach.
Meyers, R.E., R.T. Cederwall, J.A. Storch, and L.I. Klienman 1979a.
Modeling Sulfur Oxide Concentrations in the Eastern United States:
Model Sensitivity, Verification, and Applications. Preprints, Fourth
Symposium on Turbulence, Diffusion, and Air Pollution, American
Meteorological Society, Reno, NV, January 16-18, 1979. pp. 673-676.
364
-------
REFERENCES (Continued)
Mills, M.T. and A.A. Hirata 1978. A Multiscale Transport and Dis-
persion Model for Local and Regional Scale^ Sulfur Dioxi^de/Sulfate
Concentrations: Formulation and Initial Evaluation. Ninth Inter-
national NATO/CCMS Technical Meeting on Air Pollution Modeling and
Its Application. August 28-31, 1978. Toronto, Canada.
Morris, C.S., C.W. Benkley, and A. Bass 1979. Development of Meso-
scale Air Quality Simulation Models. Volume 4. User's Guide to
MESOGRID (MesoscaleGrid) Model. EPA 600/7-79-XXX. Environmental
Protection Agency, Research Triangle Park, NC, 118 pp.
Nappo, C.J. 1978. Workshop on Long-Range Trajectory-Puff and Plume
Modeling of Continuous Point Source Emissions. National Oceanic and
Atmospheric Administration, Tech. Memo. ERL-ARL-72, Air Resources
Laboratories, Silver Spring, MD.
New Orleans 1980. AMS/APCA Second Joint Conference on Applications
of Air Pollution Meteorology. New Orleans, LA, March 24-27, 1980.
Niemann, B.D., A.A. Hirata, and L.F. Smith 1979. Application of a
Regional Transport Model to the Simulation of Multiscale Sulfate
Episodes Over the Eastern United States and Canada. Proceedings, WMO
Symposium on the Long-Range Transport of Pollutants and Its Relation
to General Circulation Including Stratospheric/Tropospheric Exchange
Processes. Sofia, Bulgaria, 1-5 October 1979. WMO No. 538, Geneva,
Switzerland. pp. 337-347.
Pack, D.H., G.J. Ferber, J.L. Heffter, K. Telegadas, J.K. Angell,
W.H. Hoecker, and L. Machta 1978. Meteorology of Long-Range Trans-
port. Atmospheric Environment 12:425-444.
Pendergast, M.D. 1979. A Comparison of Observed Average Concentra-
tions of 85Kr with Calculated Values Observed from a Wind Rose Model
and A Time-Dependent Trajectory Model. Proceedings, Joint Conference
on Applications of Air Pollution Meteorology. November 29 - December
2, 1977, Salt Lake City, UT. pp. 253-254.
Pendergast, M.M. 1979. Model Evaluation for Travel Distances 30-140
km. Preprint Volume, Fourth Symposium on Turbulence, Diffusion^ and
Air Pollution. January 15-19, 1979, Reno, NV. pp. 648-651.
Powell, D.C., D.J. McNaughton, L.L. Wendell, and R.L. Drake 1979. A^
Variable Trajectory Model for Regional Assessments of Air Pollution
from Sulfur Compounds. PNL-2734, Battelle, Pacific Northwest
Laboratory, Richland, WA.
365
-------
REFERENCES (Continued)
Raleigh 1976. Preprint Volume, Third Symposium on Atmospheric Turbu-
lence , Diffus ion and Air Quality. October 19 - 22, 1976. Raleigh,
NC.
Rao, K.S., J.S. Lague, and B.A. Egan 1976. An Air Trajectory Model
for Regional Transport of Atmospheric Sulfates. Preprint Volume,
Third Symposium on Atmospheric Turbulence, Diffusion and Air Quality.
American Meteorological Society, October 19 - 22, 1976, Raleigh, NC.
pp. 3-5-331.
Reno 1979. Preprint Volume, Fourth Symposium on Turbulence, Diffu-
sion and Air Pollution, American Meteorological Society, January 15 -
18, 1979, Reno, NV.
Salt Lake City 1977. Preprint Volume, Joint Conference on Applica-
_tj.ons of Air Pollution Meteorology. American Meteorological Society,
November 29 - December 2, 1977, Salt Lake City, UT.
Scriven, R.A. and B.E.A. Fisher 1975. The Long Range Transport of
Airborne Material and its Removal by Deposition and Washout. Atmos-
pheric Environment 9_:49-68.
Santa Barbara 1974. Symposium on Atmospheric Diffusion and Air
pollution. American Meteorological Society, Santa Barbara, CA,
September 9-13, 1974.
Shannon, J.D. 1979a. The Advanced Statistical Trajectory Regional
Air Pollution Model. ANL/RER-7901. Argonne National Laboratory,
Argonne, IL.
Shannon, J.D. 1979b. A Gaussian Moment-Conservation Diffusion Model.
J. Appl. Meteor. 18: 1406-1414.
Sheih, C.M. 1977. Application for a Statistical Trajectory Model to
the Simulation of Sulfur Pollution over Northeastern United States.
Atmospheric Environment 11; 173-178.
Sheih, C.M. 1978. A Puff Pollutant Dispersion Model with Wind Shear
and Dynamic Plume Rise. Atmospheric Environment 12; 1933-1938.
Sheih, C.M., G.D. Hess, and B.B. Hicks 1978. Design of Network
Experiments for Regional-Scale Atmospheric Pollutant Transport and
Transformation, Atmospheric Environment^ 12: 1745-1753.
Sherman, C.A. 1978. A Mass-Consistent Model for Wind Fields over
Complex Terrain. J. Appl. Meteor. 18: 312-319.
366
-------
REFERENCES (Concluded)
Smith, F.B. and R.D. Hunt 1978. Meteorological Aspects of the Trans-
port of Pollution over Long Distances. Atmospheric Environment 12:
461-477.
Start, G.E. and L.L. Wendell 1974. Regional Effluent Dispersion Cal-
culations Considering Spatial and Temporal Meteorological Variations.
National Oceanic and Atmospheric Administration Tech. Memo. ERL ARL-
44, Idaho Falls, ID.
Sykes, R.I. and L. Hatton 1976. Computation of Horizontal Trajector-
ies Based on the Surface Geostrophic Wind. Atmospheric Environment
10_: 925-934.
Tong, E.Y. 1979. Characterization of Regional Sulfate/Oxidant Epi-
sodes in the Eastern United States and Canada. Paper presented at
72nd Annual Meeting, Air Pollution Control Association, Cincinnati,
OH, June 1979.
Wendell, L.L., D.C. Powell, and R.L. Drake 1976. A Regional Scale
Model for Computing Deposition and Ground Level Air Concentration of
S02 and Sulfates from Elevated and Ground Sources. Preprint Vol-
ume, Third Symposium on AtmosphericTurbulence, Diffusion, and Air
Quality. American Meteorological Society, Raleigh, NC. pp. 318-324.
WMO 1979. Symposium on the Long-Range Transport of Pollutants and
j^ts Relation to General Circulation includingStratospheric/Tropos-
pherjic Exchange Processes. Sofia, Bulgaria, October 1-5, 1979.
Wendell, L.L., D.C. Powell, and D.J. McNaughton 1977. A Multi-Source
Comparison of the Effects of Real Time Versus Time Averaged Precipi-
tation Data on S02 and Sulfate Particulate Removal in a Regional
Assessment Model. Preprint Volume, Joint Conference on Applications
of Air Pollution Meteorology. American Meteorological Society. Salt
Lake City, UT, November 29-December 2, 1977. pp. 372-379.
367
-------
-------
4.0 FINE PARTICULATES MODELING WORKING GROUP RECOMMENDATIONS
This section summarizes the discussions, conclusions and recom-
mendations of the Fine Particulates Modeling Committee, at the EPA
Workshop on Regional Air Pollution Modelings held in Port Deposit,
Maryland on October 29 - November 1. Members of the committee were:
Carmen Benkovitz
BNL
P. Coffey
New York Dept. of Environmental Conservation
Ken Demerjian
EPA-ESRL
Bruce Hicks
ANL
Carl Kreitzberg
Dr exe1 Un ive rs i ty
Robert Lamb
EPA-ESRL
Steve Lewellan
Aeronautical Research Associates of Princeton, Inc.
Paul Michael
BNL
B. Niemann
Teknekron, Inc.
Richard Pitter
MITRE
Thomas Warner
Penn State University
4.1 Background
Inhaled particles (IP) are aerosol particles which may be drawn
369
-------
into the nasal passage during respiration. They are generally spe-
cified as particles less than 15 microns diameter. (Unless otherwise
indicated, aerosol particle size refers to its effective diameter.)
Respirable particles are those which are small enough to evade the
defense mechanisms of the upper respiratory system and large enough
to be retained in the lower respiratory system. Thus, respirable
particles are between 0.5 and 3 microns in diameter. Inhaled parti-
cles greater than 3 microns are trapped in the nose and throat and
eventually swallowed. Although they do not enter the lung, their
toxicity in the digestive tract must be considered. Respirable par-
ticles may be deposited in the bronchi, bronchioles or alveoli of the
lung, where they are capable of causing respiratory problems.
Fine particulates (FP) are defined as all aerosol particles less
than 15 microns in diameter, and thus include the IP and respirable
particle categories.
Fine particulates may be important because of their carcinogen-
icity. Although the carcinogenicity of most species are not well
known, many polycyclic aromatic hydrocarbons (PAH's) are known or
suspected carcinogens, and other substances, such as chyrosile
asbestos fibers, are believed to be carcinogenic because of their
needlelike shape.
Fine particulates are important in various atmospheric proces-
ses. They may serve as catalysts for chemical reactions, including
gas-to-particle conversion. They may serve as condensation nuclei
370
-------
or ice formation nuclei within clouds. They scatter and absorb solar
radiation and affect the global atmospheric heat budget.
Future technologies may emit fine particles or their precursers
in sufficient quantity that they may have notable impact on man and
the environment because of long range transport.
A. Discussion of General Inputs and Sources of Data
The committee noted that the weakest area of the National Emis-
sions Data Set (NEDS) is particulate emissions, and that fine partic-
ulates are not well characterized. To estimate annual emissions of
fine particulates, one must presently use total emitted particulates
as a guide and specify a fraction of that amount as representative of
fine particulates. The fraction is dependent on source type.
In the near term, the EPA's Fine Particulate Data Base (FPDB)
should be an improvement over the ratioing method, and in the long
term, improvements in State's emissions inventories and NEDS should
futher improve accuracy of annual fine particulate emissions. The
committee noted the need to characterize natural emissions of fine
particulates, and in this light raised the question: What part of
(rural) fine particulates is actually the result of primary anthro-
pogenic fine particulate emissions? In other words, if most fine
particulates are secondary aerosols, and much of the remainder is of
natural origin, then primary fine particulate emissions from anthro-
pogenic sources may not need to be accurately known.
371
-------
The committee suggested that scaling factors to adjust annual
FP emissions to seasonal, monthly and daily/hourly emissions are the
most viable recourse at the present or in the near term.
The air quality data base was then considered. For annual
studies, presently-available data bases include the Florida State
University (FSU) Streaker data, the Sulfate Regional Experiment in-
tensive data (SURE II}, and St. Louis dichotomous samplers taken as
part of the Regional Air Pollution Study (RAPS). These data bases
have resolution down to 1 day, and are therefore suitable for shorter
time scales. In the near term, data from 100 urban dichotomous sam-
plers will be available. This urban inhaled particle (UIP) network
is sponsored by EPA. The committee felt that there should be a study
concerning the need to implement a similar rural inhaled particle
(RIP) network.
Trends in fine particulate concentrations are difficult to
document on a regional basis. The committee suggested that national
background monitoring sites might already have data suitable for
analysis of fine particulate trends, and that correlations between
fine particulate concentration and turbidity might be determined
in order to use long term turbidity measurements as indicators of
trends.
Field studies of chemical transformation were divided into
two time scales. The 2-12 hour time scale includes data from the
Streaker, SURE II and RAPS experiments. The 24-72 hour time scale
372
-------
includes data from the Sulfur Transport and Transformation Experi-
ment (STATE), the Northeast Region Oxidant Study (NEROS) and the
Persistent Elevated Pollutant Episode (PEPE) study. Such data sets
alone are not sufficient information concerning chemical kinetics,
but when used in conjunction with laboratory studies, the results may
yield the rates of the important physical and chemical transforma-
tions. Laboratory studies of sulfur, nitrogen and carbon roles in
gas-to-particle conversion and chemistry are important, as are
laboratory studies of fine particulate growth. Data analysis of
trajectories, plume spread (and thus cry values) and chemical
transformations are also important.
Meteorology data sets were then discussed by the committee. The
primary concern is that meteorological observations are too sparse
and too infrequent for accurate modeling of winds in the planetary
boundary layer. Current interpolation methods often disregard basic
principles of physics or oversimplify the problem in order to produce
mass-conserving flow fields, for example. The MAP3S/SURE II (radio-
sonde observations) data set is available for about one month's
duration, and data sets from Tennessee Valley Administration (TVA)
and Green River Studies are also available. In the near term,
results of EPA/FAA tetroon trajectories will be available for analy-
sis of transport.
A long term study was deemed high priority by the committee in
order to obtain higher spatial and temporal resolution for the short
373
-------
time scale. Essentially, the study considers establishing about 50
mobile rawinsonde stations (1 to 2 per state) for supplemental obser-
vations. During ten 10-day intensive study periods each in the
Northeast region and in the Southwest region, the 50 stations would
be located to provide a dense rawinsonde network with 3 hour resolu-
tion. A comprehensive 4-D data set (3 space and one time dimensions)
would then be generated for each intensive sampling period. This
data set would be useful in comparing model-generated wind fields
with the higher-resolution observed winds. The preliminary estimate
of cost for the rawinsonde stations is $80K per station for the
equipment, plus $35 per person per diem, plus $100 expendables per
observation. NASA has refurbished instrumentation on about 15
rawinsonde stations; but if these are used, additional work should be
undertaken to automate the data acquisition of these stations. The
justification for this field program should be a National Mesomet
Experiment, of which acid precipitation and visibility are parts.
The committee identified several existing sources of meteorolog-
ical data sets suitable for construction of short-term trajectories.
Presently, data sets from the MAP3S/SURE II radiosonde observations
(raobs) TVA, SDEL and Green River Studies are available, and within a
few years data from the EPA/FAA tetroon release program will be
available.
On a seasonal to annual basis, detailed meteorology or transport
data sets are not available, although the committee expressed the
374
-------
belief that a trajectory climatology on these scales either exists or
can bu generated from existing models in the near term.
B. Existing Models
1. Episode Models
Some method of prescribing the transport and diffusion of pollu-
tants is common to all episode models. There remains the question of
how accurate and precise the transport and diffusion modules of
various models really are, and a major concern of this committee was
to outline the steps necessary to evaluate this aspect.
The capability of models to represent the chemistry, conversion
and removal is highly variable, ranging from virtually no representa-
tion of any of these processes, to decay-type conversion of S02 to
sulfate and similar removal of both species, and ultimately to more
elaborate chemistry submodels with elaborate integration algorithms
to represent the non-linear roles of several species.
In all models, wind-entrained fine particulates are not directly
addressed, primarily because of the great deal of uncertainty related
to this process.
Of great concern in most episode models is the terrain and its
surface types. The complicated geometry is responsible for mesoscale
air flow patterns contrary to those one would expect with even ter-
rain. It is generally recognized that the application of models is
highly site-specific, and in that sense each application of a model
to a different region of interest represents a unique situation.
375
-------
Another concern is the method of interpolating or otherwise ap-
plying observed meteorological data, principally winds, to numerical
models. Several techniques are currently in use, and unfortunately
they more often than not ignore principles of boundary layer meteor-
ology and physics in performing the interpolation.
Finally, models need to be initialized with moisture fields or
precipitation fields if wet conversion and removal processes are to
be considered. The committee recognizes difficulties regarding the
actual prediction of precipitation and the difficulty of interpreting
observations of precipitation with respect to area affected, dura-
tion, and amount of precipitation received.
In order to evaluate regional models the committee identified
long-range tracer studies that are conducted by Savannah River
Laboratory (SRL) and plume dispersion studies conducted as part of
the Sulfur Transport and Transformation Experiment (STATE). To
evaluate the trajectory calculations of the models, the working group
identified the following types of studies which could be performed:
(1) Objective analysis: Comparison of wind fields derived from
a special dense network, such as are being produced by the severe
environmental storms and mesoscale experiment (SESAME) 79 extensive
study and by the MAP3S intensive study, with wind fields derived from
the operational rawinsonde network.
(2) Dynamic analysis: Comparison of wind fields generated by
dynamic models as part of 4-D data assimilation with wind fields
derived from the operational rawinsonde network.
376
-------
(3) Tetroon trajectory analysis: Comparison of actual tetroon
trajectories from studies such as Cumberland and Green River with
trajectories along a constant density surface using wind, pressure
and temperature fields derived from operational or special rawinsonde
networks operated at the time of tetroon release.
(4) Tower data analysis: A comparison of tower meteorological
data with conventional or special period dense network rawinsonde
data may elucidate ways in which tower data can be used to improve
the state of knowledge of the boundary layer.
Finally, the committee suggested that comparison of models with
actual episode cases should be conducted to evaluate model capabil-
ity. Two examples of such were mentioned:
(1) The SURE II experiment, consisting of several episode cases
and one non-episode case.
(2) The NEROS experiment.
2. Seasonal and Annual^ Models
The number of operational models developed for this time frame
is very small, and of these there are several techniques. EURMAP has
been developed for monthly estimates of sulfur deposition in Western
Europe, utilizing 6 hourly meteorological data and essentially per-
forming calculations of the type detailed above (for episode models)
over a 30 day time period.
377
-------
ASTRAP uses a different approach, calculating trajectories at
all starting times during the year and producing a probability dis-
tribution that the trajectory from any point will pass over any given
area.
The committee identified the following seasonal to annual model-
ing needs:
(1) Regional Episode Climatology
(a) Develop methodology
(b) Construct trajectory frequency matrices
(c) Air Quality-Visibility case studies to relate tur-
bidity and visual range to fine particulate concen-
trations
(2) Data Base Preparation
(a) Meteorology
(b) Precipitation
(c) Canadian Emissions
3. Trend (Multi-Year) Models
As with seasonal/annual models, trend models are not very com-
monplace. The committee identified three types of trend models:
(1) Residual for unknown terms
(2) Independent
(3) Stochastic
The modeling needs on this time scale were identified as fol-
lows :
(1) Intercomparison of weighted episode versus budget
approaches as means for obtaining annual averages.
378
-------
(2) Projected emissions from North America are required if
future trends are to be forecast.
(3) Comparison of multi-year wet periods with multi-year dry
periods and normal periods should be made to assess the
natural variability effects on fine particulate trends.
C. Model Development and Refinement
The Fine Particulates Modeling Committee identified the fol-
lowing questions and topics relative to continued regional model
development and refinement:
1. Transport
1.1 What is the best way to incorporate observed meteoro-
logical data into models for simulation of pollutant
transport?
1.2 Is the present observational network sufficiently
dense for long-range modeling transport needs?
1.3 Can dynamic models be used to improve the temporal or
spatial resolution of observational data in support
of regional modeling transport needs?
2. Trans format ion
2.1 We need to understand single reaction conversion
kinetics which are important to the following:
NOX nitrate
S02 -~ sulfate
volatile hydrocarbons *- nonvolatile organics
379
-------
2.2 Smog chamber studies need to be continued in conjunc-
tion with 2.1 to elucidate the behavior of systems of
chemical reactions.
2.3 We would like to know more about how gases and parti-
cles interact during conversion.
2.4 We would like to know more about the ability of
different sized aerosols to enhance gas-to-particle
conversion.
2.5 We need to understand whether adsorbed water vapor on
aerosols is important in gas-to-particle conversion or
catalytic reactions associated with non-cloud aerosols.
2.6 The evolution of the size distribution of the atmo-
spheric aerosol should continue to be investigated
through studies of aerosol dynamics.
2.7 The role of various sizes and chemical species of
aerosols to nucleate fog and cloud drops deserves
continued investigation.
3. Renoya^
3.1 We need a better understanding of the processes of dry
deposition and resuspension suitable for incorporation
into regional models.
3.2 We need a better understanding of precipitation scav-
enging of aerosols and gases applicable to regional
modeling needs.
380
-------
4. Emissions
4.1 Anthropogenic emissions of fine participates are re-
quired.
4.2 Natural emissions of fine particulates and their
precursers are required.
5. Validation Methodologies
5.1 Trajectory evaluation studies should be continued using
tracers, tetroons and observations of plume dispersion
and meander as ground truth.
5.2 Episodic evaluation studies should be conducted,
including collection and management of data bases for
meteorological observations, emission sources, air
quality observations and precipitation chemistry
observations for the region and time sequence of study.
5.3 Methodologies for comparing regional model results and
observational data need to be refined.
5.4 Methodologies for intercomparison of various regional
model results need to be refined.
D. Policy Recommendations
The following policy recommendations were formulated by the Fine
Particulates Modeling Committee during the EPA Workshop on Regional
Air Pollution Modeling:
1. Additional study of modeling of long-range transport is
recommended. Studies should focus on Lagrangian data bases-tetroons
381
-------
or appropriate tracers-where available. Models should be subjected
to well-defined sensitivity analysis and compared to observational
data as integral steps in the model development/improvement process.
2. A national Mesomet network is recommended, consisting of
about 50 mobile rawinsonde stations. Regional modeling topics would
occupy part of the network's deployment time. Regional modeling
study periods would be 10 days long, with the Mesomet grid size 20-50
km. Results should determine applicability of the operational upper
air network of the NWS to regional modeling needs and the feasibility
of higher resolution data, and the data will be used for sensitivity
and case study tests of regional models.
3. A four-dimensional data assimilation program is recommended
for the purposes of testing and validating long-range transport
models.
4. Development of dynamic models of the mixing layer is recom-
mended for future simplification (parameterization) and inclusion
into long-range transport models.
5. Homogeneous chemical reactions and homogeneous (gas phase)
systems need to be better characterized. Use of smog chambers and
chemical kinetics experiments and theory development are recommended.
Emphasis should be on NQjj, S02 and hydrocarbon species.
6. Heterogeneous chemical reactions, involving gas-to-particle
conversion or mixed phase reactions, need to be better characterized.
382
-------
7. Field study data should be analyzed to determine whether
fine particulates in rural areas are the result of primary emissions
or atmospheric conversion, anthropogenic or natural emissions, or
local sources or remote sources.
8. Improvement of emission inventories of anthropogenic fine
particulate emissions are recommended.
9. Natural emissions of fine particulates and their precursers
are not adequately known. Systematic studies of these, with the ob-
jectives of ultimately establishing emission rates, are recommended.
4.2 Specific Recommendations
The recommendations of the Fine Particulate Modeling Committee
are divided into four broad types of research: data analysis, meteo-
rological modeling, field studies, and laboratory studies.
A. Data Analy s i s
Although field studies are expensive and require considerable
resources and planning to perform, data analysis can be performed
with rather modest resources by small groups. Data analysis can
guide researchers by indicating correlations among variables, sug-
gesting simplifications to theory, and locating experimental short-
comings which should be corrected in succeeding studies.
Three recommendations were made regarding data analysis. The
first is designed to fill a void in our present state of knowledge
regarding the sources of rural fine particulates. The second study
383
-------
focuses on the dry deposition-resuspension mechanisms and their
effects on aerosol chemistry. The third study focuses on the dyna-
mics which govern the aerosol size distribution.
RECOMMENDATION 1: Mass Balance of Observed Fine Particulates
(Rural Sites). Data sets archived by large-scale studies such
as SURE II and the FSU Streaker study should be screened for
aerosol chemistry data at rural sampling sites. These data,
coupled with meteorological information for the times of aerosol
sample collection and basic emissions inventories of the most
proximite urban sources, may be useful in gaining insight into
such questions as the contributions of anthropogenic primary FP
emissions, natural emissions, and secondary aerosol formation to
total rural FP concentrations. (1 year; $75,000; Immediate.)
RECOMMENDATION 2: Chemical Species Distribution. Aerosol
chemistry data sets archived by large-scale studies such as SURE
II and the FSU Streaker study should be analyzed to study the
spatial distribution of various chemical species in fine
particulates. The length scales of various species (or
alternatively the spatial correlations) will indicate the scales
of transport and hence the relative roles of removal mechanisms
on various chemical species components of FP. (1 year;
$375,000; Immediate.)
RECOMMENDATION 3: Aerosol Particle Dynamics. The rate of
change of an aerosol size distribution is governed by several
mechanisms: sources, coagulation, dry deposition, and so forth.
Data analysis of results from studies which have collected
aerosol size distributions should be conducted. Data from
VISTTA and California are believed to be of sufficient quality
to permit rate of change analysis. Use of theoretical models
should be made to evaluate the roles of coagulation,
gas-to-particle conversion, diffusion and dry deposition on the
size distribution and mass concentration of the atmospheric
aerosol. (5 years; $750,000; long-term.)
B. Meteorological Modeling
The Fine Particulates Modeling Committee devoted considerable
attention to various aspects of numerical modeling of meteorological
processes in the planetary boundary layer. Meteorological processes
384
-------
are responsible for the transport and diffusion of all pollutants,
including FP and their gaseous precursers, and are important com-
ponents of FP removal mechanisms (wet and dry deposition).
Radiosonde observations of the atmosphere are conducted twice
daily at several hundred stations in the U.S. The stations are
spaced roughly 200-300 km apart. The spacing is determined by the
resolution necessary to resolve the synoptic features of the state of
the atmosphere. The intention is for weather forecasting for a time
period of 12 to 36 (or 72) hours beyond the observation time.
Air pollutants, including FP, are transported primarily in the
planetary boundary layer (PBL), which extends from the surface to
about 1-2 km. The PBL is a buffer zone between the surface and the
geostrophic (essentially invicid) flow aloft. The winds in the PBL
are greatly influenced by geography through shear stress and heat
flux. Because of this, observation stations at 200 km intervals are
largely incapable of satisying air pollution regional modeling needs.
The first series of recommendations focuses on evaluation and
further development and refinement of dynamic models which, it is
hoped, can bridge the time-space data gap and produce meteorological
fields with the appropriate resolution for regional modeling needs.
RECOMMENDATION 4: Generation of Mixed-Layer Wind Field by a
Dynamic Model. Use of an existing model which incorporates
dynamic and thermodynamic principles of meteorology (as opposed
to interpolation schemes or conservation of mass models) can be
made to provide wind data for the mixed layer on a fine grid (20
x 20 km). This recommendation calls for the generation of three
or four-day periods with output at hourly intervals. It does not
385
-------
include the analysis or refinement phase. (1 year; $135,000;
Immediate.)
RECOMMENDATION 5; Evaluation of Dynamic Model Accuracy.
Existing dynamic models can produce pressure, wind and
temperature fields given atmospheric observations. Several
field experiments involving tetroon tracking (Green River, FAA)
provide existing data bases for generating tetroon trajectories
and meteorological fields. The recommended study involves
comparison of (1) synthesized tetroon trajectories (isopycoric
transport) derived from dynamic model output, and (2) actual
tetroon trajectories. An evaluation procedure in this manner
can estimate the accuracy (amount of bias) and precision of
dynamic models used to produce transport winds for regional
models of air pollution. (2 years; $225,000; Immediate.)
RECOMMENDATION 6: Dynamic Model Refinement Program. In order
to improve and optimize the meteorological modules of regional
air pollution transport models, such a program should be
developed and implemented. Principal objectives are to
investigate empirical representations of more complicated
physical mechanisms, to systematically investigate the reasons
for bias and to correct these insofar as is practical, and
overall to promote meteorologically reasonable modules for
regional transport models. (2 years; $225,000; Immediate.)
RECOMMENDATION 7: Four-Dimensional Data Assimilation Testing.
Four-dimensional meteorological data sets (3 spatial dimensions
plus time) are compiled from observations and dynamic model
simulations of atmospheric processes in order to test and run
regional transport models. A major question arises in the
optimization of meteorology generation modules regarding how
well the model reproduces the real world situation. A
comprehensive program is required in order to establish the
methodology for comparing model-generated data with observed
data, realizing that observed data will not be available at
precise grid points and will contain inherent instrumental
error, and that the observed data cannot be simply interpolated
to grid points without destroying some of the information
content. (3 years; $375,000; Near-term.)
Additionally, two modeling recommendations focus on specific
meteorological problems associated with FP.
RECOMMENDATION 8; Planetary Boundary Layer Models with
Nighttime Conditions. Several models exist which simulate the
evolution of the planetary boundary layer through the course of
386
-------
the night. Several physical features of the structure of the
PBL are important to regional transport of pollutants. This
recommendation calls for a comparative study of PBL models in
order to develop model refinements and optimization. (2 years;
$150,000; Near-term.)
RECOMMENDATION 9; Chemical Deposition Modeling. Using the
results from a study proposed by Recommendation 2 and from
dry deposition field studies, a modeling effort may be able
to assess dry deposition effects on FP and to explain natural
source and sink rates of FP and their precursers. Major con-
sideration should be given to chemical fractionation caused
by airy deposition and resuspension. (2 years; $225,000;
Near-term.)
C. Field Studies
Two major deficiencies were especially noted by the Fine Partic-
ulates Modeling Committee in the area of our present understanding of
meteorological processes. One, mentioned previously, relates to our
general lack of sufficient data to characterize and understand flow
within the planetary boundary layer. The data void from lack of
adequate field studies is partially responsible for our inability to
evaluate mesometeorological models for accuracy and precision.
Specifically, better spatial and temporal resolution is required in
order to analyze meteorological variables of importance in the re-
gional transport of FP.
Second, there is a specific need for better observational data
concerning meteorology in complex terrain. Not only are sources of-
ten located in areas of geographical relief, but also the long-range
transport of pollutants is influenced by the effects of high-relief
terrain on air flow and other meteorological variables.
387
-------
Two recommendations concerning field studies cover these issues.
RECOMMENDATION 10: Implementationofa National Mesometeoro-
logical Program. A national program should be established,
consisting of approximately fifty mobile rawinsonde stations
capable of being located alternatively in the Northeast and
the Southwest for a total of ten 10-day intensive observation
periods in each region. This program should not be exclusively
dedicated to regional transport meteorology, but should be used
for a variety of research topics. This program should be con-
tingent upon implementation of Recommendation 7. The data sets
produced by high resolution observations can be used to check
both the applicability of low resolution, operational observa-
tions as input to dynamic models and the accuracy of output
from the dynamic models. (5 years; $12,500,000; long-term.)
RECOMMENDATION 11: Meteorology in Complex Terrain. The mete-
orological variables of most importance in local impact and
regional transport due to complex terrain are the temperature
and wind and turbulence fields, particularly during the night.
The use of appropriate surface observations, towers, tether-
sondes and remote sensing devices can greatly improve our
knowledge of how complex terrain affects pollutant transport.
(2 years; $1,200,000; Near-term.)
D. Laboratory Studies
In addition to the above recommendations for data analysis,
meteorological modeling and field studies, the Fine Particulates
Modeling Committee made two recommendations which primarily involve
laboratory studies.
Fine particulates are formed by several mechanisms. Most pre-
dominantly, they are formed either by condensation of fumes immedi-
ately after emission or by chemical reactions within the atmosphere.
Global emissions inventories and analyses suggest that the secondary
aerosol particle formation rate from anthropogenic gaseous precursers
is more than double the primary particulate emission rate. Further-
more, although yet unproven, major regional episodes of FP may often
388
-------
be the result of secondary aerosols, since high sulfate aerosol mass
concentrations are frequently associated with such episodes.
Both of the recommendations concern gas-to-particle conversion.
The first recommendation involves homogeneous kinetics and the second
involves heterogeneous conversion.
RECOMMENDATION 12: Kinetics of Homogeneous Gas-to-Particle
Conversion^ Gas-to-particle conversions may occur homogeneously
(all reactants of the same phase) at sufficient rates to provide
significant secondary sources of FP. The following overall re-
actions are of most current interest:
S02
NOV
S04
N03
Volatile Hydrocarbons -» Non-Volatile Organics
Studies of kinetics of homogeneous reactions should be conducted
in smog chambers, investigating overall gas-to-particle conver-
sion under controlled conditions, and using chemical kinetics
to resolve the pathways and rates of the various reactions noted
above. (5 years; $1,900,000; Long-term.)
RECOMMENDATION 13: Heterogeneous Gas-to-Particle Conversion.
This study should involve the overall reactions noted in
Recommendation 12 and should consider the roles of nucleation,
scavenging and aqueous system chemistry in the conversion of
gases to particles. The study should incorporate both labora-
tory experiments and theoretical models of heterogeneous chem-
istry in order to assess the roles of clouds and precipitation
in secondary FP formation. This latter study is especially
important because of the increasing urgency of the acid rain
(atmospheric acid deposition) problem. Current models of het-
erogeneous chemistry are largely too simplistic to explain ex-
isting experimental results. (5 years; $4,750,000; Long-term.)
389
-------
-------
5.0 ACID-DEPOSITION MODELING WORKING GROUP RECOMMENDATIONS
This section summarizes the discussions, conclusions, and recom-
mendations of the Acid-Deposition Modeling Committee, at the EPA
Workshop on Regional Air Pollution Modeling, held in Port Deposit,
Maryland during October 29-November 1, 1979. Members of the com-
mittee were:
Paul Altshuller
EPA - ESRL
Carmen Benkovitz
BNL
Ronald Drake
Battelle N.W.
Jeremy Hales
Battelle N.W.
Nick Hefter
NOAA - Air Resources Lab
Robert Hodanbosi
Ohio EPA
F. L. Ludwig
SRI, International
Paul Michael
BNL
M. Mills
Teknekron, Inc.
P. K. Misra
Ontario Ministry of the Environment
Robert Papetti
EPA - OEPER
Joe Wisniewski
MITRE
391
-------
During the Workshop the committee reviewed current national
needs in the field of pollutant-deposition research, and utilized
this information to formulate the following central research goal for
the next five year period:
To predict, on the basis of continental emission sources
and meteorological data, spatial and temporal distributions of
the atmospheric deposition of sulfate, nitrate and all other
chemical species important in the context of the overall ion
balance.
The approach of this report will be to outline the discussions
leading to this objective, following the chronological sequence of
the meeting. Research and budgetary recommendations appear in the
final sections of the section.
5.1 Background
A. Wet vs. Dry Removal
At the outset it is important to consider the individual im-
portance of wet and dry deposition processes, so that some idea of
relative emphasis can be obtained. The committee noted the following
items:
Relative importances of dry and wet deposition depend upon
pollutant species and vertical distributions in the atmo-
sphere, as well as a number of site-specific features.
Owing to measurement difficulties, the magnitudes of dry
deposition are hard to estimate.
Existing regional models predict roughly equal importance
of wet and dry removal of sulfur compounds.
Because of these features, the committee is inclined to apply
equal emphasis to both wet and dry removal processes. We note that
392
-------
an EPA-sponsored dry-deposition workshop will be conducted during
December 1979; and thus we will defer considerations involving dry
removal to that meeting. The committee asserts, however, that dry
removal processes cannot be neglected in the context of regional
pollution budgets. This is a highly important aspect of pollutant
fate, and should be emphasized strongly in any meaningful program on
regional deposition.
B. Aspects of Deposition Modeling Preliminary Analyses
1. Governing Equations and Boundary Conditions
The mathematical modeling of wet and dry deposition processes
is largely similar to general air quality modeling. Such models are
developed around material balances for multicomponent systems re-
presenting pollutant species in the atmosphere, and the resulting
differential or integral equations are solved to obtain the desired
result. There are significant additions, however, which often must
be included in modeling analysis involving deposition phenomena;
these are summarized in the following itemization:
1) Addition of disperse phases to the system. The presence of
condensed water adds a second, disperse phase to the atmo-
spheric system, and this additional phase (or phases) must
be accounted for mathematically in any model of the wet
removal process. This usually is accomplished by includ-
ing further differential equations in the model, which are
coupled with the original, gas phase equations.
393
-------
2) Interphase transport. The wet removal process is accom-
plished via physical transport of pollutant from the gas
phase to the aqueous phase (s), and additional terms must
be added to the equations governing the model to account
for this phenomenon.
3) Aqueous-phase reactions. Presence of the condensed aqueous
phase presents the likelihood of additional chemical reac-
tions, occurring within the aqueous medium. Such reactions
require further terms to be added to the governing equa-
tions.
4) Ground-level boundary conditions. Inclusion of dry depo-
sition in an air quality model usually demands that more
elaborate boundary conditions be formulated to describe
pollutant removal at the surface. While rather straight-
forward in principle, such conditions may create diffi-
culties by introducing unknown parameters and/or adding
instabilities to the system.
5) Complex flow fields. Because of condensation, complex storm
dynamics, and required resolution of micrometeorological
phenomena, considerations of flow fields become much more
complex in the context of deposition modeling. This situ-
ation requires the resolution of the conservation of mass,
energy and momentum equations.
394
-------
6) Cloud physics. Droplet formation and growth depend on
available nuclear and relative humidity. Clouds are
inhoraogeneous, time dependent entities which are difficult
to model. Precipitation patterns are highly time and space
dependent on scales much smaller than synoptic observing
networks. Modeling of wet deposition lacks, therefore, the
necessary data base for verification of small scale models.
The above entities make deposition models more complex than
clear air trans formation models. Much of the discussion of the
committee addressed the special problems related to these factors.
2. Scales of Time and Distance
Time and distance scales are important considerations in the
formulation of any numerical model of deposition processes. The
temporal and spatial resolution of a phenomenon dictates the mesh
size and time steps in numerical models which in turn determine
paramaterization and the resolution of field measurements.
The major needs discussed by this committee were:
» The temporal resolution required to address phenomena; and
The spatial and temporal resolution needed in atmospheric
deposition models.
It was concluded by the committee that the effects community
should be consulted for specification of temporal resolution require-
ments. Furthermore, it is probably unrealistic to expect concise
effects information until the experimental programs of EPA and other
funding groups are concluded.
395
-------
On the other hand, knowledge of atmospheric removal processes
indicates specific temporal resolution requirements. For example,
field and modeling results of wet removal of pollutants by frontal
storms show complex variability of rain chemistry during storm
passage. This suggests that subevent temporal resolution, of the
order of ten minutes, is required for TOO del/development purposes.
Two major points are important in this context. First, a range
of scales is required for any successful deposition modeling program.
Pertinent atmospheric processes will be investigated using finely
resolved models, which will in turn produce more efficient and prac-
tical coarser grid models. Thus, a range of time scales from ten
minutes to several hours is anticipated. Second, once temporal re-
solution is specified, spatial resolution requirements are automa-
tically defined. This is due to the relationships among transport
distance, time scale model stability and model accuracy. Subsequent
committee requirements were based on the assumption that these re-
solution requirements would sufficiently characterize the deposition
processes so that the central research goal stated in 5.0 would be
fulfilled.
C. Existing Wet-Deposition Modelj^
To achieve the goal stated in 5.0, it is important to consider
current modeling capabilities of wet removal processes. The commit-
tee therefore assembled a taxonomy of existing precipitation chemis-
try models.
396
-------
The classification in Table 5-1 consists of three major groups.
The first group comprises scavenging models which are thus pollutant
material balances and which accept specified thermodynamic and veloc-
ity fields. The second group consists of more complex models which
couple the pollutant material balances to environmental parameters
through the momentum and energy equations. The final class of models
contains the regional modeling schemes that include wet and dry re-
moval in a highly parameterized form. The models discussed elsewhere
is this workshop typically fall into this third category.
In reviewing existing models with reference to wet and dry
removal, the committee noted four deficiencies:
Limited species representation. Although there are a few
exceptions, current deposition models are limited to a small
number of pollutant species. This situation must be improved
to satisfy the overall ion balance in wet and dry removal
processes.
Inadequate characterization of transformation chemistry. At
present, wet removal models utilize highly uncertain parame-
terizations of aqueous phase transformation chemistry. The
primary reason for this situation is due to our uncertainties
concerning kinetic mechanisms and rates. This situation must
be rectified before significant modeling progress can be
made.
Inadequate characterization of interphase transport of par-
ticulate material. The interphase transport step of the
scavenging process is often difficult to characterize, be-
cause of the complexity of attachment mechanisms as well
as uncertainties involving size distribution and chemical
compositions of natural aerosols. Significant room for
improvement in current models exists in this area.
397
-------
TABLE 1: SUMMARY OF EXISTING TYPES OF WET-REMOVAL MODELS
I. Models involving pollutant material balances only
A. "Below-cloud" scavenging
1. Aerosol scavenging
2. Nonreactive gas scavenging
3. Reactive gas scavenging
B. "In-cloud" scavenging
1. Integral material balances (nonreactive)
2. Differential material balances (nonreactive)
II. Material - momentum - energy balance models
A. One dimensional, time varied
B. Multidimensional
III. Composite regional models
A. Trajectory models
B. Grid models
398
-------
Inadequate representation of complex flow fields. Wet re-
moval processes involve convective systems, frontal surfaces,
and other complex flow phenomena. Characterization of these
flows in models is very difficult, and improvements are re-
quired in this area.
5.2 Specific Recommendations
The recommendations are divided into two areas. There are a
host of recommendations related to wet deposition and a single recom-
mendation for dry deposition. The wet deposition recommendations are
divided into three types of studies: laboratory experiments, field
studies and projects involving modeling and data handling.
A. Laboratory Experiments
As detailed above, there remains much basic research to be done
in order for us to understand the chemical evolution of anthropogenic
emissions from reactive gases to aerosol particles or aqueous solu-
tions. Presently, our ability to evalate the conversion of S(>2 or
NOjj within clouds is not acceptable, primarily because we do not
fully understand the roles of metallic catalysts, dissolved gases and
free radicals in aqueous solutions. The first three recommendations
address the problems, focusing on S02, NOX and ammonia.
RECOMMENDATION 1; S02 Oxidation Mechanisms in Aqueous
Solutions.There is need to continue and expand studies
designed to elucidate SC>2 reaction rates and paths in aqueous
solutions. Modeling needs focus on representing the overall
conversion rate as a function of temperature and concentrations
of various species, including S02, 03, H202, H+, trace
metals, organics and carbonaceous materials. (5 years, $375K/
year).
399
-------
RECOMMENDATION 2; Mechanisms Leading to the Occurrence of
Nitrate in Precipitation. Studies should be undertaken to
investigate the various routes through which NO and N02 may
travel to ultimately arrive at the ground as nitrate in
precipitation. Some suggested routes are scavenging of HNO},
PAN (peroxyacylnitrate) and other organic-nitrate compounds and
nitrate aerosols, and oxidation of dissolved N02 in aqueous
solutions. Specific attention should be paid to transformation
of NO and N02 to particulate nitrate as a component of the
overall flow of nitrogen through various pathways. (5 years,
$300K/year).
RECOMMENDATION 3: Mechanisms Affecting the Presence of Ammonia
in Precipitation. Recent studies indicate that most previous
work on the role of NH3 in the conversion of S02 to sulfate
in aqueous solutions may be incorrect. Further studies are
needed to eluicade the role of NH3/NH4+ in the conversion
of S02 to sulfate, and similarly to investigate the role of
NH3/NH4+ on conversion of NOy to nitrate. (3 years,
$150K/year).
The above three recommendations are all given high priority,
since they collectively represent the major shortcomings in theory
relative to modeling needs in the area of transformation chemistry.
In the next recommendation, all four components are given moderate
priority. In some cases, existing methods may not be well-suited to
the problem at hand. In other cases, no instrumentation is presently
availab le.
RECOMMENDATION 4: Development of Monitoring Equipment.
Laboratory development, testing and calibration of monitoring
equipment designed for field applications should continue.
Several types of instruments are given higher priority for
development. These include:
a. Cloud Water Sampler: This must be able to separate
aerosols from cloud droplets while collecting enough
cloud water within a short time to allow chemical
analysis. Its need is principally to elucidate cloud
chemistry - how quickly are freshly-entrained aerosol
particles scavenged; how does cloud water solute vary
400
-------
from aerosols, chemically, and is this due to variable
nucleation, scavenging, or aqueous conversion of gases?
b. Measurement of Exotic Species: Chemical kinetics can
give reaction rates for modelers, but unless measurements
indicate that certain chemical species are present, and
in what concentrations, transformation chemistry modeling
is left in the dark. Some species, which are presently
not measured in clouds, but which are believed to be
important in precipitation chemistry, include
and HN03.
c. Size-Resolved Aerosol Chemistry Measurement;. A better
size-resolved aerosol sampler, which is compatible with
chemical analyses, is needed in order to characterize
the size distributions of individual chemical species,
and conversely to determine the chemical make-up of var-
ious size fractions of the atmospheric aerosol. Field
measurements using such samplers are needed to better
determine the sources of particulates.
d. Cloud Physics Measurements: Although there are existing
methods for measuring various cloud physics parameters,
detailed investigations related to the evolution of acid
rain require more sensitive instrumentation, such as a
better cloud water sampler (see a, above) and a reliable
condensation nuclei counter.
In total, the development testing and calibrating of var-
ious instruments is an integral part in systematically investi-
gating the physical mechanisms relevant to acid precipitation.
(3 years, $225K/year).
B. Field Studies
The greatest number of recommendations made by the Acid Deposi-
tion Modeling Committee involve field studies. These recommendations
focus on several areas. One area of interest involves collecting and
analyzing appropriate data on precipitation chemistry for long-range
transport model verification. Another area involves first-look
studies of the roles of special events (dew, frost and fog) in acid
deposition. Another area of interest involves material balance and
401
-------
chemical balance studies under both cloud-free and cloudy conditions.
The final area of interest involves cooperation with other field
studies and with agencies operating existing facilities.
1. Precipitation Chemistry Monitoring Network and Data Bank
Recently, several precipitation chemistry monitoring networks
have been established to help gain insight into the extent, magnitude
and chemical nature of acid precipitation. The recommendations made
in this area are all given high priority, because they are necessary
to insure useful, quality assured data located in a depository which
is convenient for users to access.
RECOMMENDATION 5. Quality Assurance Methodology. There is a
considerable need to develop and implement a consistent quality
assurance (Q/A) methodology for all precipitation chemistry
networks. When the various data sets begin to be merged by a
central data bank (see Recommendation 6) or by individual users,
situations of improper handling or measurement of trace species
could confound the investigation. Q/A methodology should
address the following topics:
1. Site Selection: Each site must be intensively studied to
assure that results from the station are representative of
the area.
2. Type of Collector: Acceptable collector types for
precipitation chemistry analysis should be prescribed.
3. Sample Handling: The period from the end of the sampling
time until the end of chemical analysis must be minimized
and proper handling and storage procedures must be
developed.
4. Chemical Analysis: Acceptable procedures for analyzing
various chemical components must be specified and thus
sample pretreatment must be determined.
5. Independent Laboratory Analysis: Procedures for maintaining
high analytical standards need to be developed. These
should include independent laboratory analysis of duplicate
samples on a scheduled basis.
402
-------
6. Quality Checking Procedures: Quality assurance samples
should be periodically passed into the system to assure
consistent field handling techniques, proper sample storage
and accurate chemical analysis. Development of a proper
quality assurance methodology is only the beginning of a
broad commitment in the area of precipitation chemistry
monitoring. The implementation of such techniques will
require considerable resources. (3 years $225K/year).
RECOMMENDATION 6. Precipitation Chemistry Central Data Bank.
Because of the development of several independent precipitation
chemistry monitoring networks, a central data bank is required.
There needs to be development of data storage formats and
encouragement of computer-compatible coding and checking of data
by the originators in addition to getting the project off the
ground at a central depository. (2 years, $450K/year).
RECOMMENDATION 7. Analysis of Recent Precipitation Chemistry
Network Data. In concert with recommendations 5 and 6, an
effort must be undertaken to intensify the systematic analysis
of fresh data which are being continuously generated. Studies
should include:
1. Variable pair correlations
2. Ion balances
3. Factor analysis
4. Time-series analysis
5. Material budgets
6. Modeling analysis
Much can be learned concerning proper siting, handling and
analysis by conducting systematic analyses. This recommendation
should be implemented in a manner as to provide feedback into
the operations initiated under Recommendations 5 and 6. (5
years, $225K/year).
2. Special EyentjL
It is reasonable to investigate wet and dry deposition separate-
ly since they occur under different circumstances, involve different
mechanisms, and result in different impacts on the environment.
Somewhere between wet and dry deposition there is a class of events
which requires study. These special events include fog, dew and
403
-------
frost, and may also be expanded to include drizzle and non-
precipitating clouds. Although the phenomena do not necessarily
result in deposition immediately, they may serve important roles
in the gas-to-particle or chemical conversion of pollutants.
It is not yet known how important special events are in the
overall acid deposition picture. They may, or may not, be signifi-
cant. This lack of systematic study leaves this issue open. The
recommendations below are of moderate priority, except for Recommen-
dation 8, which is high priority.
RECOMMENDATION 8. Chemical Conversion in Clouds. Field studies
to investigate chemical conversion in clouds which do not pre-
cipitate are required. This experiment can benefit from de-
velopment of some of the advanced instrumentation described in
Recommendation 4, Since 90% of all clouds do not precipitate,
we must learn more about their role in transformation of pol-
lutants, such as converting S02 to sulfate. (3 years,
$300K/year).
RECOMMENDATION 9.. Chemical Conversion and Deposition in Fog.
There is a lack of substantial data concerning how fog acts to
scavenge gases or particulates and remove them to the ground.
A field experiment is necessary in order to gain a preliminary
understanding of the matter. (3 years, $225K/year).
RECOMMENDATION 10. Deposition by Dew and Frost. A scoping
study is needed to assess the probable impact of dew and frost
on acid deposition. Again, this area is particularly devoid
of study. A small amount of work to indicate whether or not
further investigations are required. (1 year, $75K/year).
3. Material and Chemical Balance Studies
Three recommendations were made which involve field studies for
the purpose of learning the fates of pollutants which get mixed into
clouds. Although the recommendations have different foci, they could
be implemented concurrently in mutual support of one another.
404
-------
Material balance studies are designed to estimate from a limited
set of measurements, the magnitudes of inputs and sinks of various
pollutants to the atmospheric system. In Recommendation 11, the fo-
cus is on various types of storm clouds with the intent of eventually
approximating acid deposition by developing a storm climatology of
the region of interest. Additionally, comparison of different storm
type material balances will allow modelers to assess whether certain
storm types control the episodicity of acid precipitation.
Tracer studies are speical types of material balance studies,
in that tracer particles (containing environmentally rare elements)
are inserted into the air at known points and may be detected either
in outflow air or in precipitation.
A dual-doppler radar facility is an extremely powerful tool for
analysis of wind motions in clouds. Although Recommendation 13 is
geared for the development of such a facility rather than for any
specific research project, its universal utility will allow it to
contribute considerably to almost any type of field experiment
involving clouds.
RECOMMENDATION 11. Material Balance Field Studies. Field
studies of material balance should be undertaken to evaluate
the behavior and scavenging mechanisms characteristic of re-
presentative storm types (convective, cyclonic and orographic).
Design of such experiments should be coordinated with other
large scale projects, such as weather modification, in order
to optimize coverage and data acquisition. The primary inter-
est is to inventory pollutant concentrations in air entering
and exiting clouds, and also in air beneath clouds through
which precipitation wil1 fall. In concert, sampling of cloud
droplets at various points within the cloud, rain drops at
405
-------
cloud base, and rain at the ground can give data concerning
the material fluxes of pollutants through the cloud.
RECOMMENDATION 12. Tracer Studies. There is need to continue
development of suitable tracers and their release and monitoring
in the field. For cloud studies, a variety of water soluble and
water insoluble tracers are desired. For long-range transport
studies, there is need to develop relatively stable gaseous
tracers which can be detected in extremely small quantities.
RECOMMENDATION 13. Dual-Doppler Radar Facility. A dedicated
dual-doppler radar facility should be developed for use with
field studies involving clouds, such as developed in recommen-
dations 11 and 12. While this recommendation involves much
capital expense, the additional information is required to
properly interpret other data collected during the experiment.
Recommendations 11, 12 and 13 are all given high priority des-
pite their considerable expense and, in the case of Recommendation
11, difficulty in performing satisfactorily. The high priority re-
flects the tremendous need by the modeling community for field data
concerning the actual transport, transformation and removal of pol-
lutants.
4. Cooperation
This final group of two recommendations in the area of field
studies deals not so much with necessary research as it does with
optimizing field studies proposed in previous recommendations and,
stated frankly, getting more bang for the buck. These recommenda-
tions are offered as guidelines, to be pursued whenever possible.
RECOMKENDATION 14. Use of Other Existing Facilities. Effort
should be taken to maximize the use of existing, instrumented
aircraft, NWS forecasts and forecasting expertise, satellite
resources and other facilities in performing field stuides.
406
-------
RECOMMENDATION 15. Coordination With Other Field Studies.
Ongoing field studies in severe storms and weather modification
deploy a considerable amount of instrumentation in support of
those objectives. With minimal financial burden on either side,
coordinated precipitation chemistry field experiments could be
mutually beneficial.
C. Modeling and Data Handling
This final area of wet deposition recommendations deals with the
focus of the committee - acid deposition modeling. Since the long-
range transport of pollutants is intimately coupled to chemical and
physical transformations and removal processes, the entire spectrum
of models is considered. The recommendations are divided into two
groups: (1) model support and (2) model development and improvement.
1. Model Support
Two recommendations were made regarding support of regional
modeling. They are both rated high priority although they will
involve considerable efforts to perform.
RECOMMENDATION 16. Emission Inventory Maintenance. Source
terms are crucial to accurate modeling of acid deposition.
Therefore, it is necessary to maintain and continuously update
currently available emission inventories, both in the U.S. and
Canada. In support of this recommendation, the data base should
be centralized at one user-accessible depository. Pollutants
inventoried should be expanded beyond particulates and SC>2 to
NOX and other criteria pollutants.
(5 years, $150k/year).
RECOMMENDATION 17. Model Applications/Validations Data Sets.
Model testing and validating require comprehensive sets of
meteorological, air quality and precipitation chemistry data for
a specified time period. A universally available archive of such
data sets should therefore be established, and if all data sets
should be stored in a uniform format. The data sets should
include emission inventories, surface and upper air
meteorological observations, precipitation chemistry data and
air quality data. (5 years, $300K/year).
407
-------
2. Model Development and Improvement
Five recommendations were made regarding models themselves. The
recommendations reflect the spectrum of present model development.
RECOMMENDATION 18. Comprehensive Chemistry Modeling. Since
the pH in precipitation really reflects the sum of the complex
chemical make-up of the individual drops or snowflakes, a com-
prehensive chemistry model is required to investigate the ion
balance of precipitation. The model must incorporate the complex
chemistry of aqueous solutions, including solubility of gases,
catalyzed conversion rates, and dissolution factors, and should
incorporate the relevant sulfur-nitrogen and halogen-containing
species, carbonate, the major anions (Na"1", Ca"1""*", etc.) and
the hydrogen ion. (5 years, $300K/year). (Moderate Priority).
RECOMMENDATION 19. Development of a Comprehensive Regional
Deposition Model. This model is required since it is viewed as
the prime tool for estimating source impacts on specific loca-
tions. It is given high priority. The model should estimate wet
and dry deposition of all chemical species which are important
to the overall ion balance. It should be a grid model, using
structured programming techniques. (5 years, $225K/year).
RECOMMENDATION 20. Parameterized Regional Deposition Modejls.
Parallel with the implementation of Recommendation 19, there is
a need to develop simplified regional deposition models, using
approximations and parameterizations whenever such can be im-
plemented to reduce computer resources needed, while not greatly
degrading the accuracy of the results. This is given moderate
to high priority. (5 years, $150K/year).
RECOMMENDATION 21. Statistical-Climatological Regional
Deposition Modeling. An alternate method to comprehensive
regional deposition modeling and probably a method which could
be implemented more quickly if given equal support, is develop-
ment of statistical-cliraatological regional deposition models.
One method of approach is to compile an atlas of the frequency
of individual storm types and then use experimental results of
storm scavenging (Recommendations 11 and 12) and detailed nu-
merical models of storms (Recommendation 22) to estimate over-
all storm scavenging characteristics. The end result is a
statistical picture of precipitation chemistry. This is given
moderate priority. (3 years, $150K/year).
408
-------
RECOMMENDATION 22. Detailed Storm Models. Existing or modified
detailed numerical models of storm dynamics and physics can in-
corporate scavenging mechanisms and yield overall storm scav-
enging characteristics for use in Recommendation 21. Existing
storm models can also be used in conjuction with field studies
(Recommendations 11 and 12) for experimental diagnosis. (2
years, $150K/year). Moderate Priority.
D. Dry Deposition
The committee agreed to a single recommendation regarding dry
deposition.
RECOMMENDATION 23. Dry Deposition. All recommendations regard-
ing dry deposition would be deferred to a workshop sponsored by
EPA held at Argonne National Laboratory during 4-5 December,
1979.
409
-------
-------
6.0 VISIBILITY MODELING WORKING GROUP RECOMMENDATIONS
This secion summarizes the conclusions and recommendations of
the working group on visibility modeling of the Workshop on Regional
Modeling held at Port DePosit, Maryland on October 30 - November 1,
1979. The members of the working group were:
Arthur Bass
ERT, Inc.
William Eadie
BNL
Steve Eigsti
EPA - OAQPS
Mark Eltgrath
Univ. of Washington
Robert Henderson
The MITRE Corp.
Douglas La timer
SAI
James McElroy
EPA - EMSL
Michael Williams
DOE - Los Alamos
William Wilson
EPA - ESRL
6.1 Background
The Clean Air Act, as amended, requires the Administrator of the
Environmental Protection Agency (EPA) to take steps to prevent any
man-made impairment of visibility in mandatory Class I Federal areas.
To this end, models must be developed which relate visibility impair-
ment to specific sources and human activities. The working group on
411
-------
visibility modeling was convened to address the question of what
research is required for the development of such models.
Visibility impairment is primarily due to the existence in the
atmosphere of fine particulates in the size range 0.1 urn to 1.0 um.
In addition, gases, particularly N02, may contribute to visibility
impairment, however NC>2 does not exist in sufficient concentrations
to be significant except relatively near sources such as fossil fuel
plants and urban areas. The problem of modeling of visibility on a
regional scale is thus primarily the problem of modeling fine parti-
culate concentrations at large distances from sources.
The visibility modeling working group made a set of assumptions
concerning the problem of regional visibility modeling, viz:
Their concern was for modeling at distances larger than 100
km. The near source problem, i.e., the plume blight problem,
is the subject of current studies and it was assumed that
modeling of plume blight would be well developed in the near
future. The resulting plume blight models would be used to
initialize regional models.
Their concern was for modeling visibility impairment in the
western U.S. The parallel working group on fine particulate
modeling would be developing recommendations for research on
models of fine particulates in the Eastern U.S. Submodels
connecting fine particulate concentrations to visibility
could be applied equally well for Eastern or Western regional
problems. In addition it was noted that measurement of visi-
bility degradation was underway in the Eastern visibility
study.
The 1980 research budget was relatively fixed and thus their
research recommendations would apply to FY 1981 and beyond.
In addition to these assumptions the visibility working group
adopted the following guiding principles:
412
-------
The objective of visibility modeling is to provide guidance
for regulatory decisions on siting and emissions control.
This includes predictive modeling of visibility impairment in
class I areas as a result of future energy and other develop-
ments.
There is a regulatory need for visibility models in the near
future. Fully developed models which include all important
physical processes will not be available for at least the
next five years, thus, simplified models will have to be
employed. The shortcomings of these models will have to be
understood and well documented. Model development will have
to be evolutionary.
Model development is constrained by the lack of sufficient
data. Extensive field studies will be required to provide
data for model development and application.
6.2 Specific Recommendations
A. Development of a Visibility Perception Criteria Document
While the Clean Air Act requires that perceived visibility
impairment be reduced and prevented any attempt at modeling must be
based on the prediction of objective measures such as optical extinc-
tion coefficient, visual range, contrast, chromaticity, ect. Thus
some method is required to connect these objective measures with
subjective determinations of how a scene appears.
This connection between subjective analysis of visibility
impairment and objective measures of the optical quality of the
atmosphere is necessary if Federal Land Managers are to make use of
model predictions to determine the acceptability of a particular
atmospheric optical quality at a particular place. The physical
parameter which is most readily measured and thus most amenable to
use for model testing and validation is the optical extinction coeffi-
cient which is a measure of the amount of scattering and absorption
413
-------
of light by the atmosphere. The appearance of a particular scene
howevtr will depend, in addition to the atmospheric optical extinc-
tion coefficient, on the characteristics of the scene, the location
of the observer and the amount and geometry of solar insolation.
RECOMMENDATION ]: A visibility criteria document should be
produced. This document would contain photographs of selected
Class I scenes under various atmospheric conditions and illumi-
nation conditions along with associated physical parameters such
as optical extinction coefficient. This document would provide
Federal Land Managers with the data they need to determine if
objectively described atmospheric quality would lead to subjec-
tively determined visual impairment.
Tasks required for the production of such a criteria document
include:
Development of models that relate physical (objective)
measures to subjective indices,
Determination of thresholds of perceptibility and objection-
ability,
Documentation, with selected scenes, of the relationship
between physical measures and the appearance of the scenes.
C 2 years, 150K/yearJ
B. Emissions Data Base
While considerable effort has been put into the development of
comprehensive emissions data bases, additional work is required par-
ticularly if the emissions data base is to support visibility model-
ing.
A detailed inventory of the emissions of various source types,
e.g., power plants, mining operations, synthetic fuel plants, smelt-
ing and urban areas, is required and must include:
414
-------
SC>2 emissions,
NOX emissions,
primary particulates, including size distribution and chemi-
cal composition or refractive index.
soot emissions,
Hydrocarbons by species or class, and
source characteristics.
RECOMMENDATION 2: Field Studies should be performed to deter-
mine size distributions and chemical composition of primary
particulate emissions for each source type. In addition the
variability of these characteristics within each source type
should be determined and the possibility of relating changes in
the size distributions and chemical composition with more read-
ily determined factors such as operation mode or specified
source characteristics such as flow rate or temperature. (2
years, 150K/year)
RECOMMENDATION 3: A continuous refinement and updating of the
emissions inventories, with identification of data gaps, should
be undertaken, (ongoing, SDK/year)
C. Monitoring Networks
Visibility monitoring along with associated measurements of
atmospheric aerosol concentrations and size distributions are
required to determine the background visibility levels and the nature
of and the conditions leading to visibility impairment episodes.
Under the Western Energy Environmental Monitoring Study (WEEMS) a
forty station fine particulate network has been established in remote
areas of North Dakota, South Dakota, Montana, Wyoming, Utah,
Colorado, Arizona and New Mexico. This network uses a dichotomous
sampler which separates the aerosol into the size ranges of approxi-
mately 0.1 to 2.5 urn and 2.5 to 15 urn. A comparison visibility
415
-------
network of 14 stations has been established in conjunction with the
National Park Service. The data from these networks will be analyzed
to determine the spatial variability of the visibility and aerosol
background and the spatial and temporal variability of visibility
impairment episodes.
RECOMMENDATION 4: The data from the WEEMS program should be
utilized to determine the need for additonal monitoring of
visibility and aerosols in the western U.S. to support model
validation and initialization. Recommendations for additional
sites and instrumentation should be developed. (1 year,
80K/year)
RECOMMENDATION5: Analyze the WEEMS data and other available
data to determine the relationship between visibility and
meteorological conditions. The use of trajectory analysis to
determine the source of visibility impairment haze should be
performed and a climatology of visibility in the western U.S.
should be developed. (1 year, lOOK/year)
RECOMMENDATION 6: Analysis of existing monitoring data of aero-
sols should be performed to attempt to determine the relative
importance of natural sources to the regional aerosol loading of
the atmosphere. (2 years, lOOK/year)
In addition to monitoring visibility and aerosols it will be
useful to use existing meteorological measurements to determine the
nature of the wind fields and the temperature structure of the atmos-
phere in the west.
RECOMMENDATION 7: Use existing data to generate a climatology
of various levels of trajectories and a climatology of mixing
height. Also assess the need for additional meteorological mon-
itoring sites and parameters. (1 year, SOK/year)
D. Field Studies and Laboratory Programs
In addition to monitored data there exists a need for a number
of field studies to provide data required for in-depth understanding
416
-------
of various aspects of the problem. In particular it is necessary to
have a better understanding of drainage and channel flows through
complex terrain so that Lagrangian plume models can be used for the
first few hundred kilometers from the source, before the initializa-
tion of an Eulerian grid model. The field studies will also be use-
ful in evaluating the ability to calculate plume trajectories with
existing meteorological data, and the ability of reactive plume mod-
els to account for transformation, dispersion and removal processes.
In addition, the field studies will help to determine appropriate
transformation and removal parameters and how these parameters depend
on terrain, and mesoscale wind and temperature fields.
RECOMMENDATION 8: Perform a series of four week intensive field
studies, with five studies every second year. Years without
field studies will be devoted to analysis of the data from the
previous year. During the field studies tetroons will be
employed to track plume movements for at least one and a half
diurnal cycles. Meteorological data will be augmented in both
spatial and temporal coverage using National Weather Service
(NWS) type radiosondes. Mobile stations will also be employed
to obtain additional wind field data along the plume trajector-
ies. Chemical tracers should be added to the plume to help
determine transformation, and removal processes. At various
points along the plume trajectory aerosol concentration and size
distribution will be measured for comparison with model calcula-
tions. (10 years, 4000K/yearJ
In order to accomplish these field studies successfully, it will
be necessary to improve tetroon technology to permit aircraft and
satellite tracking and constant temperature level flight.
Field studies will also be required to determine the amounts of
natural emissions of HC, SOXJ and NOX and the sources, concentra-
tion, composition, and size distributions of natural aerosols (bio-
genie and soil dust).
417
-------
RECOMMENDATION 9: Perform field studies to determine the source
and characteristics of naturally occurring visibility impairing
hazes or aerosols. (3 years, 500K/year)
The dry deposition of aerosols and gases may be the primary
mechanism for the removal of pollutants from the atmosphere, particu-
larly in the Southwestern U.S. where precipitation levels are low.
For this reason, it will be important to understand the rate of dry
removal of fine particulates, and their precursors, for the develop-
ment of regional visibility models. This is a very difficult problem
and there is currently no acceptable method for measuring the charac-
teristics of dry removal.
RECOMMENDATION 10: All potentially useful dry deposition mea-
surement techniques should be employed, in a controlled manner,
during some of the intensive field studies. The method of use
of the techniques should guarantee the comparability of the
various results. (10 years, 200K/year)
A great deal remains to be learned about the chemistry of pollu-
tants in the troposphere and laboratory studies should be pursued to
help better define the reactions important to gas-to-particle conver-
sion (heterogeneous chemistry). The working group did not make any
specific recommendations for laboratory studies, since such recommen-
dations would not be specific to visibility modeling but would be the
same as those defined by the working groups on fine partlculate and
acid precipitation modeling.
E. Model Development
The modeling of the relationships between emission source char-
acteristics and regional scale visibility impairment is an extremely
418
-------
difficult task. Nevertheless, it is essential that regional haze
models be developed if rational control policies are to be adopted.
A regional visibility model is essentially a regional fine particu-
lates model with the provision that the index of refraction (or chem-
ical nature) and size distribution of particulates in the approximate
size range of .1 urn to 1.0 um are predicted quantities. Once the
nature and size distribution of the fine particulates are known a
visibility module can be employed to compute the optical attenuation
coefficient or other objective optical parameters.
The problem of predicting the size distribution and nature of
fine particulates revolves around the complex question of gas-to-
particle conversion in the atmosphere. In addition to the primary
partlculate emissions, which may remain airborne over regional dis-
tances, the gaseous emissions converted to particulates while air-
borne (e.g., SC>2 £04 particulates). Prediction of the products
of the chemical reactions which affect the gaseous pollutants will
require an understanding of the reactions and reaction rates of the
pollutants and how these are effected by temperature, humidity and
sunlight and trace atmospheric components. It must also be recog-
nized that some reactions may be non-linear under certain conditions.
In addition, reactions which may be unimportant on the scale of the
plume blight problem because of the slow reaction rate, may have to
be included in models on regional scales.
419
-------
Because of the complexity of the problem it will be necessary to
develop evolutionary models which can provide some answers to regula-
tory questions now while still being easily improved as new informa-
tion and understanding is obtained in the future.
In addition to the problems associated with pollutant transfor-
mation, a regional visibility model, particularly in the west, will
require adequate descriptions of the air flow through complex ter-
rain. If, as has been recommended by this working group, the
regional model combines Lagrangian plume models (applicable over the
first few hundred kilometers) and a regional scale Eurlerian grid
model which is initialized by the plume models, then the plume models
will, at a minimum, have to account for the transport of the pollu-
tants through channels and around obstacles. It will thus be neces-
sary to develop improved mesoscale metorological data fields to drive
these models.
In light of the complexities described above the working group
defined a series of recommendations for evolutionary model develop-
ment.
RECOMMENDATION 11: Short term evolutionary model development
should begin with a consolidation of existing chemistry and aer-
osol models into a combined plume and regional model. The plume
portion of the model should be applicable to ranges of approxi-
mately 300 km; the regional portion should be applicable to
ranges larger than 300 km and should have a grid resolution of
25 km. The model should be capable of using three dimensional
meteorological wind fields and should include chemical reactions
involving SC>2, NOX, HC, NH3, OH, 63 and photo reactives.
The aerosol dynamics should include coagulation, sedimentation
and gas-to-particle conversion. Gas and Aerosol deposition to
the surface need to be accounted for. The size distributions of
420
-------
the particles should be calculated in at least three modes. (2
years, 150K/year)
RECOMMENDATION 12: Long-terra evolutionary model development
should include constant updating as the results of field studies
and laboratory experiments are obtained. These developments
should include additional reaction mechanisms, and in-cloud and
precipitation scavenging chemistry. Also improved nucleation
and condensation dynamics should be incorporated. (8 years,
150K/year)
RECOMMENDATION 13: Improved meteorological data fields should
be developed for driving the evolutionary models. These data
fields should be developed for two spatial scales appropriate to
the plume trajectory and regional grid components of the models.
The end products should include spatially variable multi-layered
wind fields, vertical temperature structure and direct measures
of atmospheric stability. The modeling tools required for the
development of these meteorological data fields should be devel-
oped so as to constrain the flow subject to actual terrain
influences. (4 years, 200K/year)
In addition to the development of an evolutionary regional visi-
bility model the working group felt that simpler modeling approaches
should be supported both because of the near term need for some
modeling capability and the likelihood that such models could lead to
greater understanding of some aspects of the physical processes
involved in regional visibility impairment.
RECOMMENDATION 14: Simple modeling approaches to the regional
visibility impairment problem should be continued. .Approaches
such as rollback models based on a statistical analyses of air
mass trajectories and statistical models such as time series
analysis and multi-variate analysis should be supported. (3
years, lOOK/year)
A complete regional visibility model includes, in addition to a
mechanism for computing the transport and transformation of the pol-
lutants, a module for computing the optical properties of the atmos-
phere as they relate to pollutant concentrations, while techniques
421
-------
for computing the optical attenuation coefficient based on detailed
size distribution and index of refraction information are well devel-
oped such complete information may be beyond the capabilities of
regional models. Regional models may, for example, only be able to
predict the total aerosol mass loadings in a small number of (perhaps
large) size ranges. Methods for computing optical parameters from
such a limited amount of information will then be needed.
RECOMMENDATION 15: Develop radiative transfer methods for com-
puting optical parameters from reduced inputs as may be avail-
able from regional models. This should include an analysis of
the sensitivity of the resulting optical parameters to the reso-
lution and accuracy of regional model outputs and tests of the
accuracy of the models under a number of field situations using
data from the field experiments. (3 years, lOOK/year)
While the working group felt that the use of a visibility cri-
teria document is the preferred method of relating objective optical
parameters to subjective determinations of visibility impairment, a
better understanding of this relationship is needed. Radiative
transfer models should be further developed to allow computation of
such factors as contrast and chromaticity under varying illumination
conditions.
RECOMMENDATION 16: Radiative transfer models should be used to
further investigate the relationship between pollutant concen-
trations and subjective determination of visibility impairment
under a number of illumination conditions for representative
western scenes. The multiple scattering codes used in these
models should be improved and means for shortening run time
should be developed and tested. (3 years, 150K/year)
422
-------
DISTRIBUTION LIST
Dr. Paul Altshuller
Environmental Sciences Research Lab
Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Richard Anthes
Penn State University
University Park, PA
Dr. John Bachman
Office of Air Quality Planning
and Standards
Environmental Protection Agency
Research Triangle Park, NC 27711
Richard Ball
Department of Energy - Germantown
Washington, D. C. 20545
David BalIantine
Department of Energy - Germantown
Washington, D, C. 20545
Mr. Walter C. Barber, Jr.
Deputy Assistant Administrator, MD-10
Office of Air Quality Planning and
Standards
Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Arthur Bass
Environmental Research and
Technology, Inc.
696 Virginia Road
Concord, Massachusetts 01742
Dr. Gordon Bean
Atmosphere Environment Service
4905 Dufferin Street
Downsview, Ontario CANADA M3H574
Dr. William Belanger
EPA, Region III
Curtis Building
Sixth and Walnut Streets
Philadelphia, PA 19106
Dr. Carmen Benkovitz
Brookhaven National Laboratory
Upton, NY 11973
Dr. C. S. Burton
Systems Applications, Inc.
San Rafael, CA 94903
Dr. P. Coffey
New York Department of Environmental
Conservation
50 Wolf Road
Albany, NY 12223
Dr. T. L. Crawford
Tennessee Valley Authority
Muscle Shoals, Alabama 35660
Allyn Davis
Director, Air and Hazardous Materials
Division
EPA, Region VI
First International Building
1201 Elm Street
Dallas, TX 75270
Dr. Ken Demerjian
Environmental Sciences Research Lab
Environmental Protection Agency
Research Triangle Park, NC 27711
Thomas W. Devine
Director, Air and Hazardous
Materials Division
EPA, Region IV
345 Courtland, NE
Atlanta, GA 30308
Dr. Ronald Drake
Battelle Pacific Northwest Labs
P. 0. Box 999
Richland, WA 99352
Dr. Peter Drievas
Environmental Research and Technology, Inc.
696 Virginia Road
Concord, MA 01742
-------
External Distribution List
Page 2
Robert L. Duprey
Director, Air and Hazardous Materials
Division
EPA, Region VIII
1860 Lincoln St.
Denver, CO 80295
Dr. William Eadie
622 R 1200W
Battelle N.W.
Richland, WA 99352
Dr. J. Edinger
Department of Atmospheric Sciences
UCLA
405 Hilgard Avenue
Los Angeles, CA 90024
Dr. Steve Eigsti
Office of Quality Planning and
Standards
Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Stephen J. Gage
Assistant Administrator
Office of Research and Development
Environmental Protection Agency
Washington, D. C. 20460
Jan N. Geiselman, Acting Director
Air and Hazardous Materials Division
EPA, Region II
Federal Office Building
26 Federal Plaza
New York, NY 10007
Clinton Hall
EPA/ORD/RD-682
Washington, D. C. 20460
Dr. Jeremy Hales
Battelle Pacific Northwest Labs
P. 0. Box 999
Richland, WA 99352
Dr. Nick Hefter
Air Resources Lab
NOAA
8060 13th Street
Silver Spring, MD 20910
Dr. Bruce Hicks
Argonne National Lab
Argonne, I.L 60439
Dr. Glen Hilst
EPRI
3412 Hillview Avenue
Palo Alto, CA 94304
Dr. Peter Hobbs
University of Washington
Seattle, Washington 98105
Dr. Robert Hodanbosi
Ohio Environmental Protection Agency
361 E. Broad Street
Columbus, OH 43215
Merrill S. Hohman
Director, Air and Hazardous Materials
Division
EPA, Region I
Room 2303, John F. Kennedy Federal Building
Boston, MA 02203
Dr. Warren Johnson
SRI, International
Menlo Park, CA 94205
David Kee, Director
Air and Hazardous Materials Division
EPA, Region V
230 South Dearborn
Chicago, IL 60604
Dr. Carl Kreitzberg
Drexel University
Philadelphia, PA 19104
-------
External Distribution List
Page 3
Dr. Robert Lamb
Environmental Sciences Research Lab
Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Douglas Latimer
Systems Applications, Inc.
San Rafael, CA 94903
Dr. Tom Laverny
Environmental Research and
Technology, Inc.
Suite 360
2225 Townsgate Road
Westlare Village, CA 91361
Dr. Hiram Levy
Geophysical Fluid Dynamics Lab
NOAA, Princeton University
Princeton, NJ 08540
Dr. Steve Lewellen
Aeronautical Research Associates
of Princeton, Inc.
Princeton, NJ 08540
F. L. Ludwig
SRI International
Menlo Park, CA 94205
Dr. Michael McCraken
University of California
P. 0. Box 808
Livermore, CA 94550
Dr. James McElroy
Environmental Monitoring Systems Lab
Environmental Protection Agency
P. 0. Box 15027
Las Vegas, NV 89114
Dr. Paul Michael
Brookhaven National Laboratory
Upton, NY 11973
Dr. Paulette Middleton
National Center for
Atmospheric Research
P. 0. Box 3000
Boulder, CO 80307
Dr. M. Mills
Teknekron, Inc.
2118 Milvia Street
Berkeley, CA 94704
Mr. P. K. Misra
Ontario Ministry of the Environment
135 St. Clair Avenue, West
Toronto, Ontario, CANADA
Dr. Robert Neligan
Office of Air Quality Planning
and Standards
Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. B. Niemann
Teknekron, Inc.
2118 Milvia Street
Berkeley, CA 94704
Professor B. Ottar
Norwegian Institute of Air Resources
POB 130
N-2001
Lillestrrfn, NORWAY
Dr. Robert Papetti
Office of Environmental Processes
and Effects Research
Environmental Protection Agency
Washington, D. C. 20460
Dr. Courtney Riordan
Acting Deputy Assistant Administrator
Office of Environmental Processes and
Effects Research
Office of Research and Development
Environmental Protection Agency
Washington, D. C. 20460
-------
External Distribution List
Page 4
Professor H. Rodhe
Department of Meteorology
University of Stockholm
S-10691
Stockholm, SWEDEN
Dr. John Seinfeld
Department of Chemical Engineering
California Institute of Technology
Pasadena, CA 91109
Dr. Jack Shannon
Argonne National Lab
Argonne, IL 60439
Dr. Jimmy Sheih
Argonne National Lab
Argonne, IL 60439
Dr. Lowell Smith
Office of Environmental Processes
and Effects Research
Environmental Protection Agency
Washington, D. C. 20460
Dr. Joseph Tikvart
Office of Air Quality Planning
and Standards
Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Eva Voldner
Ontario Ministry of the Environment
135 St. Glair Avenue, West
Toronto, Ontario, CANADA
David A. Wagoner, Director
Air and Hazardous Materials Division
EPA, Region VII
1735 Baltimore Avenue
Kansas City, MO 64108
Thomas J. Warner
Penn State University
University Park, PA
Steven Wassersug, Director
Air and Hazardous Materials Division
EPA, Region III
Curtis Building
Sixth and Walnut Streets
Philadelphia, PA 19106
Dr. Doug Whelpdale
Atmospheric Environment Service
4905 Dufferin Street
Downsview, Ontario, CANADA M3H5T4
Dr. Michael Williams
DOE - Los Alamos Scientific Lab
Group S-2, Mail Stop 606
Los Alamos, NM 87545
Dr. William Wilson
Environmental Sciences Research Lab
Environmental Protection Agency
Research Triangle Park, NC 27711
U.S. Environmental Fr-t°ctlon Agenoy
Librn-v. Ron-, ?"' «'-?ll-A
401 M Street, 5-W.
Washington, DC 20460
------- |