MARCH 1987
ROCKY MOUNTAIN ACID DEPOSITION MODEL ASSESSMENT
Review of Existing Mesoscale Models for Use 1n Complex Terrain
ATMOSPHERIC SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 22771
-------
ROCKY MOUNTAIN ACID DEPOSITION MODEL ASSESSMENT
Review of Existing Mesoscale Models for Use In Complex Terrain
R. E. Morris
R. C. Kessler
SYSTEMS APPLICATIONS, INC.
101 Lucas Valley Road
San Rafael, California 94903
Contract No. 68-02-4187
Project Officer
Alan H. Huber
Meteorology and Assessment Division
Atmospheric Sciences Research Laboratory
Research Triangle Park, North Carolina 22771
ATMOSPHERIC SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 22771
-------
NOTICE
The Information 1n this document has been funded by the United States
Environmental Protection Agency under Contract No. 68-02-4181 to Systems
Applications, Inc. It has been subjected to the agency's peer and
administrative review, and 1t has been approved for publication as an EPA
document. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
11
-------
ABSTRACT
Existing mesoscale meteorological and add deposition models were reviewed
and evaluated for potential application to a complex terrain region within
the Rocky Mountain region. The purpose of the review was to choose candi-
date meteorological and add deposition models that might be adapted for
use as a mesoscale acid deposition model for the Rocky Mountain region.
This new model would then be used by western regulatory agencies to esti-
mate the amounts of addle deposition from proposed new sources at PSD
class I and acid-sensitive areas. Typical application scenarios for the
model will Include shale oil and gas treatment plants that emit both sul-
fur and nitrogen oxides. Thus 1t will be Important to correctly define
the source-receptor relationship of both sulfur and nitrogen deposition
over mesoscale distances 1n complex terrain.
The report Includes a review of meteorological modeling 1n complex terrain
and add deposition processes, a survey of over 60 existing mesoscale
meteorological and acid deposition models, and a discussion of the proce-
dures used to select candidate meteorological and add deposition models
for final evaluation. Evaluation of the candidate models Indicated that
no one meteorological or add deposition model 1s significantly superior
to the others; all the candidate models contained different features that
would be desirable attributes 1n an add deposition model for the Rocky
Mountain region. Hence the conceptual design of the mesoscale add depo-
sition model uses modules selected from various existing meteorological
and add deposition models.
111
-------
CONTENTS
Abstract 111
List of Figures v111
List of Tables x1v
Acknowl edgements xv11
1 INTRODUCTION AND SUMMARY 1-1
1.1 Overview 1-1
1.1.1 Background 1-1
1.1.2 Study Objectives 1-2
1.1.3 Principal Components of the Rocky
Mountain Mesoscale Add Deposition
Modeling System 1-3
1.2 Selection of the First Application Scenarios 1-4
1.2.1 First Application Scenario 1-4
1.2.2 Second Application Scenario 1-8
1.3 Data Requirements for Add Deposition Modeling 1-16
1.3.1 Meteorological Data 1-16
1.3.2 Surface Characteristics. 1-18
1.3.3 Emissions 1-22
1.3.4 A1r Quality 1-22
1.4 Report Structure 1-22
* • & dUtnmary ••••••«««••••««•••«*•»••••«•••••««••••««•••«*«•••« i—&H
2 REVIEW OF MESOSCALE METEOROLOGICAL MODELS
APPLICABLE TO COMPLEX TERRAIN
2.1 Meteorological Modeling In Complex Terrain 2-1
2.1.1 The Primitive Equations 2-1
2.1.2 Flow Patterns Over Complex-Terrain 2-3
Terrain-Driven A1r Flows .^.. 2-5
Thermally Induced A1r Flows 2-12
-------
2.2 Classification of Existing Mesoscale
Meteorological Models 2-15
2.2.1 Coordinate Systems 2-17
2.2.2 Solution Methods 2-18
2.2.3 Form of the Governing Equations 2-19
2.2.4 Simplifying Assumptions 2-19
2.2.5 Parameterization of Processes 2-21
2.2.6 Turbulence Closure 2-21
2.3 Survey of Existing Mesoscale Meteorological Models.. 2-24
3 REVIEW OF ACID DEPOSITION MODELS 3-1
3.1 Review of Add Deposition Processes 3-1
3.1.1 Transport and Dispersion 3-1
Modeling Transport and Dispersion 3-3
Modeling Implications 3-6
3.1.2 Chemical Transformations 3-7
Gas-Phase Chemistry 3-7
Aqueous-Phase Chemistry 3-10
Modeling Implications 3-14
3.1.3 Dry Deposition 3-15
Mechanistic Description 3-16
Experimental Data on Dry Deposition 3-20
Modeling Implications 3-21
3.1.4 Wet Deposition 3-28
Mechanistic Description 3-28
Cloud/Storm Climatology 3-33
Experimental Data on Scavenging Processes 3-36
Modeling Implications 3-38
3.2 Climatology and Meteorology of the Central
Rocky Mountain Region 3-47
3.2.1 Geography 3-47
3.2.2 Climate and Meteorology 3-48
Temperature and Humidity 3-48
Precipitation 3-59
Storm Characteristics 3-60
Transport Winds and Mixing Heights 3-61
3.2.3 A1r Quality and Deposition 3-79
3.2.4 Potential Emission Sources and Regions 3-89
VI
-------
3.3 Survey of Existing Add Deposition Models 3-110
3.3.1 Modeling Concepts 3-116
Eulerlan Models 3-118
Lagranglan Models 3-119
Hybrid Models 3-122
Long-Term Models 3-124
3.3.2 Existing Regional and Mesoscale Models 3-126
Model Input Requirements 3-161
Model Output Format 3-164
Transport and Dispersion 3-165
Physical and Chemical Transformations 3-168
Removal Processes 3-169
4 SELECTION OF CANDIDATE MESOSCALE METEOROLOGICAL AND
ACID DEPOSITION MODELS
4.1 Selection of Mesoscale Meteorological Models 4-1
4.1.1 Criteria for Selection 4-1
4.1.2 Technical Merit Analysis 4-2
Methodology 4-4
Analysis of Prognostic Models 4-9
Analysis of Diagnostic Models 4-9
4.2 Selection of Add Deposition Models '. 4-15
4.2.1 Desirable Attributes of a Rocky Mountain
Add Deposition Model 4-15
4.2.2 Technical Merit Analysis 4-17
4.2.3 Past Applications 4-32
4.2.4 Model Performance Evaluation... 4-32
4.2.5 Availability of the Models Reviewed 4-40
4.2.6 Available Candidate Models 4-41
4.3 Additional Selection Criteria and Selection of Final
Candidate Mesoscale Meteorological and Add
Deposition Models 4-45
4.3.1 The Final Candidate Mesoscale
Meteorological Models 4-46
4.3.2 The Final Candidate Add
Deposition Models 4-48
-------
5 EVALUATION OF CANDIDATE MESOSCALE METEOROLOGICAL
AND ACID DEPOSITION MODELS
5.1 Evaluation of Candidate Mesoscale Meteorological Models... 5-1
5.1.1 Comparative Description of the Candidate
Models 5-1
Initialization of the Wind Field... 5-1
Mass-Consistent Wind Field Adjustment.... 5-4
5.1.2 Evaluation Using a Hypothetical Terrain
Obstacle 5-7
CIT Wind Model 5-7
MELSAR 5-8
ATMOS1 5-8
CTWM , 5-15
Remarks 5-33
5.1.3 Conceptual Design of a Mesoscale Meteoro-
logical Model for the Rocky Mountain Region 5-34
5.2 Evaluation of the Candidate Add Deposition Models 5-35
5.2.1 Transport 5-36
5.2.2 Dispersion 5-39
5.2.3 Chemical Transformation 5-42
5.2.4 Dry Deposition ; 5-43
5.2.5 Wet Deposition 5-44
6 CONCEPTUAL DESIGN OF THE ACID DEPOSITION
MODEL FOR THE ROCKY MOUNTAIN REGION 6-1
6.1 Transport 6-1
6.2 Dispersion 6-2
6.3 Chemical Transformation 6-2
6.4 Dry Deposition 6-3
6.5 Wet Deposition 6-4
References R-l
-------
FIGURES
Number Page
1-1 Class I areas and add-sens1t1ve regions 1n the
western U.S 1-5
1-2 Class I areas and upper-air meteorological stations
located 1n the West, exclusive of Washington and Montana 1-6
1-3 Relationship between application scenarios and
PSD class I areas 1-7
1-4 Application scenario 11 mesoscale region containing
the Clear Creek shale oil plant and two
PSD class I areas—Flat Tops and Maroon-Bells
Snowmass Wilderness 1-9
1-5 Three-dimensional perspective of terrain showing the first
application scenario source and class I receptor areas 1-10
1-6 Application scenario 12 mesoscale region containing
the three gas treatment plants and two PSD class I
areas—Brldger Wilderness Area and F1tzpatr1ck Wilderness
Area 1-13
1-7 Three-dimensional perspective of terrain showing the
second application scenario sources and the Wind River
Mountains, which Include the class I receptor areas 1-14
1-8 Location of application scenario sources, PSD class I
areas, and currently operating NWS surface stations 1-17
1-9 Resolution of the output data from the LFM Model 1-19
1-10 Sulfur oxides emission density, 1600 GMT on a
typical summer weekday 1-23
2-1 Channeling of upper winds within a valley 2-6
1x
-------
Number Page
2-2 Gapping of airflow through mountain notches 2-6
2-3 Orographlc lifting 2-6
2-4 Superposition of gapping and lifting effects
on the mountain wind flow 2-8
2-5 Types of air flow over ridges 2-9
2-6 Envelope and cavity regions behind a
two-dimensional ridge 2-11
2-7 Mountain top Influences on the upper-level flow
component and the downward transporting of
flow momentum 2-11
2-8 Several types of possible wind flows 1n a
mountain valley 2-14
2-9 Typical Inversion destruction 1n a mountain valley 2-16
2-10 A taxonomlc tree of candidate mesoscale models
grouped by realizable model formulation characteristics 2-22
3-1 Synoptic low-pressure system and the circulation
associated with Its warm front, cold front,
and occluded front 3-4
3-2 Relationship between S(IV) (dissolved sulfur dioxide)
and the species that make up S(IV) for varying S(IV)
concentrations and pH values 3-11
3-3 Processes likely to Influence the rate of dry
deposition of airborne gases and particles 3-17
3-4 The scavenging sequence 3-29
3-5 Theoretical scavenging efficiency of a falling raindrop
as a function of aerosol particle size 3-32
3-6 Washout ratio as a function of precipitation rate 3-35
3-7 Flow chart for scavenging calculations 3-37
-------
Number Page
3-8 Acid-sensitive areas 1n the West as Identified by the
NAPAP 1984 annual report to the president and Congress,
by Omernlk, and by Roth and others 3-49
3-9 PSD Class I areas are often located 1n acid-
sensitive areas 3-52
3-10 Class I areas and upper-air meteorological stations
located 1n the West, exclusive of Washington
and Montana. 3-53
3-11 Terrain of the Grand Junction (Colorado) and Salt Lake
City (Utah) subreglon, the Denver (Colorado) subreglon,
and the Lander (Wyoming) subreglon 3-56
3-12 Average upper-air wind rose for north-
western Colorado , 3-63
3-13 Mean annual afternoon mixed-layer resultant wind
fields 1n 1981 calculated by LFM 3-64
3-14 Mean monthly afternoon mixed-layer wind fields 1n 1981
calculated by LFM 3-65
3-15 Morning wind roses over all stability classes 3-71
3-16 Afternoon wind roses over all stability classes 3-72
3-17 Annual resultant surface winds (Aug. 1979 - July 1981) 3-74
3-18 Wind direction frequency distributions for all
morning soundings 3-75
3-19 Vertical wind direction distribution profiles for
al 1 afternoon soundings 3-76
3-20 Wind speed frequency distributions for
al 1 mornlng soundings 3-77
3-21 Wind speed frequency distributions for
al 1 afternoon soundings 3-78
xi
-------
Number Page
3-22 Median yearly visual range and Isopleths for
suburban/nonurban areas, 1974-76 3-82
3-23 Comparison of predicted and measured annual
average SQ4 3-85
3-24 NAOP wet deposition monitoring sites and cities
in the central Rocky Mountain region 3-87
3-25 Comparison of predicted and measured annual average
wet sulfur deposition 1n 1981 3-88
3-26 Annual total wet nitrogen deposition
measured at NADP sites 1n 1981 3-90
3-27 Measured weekly wet-deposited sulfate and nitrate
In Manltou, Colorado—1979-1983 3-91
3-28 Measured weekly surfate and nitrogen wet deposition
1n Sand Spring, Colorado—1980-1983 3-96
3-29 Measured weekly sulfate and nitrate wet deposition
1n Rocky Mountain National Park, Colorado—1981-1983 3-100
3-30 Measured weekly sulfate and nitrate wet deposition
1n Pawnee, Colorado—1980-1983 3-103
3-31 Measured weekly sulfate and nitrate wet deposition
1n Mesa Verde National Park, Colorado—1982-1983 3-107
3-32 State emissions of SOX, NOX and reactive hydrocarbons
1n the West 3-109
3-33 Sulfur dioxide emissions 1n the Southwest, 1980-1995 3-111
3-34 Nitrogen oxide emissions 1n the Southwest, 1980-1995 3-112
3-35 Relative contributions of different sources to wet
sulfur deposition in the Colorado Rockies in 1981 3-113
3-36 Relative frequency with which trajectories emanating
from a source area end up 1n Denver 3-114
xii
-------
Number Page
3-37 Classification of long-range transport modeling approaches...3-117
3-38 The Lagranglan puff, Lagranglan plume segment, and
Systems Applications hybrid EuleHan (RTM-II) models 3-123
5-1 CIT model-generated wind fields at 50 m, 200 m, and 500 m
above ground . 5-9
5-2 MELSAR model-generated winds at 50 m, 200 m, and 500 m
above ground 5-12
5-3 ATMOS1 model-generated winds at 50 m, 200 m, and 500 m
above ground 5-16
5-4 CTVIM model-generated winds for simulation 1 at 50 m,
200 m, and 500 m above ground 5-20
5-5 CTWM model-generated winds for simulation 2 at 50 m,
200 m, and 500 m above ground 5-23
5-6 CTWM model-generated winds for simulation 3 at 50 m,
200 m, and 500 m above ground 5-27
5-7 CTWM model-generated winds for simulation 4 at 50 m
200 m, and 500 m above ground 5-30
-------
TABLES
Number Page
1-1 Chevron Clear Creek shale oil project—maximum emissions
and stack parameters 1-11
1-2 Emissions for the three gas treatment plants that
comprise the second application scenario , 1-15
1-3 LFM variables archived from October 1971 to present..........1-20
1-3 LFM variables archived after November 30, 1983 1-21
2-1 Wind model characteristics used for model classification 2-23
2-2 Existing numerical mesoscale meteorological models 2-25
3-1 Recent experiments on trace-gas deposition to
natural surfaces 3-22
3-2 Field experimental evaluation of the deposition
velocity of submicron-diameter particles 3-24
3-3 Field measurements of scavenging coefficients
of particles 3-39
3-4 Field observations of washout ratios , 3-41
3-5 Laboratory and field measurements of scavenging
coefficients of gases 3-44
3-6 Key to Class I areas 1n the West 3-54
3-7 Frequency of occurrence of atmospheric stability
at the 500-m level AGL .• 3-80
3-8 Seasonal and annual-average morning and afternoon
mixed-layer heights and wind speeds for Grand Junction,
Col orado 3-81
xiv
-------
Number Page
3-9 Summary of background air quality 1n U1nta and
Plceance Basin 3-83
3-10 Measured ambient concentrations of total
suspended participates 1n.study region 3-84
3-11 Annual total wet sulfur and nitrogen deposition
based on measurements at NADP sites 1n 1981 3-86
3-12 Existing regional air quality and deposition models 3-127
3-13 Technical attributes of existing regional and
mesoscale air quality and deposition models 3-138
4-1 Classification of qualified mesoscale meteorological
models according to taxonomlc tree given 1n Figure 2-10 4-3
4-2 Technical attribute categories, subcategorles, and
modeling methods for candidate models 4-5
4-3 Technical attributes of candidate prognostic
mesoscale meteorological models 4-10
4-4 Technical attribute scores for candidate prognostic models...4-12
4-5 Technical attribute categories, subcategorles, and
modeling methods for candidate diagnostic models 4-14
4-6 Technical attribute scores for diagnostic models 4-16
4-7 Allowable PSD Increments 4-18
4-8 Indexes of air-quality-related values (AQRV) and
possible effects of changes 1n air quality 4-19
4-9 Technical attribute categories, subcategorles,
and modeling methods for candidate add deposition models....4-21
4-10 Technical attribute scores for short-term/long-term
and long-term only models considered for application
to add deposition 1n the Rocky Mountains 4-29
xv
-------
Number Page
4-11 Past applications of the 10 highest scoring
add deposition models 4-33
4-12 Genera] Information on evaluation studies of final
candidate add deposl tlon model s 4-37
4-13 Summary of performance statistics for models ranked
highest 1n the technical merit analysis 4-39
xvi
-------
ACKNOWLEDGEMENTS
The work reported here was a team effort that Involved personnel from the
U.S. EPA and other federal agencies as well as from state regulatory
agencies. In particular we would like to acknowledge Alan Huber, the EPA
project officer, for his effort 1n focusing the goals of this project, and
Larry Svoboda and John Dale of EPA Region VIII. Many members of the
Western Atmospheric Deposition Task Force also contributed to the work,
Including Dr. Douglas Fox of the U.S. Forest Service, Mr. Donald Henderson
of the National Park Service, and Mr. Al Rlbleu of the Bureau of Land
Management. Finally, several members of Systems Applications, Inc.
participated 1n the work, Including Douglas Latlmer, Mr. M. K. L1u,
Douglas Stewart, Gary Moore, Chris Daley, and Sharon Douglas.
xv ii
-------
1 INTRODUCTION AND SUMMARY
1.1 OVERVIEW
1.1.1 Background
Acid deposition has recently become an Increasing concern 1n the western
United States (Roth et al., 1985). Although this problem may not be as
acute 1n the western United States as 1t 1s 1n the eastern United States,
1t Is currently a concern of the public and regulatory agencies because of
the high sensitivity of western lakes at high altitudes and the rapid
Industrial growth expected to occur 1n certain areas of the West. An
example of such an area 1s the region known as the Overthrust Belt 1n
southwestern Wyoming. Several planned energy-related projects Including
natural gas sweetening plants and coal-fired power plants may considerably
Increase emissions of add precursors 1n northeastern Utah and north-
western Colorado and significantly affect ecosystems 1n the sensitive
Rocky Mountain areas.
Under the 1977 Clean A1r Act, the U.S. Environmental Protection Agency
(EPA), along with other federal and state agencies, 1s mandated to pre-
serve and protect air quality throughout the country. As part of the Pre-
vention of Significant Deterioration (PSD) permitting processes, federal
and state agencies are required to evaluate potential Impacts of new emis-
sion sources. In particular, Section 165 of the Clean A1r Act stipulates
that, except 1n specially regulated Instances, PSD Increments shall not be
exceeded and air quality-related values (AQRV's) shall not be adversely
affected. A1r-qual1ty-related concerns range from near-source plume
blight to regional-scale add deposition problems. By law, the Federal
Land Manager of Class I areas has a responsibility to protect a1r-qua!1ty-
related values within those areas. New source permits cannot be Issued
by the EPA or the states when the Federal Manager concludes that adverse
Impacts on air quality or a1r-qua!1ty-related values will occur. EPA
Region VIII contains some 40 Class I areas 1n the West, Including two
Indian reservations. Several of the remaining 26 Indian reservations
1n the region are considering similar designations. State and federal
agencies, Industries, and environmental groups 1n the West need accurate
data concerning western source-receptor relationships.
1-1
-------
To address this problem, EPA Region VIII needs to designate an air quality
model for application to mesoscale pollutant transport and deposition over
the complex terrain of the Rocky Mountain region for transport distances
ranging from several km to several hundred km. The EPA recognizes the
uncertainties and limitations of currently available air quality models
and the need for continued research and development of air quality models
applicable over regions of complex terrain. Therefore, the objective of
the project reported here 1s to select and assemble the best air quality
models available for application to the Rocky Mountain area on an Interim
basis.
Such modeling 1s needed to assess the relationship between western source
emissions and receptor Impacts. To address add deposition problems 1n
the East, the EPA has launched a major effort to develop a state-of-the-
art regional add deposition model—RAOM (NCAR, 1985). According to the
current plan, this model will undergo an Intensive model evaluation during
the period 1987-1988, and thus will not be available for Impact assess-
ment, even for the 31 eastern states, before 1989. Realistically, evalua-
tion, adaptation, and application of this sophisticated model to the West
will probably not occur until 1990 or beyond. To fill this gap, a prac-
tical modeling tool with which the federal and state agencies can assess
air quality Impacts 1n the West during this Interim period 1s needed.
A1r quality modeling 1n this region 1s especially difficult due to the
complex air flow patterns over the Rocky Mountains and the difficulty of
predicting add deposition levels. Available data bases are Inadequate
for thorough model evaluation studies. Major field studies and the
establishment of a meteorological network throughout the Rocky Mountain
area would be required to collect data necessary for any thorough evalua-
tion.
1.1.2 Study Objectives
This report discusses the development of a Rocky Mountain Mesoscale Add
Deposition Modeling System,' Including the survey and selection of existing
candidate mesoscale meteorological and add deposition models. The pri-
mary objective of this work 1s to select and assemble an Interim air
quality model based primarily on models/modules currently available for
use by the federal and state agencies 1n the Rocky Mountain region. The
EPA has formed an atmospheric processes subgroup of the Western Atmo-
spheric Deposition Task Force, referred to as WADTF/AP, to develop criter-
ia for model selection and subsequent model evaluations. This group com-
prises representatives from the National Park Service, U.S. Forest Ser-
vice, EPA, Region VIII, the National Oceanic and Atmospheric Administra-
tion, and other federal, state, and private organizations. On the basis
1-2
-------
of our review of the modeling needs Identified by the WADTF/AP, the
specific requirements of the model proposed 1n this project are as fol-
lows:
Since the anticipated use of this model 1s to analyze permit applica-
tions and evaluate urban development plans, the model must be able to
process various air pollutants from both point and area sources.
The modeling areas will typically cover spatial regions approximately
200 km to 300 km on the side to assist 1n permitting new sources
within relatively short distances of Class I areas.
The temporal scales will emphasize longer time periods such as sea-
sonal and/or annual averages to obtain cumulative Impacts from both
chronic and episodic events.
The model should be able to simulate transport, diffusion, transfor-
mation, and deposition of pollutants over complex terrain 1n the
Rocky Mountain region using relatively sparse NWS upper-air sound-
Ings.
1.1.3 Principal Components of the Rocky Mountain Mesoscale
Add Deposition Modeling System
A mathematical modeling system for describing the various physical and
chemical processes associated with add deposition must consist of several
components or modules. These modules describe processes such as wind
transport, chemical reactions, plume rise, and wet/dry deposition.
Although the modeling system must be an Integrated, Internally consistent
package, 1t can be conveniently divided Into two distinct parts:
Simulation of meteorological processes
Simulation of pollutant dispersion, chemical reactions, and deposi-
tion.
This distinction can be made because chemical reactions and meteorological
processes are largely uncoupled. That 1s, at the concentration levels
typically encountered, atmospheric pollutants do not Influence the flow
field. Thus, the Rocky Mountain me so scale add deposition modeling system
consists of two principal modules—a meteorological module and an add
deposition module.
1-3
-------
1.2 SELECTION OF THE FIRST APPLICATION SCENARIOS
The primary purpose of the Rocky Mountain Add Deposition Model 1s to
determine whether acid deposition from new sources and urban growth will
adversely affect ecological systems 1n sensitive areas within complex ter-
rain 1n the Rocky Mountains. To determine this, the model must predict
the total cumulative deposition of sulfates and nitrates, as well as the
contribution of the new source. Add deposition modeling 1n the Rocky
Mountains 1s further complicated by the terrain, and the fact that many
add sensitive lakes are located at higher elevations, where orographlc
effects on precipitation Increase wet deposition of pollutants.
Before describing the selection of the mesoscale meteorological and add
deposition models (from among those models surveyed 1n Chapters 2 and 3),
1t will be helpful to Illustrate the types of application scenarios plan-
ned for the modeling system.
The two application scenarios described here represent the types of new
sources that have been proposed for energy development areas 1n the
western Rockies. They Include sources and emission characteristics
diverse enough to test the model's ability to predict cumulative and
source-specific add deposition at the sensitive areas.
One scenario Includes an oil shale plant 1n western Colorado, the other
Includes three gas treatment plants 1n southwestern Wyoming. The rela-
tionship between these source locations and the PSD class I areas and
acid-sensitive areas 1n the West 1s shown 1n Figures 1-1 and 1-2 (the
class I areas shown 1n Figure 1-2 are Identified 1n Table 3-6).
1.2.1 First Application Scenario—Colorado 011 Shale Plant
The first application scenario 1s the Chevron Clear Creek Shale 011 Pro-
ject (CCSOP), located 1n the south-central portion of the Plceance Basin
1n Garfleld County, Colorado. This site 1s approximately 40 miles north-
east of Grand Junction and 26 miles north of DeBeque. The closest PSD
class I area 1s the Flat Tops Wilderness Area, which lies approximately
50 to 55 miles to the northeast of the CCOSP site. Figure 1-3 shows the
relationship between the CCSOP and class I areas within a rather large
(750 km x 750 km) region.
For this application scenario the Flat Tops Wilderness and Maroon-Bells
snowmass areas will be the primary receptors of Interest; however, esti-
mates of cumulative and source specific add deposition need to be calcu-
lated at all PSD class I and sensitive areas.
1-4
-------
FIGURE 1-1. Class I areas (blocks) and acid-sensitive regions (hatched
areas) in the western U.S. The locations of the first two application
scenarios are indicated by the large asterisks.
1-5
-------
VA
SCALE 1:12.500.000
ki loaders
0 100 200 300 400 500 600
l""l 1
1 . 1 , , 1
0 100
1 1
1
200
• iles
1
1
300
1
rIGURE 1-2. Class I areas and upper-air meteorological stations locateu
in the West, exclusive of Washington and Montana. The locations of the
first two application scenarios are indicated by large asterisks. (Key
to locations is in Table 3-5.)
1-6
-------
00 450 500
UTM Easting Zone 12
550 600 650 700 750 600 850 900 950 1000 1050 1100 11
[00 450
500 550 600 650 700 750 300 850 300 950 1000 1050 1100 115*
UTM Easting Zone 12
FIGURE 1-3. Relationship between application scenarios and PSD class I
areas (see Table 3-6 for key). Terrain elevations are in feet above mean
sea level,
1-7
-------
The Flat Tops Wilderness Area Includes several lakes that are susceptible
to adverse effects from add deposition because of low alkalinity (Turk
and Adams, 1982). This area 1s one of the first major orographlc barriers
to air masses resulting from the predominantly southwest winds over the
Plceance Basin. Thus, emissions from the CCSOP may be carried up to the
Flat Tops Wilderness Area and precipitated out before they have been
effectively dispersed.
Figure 1-4 shows the terrain In a mesoscale region (250 km x 250 km) con-
taining the CCSOP site and the Flat Tops and Maroon-Belles Snowmass
Wilderness Areas. This region 1s dominated by a series of ridges and val-
leys leading up to the Wilderness Areas (Figure 1-5). Correct specifica-
tion of the three-dimensional wind field and spatial variability of pre-
cipitation will be an Important factor 1n modeling this scenario.
The emissions data for the CCSOP are given 1n Table 1-1. For this appli-
cation scenario the emissions from the retort combustors are assumed to
have no tall gas control units. As seen 1n Table 1-1, the extraction of
oil from oil shale through combustion results mostly 1n emissions of NOX
and CO. This application scenario will Illustrate the model's ability to
predict nitrogen deposition at source-receptor distances of about 80 km.
Because of the low solubility of NO and N02, wet deposition of nitrogen Is
usually 1n the form of nitrates or nitric acids. Thus 1t will be Impor-
tant to characterize the conversion of the emitted NQX to nitrates. The
gas-phase conversion of NOX to nitrates 1s relatively rapid and depends on
the availability of oxidizing precursors 1n the atmosphere. Nitrate
formation 1n precipitation depends on a combination of gas-phase, gas-
particle, and aqueous-phase reactions, with both reversible and Irre-
versible conversions. Thus the correct specification of the background
reactivity and precipitation patterns will be Important for this applica-
tion scenario.
One year of meteorological data 1s available for the Chevron Clear Creek
Shale 011 Project site for the calendar year 1981 (collected as part of
the PSD permitting process). These data were collected at several levels
on a 60 m tower and consist of wind speed, wind direction, temperature,
and stability. The use of on-site meteorological data to supplement the
routinely observed surface and upper-air meteorological data 1s Important
1n resolving the wind fields 1n complex terrain.
1.2.2 Second Application Scenario—Wyoming Gas Treatment Plants
The second applications scenario Includes three gas treatment plants loca-
ted 1n the R1ley Ridge area of southwestern Wyoming. These plants are the
1-8
-------
UTM Easting Zone 12
750 800
430
750 800
UTM East mg Zone 12
300
FIGURE 1-4. Application scenario #1 mesoscale region containing the Clear
Creek shale oil plant (CCSOP) and two PSD class I areas — Flat Tops (74) and
Maroon-Bells Snowmass Wilderness (76).
1-9
-------
Flat Tops
Wilderness Area
CCSOP
Maroon-B
Snowmass
Wildernc
Are
FIGURE 1-5. Three-dimensional perspective of terrain showing the first
application scenario source and class I receptor areas.
1-10
-------
TABLE 1-1. Chevron Clear Creek shale oil project—maximum emissions and stack parameters.
Source
Description
Number
of Coordinates (!•)
Stacks East North
Total
Emissions (g/a)
S02 N0_ PM 00
Stack Exit Exit
Height Diameter Velocity Temperature
(•) (») (•/.) CK)
Coal Grinding 11
Steam Superheaters 11
Retort Combustor 11
TEG Concentrator 5
Coal Feed Bins 11
Elevated Flare(17) 11
Elevated Plare(IS) 11
Start-up Shale Bin 11
Start-up Heater!10 11
Start-up Air Heater 11
Start-up Heaterfl2 11
Start-up Steam Gen. 11
719.95 4389.62
719.89 4389.51
720.01
719.88 4389.42
0.55 1.76 1.10 0.55 15.2 0.90 5.2
6.71 21.67 1.65 6.71 45.7 2.60 5.0
4389.57 93.50 1007.6 79.05 2043.8 106.7 4.90 16.6
0.02 0.05 0.04 0.02 30.5 0.60 0.7
0.002
(depends on the" quantity and composition of gases flared)
(depends on the quantity and composition of gases flared)
0.006
0.17 0.54 0.04 0.17
18.45 59.59 4.54 18.45
0.89 2.89 0.22 0.89
57.04 187.2 14.03 57.04
344
489
427
683
-------
Northwest/Mobil Craven Creek plant approximately 5 miles north of Opal,
the Exxon Shute Creek plant approximately 12 miles north-northeast of
Opal, and the Quasar East Dry Basin plant approximately 9 miles south-
southwest of B1g P1ney. The closest PSD class I areas are the Brldger and
F1tzpatr1ck wilderness areas located 1n the Wind River Range (Figure
1-3). These wilderness areas are approximately 50 miles northeast of the
Quasar East Dry Basin plant site, and approximately 80-85 miles to the
north-northeast from the Exxon and Northwest/Mobil plant sites. Other
nearby class I areas Include Grand Teton National Park and the Teton
Wilderness Area, located due north from the plant sites.
Figure 1-6 shows a mesoscale region (250 km by 250 km) containing the gas
plant sites and the Brldger and F1tzpatr1ck wilderness areas; a three-
dimensional perspective of this region 1s shown In Figure 1-7. The pollu-
tants would be carried across a relatively flat area by southwest winds
until they encounter the Wind River Range, where the air parcel would be
forced upward, possibly resulting In orographlc precipitation when the
saturation level 1s achieved.
The emissions from the three gas treatment plants are listed 1n Table
1-2. The main pollutant emissions from these sources are sulfur dioxide,
with some nitrogen oxides. This scenario differs from the first because
of the proximity of several major sources, such as the Naughton and Jim
Brldger power plants, the Opal gas plant, and other gas and chemical
plants.
The gas-phase oxidation of S02 to sulfates 1s rather slow. However 1n the
presence of hydrogen peroxide and other oxidizing agents, the aqueous-
phase oxidation of S02 to sulfates 1s almost Instantaneous. Thus, for
this application scenario the transformation of the emitted S02 to sulfate
will most likely be limited to the amount of hydrogen peroxide and other
oxldlzers available. It will be crucial to obtain an estimate of the oxi-
dizing precursors and cloud amounts 1n order to properly estimate the pro-
per S02 oxidation rate and resultant sulfate deposition.
There are several sources of supplementary meteorological data near these
gas plant sites. These Include the Fort Brldger C1v1l Aeronautics Admini-
stration 60-foot tower, the Kemmerer Coal 10-meter tower, and the Utah
Power and Light Company's Naughton Power Plant site. The period of record
for the Fort Brldger tower data are the years 1950-1954, while the Kem-
merer coal tower was operating from November 1979 through October 1980.
At present the period of record for the Naughton Power Plant tower data 1s
unknown. On-s1te meteorological data will be useful 1n determining when
air parcels are headed toward the Wind River Mountains.
1-12
-------
UTM Easting Zone 12
600 650
KemmererUO m
Tower
Nalghton Power
Plant
Bridger
600 650
DIM Easting Zone 12
FIGURE 1-6. Application scenario #2 mesoscale region containing the three
gas treatment plants and two PSD class I areas--Bridger Wilderness Area (area
between the two boxes with number 70) and Fitzpatrick Wilderness Area (69),
both in the Wind River Mountains.
1-13
-------
Wind River
Mountains
Quasar
xxon
Northwest
FIGURE 1-7. Three-dimensional perspective of terrain showing the second
application scenario sources and the Wind River Mountains, which include
the class I receptor areas.
1-14
-------
TABLE 1-2. Emissions for the three gas treatment plants that comprise the
second applications scenario (tons per year).
Plant
Capacity (billion cfd)
CO
COS*
COt*
He
HaS
Nt
S2«
SOt
•Vtffca**
TSP
vex?
Quasar
1.2
458
4,126
17,613,000
10,722
170
4,144,000
2,104
.8,745
156
194
Exxon
1.2*
264
4,126
17,007,000
10,722
106
4,618,000
1,249
5,579
92
112
Northwest/
Mobil
0.4
6,145
52
5,587,000
3,854
97
909,779
323
3,509
25
33
Total
2.8
6,867
8,304
40,207,000
25,298
373
9,671,779
3,676
15,833
273
339
— •—•—"— »-w • ^^r^r, w g W»«H w*w ^VW*»VJp •• V«> ««•*••«• *^f •#«•*•• V*W*^*|P| W
IndudM both of Exxon's plants at WMt Oiy Basin and Big Mesa,
•COS Is carbonyt aulflda.
•Assumss that all COt la vwitad.
T8P la total suspsnctod partlculatas.
•VOC Is noMnathana volatlla organic compounds.
1-15
-------
1.3 DATA REQUIREMENTS FOR ACID DEPOSITION MODELING
The types of data required for add deposition modeling depend on the
choice of the add deposition model. For example, the simplest models may
require only a single wind speed and wind direction, while more sophisti-
cated models require a three-dimensional wind field to define pollutant
transport. In this section we discuss which data are needed and which are
available for a more sophisticated add deposition modeling approach. The
required data fall Into the following categories: meteorology, surface
characteristics, emissions, and air quality.
1.3.1 Meteorological Data
Any sophisticated add deposition model will require the following
meteorological data: transport winds, surface wind speed, mixing depths,
humidity, precipitation, and cloud cover, types, and amounts. The detail
and methods of preparing the data differ for regional modeling (all states
west of the Mississippi River) and mesoscale modeling (200-300 km on a
side). Meteorological Inputs for regional modeling would come from obser-
vations and Limited Fine Mesh Model output (Fawcett, 1977; Shuman, 1978),
while the mesoscale modeling Inputs would come from a mesoscale meteoro-
logical model. In general, there are four main sources of meteorological
data:
Upper air observations
Surface observations
Limited Fine Mesh (LFM) Model output
Hourly precipitation observations
The National Weather Service (NWS) network of upper-air monitors 1s a
sparse network that covers the entire U.S. The observations consist of
twice-daily soundings (0000 GMT and 1200 GMT) of wind speed, wind direc-
tion, temperature, and humidity at several vertical levels. Figures 3-15
and 3-16 Illustrate the sparse resolution of the network over nine western
states and show the morning and afternoon wind roses at these sites. Due
to the coarseness of the network, 1t 1s able to resolve only synoptic wind
flows across the western U.S.
The NWS surface observation network 1s denser than the upper-air network
and consists of 3-hour observations of surface wind speed, wind direction,
temperature, dewpoint temperature, pressure, visibility, opaque sky cover,
cloud base heights and types, and precipitation amounts and types. Figure
1-8 shows the resolution of the currently operating surface stations 1n a
750 km by 750 km region containing the sources chosen for the first two
applications. Except for the Quasar gas plant site 1n Wyoming, there are
1-16
-------
500
UTM East me Zone 12
100 450 500 550 600 650 700 750 800 850 900 950 1000 1050 1100 11!
_l I I ! I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
xSHERIDAN MY
KMOORCROFT WY
*NORLAND MY
* JACKSON WY
-wIDAHO FA.L
MRIVERTON HY
* CASPER HY
KDOUGLAS MY
1*)?i iATELLO ID
•LANDER MY
CM
-4700
at
§ "MALAD CITY
""4650
«BIG
Qu««ar
KRAMLINS NY
HROCKSPRINGS MY
MLARAMIE MY
•CHEYENNE WY
-•SALT LAKE CITY UT
*FT COLLINS CO -
- *PROVO UT
• FRASER CO
wEAGL&eflO
"LEADVILLE CO
•GRAND JUNCTCQM CO
*USAF ACADEMJ
*COLO SPRINt S CO
XGREEN RIV
*FT CARSON CO
i i I i i i i I i i i I'M i II i i i i i i I i i i i I i i i i I i i i i I i*PUE9LCi jru. -:-n
00 450 500 550 600 656 700 750 300 650 900 950 1000 1050 1100
UTM Easting Zone 12
FIGURE 1-8. Location of application scenario sources, PSD class I areas
(boxed numbers), and currently operating NWS surface stations (*). Upper-air
stations are indicated by circles around the asterisks.
1-17
-------
no surface observation sites near any of the sources 1n the proposed
application scenarios. On-s1te meteorological data, which are available,
will be useful 1n supplementing the NWS surface observation network.
The Limited Fine Mesh (LFM) model 1s exercised twice dally to give three-
day weather forecasts across North America. The LFM model output 1s
archived every six hours of the first two days of the simulation. The
spatial resolution of the LFM model 1s approximately 127 km (Figure 1-
9). The model has shown skill 1n predicting synoptic wind flows, but less
skill 1n predicting precipitation location and amounts. The LFM model
predicts several vertical layers of wind components, temperature, dewpolnt
temperature, and relative humidity as well as surface pressure and pre-
cipitation. Tables 1-3 and 1-4 show the data available from past LFM
model simulations that are currently archived at the National Center for
Atmospheric Research (NCAR).
The hourly precipitation network has a finer spatial resolution than the
NWS surface observations network. Techniques have been developed at the
University of Michigan to grid the data from these two networks 1n a con-
sistent manner.
Another source of observed meteorological data 1s obtained from the
National Environmental Satellite Service (NESS). Such data Include tem-
perature sounding data from TIROS-n/NOAA (surface to 10 mb), water vapor
at three levels, and total columnar ozone. In addition, the SMS/GOES
satellites archive photographs from which visibility can be deduced. At
present the main use of satellite data 1n add deposition modeling has
been use of the visibility data to estimate sulfate concentrations for
qualitative model evaluation.
1.3.2 Surface Characteristics
In order to calculate the dry deposition of pollutants, Information 1s
needed concerning the structure and composition of the earth's surface.
The required Input data for most add depositions models Include surface
roughness and surface resistance or deposition velocity. These data are
usually prescribed from land use data. Land use data for the eastern U.S.
are available from the NCAR on a grid of 1/4 degree longitude by 1/6
degree latitude. This data base consists of nine land use categories
(urban, agricultural, range, deciduous forest, coniferous forest, mixed
forest, water, wetland, and mixed agricultural and range land). The data
were taken from a land cover data base that covers the entire U.S. For a
mesoscale region 1n the Rocky Mountains, this data base will have to be
adjusted to account for snow cover 1n the region.
1-18
-------
GRID POINT LOCATIONS FOR NMC LFM 41 X 36
V XT
» •»_ * » • • * \*fs' •*•*
10
Pole at I,J
FIGURE 1-9. Resolution of the output data from the LFM Model.
1-19
-------
TABLE-1-3. LFM variables archived from October 1971 to present.
VAXIABLCS SAVED ROM LIU AMD LIM-tl
LfM t P«rlorf *f Iceerd: 0000 CHT Oct. 31. lf?l - 0000 Off Aut. 31. 1977
UM-II Ptrlrt of I«eord: 1200 COT Aug. 31. If77 - 1200 CHT. JUM 10. 1911
UN (4cH-Or*«r) P«*i»4 of Ueerd: 0000 CKT JIMM 11. 1911 - Pr«*«nc
VarUbl*
Units
r«rtc««C Interval
IS It 14 30 3« 42 AS
UL. MM. (SrC-490) PEXCEMT
nieir. VATEX »/M*
B.L. MT.TEKP. DEC K
S.L. 0 K/SCC
B.L. V M/SEC
looo MB man K
•SO MS NEICHT N
700 KB HEIGHT M
500 KB KEICKT M
SOirACE TEKP. DEC
1000 KB TEKP. DEC
•SO KB TEKP. DEC
700 KB TtXP. DEC
300 KB TEKP. DEC
400 KB TOO?. DEC
300 KB TEMP. DEC
200 KB TOP. DEC
•SO KB U M/SEC
700 KB U M/SEC
SOO KB 0 M/SEC
400 KB 0 K/SEC
900MB I/SEC
200KB K/SEC
•SO KB M/SEC
700 MB K/SEC
SOO MB K/SEC
400 MB K/SEC
900 KB K/SEC
200 KB K/SEC
•SO KB • KB/SEC
700 KB M KB/SEC
900 KB « KB/SEC
PUCIP. ACT. K
SVBFACE PBCSS. {>*) KB
an. Km,) B.L. nsenrr
CB&. BUM.) i racurr
»
I
>7 -
1
u :
i
17 i
i4
+ Th**« 2 vari*bl«* are Available startlag 1200 GMT Dec. 7, 1971.
Z Thai* 8 variablaa ara available atarting 0000 OfT Oct. 1, 1972.
* Tbaaa 8 variablea ara available •tartint 0000 GMT March 6. 1974.
i Thaaa 8 variablea are available atarting 0000 GKT March 16* 1974.
4 Thaaa 62 variablea ara available etartiug 1200 GKT April 2, 1975.
$ Theaa 62 variablea ara available atarting 0000 GMT February 1» 1976.
• Theae 72 variablea are available atarting 0000 GMT Noveaber 29. 1978.
0 Theae 4 variablea are available atarting 0000 GMT November 26,. 1980.
C Theee 23 variable! are available atarting 0000 GMT Noveaber 30, 1983.
B.L. • loveat 50 ob
Layer 1 - B.L. top to approximately 720 ab
Layer 2 • approximately 720 ab to approxiaately 490 ab
-------
TABLE 1-4. LFM variables archived after November 30, 1983.
VARIABLES SAVED HUM ITU AM) LW-I1
UM (4ch-0rtf«r) P«rt«4 •( Mc»r4: 0000 CJfT Korok*r 30. 1*13 - Pmmc
Variable
ftoraeaic lnt«rvai t»*
13 IB 24 M 34 42 4|
ML. met. (src-4M) rttCEKr
PKECIP. MATH w/«z
B.L. m.TEMP. BEC K
B.L. a M/SCC
B.L. V M/SCC
looo KB incur M
•90 MB HtlCHT H
700 MB HEIGHT M
900 KB NCXCIIT H
SUXFACt TEMP. OEC X
1000 MB TEMP. DEC K
•SO MB TEMP. BBC K
700 KB TEMP. BEG K
500 MB TEJ1P. DEC K
•CO KB TEMP. BEC K
MO KB TCW. BEC K
200 KB TEMP. BEC K
•90 MB 0 M/SEC
700 MB 0 M/SEC
900 KB 0 M/SEC
400 KB 0 M/SEC
900 KB 0 M/SEC
MO KB 0 M/SEC
•90 MB V M/SEC
700 KB V M/SEC
900 MB V M/SCC
400 MB V M/SEC
MO MB V M/SEC
MO MB V M/SEC 1
•M MB M MB/SEC
700 MB v MB/SEC
MO KB M MB/SEC
KECXP. AKT. M
SLIFACE NESS. (»») MB
(IEL. MUM.) B.L. rOCDCT
(BEL. MOM.) 1 PttCtXT
010.. «OH.) 2 PKCWT
1000 KB BEV POWT BEC
•90 KB DEW POXKT BEC
700 KB BEV POXXT BEC
MO KB BEW POXKT BEC
400 KB DEW POXXT BEC
300 MB DEW POINT BEC
SEA LEVEL MESS. MB X
B.L. w MB/SEC
Total fUUs »jr lw«r 3
I
:
3
3
X
9 3
t
1
t
7 t
»
3
X
4 3
X
X
X
X
X
X •
X
X
X
X
'X
X
x .
X
X
X
X
X
X
X
X
X
X
. X
X
X
X
X
X
t . x
X
X
X
X
X
X
X
X
X
X
X
X
X
X J
7 44 3
,
[
7 4
1
4 3
1
3
3
3
3
3
3
:
s
<
E >
7 4
e
i
c
t
c
c
c
c
t
E
4
B.L. • lowest 50 mb
Laytr 1 - B.L. top to approximately 720 nb
Lay«r 2 » approximately 720 «b to approximately 490 »b
'1-21
-------
1.3.3 Emissions
Emissions from the sources Included 1n the first two application scenarios
are described 1n Sections 4.1.1 and 4.1.2; further Information will be
obtained from the PSD permit applications. For all other sources the best
data on anthropogenic emissions available for add deposition modeling are
contained 1n the 1980 NAPAP Emissions Inventory, Version 5.3 (Sellars,
1985).
The area source emissions data are provided on a grid of 1/4 degree longi-
tude by 1/6 degree latitude and cover the entire U.S. (Figure 1-10). Both
the area and point source emissions are weighted according to season, day
of week (weekday, Saturday or Sunday), and hour of day. The Inventory
contains data for 40 species, Including S02» S04, TSP, Pb, CO, HC1, HF,
NO, N02, NHo, formic add, acetic add, other organic adds, and 27 cate-
gories of VOC emissions.
The Importance of blogenlc emissions for acid deposition has not been
quantified. Blogenlc emissions consist mainly of hydrocarbons, with some
sulfur emissions 1n areas with a high rate of vegetation decomposition,
such as swamps. The best currently available blogenlc emissions Inventory
for the western U.S. .1s provided by Washington State University (1985).
This Inventory contains emissions for alpha-plnene, Isoprene, and other
terpenes by county for the entire U.S.
1.3.4 A1r Quality
The use of and reliance on existing air quality data depend on the model-
Ing approach used. Eulerlan add deposition models require air quality
data for Initial and boundary conditions, while Lagranglan models may use
air quality data for background concentrations. The largest air quality
data base 1n the U.S. 1s the EPA Storage and Retrieval of AerometHc Data
(SAROAD) network. Other networks that monitor air quality concentrations
In the West Include measurements of fine and coarse partlculate data col-
lected by the EPA Western Partlculate Characterization Study (WPCS),. and
the Electric Power and Research Institute's Western Regional A1r Quality
Study (WRAQS).
1.4 REPORT STRUCTURE
This report 1s divided Into four main sections that review existing meso-
scale meteorological models and add deposition models; select candidate
meteorological and add deposition models, and evaluate the candidate
models and present a conceptual design for a hybrid add deposition model
applicable to the Rocky Mountain region.
1-22
-------
LEGEND: SOX
5-20
i>20
FIGURE 1-10. Sulfur oxides emission density (g-mol/h/km ),
1600 GMT on a typical summer weekday. (Source: Sellers, 1985)
1-23
-------
Chapter 2 starts with a summary of complex-terrain meteorology and Its
relationship to add deposition modeling 1n the Rocky Mountain region.
This 1s followed by a classification of generic mesoscale meteorological
models, then Chapter 2 finishes with a survey of existing mesoscale
meteorological models.
In Chapter 3 a review of add deposition processes 1s presented followed
by a review of the climatology of the Central Rocky Mountain region.
Chapter 3 finishes with a survey of existing add deposition and air
quality models. The selection of candidate mesoscale meteorological and
add deposition models 1s presented 1n Chapter 4.
The selection of candidate models 1s based on a technical merit analysis
which ranks the models by assigning a numeric score to the methods used to
simulate the major processes that lead to add deposition 1n the Rocky
Mountain region. The selection of the final candidate models 1n Chapter 4
1s based on additional criteria which Includes the needs and desires of
the western regulatory agencies.
Finally, the candidate mesoscale meteorological and add deposition models
are evaluated 1n Chapter 5. Based on this evaluation, an Initial con-
ceptual design of a hybrid add deposition model for the Rocky Mountain
region 1s given.
1.5 SUMMARY
This report presents a survey and review of 65 mesoscale meteorlogical
models and 75 add depos1t1on/a1r quality models. The purpose .of this
review was to select candidate models for Incorporation Into a mesoscale
add deposition model for the Rocky Mountain region. This model will be
used by regulatory agencies 1n the western states to estimate add
deposition at sensitive receptors from proposed new sources. The selec-
tion criteria were the technical merits of the models as well as the needs
and requirements of the western regulatory agencies. Technical merit was
determined by a system of scores for model features. The needs and
requirements of the potential users were ascertained through meetings with
the Western Atmospheric Deposition Task Force and a questionnaire sent to
these agencies 1n August 1986.
Based on the results of the technical merit analysis and the comments of
the. potential users, selected candidate mesoscale meteorological and add
deposition models were evaluated to determine which models (or modules)
would be the most approplate for Inclusion Into an add deposition model
1-24
-------
for the Rocky Mountains. Based on this evaluation, a conceptual design of
a mesoscale add deposition model for the Rocky Mountain region has been
proposed.
The conceptual design makes use of a mesoscale meteorological model to act
as a driver for an add deposition model. A diagnostic wind model will be
used to generate wind fields since cost considerations preclude the use of
prognostic models. Other meteorological variables, such as boundary layer
heights, temperature, and m1crometrolog1cal parameters, will be generated
by an existing meteorological model that was designed for the Rocky Moun-
tain region. The add deposition model will use the Lagranglan puff model
formulation. The Lagranglan approach 1s the most cost effective for
estimating the Impacts of a few sources. It 1s also the approach
preferred by potential users. Transport of the Lagranglan puff would be
defined by the meterologlcal model, while the dispersion algorithm would
contain a parameterization that takes Into account the unique effects that
complex terrain has on diffusion. Dry deposition would be based on the
resistance concept, as 1s done 1n the state-of-the-art add deposition
models currently under development. Wet deposition would use the
scavenging coefficient, and would account for the different ralnout and
washout effects of rain, snow, and different storm types.
1-25
-------
2 REVIEW OF MESOSCALE METEOROLOGICAL
MODELS FOR USE IN COMPLEX TERRAIN
2.1 METEOROLOGICAL MODELING IN COMPLEX TERRAIN
The motion of the earth's atmosphere can, 1n principle, be described by
fundamental physical laws such as Newton's second law of motion and the
second law of thermodynamics (Hess, 1959). However, mathematical modeling
of atmospheric wind flow 1n complex terrain 1s difficult because of the
complex Interaction between the air flows and the lower boundary of the
atmosphere. As a basis for our discussion (1n Section 2.3) of various
meteorological models that are potential candidates for prescribing the
wind flow over complex terrain, we first summarize the fundamental laws
governing atmospheric motion, and then describe qualitatively the nature
of large-scale atmospheric flow as Influenced by complex terrain.
2.1.1 The Primitive Equations
The atmosphere 1s currently considered to be a hydrodynamlc and thermo-
dynamlc system that can be analyzed using equations from the basic laws
and concepts of physics. The three fields of physics that are most
applicable to the atmosphere are thermodynamics, radiation, and hydro-
dynamics. Using a Cartesian (x,y,z,) coordinate system and sealer nota-
tion, the basic governing equations that describe the evolution of the
state of the atmosphere can be written as follows (see Hess, 1959 or
Plelke, 1984):
« - [f— pu + ~ pv + |r pw (Conservation of mass or (2-1)
L3X ay " -J continuity
equation)
-jr « — rr •* 2nv sin* - 2ow cos* •*• Fw
at p 3X X
+s (Conservation of heat or (2-2)
J potential temperature
2-1
-------
cos* + F (Conservation of momentum (2-3)
or equations of motion)
(Conservation of water) (2-4)
^r--Kr + vaf + waz *s*m ."-1.2.—
d U I oA ojf o^ I in
(Conservation of other (2-5)
species)
e » Tv(1000/p(mb))Rd/C (Definition of potential (2-6)
p temperature)
p - p Rd Ty (Ideal gas law) (2-7)
T » T(l + 0.61 q,) (Definition of virtual (2-8)
temperature)
where the symbols 1n Equations 2-1 through 2-8 have the following defi-
nitions:
o « density;
u,v, and w » wind components 1n the x, y, and z directions
respectively;
0 * potential temperature;
S0 • source/sink term for the potential temperature;
n » angular velocity of the earth;
p * pressure;
Fx, Fy, and F2 » frlctlonal forces 1n the x, y, and z directions;
2-2
-------
qn « concentration of water 1n the solid, liquid, and gase-
ous phases for n » 1, 2, and 3, with corresponding
source/sink terms, Sqn;
xra and $x - concentrations and source/sink terms for any other
m species (m » 1, 2, ...);
Tv » virtual temperature;
T « temperature;
R
-------
Thus, we are primarily concerned with those meteorological variables used
by existing add deposition models. These variables are generally as fol-
lows:
Transport winds governing flows of air parcels over, around, and
through complex terrain.
The turbulent structure of the atmosphere.
The location, types, and amounts of precipitation.
Above the surface boundary layer, the motion of the air 1n the lower atmo-
sphere, particularly 1n mid latitudes, 1s approximately determined by a
balance between the horizontal pressure force and the Cor1oils force.*
The wind resulting from this balance 1s parallel to the constant pressure
contours or Isobars, and 1s termed the geostrophlc wind. For situations
1n which the Isobars are curved, the wind flow 1s further modified by the
centrifugal force and 1s known as the gradient flow. Deviations from the
geostrophlc (or gradient) motion can also be caused by frlctlonal force
and horizontal density gradients. For example, modification of the geo-
strophlc flow by the frlctlonal force near the earth's surface leads,
under certain Ideal conditions, to the well-known Ekman spiral 1n the
planetary boundary layer, I.e., a decrease in wind speed and a counter-
clockwise turning (In the Northern Hemisphere) of the wind direction as 1t
approaches the surface. If a significant horizontal gradient 1s present
1n the mean air density, local accelerations may also result, altering the
geostrophlc balance. This condition occurs at land-sea boundaries, along
fronts, and 1n the vicinity of thunder storms. The resulting modifica-
tions to the geostrophlc balance are typically of mescoscale dimensions.
In complex terrain, the presence of topographic features will modify the
predominantly horizontal flow of air 1n the planetary boundary layer
(which results from a balance between the pressure gradient, and Cor1ol1s
and frlctlonal forces). In the absence of any large-scale flow, topo-
graphic features can exhibit their own locally driven flows primarily by
means of thermal effects. Since add deposition Is associated with large-
scale transport, the correct specification of the synoptic winds 1s an
Important component for an acid deposition modeling system. Generally,
synoptic-scale winds can be calculated using existing upper-air meteoro-
logical measurements and meteorological model output from the National
* The Cor1ol1s force 1s an "apparent" force that originates because the
coordinate system within which the equations of motion are expressed 1s
a rotating coordinate system. This apparent force 1s most prom1nant 1n
the horizontal direction, 1s perpendicular to the motion, and has a
magnitude proportional to the velocity.
2-4
-------
Meteorological Center's Limited Fine Mesh Model (NMC/LFM). However, the
modification of these winds by complex terrain 1s not adequately resolved
using these techniques. The following subsections describe the types of
air flows found over complex-terrain and their relevence to mesoscale add
deposition modeling 1n the Rocky Mountains. Our discussion of the effects
of air flow 1n complex terrain have been divided Into two main categories
—synoptically driven terrain-Induced nonthermal flows and locally Induced
thermal flows.
2.1.2.1 Terrain-Driven A1r Flows
When large-scale horizontal air flows encounter complex terrain, the basic
law of conservation of mass (Equation 2-1) forces the flow to adjust to
the presence of the topographic features. These modifications to the
synoptic wind can be divided Into kinematic and dynamic effects. The
kinematic effects of mountainous terrain on atmospheric flow are probably
most well known. These effects consist of combinations of channeling,
gapping, lifting, or diverting.
When large-scale flows encounter valleys or canyons, the orientation of
the wind flow within the valley becomes parallel to the valley axis due to
the constraining effects of the bounding ridges. This phenomenon 1s known
as channeling and 1s Illustrated 1n Figure 2-1. The large-scale forcing
flow oblique to the valley axis at ridge height exhibits directional turn-
Ing with Increasing depth until 1t 1s nearly parallel to the axis at the
valley floor. Channeled winds are observed 1n many geographic settings
and meteorological conditions.
Wind flows may also be accelerated through mountain passes or saddles and
decelerated on the leeward side. This feature of the mountain wind field
1s known as gapping and 1s based on the conservation of mass within the
atmosphere. An Illustration of the gapping phenomenon Is shown 1n Figure
2-2.
When the horizontal air flow encounters a mountain barrier 1t may go over,
be diverted, or blocked by the obstacle, depending upon the kinetic energy
of the air parcel. When the air parcel has sufficient kinetic energy, the
result 1s orograflc lifting as shown 1n Figure 2-3. The maximum rate of
orographlc lifting occurs when the wind 1s perpendicular to the mountain
barrier. This rate 1s reduced when the wind strikes the barrier at angles
other than perpendicular. Over many mountains and hilly regions this
forced uplifting on the upwind slope causes condensation and/or sublima-
tion and precipitation and 1s an Important factor 1n the local water bud-
get. This orographlcally produced or enhanced precipitation has been
2-5
-------
Upper wind
Channeled wind
FIGURE 2-1. Channeling of upper winds within a valley.
FIGURE 2-2. Gapping of airflow through mountains notches. The flow accelerates
on the windward side and decelerates In the lee of the mountain. (Source: Perla
and Martlnelli, 1976).
FIGURE 2-3. Orographlc lifting. The maximum rate of orographic lifting occurs
the wind is perpendicular to the mountain barrier (left) and is reduced when the
wind strikes the barrier at angles other than perpendicular (right). (Source:
and Martinelli, 1976).
2-6
-------
observed 1n the Rocky Mountain region (Barry, 1973) and may be an Impor-
tant component of a Rocky Mountain mesoscale add deposition model system
since the most sensitive lakes are generally located at elevations at
which 1t occurs, and the lack of measurements 1n these areas would other-
wise preclude a knowledge of these events.
When stably stratified air encounters a terrain obstacle, the air will
either ascend or be diverted or blocked by the obstacle. Whether or not
the air flows over or around the obstacle depends on the Froude number,
Fr:
Fr « ;** u 4| > 0
where ws 1s the large-scale wind component perpendicular to the terrain
obstacle of height, H, e 1s the potential temperature, and g 1s the gravi-
tational constant. Associated with the terrain obstacle 1s the critical
Froude number, Frc, so that for Fr < Frc, the air 1s diverted around the
obstacle or blocked, and for Fr > Frc, the air will ascend over the
obstacle. The critical Froude number will vary for different terrain con-
figurations and basically represents the relative magnitude of kinetic
energy of the large-scale wind to the potential energy change needed to
move an air parcel located near the surface over the terrain barrier. The
ability of a meteorological model to correctly simulate whether air par-
cels ascend or are diverted around obstacles 1n the Rocky Mountains 1s an
Important feature 1n mesoscale meteorological modeling.
Mountain topography rarely exhibits simple geometric patterns; thus
several kinematic effects are superimposed upon one another. This super-
Imposition 1s Illustrated 1n Figure 2-4, which depicts a combination of
gapping and lifting effects due to topography. To describe the transport
of pollutants within the Rocky Mountain Region, a mesoscale meteorological
model would have to be able to simulate the kinematic effects of the com-
plex terrain. Depending on the number of available precipitation measure-
ments, 1t may also be Important that the model be capable of describing
orographlc effects on precipitation.
The dynamic effects caused by the Interaction between mountain topography
and the atmosphere are more difficult to describe than are kinematic
effects. In a stably stratified atmosphere, air forced vertically over a
ridge may oscillate 1n an Internal gravity-wave regime as 1t travels down-
wind. This phenomenon, called a lee wave pattern, may extend for miles
downwind of the triggering ridge. The wave pattern's amplitude and
character depend on the wind speed, temperature stratification, and the
size and shape of the obstacle. Following Fofchgott's (1949) classifi-
cation scheme (Figure 2-5), air flow types can be characterl2ed as
follows:
2-7
-------
FIGURE 2-4. Superposition of gapping and
lifting effects on the mountain wind flow.
(Source: Perla and Martinelli, 1976).
2-3
-------
2 LEE WAVES
(b) Mending tddy dreaming
(d) Mtaf Mraomlng
(•) rotor tfrwming
FIGURE 2-5. Types of air flow over ridges:
(a) laminar streaming; (b) standing eddy stream-
Ing; (c) wave streaming; (d) rotor streaming;
(e) rotor streaming. Dashed Hne on left Indi-
cates vertical profile of horizontal wind speed.
(After Forchgott, 1949.)
2-9
-------
Laminar Streaming—Under light winds, flow over the ridges consists
of a smooth, shallow wave; only feeble vertical currents exist and
downstream phenomena do not occur.
Standing Eddy Streaming—With stronger winds, a large semipermanent
eddy forms to the lee of the mountain, creating a larger effective
shape of the mountain with respect to the flow aloft.
Wave Streaming—With even stronger winds, Increasing with height, a
lee wave system develops downwind of the mountain ridge.
Rotor Streaming—Under extremely strong winds, which are limited to a
restricted vertical depth (on the order of the mountain's height),
severe turbulence and quasi-stationary rotating vortices occur 1n the
lee of the mountain ridge.
Of course, the air flow patterns Identified by Forchtgott are simplifica-
tions of actual flow patterns. In reality, mountainous terrain exhibits
many ridges, some oriented 1n parallel, and others 1n oblique, direc-
tions. This juxtaposition of ridge orientations leads to extremely com-
plex conditions of lee wave enhancement and suppression, depending on
atmospheric stability, wind speed, and upwind terrain characteristics.
Although the spatial scale relevant to mountain lee wave formation 1s com-
parable to the width of the mountain, flow modification also occurs on
smaller scales when the wind encounters a ridge. Upwind of the ridge, the
flow 1s accelerated; on the lee side, the flow encounters an adverse pres-
sure gradient that retards the ground-level winds. If the lee slope 1s
steep, boundary-layer separation occurs resulting 1n a reverse flow near
the ground. This leads to Intense shearing and turbulent mixing 1n the
eddy region on the lee side of a ridge. The general features of the tur-
bulent wake region downwind of a ridge are shown 1n Figure 2-6. When the
concept of boundary-layer flow separation 1s extended to a series of
ridges such as 1n the Rocky Mountains, 1t 1s clear that the description of
atmospheric turbulence over rugged terrain becomes considerably confounded
by the numerous regions of flow acceleration, deceleration, separation,
and so on. For the purposes of this application, 1s 1s Important to
describe the mean transport of air flow across these ridges and the resul-
tant turbulent structure of the atmopshere.
Another dynamic effect Involves the degree of coupling between ridgetop
winds and the winds within valleys and canyons. As an example, consider
the canyon shown 1n Figure 2-7. Above the rldgetops, the upper-level wind
1s essentially undisturbed by the presence of the canyon. However, at the
level of the ridge, part of the flow's momentum 1s transported downward,
2-10
-------
FIGURE 2-6. Envelope and cavity regions behind a two-dimensional ridge
(Source: Huber, 1976).
FIGURE 2-7. Mountain top influences on the upper-
level flow component and the downward transporting
of flow momentum. (Source: Start et al., 1976).
2-11
-------
leading to air circulation within the canyon. Depending on many factors
(e.g., wind speed, stability, geometry), the circulation within the canyon
may exhibit various degrees of coupling with the flow aloft. For example,
1f (1) the winds aloft are strong, (2) the air 1n the canyon 1s stably
stratified, and (3) the canyon 1s deep, circulation within the canyon may
be essentially decoupled from the flow aloft. In contrast, light winds
over a relatively shallow canyon may result 1n little, 1f any, secondary
circulation. Somewhere 1n between these extremes lies the canyon-wind
configuration depicted 1n Figure 2-7. The separation of flows above the
canyon from those flows within the canyon Illustrates the need to describe
the three-dimensional nature of flows 1n complex terrain. This need 1s
further emphasized 1n the following discussion of flows derived from
thermal-radiation effects.
2.1.2.2 Thermally Induced A1r Flows
In addition to the modification of the geostrophlc wind by terrain fea-
tures, modifications may also arise from thermal-rad1at1onal effects.
Local wind flows 1n valleys and mountain passes may be generated by
spatial variations 1n surface temperature resulting from variable surface
cover or vegetation and varying slope aspects (I.e., angles of Incidence
of the sun's rays). During light geostroplc wind conditions, thermally
Induced circulations (e.g., mountain-valley winds, slope winds) may be the
dominant transport mechanism near the ground; however, under neutral-lapse
and strong wind conditions, for example, kinematic and dynamic perturba-
tions to the mountain wind field far exceed thermal-rad1at1onal effects.
The thermally forced local winds that develop 1n mesoscale areas of
Irregular terrain are generally recognized to be of two types—slope winds
(upslope and downslope) and mountain-valley winds. These mesoscale per-
turbations are generated from the differential heating between air parcels
located near the ground surface and the free atmosphere at the same eleva-
tion some distance away.
Upslope winds (anabatic winds) are driven by buoyancy forces as the higher
terrain becomes an elevated heat source 1n response to solar heating dur-
ing the day, with warm air moving upslope. On the other hand, downslope
winds (katabatic or drainage flow winds) are driven by the gravity force
as the higher terrain becomes an elevated heat sink due to long-wave
radiative cooling during the night so that cold air moves downslope. When
the ground is snow-covered, upslope winds often diminish because much of
the solar radiation 1s reflected back out Into space and 1s therefore un-
able to heat the slope. Snow cover has also been known to Induce drainage
flow winds during the day, a time when such winds usually do not occur.
2-12
-------
Mountain-valley winds blow along the main axis of a valley. They are
driven by the horizontal pressure gradient caused by temperature differ-
ences along the longitudinal axis of the valley. This difference 1s a
result of the accumulation of cold air 1n the valley caused by the drain-
age flow or the evacuation of warm air caused by the upslope flow. Winds
near the surface generally blow up the valley (valley winds) during the
day and down the valley (mountain winds) during the night. Generally,
topographic gradients along the sldewall slope are steeper than those
along the valley floor, hence, slope winds tend to develop faster and
earlier than mountain-valley winds. Also, slope winds tend to be shal-
lower and stronger than mountain-valley winds.
On sunny days or clear nights, when the synoptic pressure gradient 1s weak
(e.g., under stagnant conditions), local winds are often well-developed 1n
complex terrain. The flow 1s expected to be close to the diurnal varia-
tion, as shown under the simplified terrain configuration presented 1n
Figure 2-8. The white arrows 1n this figure correspond to up- and down-
slope flows, and the dark arrows represent the 1n- and out-valley flows.
For mesoscale meteorological modeling within the Rocky Mountain region,
the correct specification of the upslope/downslope winds will be Impor-
tant. This 1s further collaborated by the recent work of Fehsenfeld and
his co-workers (e.g., Roberts et al., 1985), 1n which elevated concentra-
tions of primary emitted pollutants (NOX, hydrocarbons) were observed at
high elevations on the east side of the Rocky Mountains during upslope
wind conditions. In some cases, these eastern upslope winds occurred even
though the synoptic wind was a weak westerly. Under these conditions, the
thermally Induced anabatic winds were decoupled from the synoptic-scale
winds. As these eastern anabatic winds carry the warmer air parcels with
their associated Inversion upslope, at some point the synoptic-scale
westerly wind 1s entrained Into the Inversion, thus slowing and eventually
reversing the flow. These conditions Illustrate the necessity of obtain-
ing three-dimensional wind fields to correctly specify flow within the
Rocky Mountains.
Another dynamic meteorological event that takes place 1n complex terrain
Is the formation and destruction of temperature Inversions. The destruc-
tion of an Inversion 1n a mountain valley has been found to differ from
the well-known Inversion breakup over the plains (WhHeman and McKee,
1979). A shallow nocturnal Inversion over flat terrain 1s broken pri-
marily by the growth of a convectlve boundary layer (CBL), which develops
from the ground up during the day. In contrast, the destruction of the
Inversion 1n the valleys is a combination of the growth of the CBL and the
descent of the top of the inversion. The growth of the CBL results from
the strong solar heating of the ground that occurs over flat terrain. The
descent of the top of the Inversion, however, 1s a result of the upslope
2-13
-------
(o) Sunrise
(b) Forenoon
(c) Noon and early afternoon
(d) Late afternoon
(•) Evening
(f) Early night
(g) Middle of night
Late night to morning
FIGURE 2-8. Several types of possible wind flows in a mountain valley.
The along-valley flow is indicated by dark arrows; up- and down-valley
flow by white arrows. (Adapted from Defant, 1951).
2-14
-------
wind, which removes mass from near the valley floor. This thermally
driven divergence of mass under the CBL and the resultant strong subsi-
dence will retard the growth of the CBL unless the Inversion becomes shal-
low enough to be completely removed by surface-based turbulent convectlve
heating.
The Inversion destruction that occurs 1n the valleys 1s Illustrated In
Figure 2-9 from WMteman and McKee (1979). Such an Inversion destruction
can be affected by snow cover, which may may cause the failure of the CBL
growth as a result of the large fraction of solar radiation that 1s
reflected back out Into space. Under such conditions, the cold pool 1n
the valley may persist for days.
The prescription of the proper Inversion height 1s another Input require-
ment for most existing acid deposition models, though past add deposition
modeling sensitivity studies have Indicated that results are not as sensi-
tive to the Inversion height as to the specification of transport winds
and precipitation amounts (Latlmer et al.t 1985). However, the Importance
of prescribing the proper Inversion height 1n complex-terrain situations
has still not been quantified.
Several meteorological phenomena have not been discussed 1n this section
because they are of lesser Importance 1n describing the complex-terrain
meteorology of the Rocky Mountains. These phenomena Include land-sea
breezes, lake breezes, and the effects of tropical storms.
2.2 CLASSIFICATION OF EXISTING MESOSCALE METEOROLOGICAL MODELS
The ability of a wind model to generate winds used as Input to photochemi-
cal air quality models depends strongly on the form of the mathematical
equations that are used to represent the physical processes that Influence
wind speed and directions throughout the modeling region. In this section
we examine how wind models can be classified according to their mathemati-
cal formulation to produce a taxonomlc tree of formulation characteristics
for the candidate mesoscale models. All models under consideration are
mesoscale models with a grid size of 5 to 50 km. These models typically
extend over distances of 100 to 2000 km; therefore, mathematical terms
Important at only planetary length and time scales are removed from con-
sideration.
A number of general characteristics can be used to classify the formula-
tion of mathematical equations. These characteristics include
Coordinate system
Solution method
2-15
-------
HI
0
•«, ««,
OOWN-UkLLCV 0 L*-VALLEY
«IMO SPCCO
(a)
(b)
(c)
FIGURE 2-9. Typical inversion destruction in a mountain valley
(descent of inversion top and continuous growth of CBL) (a) potential
temperature (e) profiles from time tj (before sunrise) to tn. (after
inversion destruction), (b) inversion height h and CBL height H,
(c) wind speeds. (Source: Whiteman and McKee, 1979).
L-16
-------
Form of the governing equations
Simplifying assumptions
Parameterization of processes
Turbulence closure
We discuss the various options available under each of the characteristics
for the candidate models and also try to Indicate the types of options
that naturally go together.
Coordinate systems can be cast 1n two frames of reference: EuleHan or
Lagranglan. EuleHan systems are fixed with respect to location, while
Lagranglan systems follow a parcel trajectory (no advectlve terms). For
the purpose of this study, we focus mainly on Eulerlan formulations
because 1t 1s necessary to describe the three-dimensional wind field 1n a
mesoscale region containing the sources and receptors of Interest.
2.2.1 Coordinate Systems
The coordinate system used has a fundamental effect on the governing
equations. A slab symmetric (x, z) model with only one horizontal and
vertical dimension has a simplified form. All terms with v and
3nv/axn (1 - 1, 2, 3; n « 0, 1, 2, ...) are zero and the Cor1ol1s terms
automatically disappear. The advectlve terms are slmpHfed to two terms,
as 1s the shallow fluid continuity equation. The equations of motion are
simplified to time-dependent equations for two variables—U and W.
Single-layer (x, y) models have no terms with W and aJJ/ax", and again the
advectlve terms and the shallow fluid terms are reduced to two terms.
Two-dimensional models have limited application since only one component
of vortldty can be represented. This limits the representation of turbu-
lent transport since phenonomena such as vortex tilting or stretching can-
not occur. The two-dimensional slab symmetric representation 1s good for
representing Infinite ridges, but this limitation 1s unrealistic for
divergences created 1n typical complex terrain. For single-layer models,
the lack of vertical velocity 1s unrealistic since vertical transports of
momentum and heat are Ignored, so that the effects of boundary layer
stresses and heat fluxes must be parameterized Indirectly.
Three-dimensional coordinate systems use several'methods to represent the
vertical coordinate. Cartesian coordinates are adequate for flat terrain,
but 1n complex terrain the use of the z coordinate means that the problem
of terrain Intersections must be dealt with. A simple terrain-following
coordinate can be used, 1n which z' 1s simply set equal to z minus the
terrain height H (x, y). The effects of terrain can also be treated in a
2-17
-------
straightforward manner by using sigma coordinates: The terrain is incor-
porated using the pressure or height difference relative to a reference
pressure or terrain height and dividing by the reference pressure or total
model region thickness so that the coordinate ranges from zero at the
ground to 1 at the top, e. g.,
°
-------
2.2.3 Form of The Governing Equations
Objective analysis methods have mathematical forms that can be divided
Into two types: those that use arbitrary observation weights and those
that try to obtain an Interpolation solution that minimizes a particular
quantity. The latter method 1s generally referred to as a variational
approach. In the former method, some simple technique such as l/rn
weights for the Interpolation may be used.
Diagnostic methods span a wide range of simplifications to the governing
equations; however, all models reviewed are nondivergent. The simplest
system 1s essentially a solution for mass continuity, subject to some con-
straints. One constraint might be to solve for a wind field with minimum
vortldty or horizontal divergence. Variational methods are often used to
obtain solutions when Implementing such constraints. Horizontal diver-
gence can be entered in controlled amounts at the boundaries to account
for phenomena such as subsidence.
Diagnostic methods can also be formulated 1n terms of the equations of
motion 1n a fashion similar to the prognostic solution method. The
governing equations for both diagnostic and prognostic methods can be
written either 1n terms of the wind components, V^, or as a stream func-
tion and velocity potential. The former method of expression 1s known as
the primitive equation formulation when the equations of motion accompany
a thermodynamlc, mass continuity, and real gas law equation. The latter
method of expression 1s generally called a vorticlty/divergence equation
formulation. D1abat1c and microphyslcal processes are usually implemented
1n the framework of the primitive equation formulation, and rarely in the
vort1c1ty/d1vergence formulation. The exceptions are some frontogenesls
models not considered 1n this model selection study. While often present
in prognostic models, dlabatlc effects such as latent heating and cloud
formation are generally not simulated 1n diagnostic models since these
processes are, by nature, generally not steady-state ones. Some
prognostic models may be nonhydrostatlc and have a separate diagnostic
perturbation pressure equation. However, this equation 1s generally not
amenable to a steady-state solution in a diagnostic model.
2.2.4 Simplifying Assumptions
One of the most basic simplifying assumptions is that of anelasticity
(ap/at » 0) so that sound waves do not propogate. A further assumption
that is useful when the model region 1s compared to the atmosphere scale
height 1s the shallow fluid approximation, e.g.,
2-19
-------
• p V » 0 anelastic
v • V * 0 shallow water-nondivergent
A number of simplifications can be made to the vertical velocity equation
through the scaling of various terms. For scale motions larger than 10
km, the vertical acceleration terms are generally small compared with the
adjusting accelerations that maintain a balance between gravity and verti-
cal pressure gradient forces. When a balance between these two forces 1s
assumed, the condition 1s assumed to be a hydrostatic balance and the
vertical equation of motion reduces to the hydrostatic equation:
IE , -pg
32 Py
For smaller scales, density effects become Important, e.g., free convec-
tion and surface_dra1nage flows. These density variations can be treated
as small (p1 « pref) deviations from the reference hydrostatic balance
state. When the vertical velocity acceleration of air parcels can be
written 1n terms of gravity accelerations arising from small factional
differences 1n density, e.g.,
Dw . o* „
ot r9
then the formulation 1s commonly known as the Bousslnesq approximation.
The thermodynamic equation can be expressed in one of two ways: either
heat 1s added or subtracted from a parcel or else such thermal processes
are Ignored. The former condition 1s known as the diabatic case, while
the latter case 1s known as the adlabatlc approximation. Ad1abat1c pro-
cesses are appropriate for dry models that contain no radiative or latent
heat transfer processes.
The momentum equations can be written in terms of force-balance relation-
ships, e.g., the Ekman layer balance in which Coriolis, pressure gradient,
and friction forces are all assumed to be balanced. Another example is
the gradient wind balance in which Coriolis, pressure gradient, and cen-
trifugal forces balance. These force-balance relationships are sometimes
used 1n simplified diagnostic models for the wind fields. However, none
of the models examined rely solely on simple force-balance relationships.
2-20
-------
2.2.5 Parameterization of Processes
The simplest forcing effects are due to topography and changes 1n surface
characteristics. Changes in roughness length change the surface layer
wind profiles and the surface layer momentum and heat fluxes. Topography
Introduces drainage flow effects. Radiative forcing comes from radiative
parcel cooling or heating. An example 1s the radiative long-wave cooling
of air parcels near the surface in the evening. Latent heating 1s another
source of heating that occurs 1f water vapor and liquid water continuity
equations are present in the model formulation. Latent heat forcing 1s
generally referred to as a 'wet' formulation 1n prognostic models.
2.2.6 Turbulence Closure
None of the wind models examined are of such a high spatial and temporal
resolution that they can be called full turbulence, or large eddy, simula-
tion models. Only one of the candidate models 1s a second-order closure
model that has separate equations for momentum and temperature flux
covarlances (e.g., u'w1, w'T', V}, etc.). Most of the models considered
are first-order closure models commonly referred to as K-theory models.
The covarlances between wind components and scalars are parameterized in
terms of an eddy diffusivity times a mean gradient, e.g.,
Some of these K-theory models also assume no countergradlent fluxes by
Ignoring diagonal terms 1n the eddy diffusivity tensor, K^. A further
assumption 1s that horizontal turbulence 1s homogeneous, so that KXX -
Kyy. The simplest assumption typically made is to assume that only Kzz is
significantly nonzero.
Most of the models that were reviewed can be taxonomically classified by
the tree structure shown 1n Figure 2-10. This taxonomic tree shows which
formulation options and assumptions are realizable, I.e., one cannot
simply take any combination of options 1n the columns 1n Table 2-1 and
construct a model. For example, a diagnostic adiabatic model formulated
with governing vorticity and divergence equations does not have latent
heating. The taxonomic tree provides some insight regarding the classes
of model formulation options typical of the mesoscale models examined in
this study. We will use this taxonomic tree to classify those models that
emerge as strong candidates.
2-21
-------
Objective
Analysis
Varlattonal
Hethods
Weighted
Interpolation
Dlabatlc
roi
i
ro
Eulerian
3-DlMens tonal
Coordinate
syste*
Diagnostic
Prognostic
Primitive
Vorticity/ .
Divergence
Mass
Continuity
Hydrostatic
Mlabatlc
Dlabatlc
Adiabatlc
Horizontally Divergent
Nondlvergent
Elastic
Aiwlattlr («rf/nr DllbatlC
Boussinesq)
Adiabatlc
Primitive and m
f.ln.l.K.1 ll liul/iir _P!fM\U
boussinesq)
Vorticltu/ Dlabatlc
Divergence
Adiabatlc
Wet
Dry
Uet
Drv
FIGURE 2-10. A taxonomlc tree of candidate mesoscale models grouped by realizable model formulation
characteristics.
-------
TABLE 2-1. Wind model characteristics used for model classification.
Coordinate System
Solution Method
Form of
Governing Equations
SI uglifying
Assumptions
Parameterization
of Processes
Turbulence Closure
(x, z)
slab symmetric
(x. y)
single layer
(x. y, z)
Cartesian
(x, y, s(z))
slgma coordinates/
terrain following
(x. y, s(p))
slgma coordinates/
surface pressure
Objective analysis Interpolation formulas
Diagnostic
Prognostic
VaMatlonal relations
Continuity equation
Vortlcity/divergence
formulation
Primitive equations
Incompressible
Nondlvergent
Hydrostatic
Adlabatlc
Anelastlc
Bousslnesq
Surface effects
Topography
Radiative heating
Latent heating
K-theory (full tensor)
Second-order closure
Large eddy simulation
-------
2.3 SURVEY OF EXISTING MESOSCALE METEOROLOGICAL MODELS
This section presents a survey of currently published mesoscale meteoro-
logical models. This survey pertains only to those models that have been
published 1n available journals at this time. Some of these models have
been used by different researchers with minor modifications; 1n these
cases, the model contains model attributes most relevant to this study and
1s only listed once. Several mesoscale meteorological models that
describe single processes (e.g., the cumulus cloud mlcrophyslcal model)
are not Included 1n this survey, which focuses on those mesoscale meteoro-
logical models with the potential capabilities of simulating the
meteorology 1n a mesoscale region within the Rocky Mountains. The survey
1s most likely Incomplete and the model descriptions not totally compre-
hensive. However, the Information obtained 1s sufficient to allow selec-
tion of candidate models for describing the meteorology of the Rocky Moun-
tains.
Our examination of available meteorological literature 1s summarized 1n
Table 2-2, which presents 65 existing mesoscale meteorological models.
These models were developed by universities, national laboratories,
research groups, and private firms. Particularly useful 1n this survey
was the previous work by Plelke (1984) and Anthes (1983). The first
column 1n Table 2-2 contains the name of the model developers (and model
name, when appropriate). The second column lists available references
pertaining to development and application of the model. The third column
contains the model formulation, using the generic classification scheme
given 1n Table 2-1 (each line 1n column 3 of Table 2-2 corresponds to a
category listed 1n Table 2-1, I.e., coordinate system, solution method,
form of governing equations, assumptions, parameterization of processes,
and turbulence closure). This Information was obtained from the refer-
ences given 1n column 2; however, there may have been more recent model
developments or the model may have capabilities not listed 1n the reference,
The fourth column of Table 2-2 Indicates the operational status of the
model. "Operational status" as used here 1s a subjective decision based on
whether the model could be adapted for mesoscale meteorological modeling 1n
the Rocky Mountains without an extensive effort on the part of the modeler
that would be outside the scope of this project. If the model code or
model simulation results are deemed difficult to obtain, then the model is
classified as nonoperational. Research grade models are also termed
nonoperational in this context. Research grade models include models that
treat Idealized conditions (e.g., an elliptical mountain barrier) or that
are used primarily to Investigate the sensitivity of meteorology to
different forces or parameters.
2-24
-------
TABLE 2-2. Existing numerical mesoscale Meteorological Models.
Model Developer and Name
References
Model Type
Operational
Status
Complex
Terrain
r\j
Alpert-Neuman Mesometeorological Model,
P. Alpert, A. Cohen, E. Doron, J. Neumann
Atlantic Oceanographlc 4 Meteorological
Laboratory/Hurricane Research Division,
Robert H. Jones
Atmospheric Environment Service,
Boundary Layer Research Division,
Ontario, Canada. P. Taylor, J. Ualmsley,
J. Salnon
en Australian Regional Primitive Equation
(AttPE). J.L. McGregor. L.M. Leslie.
U.S. Gauntlett. G.A. Mills
California Institute of Technology
C1T Hind Model
U.K. Goodin. 6.J. McRae, J.H. Seinfeld
Center for the Environment and Man,
J. P. Pandolfo, C. A. Jacobs.
P. S. Brown, Jr. » M. A. Atwater
Chisholm Institute of Technology,
NUATHOS
0. ti. Koss, I. Smith
Alpert, 1981
Alpert & Neumann, 1983
Jones. 1980
Ualmsley, Salmon ft
Taylor. 1982
McGregor et al., 1978
Hills & Hayden. 1983
Leslie et al.. 1981
Mills. 1981
Goodin. McRae & Seinfeld,
1979. 1980. 1981
Brown et al., 1982
Ross t Smith. 1986
(x, y. o(p))
Prognostic
Primitive equation
Hydrostatic
Surface effects/topography/
radiation
(x, y, o(p))
Prognostic
Primitive equation
Hydrostatic
Latent heating
(x, y, 2)
Primitive equation
Nonhydrostatic
Topography
(x. y. o(p))
Prognostic
Primitive equation
Hydrostatic
Surface effects/ topography/
latent heating
(x. y, o(z))
Diagnostic
Mass continuity
Topography
(x. y. z)
Prognostic
Primitive equation
Hydrostatic
Radiation/latent heating
(x, y. o(z))
Diagnostic
Mass continuity
Topography
No
Yes
No
No
Yes
No
Yes
Yes
No
Yes
Yes
Yes
No
Yes
-------
TABLE 2-2. Continued.
Model Developer and Name
References
Model Type
Operational
Status
Complex
Terrain
Colorado State Mesoscale Model.
rt.A. Pielke. R. Aritt, R. Kessler.
H. McCumber. R.T. NcNider. J. McQueen
Y. Mahrer, A. Hiizi, H. Segal, J.L. Song
Colorado State University
3-0 Cloud/Mesoscale Model
Pielke. 1974
McNIder. Hanna ft Pielke. 1981
Mahrer ft Pielke. 1978
McNider ft Pielke. 1984
Tripoli ft Cotton. 1982
Colorado State University
CSU RAMS
R. Pielke
(This is a merging of Pielke's 3-D mesoscale
model with the Tripoli ft Cotton Model)
CSIKU (Commonwealth Scientific
Industrial Research Organization)
Division of Atmospheric Research
Aspendale. Australia
U. Physlck. J. Garratt, ft A. Troop
Institut fur Physik der Atmosphare
UFVLH. Oberpfaffenhofen. FRG
F. ScMieski, K. P. Ho ink a
Urexel University
DKEXtL/NCAR
U. Herkey, C. Kreiubery, C. B. Chang
Fandry ft Leslie. 1984
F. SoMleskl. 1984
K. P. Holnka. 198S
Maddox, Perkey ft Fritsch. 1981
Chang. Perkey ft Krettzberg.
1981
(x. y. o(z))
Prognostic
Primitive equation
Hydrostatic
Radiation/topography/ surface
effects/latent heating
(x, y. o(z))
Prognostic
Primitive equation
Hydrostatlc/Bousslnesq
Surface effects/topography/
radiation/latent heating
(x, y. o(z))
Prognostic
Primitive equation
Hydrostatic
Surface effects/ topography/
radiation/ latent heating
(x. o(z))
Prognostic
Primitive equation
Hydrostatic
Surface effects/radiation
(x. y. z)
Prognostic
Primitive equation
Hydrostati c/adiabatlc
Topography
(x, y. o(z))
Prognostic
Primitive equation
Hydrostatic/diabatic
Surface effects/ radiation/
topography/latent heating
Yes
Yes
No
Yes
No
Yes
No
Yes
No
Yes
Yes
Yes
-------
TABLE 2-2. Continued.
Model Developer and Name
References
Model Type
Operational Complex
Status Terrain
rvj
i
r\i
European Center for Hediurn-Range
Ueather Forecast (Iiuited Area Model)
£CMUF/LAM
L. Del l'l)sso
Etablissement d'Etudes et de Recherches
Meteorologiques, Centre National de la
Kecherche Heteoroloyi^ue (Toulouse,
France). Team: NC2--C. Blondln.
S. Blere. B. Bret. P. Lacarrere,
0. Tfiual. Collaborators: G. Therry.
J. C. Andre
Florida State University.
T. N. Krishnaiurtl. R. Pasch.
S. L. Nam
Frontogenesls Modeling within the Meso-
scale Analysis and Modeling Group of
the Uoddard Laboratory for Atmosphere
Sciences. NASA/Uoddard Space Flight
Center—0. Keyser, M. J. Pecnlck
Geophysical Fluid Dynamics Lab,
Movable Nested Grid Model
Y. Kurihara. M. A. Bender, R. E. Tuleya
Geophysical Fluid Dynamics Lab
1. Orlanski S B.B. Ross
Dell-Osso. 1984
Burrldge ft Haseler. 1977
Tiedtke et al.. 1979
Blondln. 1979
Krishnamurtl et al.,
1976. 1979. 1983
Keyser ft Anthes, 1982
Kurihara & Bender. 19BO
Tuleya, Bender ft Kurihara,
1983
Bender, Tuleya ft Kurihara.
19U5
Ross ft OrtansM. 19B2
Orlanski. 1981
Orlanski ft Ross. 1977
Orlanski ft Ross. 1984
(x. y. z)
Prognostic
Primitive equation
Hydrostatlc/dlabatlc
Topography
(x. y. »(*))
Prognostic
Primitive equation
Hydrostatic/1nconpresslble
or anelastlc/Bousslnesq
Radlat1on/topography
(x. y. p)
Primitive equation
Hydrostatic/radiation/latent
heat/surface effects
U. o(P))
Prognostic
Primitive equation
HydrostatIc/adlabat1c
(x. y. o(p))
Prognostic
Primitive equation
Hydrostatic/latent heating
(x. y. z)
Prognostic
Primitive equation
Hydrostatlc/anelastic/dlabatlc
No
Yes
No
Yes
Yes
No
No
Yes
Yes
No
No
Yes
-------
TABLE 2-2. Continued.
Model Developer and Name
References
Model Type
Operational Complex
Status Terrain
ro
I
CO
Geophysical Fluid Oynmamics Lab,
1. Urlanski, K. Hiyakoda. 0. Miller
Iowa State University
£. S. Takle, L. P. Chang. R. 0. Russell,
J.A. Herwehe
Iowa State University
1-U Boundary Layer Model
U. Meilraan, R. Dobosy
Meslnger A Stricter,
1982
Chang, Takle I Sanl. 1982
Hell man ft Dobosy. 1985
Lawrence Llvermore National Laboratory (LLNL) Sherman. 1978
Mass-Consistent Wind Field Model (MATHEU).
II. Walker
LLNL
Finite Element Boussinesq
R. L. Lee, J. M. Leone, Jr., P. M. Gresho
LLNL
Finite Element Anelastic
S.I. Chan, P.M. Gresho. C.D. Upson
LLNL
MAbCON
Uickerson, M. H.
Lee et al., 1982
Chan. Rodean ft Ermak, 1982
Oickerson. 1978
(x. y. o(p))
Prognostic
Primitive equation
Hydrostatic/radiation/
surface effects
Prognostic
Primitive equation
Hydrostatic/radiation/
surface effects
(x)
Prognostic
Primitive equation
Hydrostatic
Topography /radiation
(x, y. i)
Diagnostic
Mass continuity/variations!
Adiabatlc
Topography
(x. y, z)
Prognostic
Primitive equation
Nonhydrostatl c/BoussI nesq
(x, y, z)
Prognostic
Primitive equation
Nonhydrostatl c/Anelastlc
(x. y)
Diagnostic
Varlational
Topography
No
Yes
No
Yes
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
-------
TABLE 2-2. Continued.
Model Developer and Mane
References
Model Type
Operational Complex
Status Terrain
r\>
i
ro
10
Los Alamos National Laboratory
T. Yamada
Los Alamos National Laboratory
A THUS
C. U. Davis, S. S. Bunker. J. P. Mutschlecher
Mesoscale Environmental Simulations and
Operations (MESO)
Hesoscale Atmospheric Simulation
System (MASS). M. Kaplan. J. Zack.
J. Procter
National Meteorological Center
Quasi-Lagranglan Nested Grid Model
M. B. Mathur
National Center for Atmospheric Research
NCAH/MM4--R. Anthes. Ying-Hwa Kuo.
Phil Haagenson. T. Warner. N. Seaman.
J.M. Frltsch
NCAK
Hoist Mountain Airflow Model
U. Uurran, J. B. Kleinp
Mel lor ft Yamada 1974;
Yamada ft Mel lor. 1975;
Yamada ft Mel lor, 1979;
Yamada. 1978, 1979, 1980
1981. 1982. 1983
Dlckerson. 1980
Tracletlal. 1978
Davis, Bunker ft Mutschelecher
1984
Kaplan et al., 1982;
Kaplan et al., 1984;
Cram ft Kaplan, 198S;
Koch. 1985;
Koch et at.. 1985;
Kodn et al.. 1985
Proctor, 1985
Zack et al., 1983
Uccelllnl et al.. 1983
Cram. 1985
Kocln et al., 1984
Wong et al.. 1983
Mathur. 1983
Anthes ft Warner, 1978
Benjamin ft Carlson. 1986
Durran ft Klemp, 1982, 1983
(x, Y. o(z))
Prognostic
Primitive equation
Hydrostatlc/diabatic
Topography/radlatIon
Second Order Turbulence Closure
(x. y. o(z))
Diagnostic
Continuity equation/
vaNational equations
Adlabatic
Topography
(x. y. o(p))
Prognostic
Primitive equation
Hydrostatic
Surface effects/ topography/
radiation/latent heating
Yes
Yes
Yes
Yes
Yes
Yes
(x, y, o(p))/lagrang1an
Prognostic
Primitive equation
Hydrostatlc/dlabatic
Topography
(x. y, o(p))
Prognostic
Primitive equation
Hydrostatic
Surface effects/ topography/
radiation/latent heating
(x, z)
Prognostic
Primitive equation
Nonhydrostatlc/dlabatlc
Topography
Yes
Yes
Yes
Yes
No
Yes
-------
TABLE 2-2. Continued.
Model Developer and Na«
References
Model Type
Operational
Status
Complex
Terrain
i
CO
o
NCAR
Objective Analysis Procedure
S. G. Benjamin, N. L. Seaman
NCAR
Nested Grid Model
T. L. Clark. K. 0. Turley
(requires running several Models In
parallel)
NCAR
Katabatic Wind Model
U. R. Kitzjarrald
NCAR
Objective Hind Field Analysis
L. L. Sapp
National Oceanoyraphtc and Atmospheric
Administration/Environmental Research
Laboratory (NUAA/ERL): E.C. Nlckerson.
H.A. Olas. J.M. Brown; LAMP (France):
E. Richard. N. Chaunerliac. R. Rosset,
U.K. Smith
NOAA/AUML - Hurricane Research Division
H.H. WIMouuhby, H.L. Jin
S.J. Lord. J.M. PiotroMlcz
National Institute for Environmental
Studies. Jdpan
H. Ueda
Benjamin I Seaman. 1985
Clark. 1977. 1979. 1982
Clark 1 Farley. 1984
Clark & Gall. 1982
Peltier 1 Clark. 1983
Fitzjarrald. 1984
Ceselskl 1 Sapp. 1975
Nlckerson. 1979
Nickerson et al.. 1986
Ullloughby et al.. 1984
Ueda, 1983
(x. y)
Objective analyses
Interpolation formulas
(x, y. o(z))
Prognostic
Primitive equation
Nonhydrostatlc or hydrostatic/
anelastlc
Surface effects/topography/
latent heating
(x)
Prognostic
Hydrostatic
Radiation/topography
(x. y)
Objective analysis
interpolation formulas
Hydrostatic/adiabatic
(x. y. o(p))
Prognostic
Primitive equation
Hydrostatlc/radlatlon/latent heating/
topography/surface effects
(r. 8)
Prognostic
Primitive equation
Latent heating
(x. y)
Prognostic
Primitive equation
Non-hydrostatlc/Boussinesq
Tes
No
No
Yes
Yes
No
No
No
Yes
Yes
No
Yes
No
No
-------
TABLE 2-2. Continued.
Model Developer and Nan
References
Model Type
Operational Complex
Status Terrain
fxj
i
L.J
National Severe Storms Laboratory
C. E. Hane
Naval Environmental Prediction
Research Facility (000)
H. Hodur
Naval Environmental Prediction Research
Facility
P. H. Tag. U. H. Thompson
Ohio State University
2-0 Numeric*I Node I
J.I). Carlson. H.K. Foster
Uregon St«te University
J.U. Ueardorft, K. Uegoshi. Y-J Han
Pacific Morttwest Lab
MELSAR Meteorological Preprocessor
K.J. Allwlne. C.O. Uhiteman
Pennsylvania State University
Kinematic transport Node)
J.I. Warner. K.K. Mil, N.L. SeaMn
Hane. 1973
Hodur. 1982
Tag ft Rosmond. 1980
Tag. 1983
Fett ft Tag. 1984
Carlson C Foster. 1986a.b
Deardorff, Ueyosht ft Han.
1984
Allwlne ft Uhlteman.- 1985
Warner. Fizz ft SeaMn. 1983
U. l)
Prognostic
Vortlclty formulation
Nonhydrostatlc/anelastic/
latent heating
(x. y. «(p))
Prognostic
Primitive equation
Hydrostatic/radiation/ latent
heating/surface effects
(x. y. o(z))
Prognostic
Primitive equation
Hydrostatic or nonhydrostatic/
inelastic
Surface effects/radiation
(*. *)
Diagnostic
Radiation
(x. y. o(p))
Prognostic
Primitive equation
Hydrostatic/adrlabatlc
Radiation/Surface effects
(*. y. o(z)>
Diagnostic
Mass continuity
Anelastic/adiabatic
Topography
(«. y. 0(1))
Objective analysis
Interpolation formulas
Adlabattc
No
No
No
Yes
No
No
Yes
Yes
No
No
Yes
Yes
Yes
No
-------
TABLE 2-2. Continued.
Model Developer and Nane
References
Model Type
Operational Complex
Status Terrain
CJ
ro
Purdue University
Gravity Uave Node)
W.-Y. Sun
San Jose State University
R. Bornstein. S. Klotz
Savannah River Laboratory (SRL)
Drainage Flow Model (20FLOW)
Stanford Research Institute (SRI) Int.
C.M. Bhunralkar. R.E. Endllch.
F.L. Ludwiy
SKI International
Diagnostic Wind Model
R.M. Endllch, F.L. Ludwig. C.M. Bhumralkar
System Applications. Inc.
Complex Terrain Wind Model
r. A. Yocke. T. C. Myers, and M. K. Liu
Systems Applications, Inc.*
Mountain-Valley Wind Model
G. £. Moore, C. Ualy, and H. K. Liu
Sun. 1984
Bornstein. 1972. 1975
Garret & Smith 1984
Bhumralkar. 1974. 1973
Endllch. Ludwig &
Bhumralkar. 1982 '
Yocke, 1981;
Yocke. Liu t McElroy, 1977
Moore et al.. 1986
(x, y. z)
Prognostic
Primitive equation
Hydrostatic or nonhydrostatic/
anelastlc
(x, y. z)
Prognostic
Vortlclty
Nonhydrostatlc/Bousslnesq/
surface effects/radiation
(x. y)
Prognostic
Primitive equation
Hydrostatic/dlabatlc
Surface effects/topography
(x. o(z))
Prognostic
Primitive equation
Hydrostatic
Surface effects/latent heating
(x. y. o(z))
Diagnostic
Mass continuity
Adiabatic
Topography
(x. y. z)
Diagnostic
Continuity equation
Hydrostatic/adiabatlc
Surface effects/topography
(x. y. z)
Diagnostic
Primitive equation
Hydrostatic/diabatic
Surface effects/topography
No
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
No
Yes
Yes
Yes
-------
TABLE 2-2. Continued.
Model Developer and Nai
References
NodeI Type
Operational Complex
Status Terrain
i
OJ
u>
Technical University, Darmstadt
•FITHAH" (Flow Terrain with Natural
Anthropogenic Heat Sources)
United Kingdom Meteorological Office
Mesoscale Model
B. W. folding, K.N.B. Smith, R. J. Purser.
and N. Hachin
University of Alaska
Z. Sorbjan
University of Alberta
Linear Hydrostatic Model
U.S. Phillips
U.S. Forest Service
KRISSV
M. Fosberg
University of Miami
0. Boudra
University of Miami
K. Bleck
Uallbaum. 1982
Voge), Srob A WippenMnn.
1986
Pfluger, 1903
Tapp t Unite. 1976
Sorbjan. 1984
Phillips. 1984
Fosberg. 1984
Boudra. 1981
Unpublished
(x, y, o(z))
Prognostic
Primitive equation
Nonhydrostatic/anelastic/Boussinesq
Topography
(x, y. o(z)J
Prognostic
Primitive equation
Nonhydrostatlc/anelastlc
Surface effects/topography
radiation/latent heating
(x. 2)
Diagnostic
Primitive equation
Adiabatlc
(x. y. o(p))
Prognostic
Primitive equation
Hydrostatlc/adlabatlc
Topography
(x, y, o(z))
Diagnostic
Mass continuity
Topography
Prognostic
Primitive equation
Hydrostatic
Surface effects/radiation/
latent heating
(x. y, o(p))
Prognostic
Primitive equation
Hydrostatlc/adlabatlc
No
No
No
No
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
No
Yes
-------
TABLE 2-2. Concluded.
Model Developer and Name
References
Model Type
Operational Complex
Status Terrain
i
CO
University of Utah
J. Paegle, 0. U. McLawhorn. E. N. Yeh
University of Washington
One-Level Mesoscale Model
University of Wisconsin, Milwaukee
R. J. BaDentine
University of Wyoming
T. K. Parish
Yatsushiro National College of
Technology. Japan
Y. Ookouchl
Paegle & McLawhorn, 1983
Mass & Oempsey. 1985
Ballentlne, 1980, 1982
ParMsh. 1984
(x. y. o(z))
Prognostic
Primitive equation
Hydrostatlc/anelastlc
Surface effects/ radiation
(x. y)
Prognostic
Primitive equation
Hydrostatlc/dlabatlc
Surface effects/topography
(x. y. o(p))/(x. y. o(z))
Prognostic
Primitive equation
Incompressible/hydrostatic
Radiation/latent heat/topography
surface effects
(x.y, O(P))
Prognostic
Primitive equation
Hydrostat 1 c/radlat1 on/topography
Ookouchl. Uryu ft Sawada. 1978 (x. y. o(z))
Ookouchl ft Uakata. 1982 Prognostic
Primitive equation
Hydrostatic/surface effects/
radiation
No
Yes
No
No
No
No
Yes
Yes
Yes
Yes
-------
The last column of Table 2-2 pertains to the model's capability to
explicitly treat complex terrain. By explicit treatment of complex ter-
rain, we mean the complex terrain of the Rocky Mountains. For some situa-
tions, such as modeling of hurricane land fall, terrain effects can be
adequately treated by Increasing the frlctlonal effects. For the purposes
of modeling air flows over complex terrain 1n the Rocky Mountains, that
type of model would not be classified as having complex-terrain handling
capabilities.
2-35
-------
REVIEW OF ACID DEPOSITION MODELS
3.1 REVIEW OF ACID DEPOSITION PROCESSES
The physical and chemical processes that determine the fate of natural and
anthropogenic add precursor emissions are numerous, complex, and Inter-
twined. Successful modeling of pollutant deposition requires simulating
the most Important of the processes and Interactions. The most Important
often depend on the spatial or temporal scale of Interest and the geo-
graphical characteristics of the region of Interest (I.e., climate, land
form, soil, etc.).
Over the past decade a large volume of experimental and theoretical work
pertaining to pollutant deposition processes, and such related fields as
meteorology and atmospheric chemistry, has been compiled for the purpose
of understanding these processes and developing methods to mathematically
simulate pollutant deposition. Selection of the model best suited for
simulating pollutant deposition and deposition over a specific area, such
as the Rocky Mountain region, requires an understanding of the crucial
elements of the physical and chemical processes Involved, namely, pollu-
tant transport and dispersion, pollutant chemical transformations, and
pollutant removal. These processes are reviewed 1n this chapter. The
objectives of this review are to convey to the reader the complexity
associated with these processes and to Identify the key components of each
process that must be Incorporated Into mathematical models to realisti-
cally simulate pollutant deposition.
3.1.1 Transport and Dispersion
The distinction between transport and dispersion of acidic pollutants and
their precursors arises out of time and space scale considerations.
Transport 1s associated with the horizontal or vertical movement of an air
parcel's center of mass by an average wind velocity, which may be derived
by numerous methods from meteorological observations or meteorological
models. Dispersion refers to the increasing horizontal and vertical
spatial scale of the parcel's mass distribution caused by wind fluctuation
(I.e., turbulence) over temporal and spatial scales smaller than the
averaging time or space Interval.
3-1
-------
Atmospheric motion arises from a combination of dynamic and thermodynamlc
forces resulting from the atmosphere's moisture content, the lower boun-
dary characteristics (friction, heat capacity, etc.)t solar radiation, the
earth's rotation, and other factors. Scales of motion range from micro-
scale eddies that dissipate kinetic energy to planetary-scale waves that
Influence weather patterns around the globe. Wind fluctuations occurring
over short space and time scales (e.g., 1 - 500 m, 1 s - 5 m1n) are typi-
cally caused by mechanically or thermally Induced turbulence near the
earth's surface (or within clouds). Motions on these scales tend to
Influence pollutant dispersion locally (0 - 20 km from the emission
source). As pollutants travel farther from their source they are subjec-
ted to larger mesoscale fluctuations and finally larger synoptic-scale
(I.e., 2,000 km) and global-scale fluctuations. While the temporal fluc-
tuations on most scales are capable of being resolved by fast-response
Instruments, resolution of spatial variability requires Instrumented net-
works. The synoptic scale 1s the only scale resolvable by routine
meteorological measurements. This restriction 1s probably the most
crucial transport-related problem from the perspective of pollutant depo-
sition modeling. The following paragraphs briefly describe some chararac-
terlstlcs of mesoscale and synoptic flow features.
The lower troposphere 1s characterized by a mechanically and thermally
Induced turbulent boundary layer whose structure exhibits strong diurnal
fluctuations and also varies with passing weather systems and terrain
features. Under fair weather conditions during the daytime, the boundary
layer may be well mixed, with trapped pollutants transported by a well-
defined boundary-layer average transport wind, easily resolved by routine
meteorological measurements. As a typical fair weather evening
approaches, the organized transport breaks down. The mixed layer under-
goes a transition Into a stable layer that often exhibits mesoscale
1nert1al flow features, such as nocturnal jets. These shallow layers of
high wind velocity may carry pollutants great distances overnight. The
wind velocity shear associated with these jets can dissociate a polluted
air parcel, causing very rapid horizontal dispersion.
Vertical variations in horizontal winds are prevalent under stable night-
time conditions; they can also occur during the daytime as a consequence
of frictional forces near the surface or when the thermodynamic structure
of the boundary layer is baroclinlc, I.e., when a horizontal temperature
gradient exists. Baroclinity is most pronounced near fronts, although it
is a characteristic of flow patterns over all spatial scales. The wind
shear associated with baroclinity is also pronounced around mesoscale con-
vective complexes and thunderstorms (Maddox and Fritsch, 1984).
3-2
-------
Vertical shear of horizontal winds 1s usually, but not always, associated
with vertical motion. The complex three-dimensional flow accompanying a
synoptic low-pressure system and Us frontal features 1s Illustrated 1n
Figure 3-1. Three-dimensional flow characterizes numerous mesoscale cir-
culation patterns as well. Although much of the three-dimensional trans-
port occurs within the middle and upper troposphere, the vertical struc-
ture of the daytime boundary layer becomes more complicated as large-scale
or mesoscale Influences become Important. Under disturbed synoptic situa-
tions (I.e., low pressure) the boundary layer may be strongly Influenced
by frontal convergence, cloud-Induced vertical motions, and mesoscale
fluctuations associated with rain bands resulting in pollutants trans-
ported from the boundary layer Into the free atmosphere.
The presence of complex terrain also has a large effect on the mesoscale
wind flows. The mere presence of topographic features causes the synoptic
wind flow to deviate from Its largely horizontal flow creating complex
three-dimensional flow patterns. In addition, topographic features can
exhibit their own locally driven flows primarily by means of thermal
effects. These locally driven flows may be decoupled from the large-scale
flows, further complicating the wind flow pattern.
3.1.1.1 Modeling Transport and Dispersion
The most common method of determining the transport associated with
synoptic-scale eddies for regional air quality modeling utilizes radio-
sonde data for transport wind definition and simple time and space Inter-
polation methods to convert the spatial and temporal Intervals of observa-
tions Into those commensurate with the model (for a review, see Stewart
and L1u, 1982). Various definitions of transport wind Include the 850 mb
geostrophlc wind, the observed layer-averaged winds (lowest 1 km or 1.5
km, etc.), and surface winds adjusted to compensate for surface effects.
Spatial Interpolation schemes typically average nearby measurements using
various weighting functions (such as an Inverse distance-squared weighting
function), which may or may not be based on physical considerations. Tem-
poral Interpolation 1s most often linear, I.e., the three-hour value is
halfway between the zero-hour and six-hour values. Although these methods
are rather crude, they are economical and in widespread use.
More sophisticated transport wind definitions are based on dry or moist
isentropic constraints whereby an air parcel moves in such a way as to
conserve potential temperature (Fleagle, 1947; Davis and Wendell, 1976),
continuity constraints (Chien and Smith, 1973), or more complex dynamic
and thermodynamic constraints, such as those of the balance equations
(Krishnamurti, 1968). With many of these methods, spatial interpolation
3-3
-------
FIGURE 3-1. Schematic diagram of a synoptic low-pressure
system and the circulation associated with its warm front
(cross section A-A1), cold front (cross section B-B'), and
occluded front (cross section C-C1). (Source: NCAR, 1983,
and Godske et al., 1957.)
-------
0
FLOATING Id NEEDLES
FALLING IC£ NEEDLES
aOATINC FOG DROPS
"ia NUCLEI LEVEL"
SWWJ^»:^S£'^?*^^
200 400 tOO WO KM
FALLING SNOW
FALLING RAIN
FALLING DRIZZLE
' 0° C ISOTHERM
RELATIVE VELOCITY Of WARM AIR
RELATIVE VELOCITY Of COLD AIR
RELATIVE VELOCITY Of COLDEST AIR
FIGURE 3-1 (continued)
3-5
-------
1s governed by the physical constraints and, therefore, 1s not arbi-
trary. Methods also exist to Interpolate temporally without violating
constraints on the flow (see Bengtsson, Gh1l, and Kallen, 1981). With the
exception of 1sentrop1c methods, the computation Involved 1n these more
sophisticated techniques 1s considerable. Although they are potentially
capable of yielding more accurate three-dimensional synoptic flow fea-
tures, the Influences of mesoscale circulations, not resolved by the mea-
surement network, are still neglected. Techniques for resolving some of
these flow features Involve prognostic mesoscale and regional-scale
•odels, whose computational requirements are greater still. Fortunately,
the National Weather Service archives wind velocity from a prognostic
weather model, the Limited Area Fine Mesh (LFM) model. This source of
transport data to resolve synoptic scale flows 1s of great value to pollu-
tant deposition modeling.
The accuracy of transport winds calculated by various means 1s difficult
to assess because validation data consisting of properly "tagged" air par-
cels 1s sparse. Tetroon data, often selected for transport validation, 1s
often unrepresentative of an air parcel's motion (Danlelson, 1974). Gase-
ous tracers useful over regional scales have only recently become avail-
able and should prove useful for validating transport models 1n the
future.
3.1.1.2 Modeling Implications
Scales of atmospheric motion that cause transport and dispersion of add
precursors range from small eddies mechanically generated through frlc-
tlonal forces near the earth's surface to mesoscale circulation patterns,
synoptic weather patterns, and finally global planetary waves. One of the
largest uncertainties 1n modeling regional pollutant transport and pollu-
tant deposition stems from the Inability to characterize the horizontal
and vertical wind fluctuations on sub-synoptic scales from routinely
available meteorological data. Another source of uncertainty 1s the dif-
ficulty 1n selecting from the resolvable wind fields an appropriate trans-
port wind that characterizes the bulk movement of a polluted air parcel,
which, through diurnal boundary layer fluctuations and horizontal wind
shear, 1$ continuously being distorted.
To minimize the latter source of uncertainty, a modeling framework should
allow the detailed specification of wind fields, both in the horizontal
and vertical dimensions, and should not impose restrictions on air parcel
movement and deformation. In this regard, the Eulerlan framework (I.e.,
--o
-------
coordinates fixed 1n space) 1s superior to the Lagranglan framework (I.e.,
moving coordinates).*
Minimizing the uncertainty associated with resolving scales of motion
finer than those characterized by routine data 1s a problem that falls 1n
the realm of objective analysis techniques. Progress 1n this area has
been made 1n the development of weather prediction models (e.g., the Limi-
ted Area Fine Mesh model). To the extent possible, methods employed 1n
these models should be adopted for pollutant deposition modeling, as 1s
currently being done 1n the NCAR model development project (NCAR, 1983,
1985).
3.1.2 Chemical Transformations
Numerous measurements of precipitation composition 1n the northeastern
United States and abroad Indicate that precipitation acidity (free H* 1on
concentration) 1s primarily associated with the sulfate (SO.) and nitrate
(NOj) Ions. The pathways by which sulfurlc and nitric add are formed 1n
the troposphere from sulfur and nitrogen precursors has been the subject
of numerous theoretical and experimental studies. It 1s known that the
oxidation pathways of S02 and NOX are numerous and Intertwined, Involving
both gas-phase and aqueous-phase reactions. Current knowledge of reaction
pathways has been synthesized 1n recent review articles, such as those of
NCAR (1983), NRC (1983), HhHten (1983). as well as workshop proceedings,
such as the 1982 Oahlem Workshop on Atmospheric Chemistry (Goldberg,
1982). These publications therefore adequately serve as references from
which to assess those factors that, from a modeling perspective. Influence
add formation. Intricate details associated with the various reaction
mechanisms have been omitted 1n the review presented below so as not to
obscure the objectives of the review, namely, to Identify and prioritize
the factors responsible for the chemical transformation of add precursors
to adds.
3.1.2.1 Gas-Phase Chemistry
Numerous theoretical and experimental studies have Indicated that the most
Important pathways for the oxidation of precursor species (e.g., NO?, SO->)
to addle species (e.g., sulfate, nitrate) Involve the participation of
reactive Intermediate species (e.g., exdted molecules, atoms, and fr9e
radicals) formea by the absorption of sunlignt ay a variety of trace
Impurities (e.g., nitrogen oxides, hydrocarocns). In regions containing
Modeling approaches are discussed 1n greater detail 1n Section 3.3.
3-7
-------
NOX emissions, the absorption of ultraviolet light by N02 generates a
ground-state oxygen atom and NO. The oxygen atom can combine with molecu-
lar oxygen to form ozone, or can participate 1n oxidizing S02 to sul-
fate. To contribute directly to S02 oxidation, however, NOX concentra-
tions must be sufficient to maintain a high concentration of oxygen
atoms. Such conditions might be met locally within power plant plumes.
From an acid-formation perspective, the Important product from N02 dis-
sociation 1s not the oxygen atom per se, but the ozone formed by the fast
reactions of the oxygen atom with molecular oxygen.
Among other reaction pathways, ozone can reoxidlze NO to N02, photolyze to
generate an excited or normal oxygen atom, 0('D) or 0(3P), or can react
with a class of hydrocarbons (alkenes) to form highly reactive Intermedi-
ates. The consequence of the first pathway is the establishment of a
near-photo-steady-state relationship among NO, N02, and 03. The photo-
chemical reactions Involving these three species, I.e.,
Kl 3
N02 + hu -i NO + 0(JP)
0(3P) + 02 (4M) -^ 03 (4M)
K3
0 + NO -Z 0 + N0
can be written as
-------
Such radical-radical reactions produce peroxides. Most Important, the
simplest peroxy radicals, H02» react to form hydrogen peroxide, which
appears to be a major factor 1n S02 aqueous-phase oxidation (NCAR,
1983). Another consequence of the NO-to-N02 conversion via peroxy radi-
cals 1s the regeneration of the hydroxy radical. Thus the OH-ox1d1z1ng
reactions, coupled with the NO-to-N02 conversion, form a radical chain
propagation process (I.e., the oxidizing agent 1s regenerated by the pro-
cess).
The reaction of ozone with alkenes forms reactive Intermediate species
that appear to react readily with S02, though the exact mechanisms are not
certain. Some evidence suggests that the SC^-alkene-O^ oxidation mech-
anism 1s sensitive to background water vapor concentration (Cox and Pen-
kett, 1972). Mechanisms explaining the apparently slower S02 oxidation
under high relative humidity have been suggested by several chemists (see
NCAR, 1983). The sensitivity of S02 oxidation to water vapor 1s not due
to the hydroxy and peroxy radicals, whose concentrations are not affected
by water vapor. Apparently, the reactive Intermediate species formed from
the alkene-03 reactions react with water, diminishing the pool available
for S02 oxidation.
The role of ozone 1n facilitating add precursor oxidation leads to the
consideration of the Importance of stratospheric ozone. Although the fre-
quency and duration of stratospheric ozone Intrusions are still matters of
debate, their occurrence has been documented 1n conjunction with air pol-
lution studies (Mohnen, Hogan, and Coffey, 1977; Haagenson, Shapiro, and
Mlddleton, 1981). Stratospheric Intrusion of ozone 1s associated with a
folding 1n the tropopause (I.e., the Interface between the troposphere and
stratosphere), which usually occurs 1n conjunction with low-pressure
troughs during the spring and fall months. It 1s estimated that strato-
spheric ozone Intrusions and downwind mixing are responsible for a global
background ozone concentration level 1n the lower troposphere of roughly
30 to 50 ppb (Singh et al., 1980). Natural processes are therefore
responsible for a portion of the species that contribute to S02 oxidation.
A number of experimental studies (Calvert and Stockwell, 1983) have Indi-
cated that the oxidation of N02 by OH to form HN03 1s also very Important
and approximately 10 to 20 times as fast as the analogous S02-OH reac-
tion. Therefore those conditions that lead to high S02 oxidation rates
also contribute to high nitric add formation.
Another gas-phase N02 oxidation pathway Involves ozone and water vapor.
By this pathway, ozone oxidizes N02 to form N03, which can further combine
with N02, yielding N205. Reactions of N205 with water vapor and liquid
can lead to HN03. Nitric add formation by this pathway 1s expected to be
of minor Importance during daylight hours, but of greater Importance at
night.
3-9
-------
Modeling Implications. The complex pathways by which S02 and NOX are oxi-
dized 1n the gas phase require considerable simplification before they can
be treated 1n an economical manner within add deposition models. Because
a dominant pathway for both sulfuric and nitric add formation Involves
the OH radical and other reactive Intermediate species, knowledge of the
radical pool concentration should provide a strong Indication of the oxi-
dation rates.
Computer simulations of the detailed chemistry Indicate that with summer
midday Insolation under moderately polluted conditions (high OH concentra-
tions), S02 and NOX oxidation rates are approximately 3.7 ± 1.9 X/h and
34 ± 17 X/h, respectively (NRC, 1983). Midday winter Insolation under
moderately polluted conditions yields about one-fourth of the OH concen-
tration and reduces the oxidation rates of S02 and NOX to approximately
1 ± 0.5 X/h and 18 ± 9 X/h, respectively.
Deriving oxidation rates from detailed gas-phase chemistry simulations
appears to be the most realistic method of Incorporating the complex
chemical pathways Into an economical add deposition model. This method
of specifying oxidation rates has been adopted 1n several regional trans-
port models (see Sections 4 and 5) and 1s recommended for design of an
engineering model to apply to the Rocky Mountain region.
3.1.2.2 Aqueous-Phase Chemistry
Because of the Increasing awareness of acid precipitation over the past 10
to 15 years, the mechanisms of aqueous-phase sulfur and nitrogen oxidation
have received greater attention. The relative Importance of aqueous-phase
versus gas-phase chemical transformations depends on such cUmatologlcal
factors as the extent and frequency of cloud cover, relative humidity,
precipitation amount, solar Intensity, and concentration of various pollu-
tants. Recent evidence suggests that gaseous and aqueous-phase oxidation
are equally important 1n the formation of adds in the troposphere.
(1) Sulfur Dioxide Oxidation
Sulfur dioxide dissolves in water, forming the hydrate S02-H20, the bisul-
fite ion HSOZ, and the sulfite ion SOI. The dissociation is very rapid
and complete, with insignificant quantities of undissociated H^SO-i remain-
Ing. The dissociated species are referred to as S(IV) and their relative
abundance is a strong function of the solution pH (see Figure 2-2). The
3-10
-------
[s(E)]xlO-12 -
8
r9
10 II 12
FIGURE 3-2. Relationship between S(IV) (dissolved sulfur dioxide) and
the species that make up S(IV) for varying S(IV) concentrations and pH
values. Total aqueous S(IV) concentration is represented by the dashed
line and SOo-^O, HSO§ and SO? mole fractions are shown by the solid
lines as a function of pH at 25°C. 862 gas-aqueous equilibrium is
assumed. 502 Partial pressure is set at 10~9 atm.
(Source: NCAR 1983)
3-11
-------
concentration of dissolved S(IV) 1s also a function of the solution pH and
a function of the ambient gas-phase S02 concentration.
A number of proposed aqueous-phase oxidation pathways have been postulated
for the S(IV) species. These Include oxidation by dissolved 02 1n the
presence of catalysts (I.e., Iron or manganese), oxidation by dissolved
ozone, hydrogen peroxide (HpC^), and oxides of nitrogen. Figure 2-2 Indi-
cates that the primary S(IV) species at a pH level between 3 and 6 is the
bisulfite ion. Most of the postulated oxidation pathways therefore
Involve the HSO^ 1on. At pH greater than 6 the bisulfite tends to further
dissociate Into sulflte and H Ions. Note that as the pH increases, the
total dissolved S(IV) concentration Increases, provided of course that
sufficient ambient $62 1s available to the droplet. Thus, the overall
aqueous-phase oxidation rate of sulfur will be a complex function of the
droplet pH, the available ambient SC^, ambient oxldants, and their solu-
bility characteristics.
S(IV) as SO, appears to react with oxygen, producing sulfate SO* How-
ever, Scott and Hobbs (1967) demonstrated that without catalysts the reac-
tion was too slow under atmospheric conditions to be considered an impor-
tant pathway. However, when catalyzed by dissolved maganese or iron, the
oxidation rate increases dramatically. Copper has been postulated as a
catalyst as well, though there remains some debate as to its effective-
ness. Because the S(IV) speciatlon Indicates little quantity of SO! at pH
below 6, the Importance of this pathway falls off rapidly with decreasing
pH. The concentrations of dissolved catalysts are also pH-dependent. For
example, under high pH (pH > 4), laboratory studies indicate ferric 1on
concentrations fall to low levels, perhaps due to the Ion's precipitation
out of solution (Martin, 1983). This catalytic mechanism has been studied
recently and appears to involve the production of dissolved oxldants H02
and OH by reaction involving the metal ions.
Of the remaining aqueous-phase S(IV) oxidation pathways, the reactions
with dissolved Oj and HoO? appear to be especially significant for the
ambient levels of impurities commonly observed. Although the details of
these mechanisms are still not completely known, overall rates for the
following reactions have been experimentally determined:
03(aq) + HSO'(aq) -> S0*(aq) + H+(aq) + 02(aq)
H202(aq) -f HSO'(aq) * SOj(aq) + H*(aq) + H20(aq)
3-1?
-------
The 0- - HSOI reaction rate depends Inversely on the square root of H+(aq)
concentration for pH between 1 and 3 and Inversely on H+(aq) for pH 1n the
range 3 - 6.5 (Martin, 1983). The H-CL - HSOZ reaction rate, however,
appears much less sensitive to H"*"(aqJ concentrations over the range of pH
values commonly observed 1n rain and cloud water, and 1s therefore more
prominent at low pH values (Martin and Damschen, 1981).
Availability of ambient ozone to cloud or rainwater depends on the photo-
chemical production and the solubility characteristic of ozone. Although
ozone has a low solubility, Its relatively high concentrations (30 to 70
ppb) 1n rural areas ensure Us aqueous-phase availability as an oxidant.
Its ambient gas-phase availability Is likely to be greater 1n more pollu-
ted regions (e.g., either Immediately downwind of urban areas or under
highly stagnant regional scale pollutant episodes) or within regions
Influenced by occasional stratospheric Intrusion. The availability of
H202 as an aqueous-phase oxidant 1s less well understood. Solubility
characteristics are well known and Indicate very rapid Incorporation Into
the liquid phase. However, background levels are typically only a few
ppb. The gas-phase production of H202 depends on the concentration of the
hydroperoxy radical H02, which in turn depends 1n a complex way on NOX and
hydrocarbon concentrations. Computer simulations of the photochemistry
Indicate that gas-phase H202 production 1s not favored under the condi-
tions of very high NOX concentration levels (such as within a power plant
plume). However, under typical NOX and hydrocarbon concentrations 1n pol-
luted regions, the formation rate 1s higher than rural NOX and hydrocarbon
concentration levels. Aqueous-phase production of H202 also appears pos-
sible when other dissolved oxldants are present in the cloud or rain drop-
let.
The oxidations of S(IV) by nitrous add (HONO) and by N02 have also
emerged as potential pathways to sulfurlc add formation. Preliminary
estimates by Lee and Schwartz (1981) and Schwartz and White (1982) Indi-
cate that under high N02 and S02 concentrations the aqueous-phase
S0° production could be significant. Although the usually low ambient
levels of gaseous HONO render the HONO * S(IV) reaction pathway
unimportant, the potential of aqueous-phase HONO production 1s currently
receiving considerable attention.
Other aqueous-phase reactions with S(IV) are being Investigated. For
example, 1t has been speculated that formaldehyde may form a complex with
S(IV), thus prohibiting it from being oxidized (Richards et al., 1983).
The production of radicals in solution is a critical component of the
aqueous-phase oxidation processes currently receiving much attention.
With the emergence of the H202 + S(IV) pathway as a significant means of
sulfurlc add formation, all avenues of H202 production require extensive
3-13
-------
Investigation. The role of organlcs 1n producing aqueous-phase radicals,
which may subsequently produce H202, 1s a1so De1n9 actively pursued.
(2) H1tr1c Oxide Oxidation
Due to the rapid rate of gas-phase NOx-to-HN03 conversion, 1t was origi-
nally believed that the nitrate deposited 1n precipitation was primarily a
result of scavenging of nitric add formed within the gas phase. However,
recent observations suggest that gas-particle and aqueous-phase processes
are responsible for much of the formation of nitric and nitrous add. For
example, during a field study, Lazrus and co-workers (1983) found what
appeared to be continuously forming HN03 within warm frontal precipita-
tion, thus Implicating aqueous-phase transformation. Observations of very
low nocturnal nitrate concentrations and higher levels of MONO Indicate
that processes other than gas-phase chemistry are Important (see NRC,
1983, for references). However, Lee and Schwartz (1981) have determined
that aqueous oxidation of ML, alone 1s unlikely to contribute to formation
of nitrates because under all but the most extreme conditions (e.g., with-
in concentrated power plant plumes), the reaction rate constants and solu-
bility constants of NO and N02 are Insufficient to produce observable
nitrate concentrations.
Numerous pathways exist for the homogeneous aqueous-phase oxidation of
odd-nitrogen species (e.g., N203, N20*, N2°5» and N03^* Representative
rate constants of many of these reactions have been determined by several
Investigators (e.g., Martin, Damschen, and Judelkos, 1981; Epstein, Kus-
t1n, and S1moy1, 1982). However, analysis of conversion rates under typi-
cal tropospherlc concentrations suggests that, 1n all but the most extreme
conditions, homogeneous aqueous-phase reactions alone contribute negli-
gibly to formation of nitrous and nitric acids.
It now appears as though nitrate formation 1n precipitation depends on a
combination of gas-phase, gas-particle, and aqueous-phase reaction
mechanisms, and Involves reversible as well as Irreversible conversions.
Simplified reaction schemes suitable for Inclusion 1n regional add depo-
sition models are currently under development.
3.1.2.3 Modeling Implications
Several mechanisms responsible for aqueous phase oxidation of dissolved
sulfur species (S(IV)] have been Identified and are considered Important
1n add formation. These are the metal-Ion catalyzed reaction of S(IV)
with 03. the reaction of S(IV) with 03, and with H202. The aqueous-phase
reaction with H202 1s thought to be the most prominent under conditions of
3-14
-------
low aerosol or droplet pH. Since H202 (a reactive Intermediate species)
and 03 concentrations facilitate aqueous-phase sulfurlc add formation,
the same ambient conditions (relative to NOX and hydrocarbon concentra-
tions, and high Insolation) that promote gas-phase oxidation also lead to
higher aqueous-phase oxidation. However, the numerous reaction pathways
for H202 are not completely understood, nor are the potentially Important
metal-Ion catalyzed reaction mechanisms.
As with the gas-phase chemistry, detailed computer simulations have been
performed using mixed gas-phase and aqueous-phase S02 oxidation mechanisms
(Seigneur et al., 1984; Seigneur and Saxena, 1986). Deriving S02 oxida-
tion rates as a function of cloud amount, Insolation, background NOX, 03,
and hydrocarbon concentrations, etc., from these simulations appears to be
the most viable method of Including these Important chemical processes 1n
current pollutant deposition models.
Oxidation of NOX species within the aqueous phase appears to be negligible
compared with the heterogeneous reactions Involving the higher oxides of
nitrogen (I.e., N203, N204, N205, N03). Sufficient Information Is not
available for developing refined methods of mixed-phase and aqueous-phase
parameterization for these higher nitrogen oxides. No currently opera-
tional pollutant deposition model considers any NOX oxidation mechanisms
beyond simplified gas-phase mechanisms.
3.1.3 Dry Deposition
The process of dry deposition has, until recently, played a relatively
minor role 1n the overall characterization of add deposition. Only with-
in the last few years has the term "add deposition" replaced the popular
phrase "add rain" 1n environmental discussions. Part of the reason for
the slower recognition of the Importance of this process 1s the lack of
dry deposition measurements. Unlike add precipitation, which 1s rela-
tively easy to measure, no effective technique for quantifying the
partlculate and gaseous pollutant flux to surfaces 1s available for rou-
tine deployment in the field. Methods are available to measure the flux
onto surrogate surfaces, such as flat plates of different composition, but
quantifying the flux that Impinges and 1s retained by an Individual plant
or stand of trees has so far proven to be formidable.
Although there are many factors influencing the rate of dry deposition,
the end result of dry deposition can be characterized in a simple mathe-
matical formula,
V - F/C ,
3-15
-------
where
Vd = the "deposition velocity" (e.g., 1n cm/s),
P
F = the downward flux of pollutant (e.g., 1n ug/cnr per unit time) at
some representative level, and
C « the pollutant concentration at that level (e.g., 1n ug/cm3).
The mechanisms and factors responsible for determining the net flux are
extremely numerous and complex. However, as a basis for Incorporating the
dry deposition process 1n a regional or mesoscale model, these mechanisms
and factors should be elucidated.
Figure 3-3 Illustrates many of the major factors known to Influence the
rate of pollutant deposition to exposed surfaces. Broadly speaking, the
rate of dry deposition 1s Influenced by the vertical transfer of pollu-
tants within the lower atmosphere and the uptake of the pollutants at the
surface. A useful conceptual model to Incorporate these factors Into a
mathematical description of dry deposition 1s based on the analogy to
electrical flow through a series of resistors. The resistance to mass
transfer can be broadly grouped Into an aerodynamic and a surface resis-
tance that act 1n series to limit the transfer. Aerodynamic resistance 1s
often subdivided Into an additional component that Influences the pollu-
tant flux through a thin quasi-laminar layer adjacent to the surface. The
net downwind velocity of pollutant 1s related to the aerodynamic resis-
tance (ra), resistance of the laminar layer (rb), and surface resistance
(rs) by the formula Vd * l/(rfl + rb + rs).
3.1.3.1 Mechanistic Description
Turbulent transport 1s the mechanism that determines the aerodynamic
resistance. The turbulence Intensity 1s principally dependent on the
lower atmospheric stability and the surface roughness, and 1s therefore
easily determined from mlcrometeorological measurements and surface
characteristics. During typical daytime conditions, the turbulence
Intensity is large over a deep layer (I.e., the well-mixed layer), thus
exposing a large reservoir of pollutants to the process of surface depo-
sition. During the night, stable stratification of the atmosphere near
the surface reduces the intensity and vertical extent of the turbulence,
effectively diminishing the pollutant flux toward the surface. The aero-
dynamic resistance is independent of pollutant type (gas or particle)
except that gravitational settling must be taken into account for large
particles (> 2 wm). By assuming the resistance to mass transport is
3-16
-------
AIRBORNE SOURCE
LARGE PARTICLES
AERODYNAMIC
FACTORS
NEAR-SURFACE
PNQRETIC
EFFECTS
SETTLING
_L
TURBULENCE r
| THERMOPHORESIS |
'
ELECTROPHORESIS
OIFFUSIOPNORESIS
STEFAN FLOW
IMPACTION
QUASI-LAMINAR
LAYER
FACTORS
J_
INTERCEPTION
1
J
IRQWNIAN DIFFUSION \-
\
CASES
TUR1ULENCE
STEFAN FLOW
-\ MOLECULAR DIFFUSION |
SURFACE
PROPERTIES
[ORIENTATION] | STOMATA | | WETNESS |
' i , i
[ FLEXIilLITY I I WAXINE5S | | CHEMISTRY |
VESTITURE |
'
| MOTION I I EXUOATES |
| SMOOTHNESS I
_L
RECCPTOR
FIGURE 3-3. A schematic representation of processes likely
to influence the rate of dry deposition of airborne gases
and particles. Note that some factors affect both gaseous
•and particulate transfer, whereas others do not. However,
submicron particles are affected by all the factors that
inflence gases and large particles, and hence it is these
"accumulation-size-range" aerosols that present the greatest
challenge for deposition research. (Source: NRC, 1983)
3-17
-------
analogous to that of momentum transport, the magnitude of aerodynamic
resistance can be well approximated using m1crometeorolog1cal theory.
Adjacent to the surface lies a quasi-laminar layer whose resistance to
pollutant transfer 1s a complex function of pollutant characteristics.
The quasi-laminar layer 1s an Intermittent layer whose characteristics
depend on the smoothness of the surface and to some extent the variability
of the near-surface turbulence. This layer's resistance to pollutant
transport 1s very selective to pollutant type. For example, gaseous
transport across this thin layer depends on the molecular d1ffus1v1ty of
the gas—the analogy with turbulent momentum transport being no longer
valid. The gaseous transfer resistance 1s considerably larger than the
momentum transport resistance across the quasi-laminar layer.
Very small particles (< .01 pro) are transported across this thin layer by
Brownlan motion analogous to the gaseous molecular transport. However,
the particle-size dependency of the Brownlan d1ffus1v1ty coefficient
results 1n a decreasing transfer rate (I.e., Increasing resistance) with
Increasing particle size. The resistance 1s largest, and hence the depo-
sition velocity 1s smallest, within the size range 0.1-1 ym, the range
of most add particles. For particles larger than 2 ym, the particle
Inertia partially overcomes the resistance, thereby Increasing the trans-
fer rate. At still larger particle sizes (> 20 um), other effects that
may cause an Increase 1n resistance come Into play, such as particle
"bouncing". The minimal deposition velocity within the 0.1 - 1 vm range
1s fairly well documented from experimental data, Indicating the prominent
role played by rb 1n determining the overall resistance.
Other effects that can augment particle deposition near the surface
Include thermophoresls, electrophoresls, d1ffus1ophores1s, and Stefan flow
processes (see Fuchs, 1964). Thermophoretlc forces tend to cause
particles to flow away from hot surfaces and are most effective for small
particles (s 0.01 ym). D1ffus1ophoret1c effects arise near a surface
undergoing water condensation or evaporation. The flux of particles 1s 1n
the direction of the water vapor flux, and although the particle flux is
dependent on particle size, 1t 1s very small and usually negligible.
Stefan flow 1s a result of gas-phase water molecules being created or
destroyed 1n the vicinity of an evaporating or condensing surface. The
volume of molecules created (or removed) from the vicinity of the surface
causes a flow of pollutant particles away from (or to) the surface. Ste-
fan flow and diffuslophoretic effects therefore act 1n opposite ways. The
effects of Stefan flow are potentially capable of modifying the surface
deposition rates to aerodynamically smooth surfaces by an amount that is
larger than the deposition velocity for many small particles. However,
3-18
-------
the solubility of add particles counteracts this effect where dry depo-
sition to wet surfaces 1s concerned. Electrophoresls effects are
generally neglected for add particles because of the relatively low elec-
trical field strengths and low efficiency of this phoretlc effect within
the 0.1 - 1 pm size range (NRC, 1983).
Surface properties greatly Influence the rates of particle and gas depo-
sition, either directly through chemical reactions or Indirectly through
perturbing the quasi-laminar layer. The capture of gases by surfaces
depends primarily on the chemical or biological reactions that occur
between the gas and surface. Because of their reactivity, species such as
S02, N02, HNOo, 03, C02 are more readily deposited than are NO, CO, or
hydrocarbons (HC).
However, the specific locations on the plants where reactions occur often
depend on the plant's biological activity level. For example, the
diurnally variable stomatal resistance often Influences the net S02 and 03
deposition to plants (Shepherd, 1974; Wesely and Hicks, 1977; Wesely
et al., 1978). Daytime opening of the stomatal pores exposes more reac-
tive plant tissues to the pollutant, thus decreasing the plant's surface
resistance. During periods when stomatal openings are closed, the higher
epidermal (cutlcular) resistance plays a prominent role 1n overall surface
resistance.
The pubescence of plant leaves affects particle deposition (see e.g.,
Chamberlain, 1967; Martell, 1974). Deposition rates for tobacco plants
with pubescent leaves are of the order of a factor of 10 greater than for
plants with smooth, waxy leaf surfaces. Surface resistance to addle
species deposition 1s often negligible when the surface 1s covered by
water (e.g., dew or a lake surface). The high solubility of S02 and the
affinity for water of sulfate promote rapid deposition, provided the tur-
bulence of the overlying air 1s sufficient to bring pollutants toward the
surface. In the limiting case of trace constituents of low solubility,
the "wet" surface resistance 1s proportional to the Henry's law solubility
constant.
One further process linked to dry deposition 1s the process of resuspen-
slon and surface emissions. This process 1s probably an Important one 1n
arid regions and agricultural lands. The driving force causing particle
resuspenslon 1s the momentum transfer between the atmosphere and ground.
During gusty wind conditions the high frlctlonal forces are cabable of
lifting particles from the surface. These conditions are not limited to
widespread winds, but may occur locally under calm conditions 1n the form
of dust devils. The largest particles will not be frequently suspended,
nor will their movement be great because of their mass. Very small
3-19
-------
particles tend to be held to the surface by adhesive forces and tend to be
protected from resuspenslon within crevices or between larger particles.
Therefore, those particles most readily Influenced by resuspenslon pro-
cesses are those 1n the size range > 10 ym, I.e., coarse particles.
3.1.3.2 Experimental Data on Dry Deposition
Methods for measuring dry deposition fall Into three main categories:
Direct measurements using collection buckets, surrogate surfaces, and
mass budget analysis,
Laboratory studies utilizing sealed outdoor chambers, wind tunnels,
and controlled pipe-flow and flat-plate studies, and
M1crometeorolog1cal measurement methods that Include direct flux mea-
surements (eddy correlation) and flux-gradient measurements.
Direct measurements suffer from Interpretation problems because of the
unrepresentatlveness of the deposition receptacles. For example, Dasch
(1982) determined that for almost all chemically reactive particles, glass
surfaces provide the greatest flux estimates, while Teflon surfaces yield
the lowest. Neither material reproduces the variability of natural sur-
faces, which react chemically and biologically with the pollutants.
Because they are controlled, laboratory studies offer the opportunity to
Isolate the contribution made by specific resistances to total deposition
velocity. In a set of chamber experiments documented by H111 (1971), the
following ranking of trace gas deposition (relative to least surface
resistance) over oat and alfalfa canopies was determined:
Hydrogen fluoride
Sulfur dioxide
Chlorine
Nitrogen dioxide
Ozone
Carbon dioxide
Nitric oxide
Carbon monoxide
Solubility was Implicated by H111 (1971) as the controlling factor 1n
trace gas deposition to vegetation. A similar ranking of the resistance
of these species to deposition over clay and loam soils was reported by
Jude1k1s and Wren (1978).
3-20
-------
Wind tunnel data concerning particles suggests that fluxes of small
particles (0.1 - 1 ym) are limited by near-surface effects, particularly
the ability of particles to penetrate the quasi-laminar layer (Chamber-
lain, 1967; Sehmel, 1970; Sehmel, Sutter, and Dana, 1973). The deposition
of larger and smaller particles 1s aided by gravitational settling and
Brownlan diffusion, respectively.
M1crometeorolog1cal methods for measuring pollutant deposition are still
1n the development or refinement stages and are more suitable for Investi-
gating processes that affect dry deposition than for routine monitoring of
dry deposition, primarily because of the experimental care, sensitive and
costly Instrumentation, and extensive data analysis required.
The literature 1s replete with measurements or calculations of deposition
velocities and resistance for various gaseous and partlculate species over
various natural and artificial surfaces. Recent compilations of data have
been reported by Sehmel (1980) and NRC (1983). Tables 2-1 and 2-2 list
various measurements of deposition velocity (vd) and surface resistance
(rc) for S02, N02, and 03 and submlcron particles over natural surfaces.
3.1.3.3 Modeling Implications
The dry deposition of gases and particles depends on their transport
through the boundary layer toward the surface, their transport across
a thin quasi-laminar layer adjacent to the surface, and their final Inter-
action with the surface. The rate at which pollutants are delivered to
the surface depends on the nature of the pollutant (gas versus particles),
Us reactivity with the surface, and, 1n the case of particles, Its size.
Because of the complexity of the overall deposition process, regional
air quality models utilize considerably simplified parameterizations con-
sisting of a deposition velocity, vd, or a set of resistances (represent-
ing aerodynamic resistance, quasi-laminar resistance, and surface resis-
tance). The seasonal, diurnal, and spatial variability of the vegetation
over large portions of the United States dictates the need for a variable
deposition rate parameterization. For gaseous deposition, the "resis-
tance" approach appears superior to the "deposition velocity" approach
because 1t permits a greater degree of variability, I.e., the variability
due to the boundary-layer turbulence structure and vegetation charac-
teristics.
For particle deposition, the lack of surface resistance measurements
necessitates using the deposition velocity concept. Variations 1n depo-
sition rate due to the boundary-layer turbulence can be Introduced by
3-21
-------
TABLE 3-1. Recent experiments on trace-gas deposition to natural surfaces.
(Source: NRC, 1983)
Investigator*
Method
Results and Comments
SULFUR DIOXIDE
Hill (1971)
35
S02 with stable S02
carrier over alfalfa
35,
Garland et al. (1973) S02 over pasture
35,
Owers and Powell (1974) S02 over pasture
Shepherd (1974) S02 gradients over grass
Whelpdale and Shaw
(1974)
Garland (1977)
Fowler (1978)
Dannevlk et al. (1976)
Garland and Branson
(1977)
Belot (as summarized
by Chamberlain (1980)
Galbally et al. (1979)
Dovland and EHassen
(1976)
Barrle and Walmsley
(1978)
S02 gradients over snow,
water, and grass
S02 gradients, calcar-
eous soils
S02 gradients over
—Wheat
--Soybean
S02 gradients over wheat
35S02 to pine
34
S02 to pine
Eddy correlation over
pine forest
Accumulation to snow
Accumulation to snow
vrf « 2.3 cm/s,- daytime;
Implies r * 0.4 s/cm
v. « 1.2 cm/s, daytime;
u
r « 0.6 s/cm
v. « 1.3 cm/s, daytime;
v. « 1.3 cm/s> summer;
• 0.3 cm/s, autumn
v. « 1 cm/s 1n daytime;
for grass, water, snow
v. « 1.2 cm/s;
r «.0.01 s/cm
v . « 0.4 cm/s
v « 1.3 cm/s
v . »0.4 cm/s
d
vd * 0.1 - 0.6 cm/s
v. < 1 cm/s
d
« 0.2 cm/s
» 0.1 cm/s
0.2 cm/s
Continued
3-22
-------
TABLE 3-1 (Concluded).
Investigator*
Method
Results and Comments
NITROGEN DIOXIDE
Uesely et al. (1982)
OZONE
Eddy correlation over
soybeans
Galbally and Roy (1980) Gradients over wheat
Wesely et al. (1978, Eddy correlation over a
1982b) range of natural surfaces
vrf » 0.6 cm/s, daytime;
rg « 1.3 s/cm, daytime
« 1.5 s/cm, night
vrf « 0.7 cm/s;
Implies rg « 1.4 s/cm
r$ « 0.8 s/cm, daytime;
«1.8 s/cm, night
* Literature references are available 1n NRC, 1983.
3-23
-------
TABLE 3-2. Field experimental evaluation of the deposition velocity of
submicron diameter particles. (Source: NRC, 1983). Literature references
are available in NRC, 1983.
Surface
Size and Method
Results and Comments
SNOW
Dovland and Eliassen
(1976)
WeseTy and Hicks
(1979)
Lead aerosol, surface
sampling
0.05 - 0.1 un particles*
eddy correlation
0.16 cm/s in stable
stratification;
greater values in
neutral; all light-
wind data.
Net fluxes small but
upwards; vd too
small'to be determined.
OPEN WATER
Sievering et al.
(1979)
Williams et al.
(1978)
BARE SOIL
0.1 - 1.0 pm particles,
gradients
0.05 - 0.1 un particles,
eddy correlation
Wesely and Hicks
(1979)
0.05 - 0.1 urn particles,
eddy corelation
Gradients highly
variable; range of vd
typically 0.2-1.0 cm/s
in magnitude;
Including reversed
gradients in long-term
average reduces average
vd to near zero (see
Hicks and Williams,
1979).
Preliminary indications
only; vd very small,
95% certainty < 0.05
cm/s.
Surface frequently a
source; v^ very low, on
the average, but often
large for short periods.
Continued
3-24
-------
TABLE 3-2 (Continued)
Surface
Size and Method
Results and Comments
GRASS
Sehmel et al.
(1979)
Polydlspersed rhodamlne-B Average vd * 0.2 cm/s
particles with mass median
diameter 0.7 pm, de-
posited to artificial grass
exposed outdoors
Chamberlain (1960)
Hudson and Squires
(1978)
Davidson and- Fried-
lander (1978)
Radon daughters deposited
to natural grass; work
attributed to Megaw and
Chadwlck
Cloud condensation nuclei
fluxes measured by grad-
ients methods over sage-
brush and grass; particle
size probably 0.002-0.04 ym
~ 0.03 in particles,
gradients over wild oats
Weseley et al. (1977) 0.05 - 0.1 un particles,
eddy correlation
Everett et al. (1979) Partlculate lead and
sulfur, gradients
S1ever1ng (1982)
Hicks et al. (1982)
0.15 - 0.3 vrn particle
gradients over mature
rye and wheat
Sulfate by eddy
correlation
vd » 0.20 cm/s
vd « 0.04 cm/s
Average vd » 0.9 cm/s
Direction of flux some-
times changes; during •
deposition periods
vd « 0.8 cm/s, but
much lower on the
average.
vd greater for sulfur
(~1 cm/s) than for lead
from more local sources.
vd averaged 0.4 ±0.3
cm/s 1n light winds; un-
stable stratification.
vd as high as 0.7 cm/s
in daytime, about 0.2
cm/s as a long-term
average.
Continued
3-25
-------
TABLE 3-2 (Concluded)
Surface
Size and Method
Results and Comments
Weseley et al. (1982) Sulfate by eddy
vd largest for daytime
lush grass (-0.5 cm/s),
much less for short dry
grass (~0.2 cm/s);
strongly stable con-
ditions
CROPS
Droppo (1980)
Wesely and Hicks
(1979)
TREES
Hicks and Weseley
(1978, 1980)
Wesely and Hicks
(1979)
Undberg et al.
(1979)
Wesely et al. (1982)
Parti oil ate trace metals,
gradients, senescent
maize
0.05 - 0.1 pm particles,
eddy correlation,
senescent maize
Sulfate particles, eddy
correlation, Loblolly
pine
0.05 - 0.1 jin particles,
eddy correlation
Pb, Cd, S, etc. particles,
foliar washing
Sulfate particles, eddy
correlation
Vj varying widely with
element, ranging up to
about 1 cm/s
Strong diurnal varia-
tion 1n the direction
of the flux; long-term
average vd « 0.1 cm/s
Strong diurnal varia-
bility but less marked
than for small par-
ticles; average
vd « 0.7 cm/s
Very strong diurnal
variation with the
canopy a net source;
during deposition
periods vd 1s probably
greater than 0.6 cm/s
vd > 0.1 cm/s for all
quantities, on the
average
v^ not significantly
different from zero for
a winter decideous
forest
3-26
-------
defining the particle surface resistance, r_, 1n terms of the deposition
velocity (determined from the literature), the aerodynamic resistance,
?a, and the laminar layer resistance, r.. The overbars on the resistances
Indicate their values are determined under average meteorological condi-
tions, or under conditions during which the deposition velocity measure-
ments were made, I.e.,
This method 1s less restrictive than using a constant particle deposition
velocity.
Recognizing the large range of measurements due to a variety of underlying
surfaces, meteorological conditions, etc., Shelh, Wesely, and Hicks (1979)
prepared a land-use map of North America and coupled this with measure-
ments of surface resistance to S02 uptake and seasonally dependent sta-
bility conditions to yield a map of S02 deposition velocities directly
suitable for model Input.
Recent field measurements of surface resistance and gaseous deposition
velocity and high-resolution land use data can be used to calculate the
deposition velocity as a function of time of day, season, and location
directly within the model 1n a manner similar to that of Shelh, Wesely,
and Hicks (1979).
Where Insufficient Information 1s available for species other than SO?,
the following summary of properties and controlling processes (from NRC,
1983) serves as a guide 1n estimating deposition velocities for reactive
gases.
S02: Uptake by plants Is largely via stomates during daytime, with
about 25 percent apparently via the epidermis of leaves (Fow-
ler, 1978). At night, stomatal resistance will Increase sub-
stantially but cutlcular resistance should be unchanged.
When moisture condenses on the depositing surface, associated
resistance to transfer should be allowed to decrease to near
zero (Fowler, 1978; Murphy, 1976). For deposition to a
liquid-water surface, water vapor flux appears to provide an
acceptable analogy to S02 flux.
Oj: Behavior 1s like S02, but with significant cutlcular uptake
at night (r « 2 to 2.5 s/cm at night) and with surface mois-
ture effectively minimizing uptake. Deposition to water sur-
faces, 1n general, 1s very slow.
3-27
-------
NO,: Similar to 0, 1n overall deposition characteristics but with
a significant additional resistance (possibly mesophylUc) of
about 0.5 s/cm. Even though N02 1s Insoluble 1n water 1n low
concentrations, deposition to water surfaces might be quite
efficient. Chamber studies Indicate a similar overall sur-
face resistance for S02 and N02.
NO: Typical surface resistances are 1n the range 5 to 20 s/cm,
as Indicated by chamber studies and field experiments.
Nitric oxide appears to be emitted by surfaces at times, pos-
sibly as a consequence of N02 deposition and of the Intimate
linkage with ozone concentrations.
HNO*: No direct Information 1s available; however,, on the basis of
Its high solubility and chemical reactivity, substantial
similarity to HF should be expected. Consequently, the use
of rs » 0 appears to be a reasonable first approximation.
NHv Again, no direct measurements are available but 1n this case
similarity with S02 appears likely. Natural surfaces may be
emitters of NH3 because of a number of biological processes
occurring 1n and on soil.
3.1.4 Wet Deposition
Wet deposition of acidic substances 1s thought to be at least as Important
as dry deposition with regard to the overall pollutant delivery to the
ecosystem. Treatment of wet deposition on an episodic basis 1s very dif-
ficult because of the spatial and temporal variability of clouds and rain-
fall and the complexity of the m1crophys1cal pollutant attachment pro-
cesses. The following subsection briefly summarizes the mechanisms
responsible for wet deposition, discusses the roles of clouds, storm sys-
tems, and storm climatology 1n determining the distribution of wet depo-
sition, and reviews results of some field experiments. The discussion
draws from the review literature (e.g., NCAR, 1983; NRC, 1983; Stewart
and Uu, 1982), from specific publications describing mechanisms, para-
meterization methods, and monitoring efforts, and from our experience with
regional modeling.
3.1.4.1 Mechanistic Description
The sequence of events leading to wet pollutant deposition 1s represented
1n Figure 3-4 by four states of the pollutant and numerous transition
3-28
-------
w
KSUSPENSION
F^—
I
>
EVAPORATION. KSORPTION
— *
UJ
4
POLLUTANT
IN
CLEAR AIR
if
2*
o §
, iiil
-11 POLLUTANT AND
CONDENSED WATER
INTERMIXED IN
COMMON AIRSPACE
<
PI
<
ll
if
5
OLLUTANT ATTACHED
TO CONDENSED*
WATER ELEMENTS
i
i
5
<
RIACIION
jj ATTACHED
POLLUTANT MODIFIED
BY AQUEOUS-PHASE
PHYSIOCOCHEMICAL REACTIONS
1
KPOSIIION
•" POaUTANT DEPOSITED
ON
EARTH'S SURFACE
»—
^
e
^
s
i
V/l
1
FIGURE 3-4 . The scavenging sequence. (Source: NRC, 1983)
3-29
-------
steps leading from state to state. Most of these transition processes are
capable of operating 1n both directions, suggesting that the ultimate
delivery of pollutants to the surface may result from cyclic transitions.
Beginning with the first two states and their transition processes, 1t 1s
obvious that pollutants and condensed water have to commingle before wet
deposition can occur. This process of Intermixing can occur either 1n
situ or through pollutant advectlon relative to a cloud or storm system.
In situ commingling occurs when water vapor condenses on nuclei containing
the pollutant or 1n the same air space of a gaseous pollutant. Advectlon
of pollutants relative to clouds results from circulation Induced by the
cloud or the large-scale storm. Regardless of which mechanism occurs,
pollutants often remain airborne for a considerable length of time before
they encounter condensed water. Thus, over large time or space scales the
conversion process between the first and second pollutant state may be the
rate-determining step 1n the overall scavenging sequence. Note that the
evaporation and cloud detralnment (I.e., expulsion of air from a cloud)
processes can act 1n the opposite sense as well, separating the pollutant
from the condensed phase. Condensation nuclei often undergo many conden-
sation-evaporation cycles within the turbulent cloud environment before
their ultimate fate 1s determined (Junge, 1964).
Transition from the second to third pollutant state 1n Figure 3-4 Involves
a variety of attachment processes that depend on the nature of the pollu-
tant (I.e., gas versus aerosol) and Its chemical composition. This
attachment process can occur simultaneously with the 1n situ commingling
of pollutant and condensed water through nucleatlon, or can occur after
the commingling by a combination of dlffuslonal attachment, 1nert1al
1mpact1on, and phoretlc effects. Nucleatlon of cloud drops by aerosol
particles 1s a function of the affinity of the particle for water vapor,
which depends on particle composition and size. SulfuMc add, ammonium
sulfate, and ammonium nitrate aerosols are examples of Important conden-
sation nuclei over continental regions.
Dlffuslonal attachment results from gaseous or particle migration to a
water surface. The rate of gaseous absorption by hydrometeors depends on
the gas solubility and the gas molecular d1ffus1v1ty. For particles, the
rate of migration to the water surface 1s a function of the Brownlan d1f-
fuslvlty, which varies Inversely with particle size. As particle sizes
exceed approximately 1 ym, the Inertia! attachment mechanisms become
Important. These mechanisms can be viewed as the capture of smaller
particles by a large particle with a different velocity, and apply to the
capture of aerosols by cloud droplets, aerosols by raindrops, and cloud
droplets by raindrops. Impactlon occurs when the larger particle falls
relative to the smaller particle. Turbulent collision 1s a form of
3-30
-------
1nert1al 1mpact1on applicable to particle couples whose velocity differ-
ence arises because of a different Inert1al response to turbulent velocity
gradients, and generally applies to aerosol-cloud droplet collection. The
capture efficiency 1s a complex function of the size, shape, density, and
composition of the particles. Detailed discussions on these attachment
processes can be found 1n standard text books on cloud physics (e.g.,
Mason, 1971; Pruppacher and Klett, 1978).
The mechanisms of dlffuslonal and 1nert1al attachment are efficient for
removing particles less than 0.05 pm and greater than 1 mn, as Illustrated
1n Figure 3-5. Low removal efficiency 1s a characteristic of particles 1n
the "Greenfield gap" portion of the size spectrum (see Greenfield,
1957). Although electrical and phoretlc effects contribute toward collec-
tion of particles within this size range, a minimal capture efficiency
still exists relative to larger and smaller particles. This minimum 1s
associated with the size range of most secondary aerosol pollutants,
Including add particles.
The transition from the third to fourth pollutant state—the physlocheml-
cal transformation of pollutants by aqueous-phase processes (Figure 3-4)—
1s discussed 1n the previous subsection. In addition to their Importance
under precipitation conditions, these aqueous-phase reactions may occur
within nonprec1p1tat1ng clouds under fair weather conditions and signifi-
cantly enhance the overall oxidation rate of add precursors. Aqueous-
phase transformations may also affect the efficiency of the gaseous
scavenging by removing the dissolved species from the gas-Hqu1d equi-
librium reaction, hence encouraging continual absorption Into the water
droplet. For highly soluble gases, the Increase 1n absorption rate
resulting from these devo1atH1z1ng reactions 1s a function of the aque-
ous-phase reaction rate.
The final transition state—pollutant delivery to the surface (Figure
3-4)—can occur via rain and snow fall or direct 1mpact1on of cloud drop-
lets on the surface (fog). The mechanisms that ultimately govern the pol-
lutant concentration within precipitation or fog are the cloud and pre-
cipitation growth processes. Standard reference books on clouds discuss
the complexities associated with these processes (e.g., Mason, 1971; Prup-
pacher and Klett, 1978). With respect to pollutant scavenging, the Impor-
tant concept underlying droplet growth and precipitation Initiation pro-
cesses 1s that of competition between droplets (or 1ce particles) for
water vapor.
This competition affects pollutant mass acquisition as well. Cloud drop-
let growth by water vapor diffusion from nearby smaller droplets may
3-31
-------
o
z
—
5
UJ
UJ
oc
u
10° E
10-1
10-2
10-3
10-4
GREENFIELD GAP
UJ
I I I I Mill I I lllllll I I lllllll I I I I Mill I I llllll
10-3 10-2 10-1 1.0 10
RADIUS OF AEROSOL PARTICLE (//m)
FIGURE 3-5. Theoretical scavenging efficiency of a
falling raindrop as a function of aerosol particle
size. Dashed lines correspond to contributions by
electrical and phoretic effects under chosen humidity
and raindrop-charge conditions. (Source: NRC, 1983;
adapted from Pruppacher and Klett, 1978)
3-32
-------
release previously Incorporated pollutant back Into the "free" air. Addi-
tionally, the flux of water vapor between droplets may alter the thermo-
dynamlc conditions within droplets, promoting additional absorption or
degassing of pollutants. While the collection of cloud droplets by rain
or snow (attachment processes) affects pollutant concentrations within the
precipitation via mass-balance considerations, the ultimate pollutant
levels deposited on the surface will be strongly dependent on the history
of the hydrometeor and Its environment as 1t descends to the surface.
This applies equally to direct deposition of the pollutant-laden fog*
In order to Incorporate these processes parametrlcally within a regional
pollutant deposition model, the mechanisms must be either considerably
simplified or dispensed with 1n favor of an empirical approach. Insight
Into pursuing either of these pathways requires examining the character-
istics of the precipitation systems, I.e., storm types and storm climato-
logy.
3.1.4.2 Cloud/Storm Climatology
The spatial and temporal scale of precipitation systems ranges from Indi-
vidual clouds (- 5 km, 1 hr) to cyclones and their associated fronts (-
2000 km, 3-5 days). The associated precipitation can be broadly classi-
fied as convectlve or nonconvectlve, depending on cloud type. Although
storm systems typically contain both types of precipitation Individually
and 1n hybrid combinations, 1t 1s useful to make the distinction when con-
sidering pollutant scavenging.
Nonconvectlve precipitation 1s associated with clouds that grow 1n
response to gradual uplifting of the air, such as that associated with
warm fronts. Although local showery rain may occur 1n cells or bands of
convectlve origin, precipitation 1s generally of a light, uniform charac-
ter. A1r coming 1n contact with the cloud at m1d-tropospher1c levels
often has a lower tropospherlc origin and may be advected from a consider-
able distance before encountering the cloud system. A common example of
this 1s the steady southerly or southwesterly flow associated with the
warm sector of a frontal storm system. Precipitation chemistry monitoring
during this type of storm situation has shown that this "conveyor belt" of
moist air frequently contributes the largest fraction of total wet sulfate
deposition within regions of the northeastern United States (Raynor and
Hayes, 1982).
Convectlve precipitation 1s associated with clouds having more vigorous
vertical motion Induced by buoyancy forces. Thunderstorms are representa-
tive of the more Intense convectlve rainstorms. These showery precipita-
tion events are triggered Initially by cold fronts and squall lines (I.e.,
3-33
-------
mesoscale lines of Intense thunderstorms) or by Intense surface heating,
or by the uplifting of moist air masses due to elevated topographic fea-
tures such as the Rocky Mountains. Trajectories associated with convec-
tlve rain are complex and three-dimensional and depend to some extent on
the types of convectlve system. Isolated air mass thunderstorms have a
distinct low-level Inflow region providing moist air and pollutants from
medium-scale areas surrounding the storm. Frontal convectlve clouds typi-
cally are fed from low-level (I.e., boundary layer) and mid-level air, the
circulation being coupled with the larger-scale frontal circulation. Ray-
nor and Hayes (1982) have noted a second maximum 1n SO^ and NOI wet depo-
sition with cold-front-1nduced thunderstorms.
In addition to circulation and pollutant trajectories associated with
various storm systems, several different rain Initiation mechanisms depen-
dent on cloud type have a direct bearing on pollutant scavenging. These
have been Illustrated In the context of developing a parameterization of
regional sulfate scavenging for long-range transport models by Scott
(1978, 1982).
Scott (1978) classifies clouds Into three types: (a) warm or maritime
clouds whose rain Initiation does not depend on diffusion*! growth of 1ce
particles; (b) Bergeron clouds, within which rain Initiation Is dominated
by 1ce crystal growth (the Bergeron process); and (c) convectlve clouds,
whose precipitation droplet growth can be Initiated by either process and
1s strongly affected by the clouds' circulation. From theoretical argu-
ments, Scott (1978) parameterizes the washout ratio of sulfate (I.e., sul-
fate concentration 1n precipitation divided by sulfate concentration 1n
air) as a function of cloud type and rainfall rate. His results, pre-
sented 1n Figure 3-6, Illustrate the different scavenging abilities of
various cloud types, particularly evident at low precipitation rates.
For long-term predictions (I.e., annual or seasonal) of regional pollutant
deposition, 1t would appear advantageous to c!1matolog1cally distinguish
between warm frontal rain, thunderstorms, and perhaps other precipitation
types (I.e., terrain Induced). It also appears necessary to distinguish
rain Initiation characteristics cllmatologlcally because of their apparent
Influence on pollutant scavenging effldences. Unfortunately, attempts to
generate cl1matolog1cal statistics on types of precipitation for selected
regions have not been completely successful because of the difficulty of
classifying precipitation type unambiguously. This Illustrates the diffi-
culty of quantifying concepts that are established primarily for Illustra-
tive purposes. However, precipitation duration statistics have recently
been compiled for the northeastern United States for use 1n cllmatologi-
cal-type regional models (Thorp and Scott, 1982). Additionally, storm
track climatologies have been prepared by a number of methods. These
studies all reveal similar regions that are exposed to frequent storms
8701S 3
3-34
-------
10s
18*
e
5
• 1*
10*
a 01
ai
LO
PRECIPITATION RATE (mm h ')
10
100
FIGURE 3-6. Washout ratio as a function of pre-
cipitation rate. Curve 1 represents predictions
for Intense convectlve storms or from clouds
whose tops are warmer than 0°C; curve 2 repre-
sents predictions for storms where rain dev-
elops without the assistance of an Ice growth
stage; curve 3 1s for storms where the 1ce
growth process 1s necessary for Initiating pre-
cipitation; curve 4 represents observed 2*Na
concentrations 1n rainwater at Quillayute,
Washington, on 5-6 April 1970; and curve 5 re-
presents the same type of data as 1n curve 4
for 11 December 1969. Curves 4 and 5 represent
washout ratios obtained by Perkins et al. (1970)
during a series of continuous rainstorms.
3-35
-------
(Zlshka and Smith, 1980).* The full usefulness of long-term rainfall,
storm track, and trajectory statistics has yet to be realized 1n pollutant
deposition modeling.
Modeling precipitation scavenging on an event basis (short time period)
requires considerable detail of the flow field associated with clouds and
storm systems, the mlcrophyslcal characteristics of the clouds, and the
nature of the pollutant and Us Immediate environment. Figure 2-7
Illustrates the necessary components of such a scavenging calculation, as
compiled by Hales (1978). This diagram more than adequately Illustrates
the complexities associated with the problem.
3.1.4.3 Experimental Data on Scavenging Processes
Numerous field Investigations have been conducted to quantitatively assess
the entire scavenging process or certain aspects of 1t. Early studies
were conducted using radioactive fallout data. These were prompted by the
testing of nuclear weapons 1n the atmosphere and the development of
nuclear energy. Later studies focused on scavenging of more conventional
pollutants, and were prompted by a desire to compute budgets of nutrients
to soils and vegetation. The most recent experimental studies are
responses to the growing concern over add deposition Impacts.
Qualitative results from local-scale (0 - 20 km) field studies suggest
that scavenging of primary pollutants (primarily NOX and S02) 1s quite
Inefficient (Dana et al., 1976). This result 1s 1n accordance with what
1s known about low solubility of NO and N02, the reversibility of S02
scavenging, and conditions under which "equilibrium" S02 scavenging con-
ditions prevail (Hales and Sutter, 1973).*
Mesoscale studies (e.g., Hogstrom, 1974; Hales and Dana, 1979) qualita-
tively show a rise 1n pollutant scavenging due to the Increased opportu-
nity for chemical and physical transformation processes to occur 1n con-
Junction with the more efficient removal rates associated with secondary
Locations with frequent storms are discussed 1n the review of climate
and meteorology 1n the Rocky Mountain region (Section 3).
* Equilibrium S02 scavenging 1s an assumption stating that the precipita-
tion-borne S02 1s 1n equilibrium with the mixed-layer average S02 con-
centration, and hence uniquely determined by the local S02 concentration,
The conditions for which this assumption 1s valid are met at meso- and
regional-scale distances downwind from large S02 point sources.
3-36
-------
CJ
i
CO
WMssr 'LJ •s&ssxr lU A:
•UV.Tt"**' I | •«!«»» JSU.» p| .«..
•Ml M *MMMC« II. I VtMlH U.M
WM*MC1 MM t« If HMIM* M CMCttt
CUM WtIM' I- I •»•»• MMI«
s. T.i .»......-«. T.J cs'.r^rsa. i.
MW f I --'?i JBj.??y^T-t!SJ. 1^*1 •*•**• •• •••• • ••• r
»Mt § | «^»^» ••»• ^ • *••« | I »«UMIM «iM>iit» |
/PMC«*««**i«*
C«*BM M«*ftl«Kft.
*•• UHClMa*llte MBi
Ml ««MKI t«M«CM
•« J f I «*^* B|«C«W* «•••••
MH» I- *1 «C*«t*MMt
ti«M I 1 ^iMai«*iiaii
«j • ~~»1
)i| a-ggT
' •
HC^-MC^M, i
'sasg-=ir M
.-«."! -«.»
g* p*| aaa
J j ttj.u mg MTMUI J
FIGURE 3-7. Flow chart for scavenging calculations. (Source: NRC, 1983)
-------
pollutants. Finally, routine monitoring data (see NRC, 1983 for
references) suggests that pollutants are scavenged quite efficiently on a
regional scale, again as a consequence of the mixing of primary pollutants
with oxldants and cloud water and subsequent oxidation to particles.
Observed wet deposition due to scavenging within clouds (ralnout) and
below clouds (washout) can be thought of as an exponential decay process
with an associated decay constant (scavenging coefficient), or, alterna-
tively, can be expressed by a scavenging ratio (or washout ratio). The
latter quantity expresses the pollutant concentration 1n rainwater divided
by the airborne pollutant concentration.
Tables 3-3 and 3-4 11st various scavenging coefficients, A , and washout
ratios derived through field experimentation. A consensus regarding
particle wet removal rates 1s difficult to determine from Table 3-3
because of the variety of particle types, sizes, and experimental condi-
tions. Several experiments yielded a definite precipitation rate depen-
dency, 1n accordance with Scott's (1978) parameterization.* From the
experimental data, removal of partlculates by snow appears to be more
efficient than removal by rainfall. Scavenging coefficients ranging from
several percent per hour to over 1000 percent per hour have been measured,
Illustrating the efficiency and variability of the process.
Washout ratios (Table 3-4) Illustrate similar variability and precipita-
tion rate dependency. Generally, the scavenging ratios decrease with
Increasing precipitation amount.
Scavenging coefficients of gases from several laboratory and field experi-
ments 1n general show a lower removal efficiency for gases than for
particles (Table 3-5). As discussed previously, gaseous scavenging 1s
strongly dependent on solubility parameters, which are functions of tem-
perature and acidity. Thus, for extremely soluble gases, such as nitric
add vapor, the scavenging coefficient (0.2 s"1) has been estimated to be
greater than most particles (Levlne and Schwartz, 1982). The process 1s a
reversible one whose observed removal rate may not be well represented by
a constant scavenging coefficient.
* There 1s also evidence of a "Greenfield gap" 1n the particle scavenging
rate measurements, though a double minimum 1n the "Greenfield gap" size
range of 0.1 to 1 ym has been noted (Radke, Hlndman, and Hobbs, 1977).
3-38
-------
TABLE 3-3. Field measurements of scavenging coefficients, A , of particles.
J Indicates precipitation rate 1n units of mm/h. (Source: McMahon and
Denlson, 1979)
Investigator*
Particle
Type/Size
Comment
Kalksteln et al. (1959)
Georg11 (1963)
Banerjl and Chatterjee
(1964)
Makhon'ko (1964)
SMrvalkar et al.
(1960)
Makhon'ko and Dm1tr1eva
(1966)
Makhon'ko (1967)
Wolf and Dana (1969)
2 x 10-*
2 x 10'5
4 x 10-5
22 x ID'5
4 x ID'5
S04, NH4
Cl, N03
Dissolved
Inorganic
contaminant
0.4 x lO-5 Radon
2 x 10-5
< 1 x 10"5
7 x 10 "5
20 x ID'5
7 x 10'5
0.5 x 10"5 J
,-5
Fission
products
Ralnout (Makhon'ko, 1967)
Washout
Ralnout
Washout
Ralnout (Makhon'ko, 1976)
Ralnout
Washout
Atmospheric Ralnout (Makhon'ko, 1967)
dust
Fission
products
Ralnout
Atmospheric Ralnout plus washout
dust
Bakulln et al. (1970) 3 x 10
0.5
212
Burtsev et al. (1970)
Dana (1970)
Perkins et al. (1970)
Peterson and Crawford
(1970)
15 x lO-* jO.5
20 x ID-5 J0-5
13 x ID'5 J
300 x 10"5
16 x 10'5 jO.8
Pb
« 0.2
- 0.2
7.5, 3
Atmospheric
aerosol
Snow (Knutson and
Stockham, 1977)
Washout from
thunderstorm
Washout
Ralnout
Uranln and rhodamlne
particles, respectively
Ralnout
Based on Engelmann's
data (1965)
Continued
3-39
-------
TABLE 3-3 (Continued)
Investigator
Esmen (1972)
Particle
-1 Type/Size
Ap (s ] (un)
0.4 x 10"5 Atmospheric
aerosol
Comment
Includes ralnout
Rodhe and Grandell
(1972)
Acres-ESC (1974)
Graedel and
Franey (1975)
Hicks (1976)
0.7 x 10'5
snow
25-50A
rain
50 x 10"
Atmospheric
aerosol
0.4 - 1
< 1
Suggest A proportional
to rainfall Intensity
Includes ralnout
See Sllnn (1976)
Ralnout
Graedel and Franey 19 x
(1977) 18 x
28 x
43 x
65 x
92 x
Radke et al. (1977) - 150 x
~ 100 x
~ 1 x
~ 55 x
- 50 x
~ 40 x
~ 10 x
< 1 x
- 30 x
- 60 x
~ 100 x
- 140 x
i n
10" |j
10"
10-5
10"
10"5
10"5
10-5
ID'5
10
lO"5
10"
10-5
10"
10-5
10-5
10-5
10-5
0.3 -
0.5 -
0.7 -
0.9 -
1.5 -
.02 -
.05 -
0.1 -
0.3 -
0.5 -
0.6 -
0.8 -
1 -
1.5 -
2 -
2.5 -
3 -
Condensation nuclei
0.5 '
0.7
0.9
1.5
3 .
• Snow
.03 >|
.08
0.3
0.5
0.6
0.8
1
1.5
2
2.5
3
4 J
> Rain
* All literature references except Radke et al., 1977 are available 1n McMahon
and Oenison, 1979.
3-40
-------
TABLE 3-4. Field observations of washout ratios. (Source: McMahon and Oenlson, 1979)
CO
I
Investigator*
Hinzpeter (1958)
Chamberlain (1960)
Small (1960)
Pierson and Keane (1962)
Pierson and Cambray (1965)
Pelletier et al. (1965)
Gatz (1966)
Ratio
(Mass basis)
1250
710
400
1100
620
290
230
130
430
4000
470
1100
560
520
480
500
420
600 - 800
475 - 2100
1100 - 9200
Contaminant
0
0
0
0
0
0
137Cs
95Zn
2inpb
0
0
0
137Cs
ilnCo
Ba
i-nr
I
137Cs
0
Pollen
Precipitation
0.1
1.0
10
0.15
1.0
10
Not Provided
Not Provided
Not Provided
Not Provided
Not Provided
Rain
0.15 - 3.6 mm/h
Comment
Rain (mm per day)
Snow (equlv.
water mm per day)
Air cone, at 1200 m
Air cone, at 1200 m
Air cone, at 1200 m
October 1956
September 1959
Average (3 years)
Air near ground
Air near ground
Air near ground
Air near ground
Air near ground
Annual means
January 1963-June 1964
Surface area
Continued
-------
TABLE 3-4 (Continued)
CJ
I
Investigator*
Georgll and Belike (1966)
Crawford (1968)
Van de Uesthulzen (1969)
Health and Safety
Laboratory (1970)
Perkins et al. (1970)
Gatz (1972)
Peirson et al. (1973)
Gatz (1975)
Ratio
(Mass basis)
190
19
100 - 2700
W = 900 p~°-59
160 - 18,000
1500 - 5500
751
951
169
1212
698
671
380 - 2900
375
110
125
150
140
250
Contaminant
S02
so2
131,
137Ca
Pb
38C1
Cu
Fe
Pb
Mg
Mn
Zn
23 elements
Al
As
Cd
Cr
Cu
Fe
Precipitation
0.3 mm
11 - 20 mm
Not Provided
Not Provided
Rain
0.1 - 8.0 mm/h
Not Provided
Not Provided
Not Provided
Comment
Rain storm
Rain storm
Snow
P = mm rainfall per
3 months
Sampled rain
Sampled rain
Sampled rain
Sampled rain
Sampled rain
Sampled rain
days
days
days
days
days
days
Continued
-------
TABLE 3-4 (Concluded).
CO
CO
Investigator*
Prahtn et al. (1976)
Gatz (1977)
Krey and Toonkel (1977)
Ratio
(Mass basis)
125
76
325
110
4000
2400
457
548
352
370
253
179
76
970 p-°-17
1400 P'U1
Contaminant
N1
Pb
T1
V
S
Na
Mg
K
Ca
ta
Fe
In
Pb
90Sr
Pb
Precipitation Comment
Not Provided Includes both wet and dry
deposition
Not Provided Netromex, 1971-72 scavenging
. ratios vary with particle
size— see also Gatz, 1975
Not Provided
P » monthly precipitation (cm)
Literature references are available In HcNahon and Den1son, 1979.
-------
TABLE 3-5. Laboratory and field measurements of scavenging
coefficients of gases.
Investigator
Gas
LABORATORY RESULTS
Belike (1970)
FIELD DATA
Dana, Hales, and
Wolf (1975)
17 x 10'5 J°*6
NO, » (1/4)
Makhon'ko (1967) 6 x 10
Hales et al. (1970) 2 x 10
-5
,-5
0.4 x 10'5
1.3 x 10
-5
S02 (J • rainfall mm/h)
N0
S0
S02 small-scale experiment
SOo large-scale experiment
(lower value of A.
attributed to desorptlon
of S02 from water drops)
S0
3-44
-------
3.1.4.4 Modeling Implications
The process of wet deposition of gaseous and participate pollutants 1s an
Important natural cleansing mechanism for the atmosphere, resulting 1n the
delivery of airborne pollutants to the earth's surface. In general, the
following three pathways of delivery can be Identified:
(1) Gaseous pollutants are scavenged by cloud droplets and precipi-
tation and reach the ground'as dissolved gas.
(2) Partlculates are removed by ralnout and washout processes.
•
(3) Gaseous pollutants are scavenged and then chemically transformed
within clouds or raindrops before delivery to the surface.
In the first pathway, direct scavenging of gaseous pollutants by precipi-
tation 1s governed by two factors: (1) transport of the gas to the
hydrometeor (cloud droplets or precipitation drops) and (2) the solubility
of the gas. Transport of the gas occurs by means of molecular and turbu-
lent diffusion and 1s Independent of species type. Solubility, on the
other hand, 1s dependent on the particular chemical species.
Because nitric oxide has a very low solubility 1n water, 1t 1s not effec-
tively scavenged by precipitating systems. Nitrogen dioxide dissolves 1n
water to a limited extent; Its solubility, like that of S02, Is a function
of droplet pH. However, evidence to date Indicates that the scavenging of
N02 1s only about one-quarter that for S02 (Belike, 1970). Since nitric
acid vapor 1s highly soluble and shows little tendency to desorb back Into
the atmosphere, 1t 1s probably scavenged more efficiently than S02. Other
gases can be described according to the same reversible scavenging theory,
with allowance made for their different solubility characteristics.
Dissolved S02 reacts rapidly to form sulflte and bisulfite Ions. This 1s
a reversible reaction, whose equilibrium depends on the aqueous-phase S02
concentration, temperature, and hydrogen 1on concentration. The flux of
S02 to, or from, the hydrometeor depends on the concentration gradient
between the droplet surface and the ambient air. Under conditions of low
droplet pH, the equilibrium shifts toward more S02 1n solution; thus,
unless the ambient S02 concentration 1s sufficiently high, desorptlon of
the gas occurs and the scavenging 1s Inefficient.
Incorporating reversible scavenging Into a model requires consideration of
whether or not the aqueous-phase S02 concentration 1s 1n equilibrium with
the gas-phase concentration. For S02, the rate-determining process 1s the
flux to, or away from, the hydrometeor. If the vertical distribution of
3-45
-------
S02 Is highly variable or the hydrometeor 1s falling rapidly, the gas-
phase/aqueous-phase concentration may not approach equilibrium and thus
may require a more complex parameterization like that of Dana, Hales, and
Wolf (1975). Under certain conditions, however, the scavenging process
may be 1n equilibrium, resulting 1n a direct relationship between aqueous-
phase concentration and local airborne S02 concentration. For regional
model applications, where vertical pollutant distributions are more nearly
uniform, this assumption appears valid for distances several kilometers
downwind from major point sources (Hales, 1978).
Caution should be exercised when applying "washout coefficient" theory to
gaseous scavenging. This Irreversible theory requires that aqueous-phase
concentrations continually Increase as the hydrometeor 1s exposed to air-
borne pollutants. Application of this theory generally leads to overpre-
dlctlon of S02 deposited by rain.
The second pathway Involves Incorporation of particles Into clouds or
raindrops. Because particles exist 1n a variety of sizes, the Interphase
transport mechanisms are far more complex. Since sulfate particles are
hygroscopic, they can serve as condensation nuclei and can be scavenged by
a process known as ralnout, or they can be collected below the cloud base
by the washout process. Scavenging of nitrate aerosol 1s thought to occur
1n a similar manner. In both processes the scavenging 1s Irreversible;
partlculates remain 1n solution until the droplet evaporates.
Thus, aqueous-phase concentrations depend on the history of the hydro-
meteors. Cloud types are formed 1n response to different conditions, and
each type 1s characterized by various droplet growth mechanisms, and
hence, various droplet growth histories. The ralnout process for several
cloud types has been Investigated by Scott (1978). His study suggests
that during light rain the washout ratio (defined as the pollutant concen-
tration 1n precipitation divided by the pollutant concentration 1n air)
may vary by two orders of magnitude, depending on cloud type, whereas for
more Intense rainfall the dependency diminishes.
Particles are also scavenged by below-cloud processes. The primary Inter-
phase transport mechanisms are 1mpact1on, Intersection, and diffusion.
Although modeling these mechanisms 1s fairly straightforward, the sensi-
tivity of the overall pollutant scavenging rate depends on aerosol and
raindrop size distributions. Uncertainty 1n these distributions will lead
to washout rate uncertainty.
The third pathway establishes a close relationship between aqueous-phase
transformations and wet deposition. The effectiveness of aqueous-phase
oxidation pathways 1n promoting S02 transformation 1n real cloud systems
1s very uncertain, but the third pathway 1s Important 1n Interpreting the
3-46
-------
nitrates typically observed 1n rainfall samples. Although nitrate aero-
sols may be Incorporated 1n rain 1n a manner similar to sulfate Incorpora-
tion, the high reversibility of reactions Involving secondary nitrogen
pollutants (1n both gas-phase and aqueous-phase transformations) can
significantly affect the nitrate concentration 1n rain.
3.2 CLIMATOLOGY AND METEOROLOGY OF THE
CENTRAL ROCKY MOUNTAIN REGION
3.2.1 Geography
The central Rocky Mountain region straddles the highest mountains of the
Continental Divide. It Includes most of northern Colorado, a small por-
tion of eastern Utah, and the southern quarter of Wyoming. The average
altitude of the region 1s over 6,800 feet MSL, and 1t has over 40 moun-
tains that are 14,000 feet or higher. Emerging gradually from the plains
of Kansas and Nebraska, the high plains slope gently upward, broken by
occasional hills and bluffs, to the base of the foothills of the Rocky
Mountains. Elevations on the plains range from less than 4,000 feet 1n
the northeast to 6,000 feet 1n the southeast.
At elevations of 5,000 to 6,000 feet, the plains give way abruptly to
foothills of 7,000 to 9,000 feet. In Colorado the higher mountains of the
Front Range rise to the west of the foothills, and reach heights of over
14,000 feet. To the west of the Front Range are many additional ranges,
generally extending north and south, but with many spurs and extensions 1n
different directions. These ranges enclose numerous valleys and parks.
In the western and northern portions of the region the landscape 1s domi-
nated by rugged plateaus, high mesas, and deep canyons. An exception 1s
the U1nta Mountains, an east-west oriented range 1n northeast Utah with
peaks reaching 13,500 feet.
Most rivers 1n the region rise within Us borders and flow outward, with
the exception of the Green River, which rises 1n the Wind River Mountains
of Wyoming and flows southward through extreme northwest Colorado and
western Utah. Three of the nation's major rivers have their source 1n the
region: the Colorado, the Arkansas, and the Platte.
Soils of the central Rocky Mountain region fall Into two major orders:
1ncept1sols and aHdlsols (Greenland and de 811 j, 1977). Inceptlsols are
found In the mountainous areas of the region, and Include alpine turf
soils, brown forest soils, and wet alluvial soils (Butzer, 1976). Natural
vegetation found on these soils Includes forest and alpine tundra.
Extreme weathering and significant leaching are not evident. Bedrock
usually lies close to the surface and slopes are often steep, creating a
3-47
-------
weatherlng-Hmlted situation 1n which removal of debris 1s more rapid than
the rate of weathering. Such soils provide very little buffering of
acidic Inputs. AHdlsols are found 1n the plateau and plains areas, and
are soils of arid or semlarld climates with a thin or light colored A
horizon (Butzer, 1976). They Include desert soils, red and grey semi-
desert soils, and white and associated black alkali soils. Natural vege-
tation 1s usually a thin cover of shrubs and grasses.
Acid-sensitive areas of the West have been Identified by a number of
Investigators (see Figure 3-8). Although there are differences 1n these
three maps, there 1s consistency 1n the Identification of several sensi-
tive areas 1n the Rocky Mountain region. These areas Include the High
Ulntas east of Salt Lake City, the Flat Tops and Mt. Zlrkel wildernesses
and Rocky Mountain National Park 1n Colorado, and the Wind River Range of
western Wyoming. All of these sensitive areas, except the Ulntas, are
designated Prevention of Significant Deterioration (PSD) class I areas,
which are offered special protection under the Clean A1r Act. Class I
area protection 1s a primary objective of the regional add deposition
model that will be developed as the chief product of this project. Figure
3-9 shows that class I areas are often ac1d-sens1t1ve areas. Figure 3-10
and Table 3-6 provide the locations and names of these areas.
3.2.2 Climate and Meteorology
The central Rocky Mountain region Includes most of northern Colorado, a
small portion of eastern Utah, and the southern quarter of Wyoming (Figure
3-11). The climate 1s broadly classified as continental, of a highland or
mountain nature. The area encompasses numerous mountain ranges, parks and
valleys. Such elevatlonal complexities produce widely varying climates
over small geographic areas.
3.2.2.1 Temperature and Humidity
Temperature and humidity regimes are profoundly affected by differences 1n
elevation. Mean annual temperatures range from less than 32°F on high
mountain summits to 53°F at Boulder and Grand Junction 1n the east and
southwest, respectively.* Summer temperatures are very warm at lower ele-
vations. July maxima 1n the eastern plains and western valleys (4,000-
6,000 feet) often exceed 95°F, but 100° 1s exceeded only occasionally.
* Climate statistics from NOAA's Climates of the States. Vols. 1 and 2,
Gale Research Company, Detroit; and Guide to Colorado's Weather and
Climate. Enrnap Corporation, Boulder, Colorado.
3-48
-------
TOTAL ALKALINITY OF
SURFACE WATERS*
Wnltm
( Oran Map) i
50-100
O 100-200
O 200-400
CD >4OQ
FIGURE 3-8a. Acid-sensitive areas in the West as Identified in the NAPAP
Annual Report 1984 to the president and Congress.
3-49
-------
CD
Total Alkalinity'
(/i.q./l.)
<200
200-399
400 - 599
600-999
1000-1999
>2000
FIGURE 3-8b. Acid-sencitive area:
in the West as identified by Omernik,
"Alkalinity of Surface Waters" (1982),
Corvallis Environmental Research Lab-
oratory, U.S. Environmental Protection
Agency.
3-50
-------
FIGURE 3-8c. Acid-sensitive areas in the West as identified by Roth
et al., "The American West's Acid Rain Test" (1985). (Note: Dashed
lines Indicate areas with annual mean precipitation pH < 5.0.)
3-51
-------
FIGURE 3-9. PSD Class I areas (boxes) are often located 1n
add-sens1t1ve areas (shading).
3-52
-------
Vandenburg
San
SCALE 1:12.500.000
k1loneters
0 100 200 300 400 500 600
El Paso
100 200
• i 1 es
300
FIGURE 3-10. Class I areas and upper-air meteorological stations located
in the West, exclusive of Washington and Montana. (Key to locations is
in Table 3- 6.}
-------
TABLE 3-6. Key to Class I Areas located in the West.
10 NAME
1 Mount Hood Wilderness
2 Eagle Cap Wilderness
3 Bells Canyon Wilderness
4 strawberry Mountain Wilderness
5 Mt. Jefferson Wilderness
6 Mt. Washington Wilderness
7 Three Sisters Wilderness
8 Diamond Peak Wilderness
9 Crater lake National Park
10 Gear hart Mountain Wilderness
11 Mountain Lakes Wilderness
12 Kalmiopsis Wilderness
13 Pedwood National Park
14 Marble Mountain Wilderness
15 Lava Beds Wilderness
16 South Warner Wilderness
17 Thousand Lakes Wilderness
18 Yolla-Bolly - Middle Eel Wilderness
19 Lassen Volcanic National Park
20 Caribou Wilderness
21 Desolation Wilderness
22 Mokelurome Wilderness
23 Point Reyes wilderness
24 Emigrant Wilderness
25 Yosemite National Park
26 Boover Wilderness
27 Minarets Wilderness
28 John Muir Wilderness
29 Kaiser Wilderness
30 Kings Canyon National Park
31 Sequoia National Park
32 Pinnacles Wilderness
33 Ventana Wilderness
34 Dome Land Wilderness
35 San Rafael Wilderness
36 San Gabriel Wilderness
37 Cucamonga Wilderness
38 San Gorgonio wilderness
39 Agua Tibia Wilderness
40 San Jacinto Wilderness
41 Joshua Tree Wilderness
42 Chiricahua Wilderness
43 Saguaro Wilderness - East
44 Saguaro Wilderness - West
45 Galiuro Wilderness
46 Superstition Wilderness
47 Mt. Baldy Wilderness
48 Sierra Ancha wilderness
49 Mazatzal Wilderness
50 Pine Mountain Wilderness
Continued
3-54
-------
TABLE 3-6 (concluded).
MO NAME
51 Petrified Forest National Park
52 Sycamore Canyon Wilderness
53 Grand Canyon National Park
54 Zion National Park
55 Bryoe Canyon National Park
56 Capitol Reef National Park
57 Canyonlands National Park
58 Arches National Park
59 Jarbridge Wilderness
60 Craters of the Moon Wilderness
61 Sawtooth Wilderness
62 Selway Bitterroot Wilderness
63 Bed Bock lakes Wilderness
64 Yellowstone National Park
65 North Absaroka Wilderness
66 Washakie Wilderness
67 Teton wilderness
68 Grand Teton National Park
69 Fitzpatrick Wilderness
70 Bridger Wilderness
71 Mount Zirkel Wilderness
72 Bawah Wilderness
73 Pocky Mountain National Park
74 Flat Tops Wilderness
75 Eagles Nest Wilderness
76 Maroon-Bells Sncwmass Wilderness
77 West Elk Wilderness
78 Black Canyon of the Gunnison Wilder,
79 Great Sand Dunes Wilderness
80 La Garita Wilderness
81 Weminuche Wilderness
82 Mesa Verde National Park
83 Wheeler Peak Wilderness
84 San Pedro Parks Wilderness
85 Bandelier Wilderness
86 Pecos Wilderness
87 Bosgue del Apache Wilderness
88 Gila Wilderness
89 White Mountain Wilderness
90 Salt Creek wilderness
91 Carlsbad Caverns National Parks
92 Guadalupe National Park
3-55
-------
I
in
CJl
467
4630
4590
370 410
450 490 530 570 610 650 690 730 770 810 85C
530 570 610 650 690 730 770 810 850
670
423
- 4630
- 4590
4550
•* 4510
i 4470
3 4430
5 4390
4350
4310
4270
4230
FIGURE 3-lla. Terrain of the Grand Junction, Colorado and Salt Lake City, Utah
subregion. (Boxed numbers refer to Class I areas; see Table 3-6 for key).
-------
466
20 360 400 440 480 520 560 600
4620
410
360 400 440 460 520 560
EASTING (KM)
•n4660
- 4620
- 4580
-. 4540
•= 4500
- 4460
k- 4420
- 4380
- 4340
- 4300
- 4260
- 4220
- 4180
•: 4HO
100
FIGURE 3-llb. Terrain of the Denver, Colorado subregion.
3-57
-------
501
380 420 460 500 540 580 620 660 700 740 760 ^i^mSOlO
X iiiiiiiiiii««iiiiainiiiiiiilFlliiiill¥llllHnillllllllllllllllll IT1 I II1 I lilBT ITU -"•• * v
I
en
00
4970
imilViiiniiilM-M.iiiiiliiiiiiilTtwil/11
3m«n mil 11 !!• i-yi »inyiniiii«iiiiin«iiii
80 420 460 500 540
i 4970
4 4930
-. 4890
* 4850
-_ 4810
- 4770
-. 4730
4690
465
580 620 660 700 740 780 820 860 900 940 980 1020 1060
EASTING (KM)
4650
FIGURE 3-llc. Terrain of the Lander, Wyoming subreglon.
-------
Although temperatures are high, low relative humidity makes most days com-
fortable. For example, the July mean relative humidity at 1700 LSI 1s
21 percent at Grand Junction and 36 percent at Denver. July temperature
minima are generally 1n the 50's and 60's. At the 6,500-9,000 foot level,
which Includes most areas of southern Wyoming and central Colorado, July
maxima are usually 1n the 70's and 80's, while minimum temperatures range
from 40°F to 50°F. Even high mountain areas enjoy mild summers; Lead-
vine, at 10,158 feet, has mean July high and low temperatures of 72°F and
42'F, respectively.
Winter temperatures vary widely, and are Influenced by topographic loca-
tion. Mountain parks and protected valleys often have very cold tempera-
tures when skies are clear and nocturnal Inversions are we11-developed.
Such areas Include North Park and the Yampa and White River Valleys of
Colorado, the U1nta Basin of Utah, and the Green and North Platte River
valleys of Wyoming. Climate stations 1n these areas report mean January
minima of -10°F to 5°F, while 5°F to 15'F 1s more typical of the region.
Mean January maxima of 20°F to 35°F 1n these valleys are similar to those
experienced over much of the region. The lowest official temperature
recorded In the region, -60°F, occurred at Maybell, on the Yampa River 1n
northwestern Colorado.
Winter 1s characterized by rapid and frequent changes between mild and
cold periods. Usually there are less than 10 cold waves during a winter.
and most of these occur as shallow outbreaks of continental polar air that
slide southward along the eastern slope of the Rockies. Despite the
occasional cold spell, cities 1n the high plains, such as Boulder and Den-
ver, enjoy the mildest winter temperatures of the region. The mean
January maximum and minimum for Boulder are 468F and 22°F, respectively.
The mild temperatures are due to a phenomenon known as the "chlnook".
CMnooks occur when strong westerly winds aloft are mixed to the surface
and warmed by their rapid descent from the Continental Divide. Strong
Chinook winds may warm temperatures 25-35°F 1n only one to two hours.
Accompanying humidities of less than 10 percent promote the rapid evapora-
tion of snowcover. Reg1onw1de, wintertime relative humidities are 30 per-
cent to 40 percent during the day and 45-60 percent during the early morn-
ing, except 1n protected valleys, where cold air Inversions produce morn-
Ing humid1tes of 75-90 percent.
.3.2.2.2 Precipitation
Precipitation 1n the central Rocky Mountain region 1s Influenced by dif-
ferences 1n elevation and the orientation of mountain ranges and valleys
with respect to large-scale air flows. Prevailing westerlies carry storms
over the region from the Pacific during the fall and winter, but much of
3-59
-------
the moisture 1s lost to mountain ranges of the region. A large part of
the remaining moisture 1s deposited orographlcally on high mountain peaks
and western slopes 1n the region, leaving valley locations and eastern
slopes relatively dry. This systematic variation 1n precipitation 1s
reflected 1n the presence of a lower treellne below which Insufficient
moisture 1s available to support tree growth. The lower treellne rarely
occurs below 6,000-7,000 feet and 1s lowest on western slopes. Locations
below 6,000 feet west of the Continental Divide generally receive less
than 15 Inches of precipitation (liquid water equivalent) per year, while
areas below 5,000 feet often receive less than 10 Inches. In contrast,
the mean annual precipitation of higher mountain areas may exceed 50
Inches. Precipitation west of the Continental Divide 1s fairly evenly
distributed throughout the year. East of the Divide, Influxes of moisture
from the Gulf of Mexico during the spring produce precipitation maxima
during March-May. Mean annual precipitation at communities east of the
Colorado Front Range varies from 12 to 20 Inches, with the largest amounts
found nearest the mountains.
3.2.2.3 Storm Characteristics
Storm types that significantly affect the region are Pacific frontal sys-
tems that move from west to east 1n late fall, winter, and early spring
(western slope areas are affected most); Influxes of Gulf of Mexico mois-
ture from the southeast 1n the spring (eastern slope areas are affected
most); and thunderstorm activity triggered by Pacific or Gulf of Mexico
moisture during late spring and summer (entire region 1s affected).
Pacific frontal systems produce large amounts of precipitation (largely
snow) 1n high mountain areas and light amounts 1n western valleys, pri-
marily during winter. Heaviest precipitation 1s associated with low-pres-
sure centers moving over the region. Snowfalls outside of mountain areas
are usually less than five Inches, but occasionally will exceed 10 to 15
Inches. Mountain areas may receive a foot or more from each storm.
Pacific storms often produce strong, gusty west-southwesterly winds over
southern Wyoming, piling a few Inches of snowfall Into drifts several feet
deep. Even 1n the absence of any major frontal systems, westerly winds
aloft easily mix down to the surface of Wyoming's elevated and exposed
plateau areas, necessitating elaborate snow drift control programs to keep*
highways and railroads passable.
•
Locations east of the Continental Divide experience the heaviest rain or
snow falls during occasional Influxes of moisture-laden air from the Gulf
of Mexico 1n the spring. Maritime tropical air 1s transported from the
southeast by low pressure systems moving to the south and east of the
region. Lows that stall and Intensify over southeastern Colorado or
3-60
-------
northern Texas are most effective at bringing moisture northward and west-
ward over the high plains. Occasionally such a scenario will occur 1n
concert with an outbreak of continental polar air from the north. Over-
running of the cold air by the warm, moist air can produce spectacular
snowfalls on the eastern slope of the Colorado Front Range. The greatest
snowfall ever recorded 1n North America 1n a 24-hour period, 76 Inches,
occurred during such an episode at Silver Lake on 14 and 15 April, 1921.
Since the cold air mass 1s relatively shallow, precipitation does not
extend west of the Divide.
Thunderstorms occur over the region from April through September, and pro-
vide nearly all of the region's precipitation during the growing season.
They are produced primarily by local surface heating, and their number and
Intensity are dictated by the amount of moisture 1n the air mass. The
most severe thunderstorms occur during southerly flows from the Gulf of
Mexico and areas of hurricane development off the west coast of Mexico.
Thunderstorms are most prevalent on the eastern plains, and the frequency
of hall damage to crops from severe thunderstorms 1s very high. Tornadoes
almost never occur 1n the mountains and 1n the west, but do occur occa-
sionally 1n the eastern plains of Colorado. From June through August,
thunderstorm activity In the mountains 1s an almost dally event. A typi-
cal summer day dawns bright and clear. By late morning, cumulus formation
begins over peaks and ridges. By early afternoon, thunder may be heard
and short periods of rain, hall, or graupel (at higher elevations)
occur. Lightning 1s frequent, and puts hikers and campers at risk,
especially above the treeline. Each summer one or more persons are killed
by lightning 1n the Colorado Rockies. Thunderstorms forming along the
Continental Divide develop quickly and move 1n a generally eastward direc-
tion over the foothills and out Into the plains. Large, slow-moving
storms may produce severe flooding. On the night of 31 July, 1976,
Colorado's greatest natural disaster occurred along the B1g Thompson Can-
yon 1n the Front Range when a flash flood, fed by torrential rains measur-
ing as much as 10-12 Inches within several hours, caused over 139 fatali-
ties.
3.2.2.4 Transport Winds and Mixing Heights
Perhaps of most Interest regarding add deposition 1n the Rocky Mountain
region 1s the general climatology of transport winds and mixing heights.
Transport winds are the upper-level winds responsible for transport of
local and regional air pollutant emissions and species formed from these
emissions 1n the atmosphere. Transport wind directions are Indicative of
the direction of pollutant transport, while wind speeds are Indicative of
dilution. The mixing height also affects dilution of emissions.
3-61
-------
Figure 3-12 shows the average upper-air wind rose for the northwestern
Colorado region. This wind rose was calculated by combining data from
four National Weather Service upper-air meteorological stations—Denver,
CO; Grand Junction, CO; Lander, WY; and Salt Lake City, UT) for the years
1978 and 1979. These winds were determined at approximately 1800 m (6000
ft) above ground level, thus they do not reflect the Influence of local
terrain and are generally from the west to the east, characteristic of the
synoptic or synoptic-scale winds of the region. Figures 3-13 and 3-14
show the upper-air resultant winds for the entire southwestern U.S.,
Including Colorado. These upper-air winds were calculated by vector-
averaging the afternoon winds for each day 1n 1981 using the wind fields
predicted by the National Meteorology Center Limited Fine-Area Mesh (LFM)
model. Winds each afternoon were taken from the midpoint of the mixing
height, which averages 2600 m 1n Colorado. Thus, the average height of
these winds 1s 1300 m above ground (2600/2), still well above the effect
of most terrain. Note from Figure 3-12 that the resultant wind 1n
Colorado 1s consistent with the upper-air wind rose shown 1n Figure 3-12,
showing generally west-to-east transport. The monthly breakdowns 1n
Figure 3-14, however, show considerable variation about this average, with
transport from the south and southwest 1n some of the spring and summer
months.
At lowe.r levels—which are relevant to transport over relatively short
distances (< 100 km) before emissions have been thoroughly mixed 1n -the
vertical to the top of the mixed layer—the complex topography of the
region has a more dramatic effect on wind direction.
Since the region 1s relatively dry, there are many periods with strong
Insolation during the day and clear nights. This allows rapid heating of
the ground surface during daylight hours and rapid cooling of the ground
surface at night. This strong heating and cooling cycle tends to create
strong upslope flows during the day and downslope drainage flows at
night. The upslope/downslope cycle occurs at all geographic scales; that
1s, the tendency can be noted 1n canyons, river valleys, mountains, and
mountain ranges. The larger the horizontal scale of the geographic fea-
ture, the greater 1s the volume and thickness of air that 1s moved by this
diurnal cycle of heating and cooling. Since geographic features of many
scales overlie each other, the cyclic air movements are also complex, with
thin layers of air embedded within thicker ones and all of these embedded
within the synoptic upper-air flow discussed above. Figures 3-15 and 3-16
show morning and afternoon winds, respectively, at 300 m above ground
level (agl). This level 1s considerably below the 1800 and 1300 m agl
winds summarized previously, so these winds are more likely to exhibit the
Influence of terrain features. The diurnal effects of heating and cooling
can be seen by comparing these two figures. For example, 1n Grand Junc-
tion 1n western Colorado, winds are typically southeasterly 1n the morning
3-62
-------
METERS PER SECOND
km\mm
SEEilpSIB*
-:.. : :|.:i...--.-" ••••• •
::H::::::gH
B*:
•.«•••• ••:•;- "»
HNM
NSM
SE
ENE
3-5 5-7 7-9 8-11 13 15 17
ESE
3SE
FIGURE 3-12. Average upper-air wind rose for northwestern Colorado,
(Source: Latlmer et al., 1985)..
3-63
-------
UJ
C:
*>
0 5 10
•eter«/tec
X*
\ *
I »
/
I
mil
ri i i\i i *
i i i i \\\ i
ili 111 \ i \
,m|j ill]
/ //nil 11'
\\
FIGURE 3-13. Mean annual afternoon mixed-layer resultant wind fields 1n 1981 calculated
by LFM. (Source: Latlmer et al., 1985).
-------
/ILL
/ / ///
' ' //////
/' / / ///////
(a) January
(b) February
FIGURE 3-14. Mean monthly afternoon mixed-layer wind
fields in 1981 calculated by LFM. (Source: Latimer et al.,
1985).
3-65
-------
I I II
•fltrfttC
(c) March
77711
(d)
FIGURE 3-14. Continued.
3-66
-------
77/111
(e) May
(f) June
FIGURE 3-14. Continued.
3-67
-------
(g) July
I""!""!
t i 10
(h) August
FIGURE 3-14. Continued.
3-68
-------
/ /
(1) September
11 in
(j) October
FIGURE 3-14. Continued.
3-69
-------
(k) November
U * '
\ I ///
(1) December
FIGURE 3-14. Concluded.
3-70
-------
>5 M/S
2.5 - 5 M/S
0 - 2.5 M/S
I i I
50 100
PercontaQe
FIGURE 3-15. Morning wind roses over all stability classes,
(Source: Latimer and Hogo, 1983).
3-71
-------
I I I I I
0 50 100
Percontaoe
FIGURE 3-16. Afternoon wind roses over all stability classes
(Source: Latimer and Hogo, 1983).
3-72
-------
and northwesterly 1n the afternoon, reflecting the effect of the downslope
and upslope winds 1n the Colorado River basin, which runs from the south-
east to the northwest 1n Grand Junction. The reversal of winds 1n Salt
Lake City and Denver 1s also the result of the diurnal heating and cooling
cycle. Figure 3-17 shows resultant surface winds for the two-year period
from August 1979 to July 1981. Note that transport 1s more southerly 1n
Utah than that shown 1n Figure 3-12, Indicating the channeling effect that
the north-south oriented mountains In that state have.
Figures 3-18 and 3-19 compare the morning and afternoon wind directions at
four locations 1n northwestern Colorado and eastern Utah. This region 1s
of particular Interest because of the potential for synfuel and other
types of energy development. These figures show wind directions at dif-
ferent elevations above ground (I.e., 150, 300, 500, and 1000 m agl). In
the morning, there 1s a great deal of wind shear at all sites. The shape
of the wind direction distribution for each site 1s unique at the lower
levels, reflecting the orientation of each valley, but at the higher
levels the distributions reflect the synoptic, west-southwesterly flow 1n
the region. In the afternoon there 1s much less wind shear and the lower-
level winds are also generally westerly, suggesting that the.daytime mix-
Ing has coupled these layers with the synoptic flow aloft.
The diurnal variation 1n wind direction at these sites can be observed by
comparing these two figures (Figures 3-18 and 3-19). The C-b site, the
proposed location for the Cathedral Bluffs oil shale development, does not
have a well-defined drainage flow because of Us relatively high eleva-
tion, and typically has shallow drainage flows 1n the morning. Above the
first 150 meters, both morning and afternoon winds are most often from the
south to the southwest.
The U-a/U-b site, the proposed location for the White River oil shale pro-
ject, 1s located 1n the middle of the U1nta Basin of eastern Utah. The
entire basin slopes downward toward the north, but winds are not particu-
larly confined by the terrain. Only a shallow morning drainage flow
exists at the 300 m level and below. Weak winds from the south and south-
west predominate. In the afternoon, northwest upslope flow at the surface
and south-to-southwest synoptic-scale winds at higher levels (>150 m) are
frequent.
Craig and Grand Junction, Colorado, are located 1n valleys where drainage
flows 1n the morning are much more strong. At Craig, an easterly drainage
flow 1s observed up to 300 m, and at Grand Junction, a southeasterly
drainage flow 1s observed at 500 m.
Figures 3-20 and 3-21 summarize the wind speeds at these four sites 1n
both the morning and afternoon, respectively. As one would expect, wind
3-73
-------
FIGURE 3-17. Annual resultant surface winds (8/79-7/81).
(Source: Cahill et al., 1983).
3-74
-------
M1NO 01KCCTI0N
WIND OIKECTIBN
i8
bJ
•
m*
bJ
*sj
1
JjjJ
u.2-
V
11
hJ
§0-
(••NO JCT
/'I
/
> »
/ I"-
• .'1 '••••••.
••:." ••-•/ ••' %''. •• /••
* • . . . • • %. • / \
1 ' la ' t Mr1' k ' L i ! . i...
M1NO Dl»fECTlBN
FIGURE 3-18. Wind direction frequency distributions for all morning
soundings. (Source: Systems Applications, Inc., 1982).
3-75
-------
_iso urn** t:
.JOO Nfff*3 bJ
...SOO Nf Tf»1
. lOOONtUM
WIND OIKECTIBN
WIND OIRECTJ0N
J o_ w«w Jet
• i ^
Ui
LJ
CJ
u
5,
WIND D1RECT10N
\i M ' ^ ' I ' '«.'
MIND OIKttCTIBN
FIGURE 3-19. Vertical wind direction distribution profiles for all afternoon
soundings. (Source: Systems Applications, Inc., 1982).
3-76
-------
WIND srcto
. JOO «Tf M
..»0fl WtfdJ
. iooo«nr«i
E
'•'•..
.' '*.•
i*iy... ...y\v---
81 : .' 1 \ V •
>- • • X. \ v
i)
«.j
ci
UJ
So
M U V \
"•»•..... •.
WIND SPEED
M1NO
c
£°-p
11 ^
g
£
t'
xt
\
N^
k \i ^i i»''»',*
MIND SPEED
FIGURE 3-20. Wind speed frequency distributions for all morning soundings
(Source: Systems Applications, Inc., 1982).
3-77
-------
II II II 17 I* 11
WIND SPEED
.JOO HCTMl U°-T
. ADO "ttr»j }J
. J000«f TfHI UJ
2 ' t i i rVsfe^^^ri
5° i J s J J i it ii i* Vf iq i
HIND SPEED
X i • . i i i *W*"»^3»r*^.l..
got M \ \ ? » ll ll |T|7 II ,I1 2
MJNO SPEED
2)
UJ
So.
UJ fM
Cf
LI
UJ
2°
I. t L i I. I. rV-iyJ — I — '
i ii u 11 i' it ;i 2)
HIND SPEED
FIGURE 3-21. Wind speed frequency distributions for all afternoon soundings.
(Source: Systems Applications, Inc., 1082).
3-78
-------
speed typically Increases with height above ground. Winds are weaker 1n
the morning than 1n the afternoon, particularly at low levels. The 1000 m
wind speed does not change significantly between morning and afternoon.
The frequency of occurrence of morning and afternoon stability categories
at the 500 m level (determined from temperature lapse rates) at these four
sites 1s shown 1n Table 3-7. At the sites where drainage flows are
strongest (I.e., Craig, Grand Junction, and U-a/U-b), the atmosphere 1s
most stable. At the C-b site, where drainage flows are not as organized,
neutral conditions predominate 1n the morning. In the afternoon, the
atmospheric stability 1s most often neutral at all four sites.
The morning and afternoon mixing depths for Grand Junction are listed 1n
Table 3-8. Because of Increased surface heating, summer afternoon mixing
depths are more than three times the average winter value—3940 m versus
1160 n. The atmosphere 1s most stable and least well mixed 1n the win-
ter. In winter, drainage flows at some sites can persist all day, 1n con-
trast to their normal occurrence at night and 1n the morning.
3.2.3 A1r Quality and Deposition
The Rocky Mountain region has nearly the best air quality of the contigu-
ous 48 states, as evidenced by Figure 3-22. Median visibilities In the
region are 70-80 miles. Except for partial late concentrations, ambient
air concentrations of the criteria pollutants are well below ambient air
quality standard levels (Table 3-9). Parti oilate concentrations are often
above the existing TSP standards because of severe levels of windblown
dust and ozone concentrations that are often near or slightly 1n excess of
the standard, perhaps because of stratospheric ozone Intrusions (Table
3-10).
Figure 3-23 shows annual average sulfate (SOT) aerosol concentrations 1n
the southwestern U.S. for 1981 based on measurements and regional model
calculations. Both the measurements and model calculations suggest that
sulfate concentrations 1n the Rocky Mountain region are generally 1n the
range from 0.5 to 1.0 ug/m3, roughly an order of magnitude below concen-
trations observed In-the eastern U.S. and 1n southern California.
Nitrate (NOZ) concentrations 1n the region average 0.25 pg/m3 (Latlmer
et al., 1985), although there are fewer confirming measurements.
*
Although there are no measurements of dry deposition, the National Add
Deposition Program (NADP) has a number of wet deposition monitoring sites
1n the West (see Table 3-11 and Figure 3-24). Figure 3-25 compares model
calculations and NADP measurements of wet sulfur deposition fluxes 1n
3-79
-------
TABLE 3-7. Frequency of occurrence of atmospheric stability at the 500-m level
(AGL) (percent).
Morning
Unstable Neutral
(A.B.C) (D)
Cathedral
Annual
Winter
Spring
Sunnier
Fall
Bluffs
10
2
11
13
10
(C-b)
55
48
61
51
61
Stable
(E)
34
50
.28
36
27
Very
Stable
(F)
0
0
0
0
2
Unstable
(A.B.C)
19
6
34
45
17
Afternoon
Neutral
(D)
55
69
54
-32
63
Stable
(E)
27
25
12
23
21
Very
Stable
(F)
0
0
0
0
0
Craig, Colorado
Annual
Winter
Spring
Sunnier
Fall
2
0
5
0
2
White River (U-a
Annual
Winter
Spring
Summer
Fall
1
0
0
0
2
Grand Junction,
Annual
Winter
Spring
Summer
Fall
*
2
6.
. 1
0
1
39
43
56
21
37
? U-b)
25
33
29
14
24
Colorado
52
33
63
57
49
59
57
39
80
59
74
65
71
87
74
44
56
31
43
49
1
.0
0
0
2
0
2
0
0
0
2
6
0
0
1
16
6
31
14
13
16
6
26
29
10
1
3
0
0
0
64
65
54
66
69
52
40
64
54
53
84
51
96
98
91
20
29
14
20
18 .
31
50
10
17
36
13
40
3
2
9
0
0
0
0
0
2
4
0
0
2
2
7
0
0
90
3-80
-------
TABLE 3-8. Seasonal and annual-average morning and afternoon mixed-layer
heights and wind speeds for Grand Junction, Colorado. (Source: Holz-
worth, 1972).
Morning
Season
Winter
Spring
Summer
Autumn
Annual
Height (m)
329
628
307
273
384
Wind Speed (m/s)
3.4
5.4
4.7
3.9
4.3
Afternoon
Height (m)
1160
3166
3940
2133
2600
Wind Speed (m/s)
3.4
6.6
6.1
4.6
5.2
3-81
-------
IMS
P; Itsed on
phot CMC try
M: Iatc4 o* nephcloKtry tfaU
•: »*ied en uncertain e«tr«pol«tlo« of
wiitblllty frequency tfistrlbutton
t$
FIGURE 3-22. Median yearly visual range (miles) and isopleths for suburban/non-urban areas,
1974-76 (Trijonis and Shapland, 1978).
-------
CO
OJ
TABLE 3-9. Summary of background air quality in Uinta and Piceance Basins (concentrations in pg/m3).
National Aablent Air
Quality Standard* (NAAQS)
Pollutant/Averaging Tiae Primary Saoondary
Sulfur Dioxid* (SO.)
3-hr Mxleue — 1300
3-hr 2nd hlgheet
24-hr •exiaua 363 —
24-hr 2nd hlgheet
Annual average . 80 —
Total Snap. Partic. (ISP)
24-hr awUauB 260 150
Annual average 75 60
Nitrogen Dioxide MDj
Annual average 100
Carbon Monoxlda (CO)
1-hr aaxlaua 40,000
1-hr 2nd higheet
8-hr •axlaua 10,000 —
8-hr 2nd hlgheat
Ozone (Oj)
1-hr •exiaua 240 —
1-hr 2nd higheet
Annual average —
loeco-
1980 1975
4.5 IS
15
2 10
10
1 3
83.5 74.7
19 24.5
5.5 5
7400 3000
2000
4500 1300
700
160 150
150
70
Whit. ««•* ***• •"•><* Cathedral Chevron Naval Oil
(O-o. 0-b. A6) Rio Blanco (C-a) Bluff. (C-h) Sit. SK.I. B~.~.
1976
5
5
0
0
0
101.2
23.5
' 5
3000
2700
1800
1700
140
140
6)
1977
n
10
to
5
1
58.3
22.2
0
1000
900
600
600
160
160
61
1978
27
25
14
13
3
62.7
15.0
1
700
600
400
400
137
135
n
1979
113
47
14
6
0
52.9
12.5
1
1800
900
500
500
151
149
73
1980 1975-1976 1978
16 234, 130 78, 286
16
9
a
1 26, 26 26, 26
127 211, 102, 59, 60,
281 160
19.6 9, 9, 15 14, 13, 26
2 17, 11* 11*
2500 4800, 6900 _
1900
1500
1400
143 136, 178 176, 144
143
68 68,62 114, 76
1978 1980-81 1980 1981
•8 21 44 118
43 13 69
1 3
178 30 37
11
10* 4
2800 5400
1700
160 148 206 265
Total
concentrationa (NO +
-------
TABLE 3-10. Measured ambient concentrations of total
suspended participates (TSP) 1n study region.
Site
Colorado
Frulta
Palisade
Rifle
Glenwood Springs
Meeker
Rangeley
Craig
Utah
Green River
Vernal
U-a, U-b
Tosco
Federal and State
Ambient A1r Quality
Primary
Secondary
Year
1979
1980
1979
1980
1979
1980
1979
1980
1980
1980
1980
1979
1980
1978
1979
1980
1978
1979
Maximum
24-hr
173*
166*
13°*
163
694*
510*
188*
203
212*
273*
382*
196*
163*
105
106
80
63
53
84
Annual
Geometric
Mean
69*
43
47
128*
156T
57,
68
66*
70*
86*
64*
53
31
35
32
15
13
19
Standards
260
150
75
60
Exceedance of secondary standard.
* Exceedance of primary and secondary standards.
3-34
-------
FIGURE 3-23. Comparison of predicted and measured annual average $04
(ug/m3). (Values measured at MFCS sites are shown in solid boxes;
values measured at WRAQS sites are shown in dashed boxes.)
3-35
-------
TABLE 3-11. Annual total wet sulfur and nitrogen deposition
based on measurements at NADP sites 1n 1981. (Source:
Latlmflr ft. al., 1985b).
Site Name
Toofcetone
Oliver Knoll
Grand Canyon NF
Organ Pipe Cactus HI
Bishop
Bopland
Sequoia NP
Channel Islands NP
Davis Site
Yosemite NP
Alanosa
Mesa Verde NP
Sand Spring
Bocky MtnNP
Hanitou
Pawnee
Lost Creek Dan
I-Bar
Udar Mountain
Met
Sulfate
kg S/ha
4.520
2.660
2.984
1.304
0.326
0.950
1.243
1.784
1.359
1.224
1.268
2.502
2.017
1.390
2.301
2.103
0.937
1.486
1.357
Wet
Nitrate
kgN/ha
1.216
0.677
0.841
0.384
0.123
0.562
0.804
0.703
1.043
0.916
0.498
1.067
0.824
0.854
1.437
1.145
0.410
0.613
0.569
Data
Recovery
0.929
0.955
0.955
1.000
0.883
0.983
0.887
0.762
0.833
0.833
0.967
0.929
1.000
0.979
0.962
1.000
0.837
0.983
0.925
3-86
-------
•ROCK springs MY
BSali Lake City UT
•NflDP Sites
Cedar Noun
'Pawnee CO
••Sand Spring CO _ . „
* Rocky Mountain National
Park CO
nn UT
BGrand Junction CO
Maaa Verde CO
• Boulder CO
eDenver CO
»Hanitou CO
BColorado Springs
Rlemosa CO
ery CO
FIGURE 3-24. NADP wet deposition monitoring sites (A) and cities (D)
in the central Rocky Mountain region.
3-87
-------
FIGURE 3-25. Comparison of predicted and measured annual average
wet sulfur deposition (kg/ha) in 1981. (Values measured at NADP
sites are shown in solid boxes.)
3-88
-------
1981. In the Rocky Mountain region, measured deposition expressed as sul-
fur ranged from 1.4 to 2.5 kg/ha. (Deposition rates expressed as sulfate
are three times larger than rates expressed as sulfur because the molecu-
lar weight of sulfate 1s three times that of sulfur. Thus the deposition
rates of 1.4 to 2.5 kg/ha, when expressed as sulfate, are 1n the range of
4.2 to 7.5 kg/ha.) As shown 1n Figure 3-26, nitrogen deposition rates are
somewhat lower, ranging from 0.5 to 1.4 kg/ha (these deposition rates,
when expressed as nitrate, are 4.4 times larger, I.e., 2.2 to 6.2 kg/ha).
Approximately one-half of the annual wet deposition of sulfur and nitrogen
1n Colorado occurs during the summer season. In other words, half of a
year's deposition takes place In one quarter of the year. In 1981 the
average annual sulfur and nitrogen deposition rate 1n Colorado (averaged
over six NADP sites) was 1.9 kg/ha and 1.0 kg/ha. Summer sulfur and
nitrogen deposition fluxes were 0.9 and 0.5 kg/ha, respectively, or 47 and
55 percent of the annual deposition fluxes. Very little wet deposition
occurs during the winter, and the remaining annual deposition 1s roughly
evenly split between the spring and fall seasons. The annual variation 1n
deposition rates could be the result of a combination of factors. First,
photochemical reactivity 1s enhanced 1n the summer compared to winter, so
that atmospheric sulfate and nitrate formation may be faster 1n summer
than 1n winter. Second, atmospheric mixing 1s deeper 1n summer than 1n
winter, perhaps allowing sulfate and nitrate to be transported Into the
Rockies better during the summer than 1n winter. Figures 3-27 through
3-31 show weekly deposition rates over the available period of record at
five NADP sites in Colorado (I.e., Manltou, Sand Spring, Rocky Mountain
National Park, Pawnee, and Mesa Verde National Park).
Although most of the wet deposition of sulfur and nitrogen 1n the Rocky
Mountains occurs during the summer, winter deposition 1s also Important.
During winter, the wet and dry deposited pollutants tend to collect within
the snowpack. During early stages of snowmelt, the runoff may contain a
higher concentration of add than 1s normally found 1n the bulk snow
because of the freezing point depression caused by various snowpack
Impurities, Including adds. This phenomenon of a highly concentrated
flow of add 1n the early stages of runoff 1s frequently referred to as
"add shock." During this period a five-fold Increase 1n lake acidity has
been observed by Se1p (1980). The add shock phenomenon has been exten-
sively studied In Scandavanla (Johannes, Galloway, and Trout, 1980), and
has been observed 1n the U.S. 1n the Sierra Nevada Mountains (Ashbaugh et
al., 1985), and the Rocky Mountains (Baron, 1986).
3.2.4 Potential Emission Sources and Regions
Currently, emissions 1n the Rocky Mountain region are relatively low
(Figure 3-32). Emissions are relatively low 1n the western states, with
3-89
-------
FIGURE 3-26. Annual total wet nitrogen deposition (kg/ha)
measured at NADP sites in 1981.
3-90
-------
-, 100
e
en
* 80
c
0
£ 60
o
a
0 40
o
«
<*-
a 20
CO
^
4>
3 0
-
-
I'll
1 Fl—
S le-JJ »l 1 H. n
<••
(
1
1
Jp*
r
ill
If
,
•!
*
7
i
''
ill
'
i
9 U
i
1
PH
ll
!:
it!
'"
>,''
I b
JL
1 b
W
V
!
if!
ii
jj
C
Mi
rl»
.!'
!i;
I
'lj
'
p
ji
1
',
„
1i
jr
'II
"jjl
tt
)J
',
1
'i
ii
•M
'
''
!!
IK
•™
-
_
! ~
'"I
ft ™"
i
rr'l' fsi
May Jun Ju) Aug Sep Oct Nov Dec
100 _
CM
E
80 •«
C
O
60 t!
40
o
Q.
0)
0)
20 3
03
-. 100
-------
Sulfate
_ 100
(M
E
cn
£ 80
c
o
£ 60
0)
o
o.
V
0 40
V
^J
•0
i|
a 20
00
t)
3 0
—
—
-
-
i
i
•*!
,! j
'i
M
« nhliit
"
1
iiii
Til
:::!
i:i:
(
1 J
!"
<
Jan Feb Mar Apr May Jun
r
•
_
-
tjl!
'I*
i
1.
«-
'iii
'
i,
"
'i
'»
\
,
—
•
™
—
n ft illsM n • • IliiiU • • iii> » « • • •
Jul Aug Sep Oct Nov Dec
100 _
CM
E
-80 f
C
O
-60 t!
-40
o
o.
V
o
93
-20 ^
~ 100
(M
e
£ 80
c
o
5 60
0)
o
Q.
V
0 40
V
10
t 20
z
3 n
-
—
-
-
"*
-
;._..n _„
n
rm
f1
'''
i!i
_
mm! Im
~
i;|:
:i:;
;i!i
'!''
t
m
m
M»
1
1.
"F
i
''
('
-
-
-
_
-
^
-
Ft n
iUU ^
(\J
e
80 f
c
o
60 5
CO
0
a.
u
40 °
V
"5o
20 ^
z
OJ
n 3
Aug Sep Oct Nov Dec
FIGURE 3-27b. In 1980 (38.5% valid data).
3-92
-------
~ 100
CVJ
e
f 80 -
c
o
.t 60-
o
a
40 -
3 20
00
Sulfate
-
-
-
~
—
^^.JLJL™
Jan Feb Mar Apr
i :
'"
f"^
il
!
r
f
I1
;r
jj
ti
if
i'
_.
NK
:
'A
1"
1
'!
t
'",
KM
1
'
,!
nt
Nay Jun Jul
^-
!|.;
,|!
PW
„
I"'
t
'li
It"
,'h
II
,,»
I1
i;
jl
1
•'1
Aug
IM*
j
. I :
1
"
i
Tn.juL.jiJ.
Sep Oct Nov Dec
—
—
-
100 _
CM
80 .*
c.
o
r-
60 t!
(0
o
o.
4>
40 °
D
<0
20 |
o>
0 3
Nitrate
~ l\JU
fVl
vM
e
V
f 80
c
0
i 60
(0
o
a
v
Q 40
ti
«u
^ 20
-u
0)
2 0
"
-
-
•
•^
H H H »(-1 mr-i
'••
j^.nlT
i
it:
f
r
j '
i!
(
t
ft
i
;,!'
j!
n
~
1
>>i
.«
•i'
'^
.
H
'!'
„"
i1!'
r
^
1 I1!
i
ii
ji
i!
ij
V
,
i
i
1
PW|
I
1
i
"IT;
i
i
,1,
'i
I
i
i
i"
-
—
-
^m
-nii n^-. 1 1. » iiflo mfl
100 _
<\f
e
80 f
c
o
60 5
(/)
o
o.
u
40 °
*>
20 5
z
j->
n s
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov . Dec
FIGURE 3-27c. In 1981 (69.25! valid data).
3-93
-------
~- 1 vJU
C\J
E
£ 80
c
o
5 60
CO
o
Q.
4)
0 40
0)
<^>
£ 80
c
0
5 60
(0
o
o.
0)
Q 40
0)
<0
L.
^ 20
3 0
M*»
,
mm
u.
May
Jun
mn
j
1
i
,
K*N
I
I
'i
,'
t
A
Jul
r
tm
-
j
Aug
,
mm
m
-
H
Sep
^B
.rJL. . .. .J
1UU „
(\J
E
80 £
c
o
60 *!
in
o
Q.
40 °
0)
^
<0
>4-
20 3
00
« -r
Oct Nov Dec U
Nitrate
_
—
-
—
M
—
-
ml""!™ r«r~I« n * B(—I m ( — ]
73
,
-
••
_
-1,
1 -
1 1
•I
,
_
—
,
.
T
-
-
—
_
—
-
.Ar.^0 -
100 _
eg
E
•s.
80 f
c
o
60 t!
(A
O
Q.
40 Q
0)
«0
20 £
z
u
n 3
Jan- Feb Mar Apr May Jun Jul Aug Sep Oct
Nov Dec
FIGURE 3-27d. in 1982 (65.4% valid data).
3-94
-------
~ 100
E
f 80
c
o
£ 60
CO
o
Q.
1)
0 40
^>
>o
5 20
CO
«+^
4J
3 0
Sulfate
-
-
™
-
*" •
-
p^pw
1 ' I
• ET.E..
FW
I
!
1?
, !
I
••H*
rT*"i
1 1 'i
i , i
Jili,
ii
1
i
!
:'
:»'
1
a.
IT
V
't !
i
1
nw
!
-------
CVJ
E
CT>
£ 80
c
o
t! 60
W '
0
a.
0 40
«
£
s 20
00
2 0
_
-
-
— _
—
>rj!p] jrT_F7n_PHr-u-,»r-
•
Jan Feb Mar Apr Ma;
r
n n n n nl.
1 Jun
_
<;
1 rvf
IK
iii
-.',::
j«!
i:j:
,jl.,P
~
—
-
_
-
—
n
'• !• • n n n_^ni— in
1 UU r~
£
CT>
80 £
c
o
60 t
V)
o
CL
40 °
(D
W
20 -5
CO
D
n 3
Jul Aug Sep Oct Nov Dec
Nitrate
_ 100
f 80
c
o
5 60
(0
o
Q.
4>
0 40
tt
80 f
c
0
60 ^
(0
0
Q.
0)
40 °
0)
-------
100
^•s * w w
(^J
e
* 80
c
o
5 60
0
Q. •
4)
0 40
93
+}
ID
i^
3 20
00
**
3 0
-
-
~
-
~
—
V
JLJLJLJI • • uNL
n1
1;
,
'
'
if!
•
•*
,.
ill 1
1
•III
T<
1 t
31
V
.e
;;!.
»'
f
t
,'_
T
•1
'
!''!
*••
1
1
-
m*
t '
_
-
-
-
n
•J..L.m.pJL
100 _
(M
80 J
,
c
o
60 t!
in
o
Q.
4)
40 °
0)
"Jo
20 3
^
D
n 2
Apr May Jun Jul Aug Sep Oct Nov Dec
100 Nitrate
rg
e
V
* 80
c
0
5 60
o
a
Q 40
c.
t: 20
3 0
-
"
-
_ Fl
!t!: ^^ ' l '
m- *>' • P"
rLfT-i 1^1 ii:"'. j In n
it » » n • • ulLJ l-i 1 !— f~f I— i 1 | in i 1
Jan Feb Mar Apr May Jun Jul
Ln-T-T
T
IT
Aug Sep
_
_
-
-
L-
..m.r-v
Oct Nov . Dec
1UU _
<\j
e
80 f
c
o
60 £
(0
o
a.
u
40 °
«
20 ^
z
0 3
FIGURE 3-28b. In 1981 (76.9X valid data).
3-97
-------
_ 1UU
C\J
E
CT>
£ 80
c
o
5 60
ff>
o
Q.
A)
0 40
o>
|Q
3 20
00
0}
3 0
~ 100
CsJ
c
en
£ 80
c
o
^ 60
(/>
o
Q.
V
0 40
o>
*j
L.
£ 20
z
2 0
_
-
_
—
-
_
-
Jan Feb
ill!
11
. ' I
~n
1 1
"
Mar Apr May Jun
n n
Jul
-
•
••
•R
Aug
!.
!
• ,
!'
»
"'
It
1
u
Sep
•
T-T
Oct
-
—
-
-
""
—
JT_ p.
Nov Dec
Nitrate
_
-
—
-
_
•~
?W_n.^
Jan Feb
1
_. Jl. P f\
iT
1
Mar Apr May Jun
m n
Jul
n
fl.
Aug
r
UM
w
1
Sep
1
rl
Oct
P
_
-
—
—
_
—
-FHJl"
Nov Dec
i uu ^
(SI
E
cn
80 £
C
o
60 ti
w
o
Q.
ID
40 °
D
<0
20 3
0)
0 3
100 _
CM
e
en
80 £
c
o
60 £
V)
o
a
ft)
40 °
A)
**
2
20 t!
u
0 3
FIGURE 3-28c. In 1982 (84.655 valid data).
3-98
-------
- 100
(M
e
en
f 80
c
0
• —
£ 60
W
o
Q.
V
0 40
V
•0
l4_
3 20
CO
^^
o
3 0
_
-
—
"•
_
r""i
i-*d3ditrrTH&
••
'
.
MM
Jan Feb Mar Apr May
Sill
iTI
1 1
111
Jun
fate
flfl
'l
i!'
:;
'
1
•MM
.F
.!'
1
!
i
irf
B!
'!'
j
1
Ij
•
Jul Aug
[W
"it
H"
!3L
"Jjl
i'j
H
Sep
••V
-
-
—
p—
1 1 1 1
1 kJH'h
Oct Nov Dec
100 _
CM
80 f
c
o
60 ^
«
o
Q.
(D
40 °
*J
•0
2.0 3
CO
V
0 3
^ lOOr—
esj
e
X
• 80
c
o
•^-
^ 60
(0
o
Q.
U
0 40
0)
^j
•0
5 20
^%
3 0
-
«•
•
~ _^ ^^^
MM
I
<1tr
f*
! 1 ''
U'
al
T
:e
•r
F
1
>r
P*P
R*
!;:
',.'.':
iliii
-
«•
-
-
fWl
r^JWh n-FL-i-n,"
1 UU ^^
CM
e
^
80 *
c
o
60 5
(0
o
Q.
0)
4.0 °
0)
*»
20 .t:
z
4J
n 3
Apr May Jun Jul Aug Sep Oct Nov Dec
FIGURE 3-28d. in 1983 (84.6% valid data).
3-99
-------
Sulfate
- 100
CM
E
^
CT»
£ 80
c
o
5 60
(0
o
a
0)
0 40
0)
"Jo
~ 20
(X)
a>
3 0
- 100
CSJ
E
£ 80
c
o
5 60
o
a .
0 40
0)
<0
5 20
u
2 n
_
-
_
-
—
^
-
—
rf
. . . . . .n-n-rHunJ 1
Jan Feb Mar Apr
_
-
M
^
•^
-
—
•—,!— iT"! f— 1
. .n. . ..rn'rl- L-T-. !
Jan F«b Mar Apr
iff*
•«•
May
F
[
1
F
May
_
fl,
i
n
!f
-,
I
Jun Jul
Nitrate
In.
sn
Hi
||
!|!i
n
pm
i
i
i
t
,'
'
!
T
,
••«
'*
Jun Jul
•M
!
w-
•!
m
I
"JL
,
Aug
1
n n
Aug
""
-
™
—
™
_
-
_
^^ • n n "™
llJljj' m Ln. . .PI.
Sep Oct Nov Dec
••
••>
^
11 Fl n "
TLru-l Ln-n.n. ..n™
i UU ^,
fM
E
cr>
80 £
c
o •
60 "5
(ft
o
Q.
40 °
0)
<0
»l^
20 "5
0 3
100 _
e
80 £
§
60 5
w
o
S"
40 °
(U
•0
L.
20 ^
z
H)
n 3
Sep Oct Nov Dec
FIGURE 3-29a. Measured weekly sulfate and nltrat wet deposition 1n Rocky
Mountain National Park, Colorado—1981 (71.2% valid data).
3-100
-------
« 100
Q 40
o
t*.
3 20
s
3 0
_
-
_
—
L,.in_. .Hfe. . Jl li^Jl
Jan Feb Mar Apr h>
T
i!
T
1
bui fa
mm
*™
K"
r,
1
-------
Sulfate
~ 1UU
C\J
e
^r
£ 80
c
0
5 60
0)
o
Q.
0)
0 40
0)
<0
i 2°
» 0
~ 100
(M
e
on
£ 80
c
o
£ 60
(D
O
Q.
0)
0 40
«
«o
u
.t 20
2
^
D
3 0
_
-
-
-
"
-
Hi
"I-
.
Jan Feb Mar
k
r
Apr
k
PW
•4n
i
May
,
!
Fh
Jun
im
i
!
I
Jul
• '
.
t
Aug
•
_
—
-
_
-
—
_JL "
— pi. — | | —i | J^| | y- p^j— j t
Sep Oct Nov Dec
Nitrate
-
-
-
.
•
-
m l» n r-i s . J 1
1
vm
'.'.:
"L
t
_
PP^
i.l
]
-FT
i
-
r
MM
•""
-
—
-
-
—
-
jq , ,., , -rrT]T
1 UU ^
C\J
E
80 f
c
o
60 t
o
a.
0}
40 °
D
W
20 3
-------
-100
(VI
e
\
cr
= 80
c
o
•r»
£ 60
o
a
4)
Q 40
*:
0)
n a:
Apr May Jun Jul Aug Sep Oct Nov Dec
_ 100
e
f 80
c
o
•••
tf 60
(0
o
a
V
Q 40
^>
*0
L.
^ 20
*^
u
2 0
Nitrate
^
-
•»
^^n.n- .
Jan Feb
M
i:!:
n
"!' »" L» »
„'
Pi
*
I
*
ct
'H
P-l
ii
"
•f'
i
ii
H
>P«
"
•MM
n
»rf'L
1
,
i
Mar Apr May Jun
-
N
1
t
11
ll
!
;
1
\
•
m.
T
j
Jul
•••
*.
•H
V
fffl ffl-1 __
Aug Sep Oct Nov Dec
100 _
(M
•s.
cn
80 =
c.
o
60 5
0)
0
CL
0)
40 °
0)
*^
20 £
OJ
0 3
FIGURE 3-30a. Measured weekly sulfate and nitrate wet deposition in
Pawnee, Colorado—1980 (48.IX valid data).
3-103
-------
CM
E
^>
* 80
c
o
5 -60
f 80
c
o
5 60
(0
0
o.
OJ
0 40
u
^>
t.
** 20
*f ~ •
z
3 0
_
-
—
-
-
Jlrffl"
:::•
n
Jan Feb Mar Apr
M*
— ,
,
HP
'
••
n
1
,
May Jun
Ml
}
-
'f
NH
[
1
,
"
M«
I
i
Ju
mm
1
i
[
(W
}
•V
1
,!,
''
,
Aug
jD
n
y
,
•w
it
i
Et
T,
ill!
ijj!
-
—
-
m
—
—
1 ,. "
Sep Oct Nov Dec
Nitrate
_
—
—
—
••
B-w
n « m » • • « «r—i UFf ' 1
•9
Jan Feb Mar Apr
n
n
„
f
ff^
T
FW*1
-
.1
May Jun
-
-
1
•"
MM
•*
rM
Ju Aug Se
•"
p Oct
_
-
_
—
-
Nov Dec
1 UU —
•o
t|
20 3
00
0)
0 3
100 _
CM
e
cr>
80 f
c
o
60 t
in
o
a.
0)
40 Q
u
•0
on *J
0 *
FIGURE 3-30b. In 1981 (59.6% valid data).
3-104
-------
_ 100
en
f 80
c
o
£ 60
(0
o
a
v
3
CO
40
20
Sulfate
••
-
-
»
!i!l
1
111
ii!
l|
ti
i1'
!i
T r-
'LF
•fl'
j[
J
I
,,
KF
1
j1
• '
H
»t
1 I
11,
i
Jan Feb Map App May Jun
()Mj""j
ill
,i;
i
'1!
Jul Aug
••
-
"
-
ra
1 "InvM [it i n ~
^I'rfTy LTL M
Sep Oct Nov Dec
100 _
CM
E
"V
80 J
c
0
60 "E
0
Q.
0)
40 °
0)
•0
^^
20 3
V)
0)
0 2
^ 100. —
80 f
c
o
f f\ ^^
60 ,t-
-------
Sulfate
_ 1 UU
(M
e
cr>
f 80
c
o
2 60
(0
0
Q.
ft)
0 40
ft>
I
1
^^
_
^l_j-T
n
mm
;
1.
j
N«*
r
'
••
•'
J
"•!
, ;
n
—
i
kfr
»
"•
•
_
—
—
_
-
'
-
B BBP^
I UU ^
fM
E
tr>
80 £
c
o
60 tf
0)
o
a
0)
40 °
4)
f 80
c
o
5 60
•0
t.
i 20
ft>
2E Q
••
-
-
_
-
•
••
Jan Feb
n
:•:•
"^
Mar
_
>tn
m
—
Apr
•
1
m
'
j!
t
-T
m
II':
I
i
"f
May Jun
n
t
—
Jul
Mi
Aug
—
-
—
_
~
—
1 f!l 01 t^ 1 * * i t (I B •* *** "^-.j
IUU „
N.
80 f
c
o
60 ^
(0
o
Q.
40 °
ft>
4.)
10
20 ^
n 2
Sep Oct Nov Dec
FIGURE 3-30d. In 1983 (65.4« valid data).
3-106
-------
« 100
(V)
e
^
J! 80
c
o
~
w 6°
o
a
o
Q 40
V
+>
ID
it-
a 20
*>
* 0
Sulfate
-
~
-
L,.Rlh-
,
i
•?
ww
••
•P
a
,
* *
Jan Feb Mar
3^
mm
t
1
s
f
•V
d
• • • ">^«
M*
' I
1
"t:
Apr Nay Jun
m
~*
i
,i
1
i
'.
i
!
Jul
••
1 (
7
,.
i
:j|
i{j
'
jl
I
I
f
'l
m
'i
MV
i
Aug Sep
Oct
—•—I
tj
1 '
t
JT
Nov
!W
ii
-
-
J
Dec
100 _
CsJ
E
cn
80 ;*
c
o
60 tf
(A
0
a
0)
40 °
^
^j
-------
Sulfate
„ 1 UU
<\J
E
? 80
c
o
5 60
(0
o
Q.
1)
0 40
^
"4-
3 20
CO
0)
3 0
_ 100
CM
E
CT>
£ 80
c
o
5 60
(0
o
o.
u
0 40
V
^ 20
z
0>
2 0
-
-
-
-
p
pn . i
" (Ill 1|L
• «|lfi|l!if^i!iill;;ii
MB
i
:;;;
•••
Jan Feb Mar
ftiH 1 !• ni .In !....]• «flii
Apr Nay Jun
HH
'"
t
i
'•
Jul
-
-
—
J
j
i . —
j •
i "•" p
i
! i-:"
__^ ' ' n
nl In n_J ' n i nlili ni _r- ''
Aug Sep Oct Nov Dec
Nitrate
—
—
—
—
~ F)
" rmL M
m n| 1 1 | fT_ [
-
,_,-„
Jan Feb Mar
_r
"
-
•
mm
?
:
':
f
1
i
'•{
\ rr
PI FT— 1 m
Apr nay Jun Ju Auq Sen Det wnu n.r
1UU _
(\J
E
80 ,f
c
o
60 5
-------
;aoo Scale
•o in : kiloton/yr
| Point sources SOx
N Area sources SOx
Point sources NOx
Area sources NOx
Point sources RHC
Area sources RHC
FIGURE 3-32. State emissions of SOX. NOX and reactive hydrocarbons (RHC) in
the West (Based on 1983 NEDS inventory).
3-109
-------
fairly significant exceptions 1n California, Arizona, and Texas. Large
point sources (I.e., smelters and power plants) and urban areas are shown
1n the southwestern U.S. for the years 1980 and 1995 (see Figures 3-33 and
3-34). In the Rocky Mountain region, existing and future emissions
sources of S02 and NOX Include (1) power plants 1n northwestern Colorado,
southwestern Wyoming, and central Utah and (2) the urban areas of Denver/
Ft.CollIns/Colorado Springs, Grand Junction, and Salt Lake City. Signifi-
cant emissions sources elsewhere 1n the Southwest Include large coal-fired
power plants In the Four Corners area; large copper smelters 1n Utah,
Nevada, New Mexico, Texas, and mainly 1n Arizona and Mexico; and the urban
areas of California and Arizona.
Preliminary regional model calculations of source contributions to wet
sulfur deposition in western Colorado suggest that approximately 90 per-
cent of the sulfur 1s transported from outside the state. Estimated
sources of wet-deposited pollutants 1n the Colorado Rockies are shown 1n
Figure 3-35.
Trajectory analysis can be used to estimate the source regions' contribu-
ting to deposition 1n a given location. Figure 3-36 shows the relative
frequency with which trajectories emanating from a number of areas 1n the
United States pass over the Denver area 1n summer and winter. This calcu-
lation was based on the workbook by Draxler and Heffter (1981), which
Includes the climatology of regional-scale dispersion throughout the
United States.
3.3 SURVEY OF EXISTING ACID DEPOSITION MODELS
From the review of the pertinent chemical and physical processes governing
the fate of atmospheric pollutants presented 1n Section 3.1, 1t 1s evident
that modeling approaches must address various time and space scales. Any
model developed for short time scales, such as one providing hourly to
dally average concentrations, needs to consider short-term fluctuations of
transport winds, such as the complex flow over complex terrain and near
fronts as well as the effects of small-scale and mesoscale turbulent
flows. The diurnal characteristics of the boundary layer also exert a
great Influence on short-term pollutant distribution patterns. Similarly,
the time-dependent nature of chemical transformations and dry deposition
and the temporal variability of precipitation 1s Important. A model
capable of high spatial resolution needs to provide detailed treatment
of the vertical and horizontal pollutant distributions near major sources,
which requires specification of plume rise and near-source transformation
rates.
3-110
-------
(a) 1980
(b) 1995
FIGURE 3-33. Sulfur dioxide emissions in the Southwest, 1980-1995.
The circled areas are proportional to emission rate.
3-111
-------
(a) 1980
(b) 1995
FIGURE 3-34. Nitrogen oxide emissions in the Southwest, 1980-1995
The circled areas are proportional to emission rate.
3-112
-------
362 Copper Shelters
Power Plants
22 Other Industry
42 Urban Areas
102 Oil Industry
202 Transport fro* Outside Kegton
FIGURE 3-35. Relative contributions of different sources to wet
sulfur deposition 1n the Colorado Rockies 1n 1981. (Source:
Latlmer et al., 1985.)
3-113
-------
*>,
I
»—•
t~>
t*
FIGURE 3-36a. Relative frequency with which trajectories emanating from a source area
end UP In Denver: winter. (The lighter the shading, the fewer the trajectories.)
-------
FIGURE 3-36b. Summer.
-------
If long-term averages are to be provided by a model, 1t 1s necessary to
account for short-term variations where they are Important. Many models
simply neglect these variations without proper justification. In some
Instances, such as the specification of long-term dispersion, the effects
of synoptic motions dominate the effects of small-scale dispersion, thus
justifying the neglect (Munn and Bolln, 1971). However, because of pos-
sible temporal correlations, processes other than transport and dispersion
may not be treatable 1n as simple a manner. For example, Stewart and co-
workers (1983b) showed that the temporal correlation of high sulfate con-
centrations with high scavenging coefficient (I.e., large precipitation
rate) leads to greater wet deposition than occurs with a time-averaged
scavenging coefficient.
Models that Invoke homogeneous assumptions, such as constant wind velocity
and precipitation rate, need to be concerned with the effects of spatial
correlations on region-wide averaging. For example, those models that
specify a uniform precipitation scavenging rate are likely to underpredlct
wet deposition on regions of elevated terrain due to the effects of oro-
graphlcally enhanced precipitation. This may be an undesirable feature
for a Rocky Mountain add deposition model since most sensitive lakes are
located at these higher elevations.
3.3.1 Modeling Concepts
A variety of regional and mesoscale models have been developed 1n the past
decade. For convenience, these models can be classified according to
modeling approaches. As shown 1n Figure 3-37, the models can first be
divided Into those designed primarily for episodic conditions and those
for long-term applications (averaging one month or longer). Although an
episodic model can become, 1n effect, a long-term model 1f exercised over
periods longer than one month, computational or Input data requirements
often limit the use of some episodic models to periods of less than a
month.
The episodic models can be further divided Into three categories according
to model framework: the fixed-coordinate EuleHan models, the moving-
framework Lagranglan models, and hybrid models, which merge aspects of
both the Lagranglan and Eulerlan concepts. On the other hand, long-term
models can be classified Into two basic categories. In the first, the
long-term averages are derived from short-term concentrations computed
from an episodic model 1n a deterministic fashion. In the second cate-
gory', a representative transport and dispersion pattern 1s first obtained
from either Individual trajectory calculations or from climatic data. The
subsequent diffusion calculations provide long-term concentrations only
1n a statistical sense.
3-116
-------
MEGIONAL/MESOSCALE MODELS
EPISODIC MODELS
LAGRANGIAN
LONG-TERM MODELS
HYBRID
EULERIAN
i
»-
i-
—i
•
1
1
[SOURCE
(Forward
ORIENTED
Trajectory)
PLUME
PUFF
I PARTICLE
RECEPTOR ORIENTED
(Backward Trajectory)
BOX
UPWIND SECTOR
ANALYTIC
DETERMINISTIC MODELS
STATISTICAL MODELS
BOX I GRID
2-DINENSIONAL
3-OIMENSIONAL
FIGURE 3-37. Classification of long-range transport noddling approaches.
-------
For the purpose of classifying regional and mesoscale models, we will
define episodic models as those models that yield results with temporal
resolution equal to or shorter than one day. This should not Imply that
these models cannot be used to generate longer-term averages. Indeed, a
number of episodic models are used to calculate monthly, seasonal, and
even annual concentrations, but their underlying concept 1s based on simu-
lating the direct Interaction of emissions and short-term wind fluctua-
tions.
3.3.1.1 Eulerlan Models
The modeling approaches discussed 1n this chapter focus on the atmospheric
diffusion aspect of long-range transport. These kinematic models examine
the effect of the atmosphere's motion on pollutant distribution rather
than the forces that govern the atmospheric motion. Models that treat the
latter are referred to as dynamic models. Their use 1s widespread for
weather prediction and, to a lesser degree, air pollution studies. The
dynamic model of Kreltzburg and Leach (1979), for example, has been used
to determine three-dimensional trajectories near synoptic low-pressure
systems for atmospheric cleansing studies. Transport of pollutants under
sea breeze conditions has also been Investigated using a dynamic model
(Dleterle and Tingle, 1976). The dynamic model of Plelke (1974) was
recently used 1n connection with pollutant transport studies 1n mountain-
valley flows (McNIder, Hanna, and Plelke, 1981).
Nevertheless, kinematic models are more popular for air pollution studies
because they are far simpler, consisting predominantly of the mass conser-
vation equation for one or more pollutant species. Momentum and thermo-
dynamlc equations are not required. Instead, atmospheric flow fields are
continuously updated through model Input. Although this results In data-
Intensive models, the data are readily available and usually processed for
model Input 1n an objective manner. A careful examination of existing
Eulerlan episodic models reveals a hierarchy based on the degree of com-
prehensiveness 1n the treatment of various atmospheric transport and dis-
persion processes. The hierarchy, as defined by Stewart and L1u (1982),
1s listed below from least to most comprehensive.
Two-dimensional grid; constant mixing layer.
Two-dimensional grid; variable mixing layer; more detailed treatment
of the surface layer.
3-118
-------
Three-dimensional grid; two-dimensional winds; some vertical resolu-
tion, Including variable mixing layer and vertical d1ffus1v1ty pro-
file.
Three-dimensional grid; three-dimensional winds; vertical detail;
large-scale vertical mass transport through convergence/divergence.
3.3.1.2 Lagranglan Models
The majority of Lagranglan regional and mesoscale models designed for epi-
sodic applications are sequential trajectory models, which can be cate-
gorized according to two modes of operation. As shown 1n Figure 3-37,
they are either source-oriented models, which use forward trajectories, or
receptor-oriented models, which use backward trajectories. The former
approach 1s primarily used to examine the fate of emissions emanating from
Individual source or source regions, while the latter focuses on the
resultant pollutant Impact on Individual receptor or receptor regions from
distant upwind emission sources. The short-term nature of these Lagran-
glan models requires simultaneous simulation of the transport-dispersion
and kinetic processes. No attempt 1s made to Isolate these two processes,
as Is characteristic of the statistical long-term approach.
Although the sequential trajectory approach 1s often used for long-term
simulation, the short-term application requires some parameterization
of the plume meandering and the plume spread. The guideline used for
depicting this small-scale dispersion assumes that under steady-state
homogeneous conditions the concentration distribution of an Inert species
has a Gaussian form.
Source-oriented Lagranglan models typically treat the plume elements 1n
either a puff or segmented format. The puff superposition model repre-
sents a continuous plume by a series of discrete, sequentially emitted,
expanding puffs. Alternatively, one may visualize the configuration of a
continuous plume as a series of discrete but contiguous plume segments,
the length of which depend on the wind speed and time step. The formula-
tion of dispersion 1s different for each type. The plume segment approach
treats only dispersion perpendicular to the wind direction, making the
justifiable assumption that pollution advectlon dominates dispersion along
the transport vector. On the other hand, puff models treat horizontal
dispersion both along and perpendicular to the transport vector. The dis-
persion parameters are usually specified by the relationship oh « ax « ay,
where the subscript h refers to the horizontal direction, and subscripts x
and y refer to the component along the wind and across the wind respect-
ively. Dispersion along the transport vector 1s required because of dis-
continuous emissions.
3-119
-------
There are advantages and disadvantages Involved with each type of plume
representation. Several years ago a workshop was held specifically to
address this question of plume representation (Nappo, 1978). Although the
Issue of the best method 1s still unresolved, the workshop analyzed each
method quite comprehensively. Advantages of the puff approach over the
plume segment approach that were noted at the conference Include
A better physical representation of the plume under curvilinear flow
or stagnant conditions.
A conceptually simpler approach to treatment of plume rise and to the
Interaction of pollutant mass with boundary layer.
However, advocates of the plume segment approach noted several drawbacks
to the puff approach:
Variable puff release Intervals are needed so that the puff approach
approximates a continuous emission release near the source, where
puffs are smaller; further downwind, where puff dimensions are large;
as well as 1n nonstatlonary flow conditions.
The treatment of plume chemistry requires a knowledge of the volume
within which chemical transformations occur and the rate of dilution
of the volume. Treatment of continuous emissions by discrete puffs
that dilute Independent of each other, particularly 1n the downwind
direction, does not adequately represent the diluting volume in a
continuous plume.
The a priori assumption that the dispersion along the trajectory
equals the dispersion 1n a transverse direction 1s not Justified.
Dispersion along the wind direction should be dependent on the puff
spacing, becoming negligible as the puff release-Interval decreases.
Most source-oriented sequential trajectory models are conceptually Identi-
cal (Stewart and L1u, 1982). Differences 1n the models are found predomi-
nantly 1n the prescription for the horizontal and vertical dispersion
algorithms.
An alternate technique for simulating emissions transport, dispersion, and
removal, from a source-oriented Lagranglan point of view, 1s to track a
series of discrete particles from each source. Advectlon 1s simulated 1n
much the same manner as In other source-oriented sequential trajectory
models. After the advectlon step, the particle's position 1s redefined 1n
a quasi-random manner. The particle 1s displaced by a representative dis-
persive length, determined from scale considerations, typical horizontal
3-120
-------
eddy diffuslvltles, and the time step. Particles are released at frequent
time Intervals (typically three hours) and their horizontal positions are
later mapped onto a grid to calculate concentration distributions. Chemi-
cal transformation and removal processes may be Incorporated Into the
model as stochastic events during each time step or as first-order decay
processes.
Another type of source-oriented model uses an approach that 1s completely
different from that of the plume segment, puff, or particle models. A
representative model of this type, developed under sponsorship of the
Scandinavian Council for Applied Research by Nordlund (1975), treats
transport and removal of S02 from the reference frame of a series of con-
tiguous horizontal cells moving with an average wind velocity. The cell
dimensions are equal to the dimensions of the emissions grid and are
Initialized at the upwind boundaries of the region of Interest. A tem-
porally and spatially variable wind field advects the Individual cells
according to the local wind. So that the cells maintain common boun-
daries, the cells are allowed to deform along both axes. Horizontal dif-
fusion 1s neglected by restricting mass flux between boundaries, although
a smoothing function 1s passed over the concentration fields at regular
Intervals. Emissions are Injected Into the cells as they pass over the
grid and are assumed to mix Instantaneously throughout a representative
mixing depth. Within each cell, SO? concentrations are affected by oxida-
tion and dry and wet deposition. These'processes are parameterized 1n a
lumped fashion, I.e., through the specification of a single decay rate.
The development of receptor-oriented sequential models has generally pro-
ceeded along two different lines of thought. The objective of both
approaches 1s to calculate the concentration or deposition of a pollutant
at a selected receptor over a given period of time. One method uses a
sequential trajectory approach, and 1s typically exercised 1n two
phases. Input to the first phase consists of spatially and temporally
varying representative wind fields and receptor locations. Output from
this phase consists of a series of trajectory segment endpolnts Identified
by the appropriate receptor and a representative advectlon time. This
Information 1s supplied along with precipitation fields (1f appropriate),
mixing depth fields, and a gHdded emission field to a concentration/
deposition program that Integrates the pollutant mass balance equations
for a parcel moving along a given trajectory toward the receptor. Emis-
sions are Injected and uniformly mixed as the parcel passes overhead.
An alternative type of receptor-oriented Lagranglan model, applicable
to both long-term and episodic concentration and deposition estimates,
utilizes average single-layer or multiple-layer back-trajectory endpolnts
to define an upwind sector from which air parcels passing over a sensitive
receptor originate. These trajectory endpolnts are determined at specific
3-121
-------
time Intervals. The primary conceptual difference between this model and
trajectory box models 1s that the "box" or air parcel 1s represented by an
elliptical puff encompassing the area within which emissions are potential
contributors to the receptor concentration or deposition. As one looks
farther back 1n time (toward the trajectory origin), the ellipse Increases
1n horizontal dimensions by an amount proportional to the x,y distribution
of sublayer trajectory endpolnts. The variables determined within this
"puff are the probabilities of contribution to a normalized concentration
and deposition for both SC^ and sulfates.
A moving coordinate system 1s fixed to the puff as 1t approaches the
receptor so that the distribution of the probability of source contribu-
tions can be Influenced by dry and wet deposition. The output of the
model 1s a spatial field representing the probable contribution per unit
area for a given averaging time. Multiplication of this field by a
spatially varying emission field yields a predicted receptor Impact (con-
centration or deposition) and the absolute contributions per unit area
from the entire emission region. A representative model of this type 1s
the UMACID model (Samson, 1980, 1981).
3.3.1.3 Hybrid Models
The final type of episodic model discussed here 1s the hybrid approach,
which characterizes the transport and diffusion processes by combining
EuleHan and Lagranglan frameworks. The RTM-II model (Durran et al.,
1979; L1u, Stewart, and Henderson, 1982) 1s one example of the hybrid
approach. This model uses the Lagranglan plume segment approach to treat
near-source transport and diffusion 1n order to avoid unrealistic Initial
dilution of the emissions. When the plume segment expands to horizontal
and vertical dimensions commensurate with a grid cell, the pollutant mass
1s transferred to the grid cell for subsequent calculations.
The hybrid model RTM-II and the popular Lagranglan puff and Lagranglan
plume segment models are schematically Illustrated 1n Figure 3-38. The
primary advantage of the hybrid approach over the Lagranglan approach 1s
the constant spatial resolution at large downwind distances. In the
Lagranglan models spatial resolution Increases with travel time.
Another hybrid model, ADPIC, developed at Lawrence Llvermore Laboratory
(Lange, 1978), uses a combination of a three-dimensional wind and diffu-
sive wind to transport a large number of particles 1n a Lagranglan
fashion. Each particle represents a quantized emission and 1s tagged to
-permit chemical transformation and removal processes to be parameterized.
The model 1s considered hybrid 1n nature because the diffusion process 1s
parameterized 1n a Eulerlan manner. During each time step, concentrations
3-122
-------
Lagrangian
Puff Concept
Lagrangian
Plume Segment
Concept
Eulerian-
Lagrangian
Hybrid
Concept
FIGURE 3-38. The Lagrangian puff, Lagrangian plume segment
and Systems Applications hybrid Eulerian (RTM-II) models.
3-123
-------
1n adjacent cells are obtained by counting particles. A diffusive wind
1s then calculated using the concentration gradient and a spatially and
temporally variable eddy d1ffus1v1ty. The diffusive wind 1s vectorally
added to the transport wind for the Lagranglan advectlon step. Because of
the large number of particles released to simulate a continuous plume, the
model 1s most economically applied to a single source over small and
Intermediate spatial scales. Extension of this modeling technique to
multiple sources over mesoscales and regional scales 1s, 1n theory,
straightforward.
3.3.1.4 Long-Term Models
From the standpoint of pollutant budget and control strategies, It Is
of Interest to determine the long-term average of airborne pollutants 1n
various regions. Long-term regional models are designed to provide esti-
mates of monthly, annual, or seasonal concentrations and deposition 1n an
economical manner. Although such models use many different modeling con-
cepts to describe transport, dispersion, and removal, the majority fall
Into two categories.
The first category comprises those models that determine long-term aver-
ages from shorter-term concentration and deposition estimates computed 1n
a deterministic fashion by an episodic-type model. Such models are simi-
lar to those discussed In the previous section, although parameterization
may be simplified for computational economy.
Long-term models 1n the second category adopt a statistical approach.
Individual trajectory segments are calculated and their endpolnt coordi-
nates combined to obtain an average location and standard deviation of the
pollutant distribution as a function of time. Transformation and removal
processes are considered statistically Independent of the distribution
function. To reduce the computational burden, many long-term models of
this category eliminate the time-consuming and data-Intensive trajectory
calculations and Instead specify the pollutant distributions resulting
from transport and dispersion by a simple analytic form using c!1matolog1-
cal wind data. Examples of both types of long-term models are discussed
below.
Many types of deterministic models have been developed. One of the.
simplest approaches 1s Illustrated by a model developed under sponsorship
of Environmental Canada (McMahon, Denlson, and Fleming, 1976). This
model, the Long-Distance A1r Pollution Transport Model (CCIW/LDAPTM),
treats transport, diffusion, transformation, and deposition of sulfur and
nitrogen species during a time Interval of 24 hours, using the concept
often referred to as the Eulerlan "sector box" approach. A receptor 1s
3-124
-------
Impacted by emissions 1f 1t lies within an expanding sector downwind of
the emissions. Each sector 1s defined by a 24-hour average wind velocity
and an assumed spread angle, which 1s a function of the 24-hour average
stability and a representative wind meandering angle. Pollutant trans-
formation and dry and wet deposition are calculated by evaluating a set of
exponential decay functions using the travel time between source and
receptor as the Independent variable. Use of exponential decay functions
Implies that all transformation and deposition rates are constant over the
24-hour period.
Other models characterized as deterministic Include those Lagranglan epi-
sodic models that can economically be run for long time periods. Economy
can be achieved by simplifying the dispersion, transformation, and removal
parameterization and by tracking fewer puffs per day. Alternatively,
the economy may be achieved by neglecting the horizontal dispersion of
sequentially released plume elements. This method assumes that the varia-
bility of trajectory position 1s sufficient to describe the long-term con-
centration and deposition pattern. In the absence of systematic trajec-
tory biases, the long-term grldded concentration and deposition fields
should not reflect this artificial restriction on the pollutant disper-
sion.
Long-term regional models that specify the mean pollutant transport and
dispersion from trajectory endpolnt statistics are classified as statisti-
cal long-term models. The modeling concept adopted by these models 1s
similar to .the spurce-orlented puff approach except that the puff location
and horizontal dimensions are governed by trajectory statistics as pro-
posed by Bolln and Persson (1975) Instead of by small-scale dispersion
parameters. Because these puffs are entitles only 1n the time-averaged
sense, the explicit Interaction between the pollutant mass and the pre-
cipitation scavenging mechanism, both of which vary temporally and
spatially, 1s absent. This limitation has been partially overcome by the
methods used 1n the ANL/STRAP and ANL/ASTRAP models, discussed 1n detail
by Stewart and L1u (1982). However, most statistical long-term models
assume that over long time periods the Interactions are adequately simula-
ted by time-mean Interaction term.
Certain long-term models of the statistical type eliminate the time-con-
suming, data-Intensive trajectory calculations. One such model has been
proposed by Fay and Rosenzwelg (1980). This model 1s based on the assump-
tion that, over sufficiently long time periods, the mean horizontal trans-
port and diffusion can be represented by a steady-state solution to the
diffusion equation. If temporally and spatially constant transport, dis-
persion, transformation, and removal processes are specified, the solution
1s given by an analytic expression. Transport 1s specified by a mean wind
velocity; dispersion 1s governed by a representative horizontal eddy d1f-
fuslvlty. Transformation and removal are specified by bulk decay con-
stants, whereas a uniform vertical concentration distribution 1s specified
beneath a cUmatologlcally representative, spatially uniform mixing
height.
-------
Another model of this type 1s the one developed by Fisher (1975) for
examining the European SC^ budget, and modified to Include the sulfate
budget (Fisher, 1978). The model 1s a steady-state statistical model
1n the sense that long-term average concentrations and depositions are
determined from frequency distributions of various parameters. In the
earlier version the direction from which horizontal transport occurs was
treated by wind rose data; 1n the updated version trajectory roses are
utilized. These latter data Indicate the frequency with which trajec-
tories originate 1n different upwind sectors from a given receptor. The
downwind concentrations and depositions are obtained from the product of
(1) an expression specifying normalized concentration as a function of
downwind distance, height, depth of the mixed layer, deposition velocity,
wind speed, eddy d1ffus1v1ty, and stability, and (2) a frequency distribu-
tion of these variables. Wet deposition, also Incorporated Into the
model, 1s specified by assuming random transition from wet to dry periods
as prescribed by Rodhe and Grande!1 (1972). In the updated model Lagran-
glan wet and dry periods are specified for Individual geographic regions
1n order to better Incorporate the effects of scavenging by orographlc
(I.e., terrain Induced) precipitation.
3.3.2 Existing Regional and Mesoscale Models
Existing regional and mesoscale transport models that are readily Identi-
fied 1n the open literature are listed 1n Table 3-12. From this survey,
candidate models suitable for application to the add deposition Issue 1n
the Rocky Mountains will be selected. Although some of these models are
exclusively designed for a-mesoscale (250 - 2500 km) applications, most
are capable of simulating pollutant transport over scales ranging from 25-
250 km (the a-mesoscale range). This survey 1s not meant to be exhaustive
since current models are continually being modified or updated as new
Information or parameterlzatlons become available.
Because there 1s no convention for naming or renaming models as Improve-
ments or new applications arise, some models 1n our 11st are predecessors
to others. Model Information has been derived from publications and
updated through direct contact with model developers where appropriate.
The variety of model Input requirements, output Information, transport and
dispersion algorithms, physical and chemical transformation, and dry and
wet deposition parameterlzatlons are discussed 1n this section. High-
lights of each model's methods of treating these processes are listed 1n
tabular format 1n Table 3-13. The discussion below 1s based on Informa-
tion contained 1n the table.
3-126
-------
TABLE 3-12. Existing regional air quality and deposition nodels.
Model Name and Developer
References
Model Type
Long- or Short-
Tern Assessment
ro
ACS/ESC
Acres Consulting
Services Ltd.
AES/LRT
Atmospheric Environment
Service; Canada
ANL/STRAP
Argoime National
Laboratory
ANL/ASTRAP
Argonne National
Laboratory
ARL/ATAD
NOAA Air Resources
Laboratory
ARL/BAT
NOAA Air Resources
Laboratory
ARL/MTDF
NOAA Air Resources
Laboratory
McMahon, Oenlson, and
Flenlng (1976)
Voldner et al. (1981)
Olson and Voldner (1981)
Shelh (1977)
Shannon (1981)
Shannon (1979)
Heffter (1980)
Heffter (1983)
Draxler (1977, 1979)
Lagranglan
(source oriented)
Lagranglan (receptor-
oriented)
Short tern
Long tern
Lagrang1an-stat1st1cal Long tern
Lagrang1an-stat1st1cal Long tern
Lagranglan (source-
oriented)
Lagranglan (source-
oriented)
Lagranglan (source-
oriented)
Short tern and
long tem
Short tern and
long tem
Short tern
Continued
-------
TABLE 3-12 (continued)
ro
oo
Model Name and Developer
References
Model Type
Long- or Short-
Term Assessment
BNL/AIRSOX
Brookhaven National
Laboratory
CAPITA/MCARLO
CAPITA - Washington
University
CCIW/LDAPTM
Canada Center for
Inland Water
CEGB/SS
Central Electricity
Generating Board, U.K.
CEGB/TD
Central Electricity
Generating Board, U.K.
CERL/LTSDM
Central Electricity
Research Laboratory
England
CIT/HN03
California Institute
of Technology
Meyers et al. (1979)
Klelnman (1983)
Patterson et al. (1981)
Husar and Patterson
(1981)
McMahon (1976)
Maul (1977)
Maul, Barber, and
Martin (1980)
Maul (1977)
Maul, Barber, and
Martin (1980)
Fisher (1978)
Lagranglan (source- Short tern and
oriented) long term
Lagranglan (particle Short term and
model) long term
Eulerlan box model
Russell and Cass (1986) Lagranglan (source-
oriented)
Long term
Lagranglan (steady- Short term
state analytic)
Lagranglan (source- Short term and
oriented) long term
Eu1er1an-stat1st1ca1 Long term
Short terra
Continued
-------
TABLE 3-12 (continued)
VO
Model Name and Developer
References
Model Type
CIT/UAPM
California Institute
of Technology
CSU-EPA/Model B
Colorado State Univer-
sity and U.S. Environ-
mental Protection Agency
DM/RADH
Danes and Moore
DMI/LRPOH
Danish Meteorological
Institute, Denmark
EI/EDRAB
Envlroplan, Inc.
EPA/ROM
EPA Meteorology
Laboratory
McRae, Goodln, and
Seinfeld (1982)
Henml and Relter
(1979, 1981)
Runchal (1980),
Goodln et al. (1980)
Austin et al. (1981)
Prahm and Chrlstensen
(1976, 1977)
Dlttenhoefer et al,
(1983)
Lamb (1983)
Schere and Posslel
(1984)
Eulerlan (3-D)
grid model
Lagranglan (source-
oriented)
Eulerlan (2-D)
pseudospectral
Long- or Short-
Term Assessment
Short term
Short term
Lagranglan-partlcle Short term
model
Short term
Lagranglan (receptor- Long term
oriented)
Eulerlan (3-D) grid Short term
model
Continued
-------
TABLE 3-12 (continued)
Model Name and Developer
References
Model Type
Long- or Short-
Term Assessment
CO
I
GO
o
ERT/ADOM
Environmental Research
& Technology
ERT/ATM
Environmental Research
and Technology, Inc.
ERT/DISCOEP AND
DISCDEP 2*
Environmental Research
and Technology, Inc.
ERT/SURAD
Environmental Research
and Technology, Inc.
ERT/MESOGRID
Environmental Research
and Technology, Inc.
ERT/MESOPLUME
Environmental Research
and Technology, Inc.
Venkatram, Sclre, and
Plelm (1984)
Rao, Lague, and Egan
(1976)
Venkatram and Plelm
(1985)
Yamartlno, Plelm, and
Lung (1983)
Lavery et al. (1978)
Morris, Benkley, and
Bass (1979)
Benkley and Bass
(1979a)
Eulerlan (3-D) grid Short-term
model
Lagranglan (receptor-
oriented)
Lagranglan-stat1s-
tlcal
Short term
Long term
Eulerlan (3-D) grid Short term
model
Eulerlan (3-D) grid Short term
model
Lagranglan (source- Short term
oriented)
The ERT/ESEERCO model 1s a combination of the trajectory phase of the ARL/ATAD model and the
ERT/DISCDEP2 model. Therefore, the ERT/ESEERCO model 1s not listed 1n this table or 1n
Table 3-13.
-------
TABLE 3-12 (continued)
to
I
•—•
CO
Model Name and Developer
References
Model Type
Long- or Short-
Tern Assessment
ERT/MESOPUFF
Environmental Research
and Technology, Inc.
ERT/MESOPUFF-II
Environmental Research
and Technology, Inc.
GDR/LRT
Research Institute for
Chemical Toxicology of
the DDR Science Academy
IAP/LRT
Institute for Atmospheric
Physics, Hungary
KNMI/MAPM
Royal Netherlands
Meteorological Institute,
De Bllt
LANL/ATMOS2
Los Alamos National
Laboratory
Benkley and Bass
(1979b)
EPA (1982)
Lagranglan (source-
oriented)
Sclre et al. (1983) Lagranglan (source-
Sdre and Lurmann (1983) oriented)
Renner et al. (1985)
Szepesl (1978)
Van Dop et al. (1982)
Van Dop and de Haan
(1983)
Davis, Bunker, and
Hutschleeher (1984)
Lagranglan (source-
oriented)
Lagranglan (receptor-
oriented)
Eulerlan (3-D)
grid model
Eulerlan
Short term
Short term
Short term
Short term
Short term
Short term
Continued
-------
TABLE 3-12 (continued)
CA>
f\J
Model Name and Developer
References
Model Type
Long- or Short-
Term Assessment
LLL/ADPIC
Lawrence Llvermore
Laboratory
LLL/LIRAQ
Lawrence Llvermore
Laboratory
MEP/TRANS
Meteorological and
Environmental Planning,
Ltd.
MIT/MITMADE
Massachusetts Institute
of Technology
MIT Model
Massachusetts Institute
of Technology
MSPA/MGM
Minnesota State
Planning Agency
NCAR/RADM
National Center for
Atmospheric Research
Lange (1978).
Lange and Knox (1974)
Lagranglan/Eulerlan
hybrid (3-D)
MacCracken et al. (1978) Eulerlan (3-D)
Duewer, MacCracken, and grid model
Walton (1978)
Uelsman (1980)
Fay et al. (1985)
Fay and Rosenzwelg
(1980)
Ritchie, Bowman and
Burnett (1983)
NCAR (1986)
Lagranglan-stat1st1cal
short term and
long term
Eulerlan (2-D)
analytic
Eulerlan (2-D)
analytic
Lagranglan (source-
oriented)
Eulerlan (3-D) grid
model
Short term
Short term
Two versions:
Long term
Long term
Short term
Short term
Continued
-------
TABLE 3-12 (continued)
Model Name and Developer
References
Model Type
Long- or Short-
Tern Assessnent
CO
I
to
u>
NIAR/LRT
Norwegian Institute for
Air Research, Norway
NIPH/PUFF
National Institute
of Public Health.
The Netherlands
NIPH/GRID
National Institute
of Public Health,
The Netherlands
NHI/PUFFCHM
Norwegian Meteorological
Institute
OME/STAOMOD
Ontario Ministry of the
Environment, Canada
ORNL/AIRDOS
Oak Ridge National
Laboratory
ORNL/PHENIX
Oak Ridge National
El1assen and Saltbones
(1975)
Van Egmond and
Kesseboon (1983a)
Van Egmond and
Kesseboon (1983b)
Ellassen et al. (1982)
Hov, Schjoldager, and
Uathne (1982)
Venketran, Ley, and
Hong (1982)
Ley (1981)
Ellenton, Ley and
Mlsra (1985)
Moore et al. (1979)
Murphy (1981)
Lagranglan (source- Short tern and
oriented) long tern
Lagranglan (source- Short terra
oriented)
Eulerlan (3-D) grid Short tern
model
Lagranglan (source- Short tern
oriented)
Lagrang1an-stat1st1ca1 Long term
Lagranglan
(statistical)
Eulerlan (2-D)
pseudospectral grid
Long tem
Long tern
Continued
-------
TABLE 3-12 (continued)
Model Name and Developer
References
Model Type
Long- or Short-
Term Assessment
CO
I
co
PNL/ANDEP
Pacific Northwest
Laboratory
PNL/LRT
Battelle Pacific
Northwest Laboratory
PNL/MELSAR
Pacific Northwest
Laboratory
PNL/RAPT
Battelle Pacific
Northwest Laboratory
PNL/STRAM
Battelle Pacific
Northwest Laboratory
RI/MTODIS
Rockwell International
SAI/AIRSHED
Systems Applications,
Inc.
SAI/CCADM
Systems Applications, Inc.
Weber, Buckner, and
Weber (1982)
Wendell, Powell, and
Drake (1976)
Allwine and Whlteman
(1985)
Powell et al. (1979)
McNaughton (1980)
Hales, Powell, and
Fox (1977)
Wang, Waldron, and
Bushey (1980)
Wang and Waldron (1980)
Reynolds (1977)
Ames et al. (1978)
SAI (1987)
Lagranglan
(statistical)
Lagranglan (source-
oriented)
Lagranlan (source-
oriented)
Lagranglan (source-
oriented)
Lagranglan (source-
oriented)
Lagranglan (source-
oriented)
Eulerlan (3-D)
grid model
Lagranglan (source-
oriented)
Long term
Short term and
long term
Short term
Long term
Short term
Short term
Short term
Short term
Continued
-------
TABLE 3-12 (continued)
CA>
to
01
Model Name and Developer
References
Model Type
Long- or Short-
Tern Assessment
SAI/RIVAO
Systems Applications,
Inc.
SAI/RTM-II
Systems Applications,
Inc.
SAI/RTM-LT
Systems Applications,
Inc.
SAI/RTM-III
Systems Applications,
Inc.
SCAR/LRTM
Scandinavian Council
for Applied Research
Finland
SRI/ENAMAP-1(1A)
SRI International
SRI/ENAMAP-2(S)(N)
SRI International
Latlmer et al. (1984)
Durran et al. (1979),
L1u, Stewart, and
Henderson (1979)
Stewart et al. (1983a)
Stewart et al. (1983c)
Wojdk et al. (1978)
L1u, Morris, and Killus
(1984)
Stewart, Morris, and
Reynolds (1987)
Nordlund (1975)
Bhumralkar et al.
(1980), Meyerhofer
et al. (1982)
Johnson (1983)
Lagranglan (source-
oriented)
Eulerlan (2-D)
hybrid model
Eulerlan (2-D)
grid model
Eulerlan (2-D)
grid model
Lagranglan
Lagranglan (source-
oriented)
Lagranglan (source-
oriented)
Short term and
long term
Short term
Short term or
long term
Short term
Short term
Long term
Short term
Continued
-------
TABLE 3-12 (continued)
Model Name and Developer
References
Model Type
Long- or Short-
Term Assessment
to
I
co
Ot
SRI/EURMAP-1
SRI International
SRI/EURMAP-2
SRI International
SRL/SPM
Savannah River
Laboratory
TRI/REGMOD
Teknekron Research
Institute
UI/RCDM-3
University of Illinois
UK/STEM1
University of Kentucky
UK/STEM2
University of Kentucky
Johnson et al. (1978)
Mancuso et al. (19.79)
Kern (1975)
Mills and Hlrata (1978)
Nlemann and Pechan
(1981)
Carmlchael and Peters
(1979, 1984a, 1984b)
Dronamraju (1986)
Carmlchael, Peters, and
Kltada (1986)
Carmlchael and Peters
(1984a, 19845)
Shim et al. (1986)
Hong and Carmlchael (1986)
Kltada, Carmlchael, and
Peters (1986)
Cho (1986)
Lagranglan (source-
oriented)
Lagranglan (source-
oriented)
Lagranglan (source-
oriented)
EuleHan (2-0)
pseudospectral
Eulerlan-analytlc
Eulerlan (3-D) grid
model
EuleHan (3-D) grid
model
Long term
Short term
Short term
Short terra
Long term
Short term
Short term
Continued
-------
TABLE 3-12 (concluded)
Model Name and Developer
References
Model Type
Long- or Short-
Terra Assessment
UMACIO Sanson (1980. 1981)
University of Michigan
Trajectory (receptor-
oriented)
Short tern and
long tern
u>
I
to
UM/STOCHACID
University of Michigan
US/RDDM
University of Stockholm
Sweden
USSR/LRT
UWATM-SOX
University of Wisconsin
WU/GIL
Washington University
Snail (1982)
Bolln and Persson
(1975)
Veltlscheva (1979)
Wllkenlng and Ragland
(1980)
Ragand and Wllkenlng
(1981)
Glllanl (1978)
Lagrang1an-stat1st1cal Long tem
Lagranglan-statlstlcal Long tem
Eulerlan (3-0) grid
model
Eulerlan (3-D) grid
nodel
Short tern
Short tern and
long term
Lagranglan (source- Short term
oriented)
-------
TABLE 3-13. Technical attributes of existing regional and mesoscale air quality and deposition models.
CO
co
o
HMtel
ACS/CSC
ACSAMI
AM. /SIMP
ttodal Input
Source Mission rate
toss fund ion
Average Mnd epeed
Dispersion angle
m»nuj depth
Source-receptor distance
AverauB «ind bearing
Avereoe source to recep-
tor bearing
J-0 Mnd field froa
objectively ana ly led
radiosonde data
( 1 to 6-hour interval*)
Diurnal and seasonally
varying Biting depth
Cridded eaieaions
inventory
24-hour total precipi-
tation fields
Hind velocity data
obtained froa radioaonda
ascent at IZ-nr interval*
Hourly precipitation
data used to produce
t-nour totals
Source local ions and
ee4 salon alrengtns
ttidsl Output
acceptor concen-
t rat ion
Dry deposition
Mat deposit ion
Spatially varying
average SOj and SO,
concent rat ion*
Spatially varying
SOj and SO, dry
and tait deposition
and precipitation
PH
terimtal field
af SOj and SO, con-
cent rat ione (at Z a)
for i ant a 1 field of
SOj and 50% total
deposition
Iranaport and fhy*ic*l and
Dispersion Algoritha* Dwaical transferee* Ion
Cbnstant wind rhlf-llfa pollutant
decay
Circular unifons puff
Unlfora pollutant distribution Saasonslly varying capa-
assuaed eith each trajectory bility, but conatant SOf
oiination rat* at 1 Vh
At each receptor, concent ret ions reported
froa } trajectories are everaqed
MOj* MOj react ion
Itorimtal puff diamaian const snt (MO » NOj inatsntaneouo)
as puff travels to each receptor,
picking up aaissions along
trajectory
Minds averaged over lout* t 2 ba Lines', decay conatant far
SOj equivalent t* ).t Vh
Irajectoriea influenced by ore- but user apecified
cipitation are cuefained ta fora
ststisticsl dispersion reletionahip
Vertical distribution derived by
by puff-on-csll aethod
Meaoval Processes
Dry
Deposition velocity
CtMCffpt.
Vd « O/Z cWs, particle*
» 1.* cai/s, S02
Const snt deposition
velocity:
V » 1JI oa/s, SO
* O.I ca/s, SO^
Different psrsseter
values used in dif-
ferent studies
Deposition velocity
concept l
V s I'.O csv%, SO
d Z
V^ « 0.1 ce/s, 50^
set
Huhout
coefficient
X » J , to" II •'', particle*
s 4 « 10"*« *"\ SO
= l/« SO , NO
R : rainfall rats {mm h~')
so, .-so,
reanval rate
linearly depend-
ent on precipi-
tstion rste
Scavenging ratio
is constant
feaovsl of
trajectory aass
upon encountering
precipitation
Continued
-------
IAB.C S-l) (Continued)
U3
Model Ibael Input Mtdel Output
AM./ASIMP Objective uind field Htan paaitian and die-
deteraioetlun Persian ot at let ice
froa t reject arieo
Source lucat tana and
eolaeion etrengths Cridded SO,. SO^
(diurnal l» varying height- concentration and
dependml aaiaaian retea) depaaltion pat tome
Precipitatlan fialdo Flu« af S through
object ively determined grid beundariaa
froa a-hour precipitation
tat ale
AM./M*O Subjectively arobjec- 9wH- ar long-tore
lively analynd ulnd pollutant concent re-
fieldo tiona and depaeitlana
in nap ar tabulated
AveraoB terrain height fora
Sourc*. locat iona and frajectory atatiotice
eoieauna at canal ha
nft/Ml l|>|ier-air tabulated concent ra-
atation ulnda tiana at the racaptor
aitaa
Source local igne and
Mueoion ratee Goncantratian aapa
trajectory aapa
Iranapart and Pftyaical and •anoval froceaeeo
Oiaparaian MoaritbM Owoicel Iranafaroat ion Dry kat
Hariaontal apraad of pollutant SVSO* ('•"•faraatian Steaanal and diurnal SOj, S04 eat
determined froa anioatile atatiatica rot oa very diurnal ly and varying depoait ion dapoail ian varies
af trajectory ondpeinta aaoaenellyl valacttiaai a* aauore root of
•J-).> IA (auaaer) n>q. »d > rainfall rate
Vortical diffusion calculated by 0,1-1. » S/k (uintar)
M-layar 1-O admodol (Cauaaian- - O^i ca/a
Hiaent, Caneervatian oathad) (auaaer)
K( varie. diumally and (uinter)
2 4
Oy < O.M •adiaactiva decay Depnitan velocity Scavenging ratio
concept, »d ousted cunceptt
Cauaaian diet ribut ion aaauard oe t ca/a £«».*» 10
by voluae
a * {IK t)1/Z. Owre
K « > oVa
Alternatively, a uniforo oiilnej
condition can be uaed
Cauaaian or unifara pollutant
diet ribut ion
Modified Cuter advactian Cbnatant decay Qmatant reaoval Cbnatant reaoval
rate rate rate
forvard and backvard t reject arieo
that can branch due ta vertical
vMdaheer
Caueeian hariiental distribution!
o > n.st
Uiifera vertical ailing uithin
Gmtinuvd
-------
IMLC 1-11 (Continued)
CJ
I
Itodel Nidel Input Itodel Output
AM./NIDT Hediaaunde date at upper Concentrations at
air aitaa specific receptor
locations for speci-
Muirly surface date fled ssspling tiaas
Source lucal luna and
eoiaaion strengths
Pssquill-Gifford
at abilities
Precipitat tun rates
(b-rtour internal)
Bnt/AIRSut Spatial 1, and teeporelly DieUioution of SO,
•sr>i»9 Mnds (torn and S04 concantrstion
radioaundea
transport Minds detereuned
by averaging mnda through
the ai>ed layer
Mitinq depthe-diurnally
varying
Source local lone and
e*is«i*M» at renal ha
CAPUA/ Gridded 2-O «ind field Spot to 1 field of SOj.
HCMLO frue. audi'ied aurface SO, concent ral ion
•indobeer.ationa ot field and deposition
4-nuur intervals f is Ids
Cridded eauaaion inven-
tory for quant tend eees
deteroinat tone
Iraneport and Physicsl end
Dispersion Algorithms Chaoical Iranaforoat ion
o . O.H lladioective dacsy
Csueeisn pollutsnt distribution
in horirantsl direction
Midth of pluae (for grid
intercept ion) is I Oy
1-0 diffusion aquation solved uith
finite difference (forasrd in tie*.
cantered in space) schist to detar-
une concentration diet ribut ion
Kz profile depends on stability
Gaussian distribution sseuaed linear SOj/SO, ^
around trajectory andpoint transfereetien
Dif fusion calculated using IX-lsyer Diurnal ly varying
I-O subeudel rate
(0.» tflt •verage)
Diurnal ly varying eddy diffusivitiss
5-hour advert ion tie* step Onetent end unifore,
SOySO4 onidotian rate '
Sjndoo horijontel diaplacoaant ef O.o S/h (vinter)
•t lei (auaulatmg affect a of eddy 1.S* Vh (ouaaer)
diffusion) every 1 hours
Probebillty det mined by
Hiving height eeeaonelly and the oiidation rate
diurnal ly vsriabla
yn ^ un + IMO
f f )
Uni'ora vertical distribution end PAN production
duringdeytiaw. No diaperelon
at night
•eenval Processes
Dry
Dapositlon velocity
concept
Pollutant raouval
froej lows! layer of
1-11 diffusion aodel
Deposition velocity
concept (diurnal ly
varying)
total deposition prob-
ability deteroinad nitti
folloMing dspseition
SOj a 5.» Vh
S04 e 1.» S/h
Dry SO, reeovsl Is
•» of totsl
Dry S04 reeovel is
IK of total
Hit
Scavenging ratio
concept i
E s «.2 » 10
by voluae
Deposition rate
proportional to
rainfall aaount
• : OJ17 P. S02
• i 0.10S P. S04
P : rainfall rate
aa/h
set SOj rasoval
la « of total
Met S04 raaoval
ie 8OX ef total
Continued
-------
IMU •-!» (Continued)
Modal
CCIK/UMrMM
CtGB/SS
ceo/ID
an A ISDN
CII/MMi
NHfel Input
Daily eean «ind sp«id and
direction at at at lone
Daily oliing depthe and
eiabilitiee end precipi-
tation at alatun
local lone
Source local ions and
eniaaion strengths
land apaed and direction
Bnuidary layer height
Cnission rates end
release height
Deposition velocity
Surface pressure, rain-
fall, ximt speed/direc-
tion, aolar radiation.
surface roughneaa
Source aniaaiona
Criuoed eaiaaians aourca
intent ocr
frequency of occurrence
of various Mtnd speeds.
directions, ainng depths.
and diffusivity categariea
Ourat ion of Met and dry
periods
Hourly utnd field
Ikiurl, N0m. INK. CO,
Ml . aiHl let eeussian*
Model Output
ntooepneric concen-
tretlan of SO^, SO^
at selected receptor
SO, loading at
selected receptor
SOj and SM
concent rat ion*
•at depoeUion
Dry deposition
Concent ret ion of SO,
at receptors, Met
Deposition of SO,
long-tar* average dry
and -at depooition
fie Ida and ground-
level concentration
field* far SOj. SO,
•IBCflfeCOsr 0*>t ^^»»
MMO}. MM,, SOj. HIM,
eulfete and nitrate
aerosols
Iraneport end
Dieperaion Algorithms
Concentration unifor* scroea •
eecter eoanating fru* oourcee
Angle of aactar spread ia Of plua
aaoll angle dependent an etability
Sector bo. concent ret ion cstculsted
on dsily basic
Stesdy^tote
Vert ice 1 dieperaien calculated by •
Croon' e functiane apprmi**tian to
eolutiano of the one-dioeneionol
diffusion equetian
Htriamt*! treetomt not oMtianed
aetietlcel eadel
Airborne end deposited SO^, 50^
depend an esigltting feet or*
der ived fro* wind end stixing
depth stabilities aa noil aa
pollutant raaoval rete eetioatee
Physicsl snd «aaov*l 1
Oiasicel Irsnafaraetion Dry
Ms **>lieit SOj * S04 Oepoaitien velocity
traneferaotion treated
». . • 0* c*/e
d,S
•eeoval paraaetera ars
adjusted t* reflect average
SO-/SO level* baaed on
half-life, transport
distances, and tine
Uneer S0? * SO^ *t Oepooitien velocity
Jv.ltr*/e concept
V. . 0.1 cWa
d
Dmetant rat* *f Deposition velocity
SO* eulfete concept (constant)
W of SOj converted
neer the source
Linesr SO/SO^ Dry deposition
tranaferaation
V < O.S co/e
d.SO
10* of SO, converted Z
neer aource
N> out fete reaov*!
0.14 *A> *e*y fro* *ourc*
Rntochanic*! Kinetic None etated
Mfcnaniaa (llusaell at el.
I9S1, 198%)
Ftuceae**
Met
Nsahout coefficient
*s « 4 « 10"* j/e
(J * precipitation
rat*, oo/h)
Sccvenging
coefficient
A « J«1o"*«f SO,
2
Scavenging
coefficient
proportional to
precipitation
rote
Scavenging
coefficient i
AS02 < ASO,
A < 10'Vs
Mm*
layer
Dmtinucd
-------
IMLE l-ll (Continued)
Model
. Mxhl Input
Output
tronaport and
Oiapereian AlgarlthM
Phy*ic*l and
Cneaicel IreMforaat ion
ftOC»«»«»
Dry
Met
CII/lkVN Rwtolttic ret* cunetant* Concentration fialde ttawrtcil eolution of the atao-
•eheric diffusion equation, K-
Iheroucheaicel rate con- theory clo
•tente
HyJfucetbon lueping
piucvdurea
Cbaplai BhotodiMlnl
•rdujniai
Otpocltlon Mloelty
dcpWMtont on •lability,
•ind apaad, Scteidt
and Prandt 1 nurixra,
tonin-Obuldwv length
J-O «ind field, depaa*-
lion velocitiee, ailing
height, relative huiidity
teipefeture.
Iwbulcnt
•ulac iMO
f*co touqhncM, ultra-
violet flux.
CJ
I
Initial end boundiry
condition*
(nieeiuna
CSU-CPA/ Cliaatologicel oean day-
Itodel O t we suing height and
nightt we aouivalent ail-
ing height from felivarth
•ediueunde deta uaed ta
calculate traneport uind
Mthin nocturnal layer
end dett we layer
Average gridded SOj
and SO^ concentration
at aurfaca
Average gridded SO,
and SO^ above night-
time oteblo layer
ta precipitation acav-
ening
Pluaa trajectories oeterained by
conatant velocity aethod
Average pluae uidtlt IN ia BUB of
dieparaion over each tiae aegaant
Ov^. uhich in turn depende on vorti-
cal mnd eheer of * and y coaponant*
af uind
traneport accuro aver t«o layer* *t
night and one loyer during the day
linear
tion
tranoforaa.
k > I.Mt • to*4
e»p (OJK1 M RH)
RH ia relative timidity
Drpnitian velocity Scavenging velocity
defined by
2 ca/e
(day) t > i » 10* for SOj
1 ca/e and SO^
(night) P « precipitation
(day)
: O.I cay*
(night)
fatal SOj and S04 re
by dry and «et pracenee
rate
•eaoval rate de-
fined by
f»
uhere f ie prob-
ability of precipi-
tetion, h i* ailing
depth
Continued
-------
HMt I-U
HMkll
MMtel Output
Iranaport and
Dioperaion Algarlthoa
Physical and
Cneateal Iranafonat ion
al
Dry
avt
ON/MM Grlotted mnd field
CridJed aurfaca raiajiiess
valiiea and topography
Cridded oi.ing height
field
Qmetant ar variable
diffuoivlly lenour
Particle poaltlon
Concent ret ion of SO
or other pel latent
qiven ot selected
Legrangien rendoB-oslk advert Ion and Radioactive decay rote, er
diffusion (deterainietle edvectian, linear SO, » SO, tranafor-
randoa dispersion) not ion rote
Vertical profiles of Kind apead and
ditTustvlty deteroined froo at ability
Deposition velocity
concept
vv"1
*d far other
pollutants are
apectried by uaar
Han* treated
CO
Cloud cover, ceiling
neignt, and other aetoet-
o logical data
DM/1 OK
M Spatially and toaparslly Griddi
' varying mutt field froa and SI
Uiifaro and conatant
oi»tng deptha
id overage SOj Paint aaurces *inatontanaoualy"
^ concent rot iana aimed into coll
Uiiforo) vertical palletont
dietributlaoj
Linear SOj/SO, Oepoaition reaavsl
tranafaraotlon. 0.7 Vh rateai
SOj x ».» «/h
SO, > 1.* Vh
Ha tree! ed
CaoatM MI Mien rain
Peeiataopectral aethsd used to
salve 2-O cooeorvotian eousHona
CO
Conatant and unlfaro
diffuaivities
Ma vertical dtrfueion
CNVAI
o>per air oind. teoptre-
ture and tumidity obeer-
Criddad aataaiona
•ack t reject or tea Iw aaoe as MB/HnD. emcapt diurnal Linear chooicsl Irene. Linear deposition
floM rogioea are accounted far ay faraatian ratos (deposit Ian
l**>uol overage t tuitchinq bet wen a ceovoctive aiied velocity concept)
deposit ion source- layer transport node and a stable
receptor rolotlanahipa isantroplc aode
linear
frVk/MM
Griddad Sdj. SO,. NO, Paint aourcea treated as a
I uind MJ^. o,, HC concentre, aubqrid production of pollutant
fielda and M«ing depth tiono for al«ed layer
fielda (I layore). and invar- tiquintic polynomial advertion
sion cloud Isyar
Surface roughness Variable tia> atep
(Mailed M0|l HC chaaical
kinetics
lly
Suurn liiral luna ami
Verticsl diffusion
Variable K( profile
Dry deposition baaed tfeahout and
on diffusion through rsinout
aurfece layer and coefficient a
adsorption of our face
Specira dependent
Diurnal variablee
Gmtinurd
-------
IMLC J-U (GuntuwBd)
Input
Midel Output
Iranaport end
Oiepereion Mgoritnao
rhyeicol end
Oieaicel Irene format ion
ovel Proceeaea
Dry
Met
(PA/MM Selellito photo* fM
(continue) eatuut ing cloud cover
ttourly precipitation data
CO
I
}-duK«siunal idnde from Gridded
IMM> ' CMC dtnmaic primitive concent rot lone, end
eujuw tun vet and dry deposi-
tion of all epeciee
Surface fluiea of heat,
•ater vapor, and eom-
anl ua from e boundary
Object ively enalyxed, grid-
ded eurfece data of pre-
cipitation type end eeeunt,
eno> co»«r, eoil charec-
tertetice, cloud obeeroe-
Itoii* and vieibility
«w •dvMtion aquation 1* «oly«d
Ming th» Blwtaan cubic •plin*
nlutian technique
Hirimtsl dtrrucioa is MilMd
Ming the Crmk-Nicnalaon inte-
gration i
Vtrticel diffueion ia obtained uaing
either the CrankJUcholaon, or the
higher-order fully implicit Oenk-
Mcholeon echm idien the vertical
diffuaivitiea are Urge
RHC. end S0
Detailed NO^
gae-endequ
iatry (*kinaon and tloyo)
Dry depoaltion of goa-
eoua pollutante ie
baaed on the aua of
the eerodynaaic, sub-
layer, and eurfece
reeiatancee
Dry deposition of
psrticlea uses the BUB
of the turbulent flu*
end the flu> duo to
gravitational Bottling
Speciia varying
Diumel and seeaonelly
varying
Detailed cloud
atcrophysics
using formulation
developed by
Keealer
vertical profilea of teop-
ereture* huiiind epied at 1,000
fl u*»d for conetructing
t reject oriee
Hi«im| dopth informal Ion
aloiMi 1 tajvctory
Oncontrotion of
SO,, S04 along
t reject or lee
hot ant an
Me horlej
M to m 1
eoua vertical mining linear SO^/SO^
conversioni
ntal dtaperaloni 1 S/h (winter)
a to parcel adverted 1 tftt (aimmer)
along trajectories
Linear decoy Not t noted
corresponding tot
*d a 1.0 cm/a. S0{
> O.I cm/a, SO^
nmtlnued
-------
U 1-1}
OJ
I
cn
IfcVMpDfl WW nt^pBICttl WB RtMOvAl ft OCV ••• •
Modal NMkU Input Modal Output Oioparaion Algoritnoa Oiaoical Ironofonaat ion Dry
UI/OISOK* m»r-air Mnd ond MmtMy ond doily Iteao AI40 to ealoulata the lha tronororootlom Oapooition rotoo oro
ond taoparatia-a doto ovorogad dapaoition trojactoriaa ' S0( * ouUotoo atotiatially
DISCDCF {• riu«aa at raeaptaro. xoi » nitrotoo paraaataruad
SiMrca locotrano froa aactt oourco Cauaaian norinntol oro ainulatod on o lono-
grap (imt ond dry) proTiloi torn owroaa atatiatical Hit/dry duration
booio • pariodi utilitad
Ikiirom vartical profila
KM niiing hoiglit voriaa in tlno
in OISOICP. ond 10 canatmt far
oodi trojoctory in OISOXF2
(onuol to noiiouo miing dopth
oncountorod during firat 24-ter
poriod)
C«I/S«M> Ok>ctiM aloaritho to Eriddad OMroaa Point oourcaa diffuoa Ovidotion of SOj » SO4 Oapooitian valacity
d.t.c.10* apatially and (I*-hour) SO, ond -inatantanaoualy* into grid boood on chanieol kinatico coneoptt
taoporally Mrytng umdo, SO^concontrntion oipirioant idth «oriouo "d.sn **r'M
dirfuoi.itioa, dapaoition CfTactlvo otodc hoight datar- HC. NO^ lava la Z
valocitiaa, and oinao layar racaiving oaiaoiana apat tally ond
tranafo»atian rotoo Tanporal ond opatial tooporolly accordlnq
nariantol oJwctiM calculatad «oriotion to atablllty and
Surfaca data uood nth by SHUSH rougnnaao .
offer air data for niung
daptha Kj froTilaa in ourroeo and Oman layar Vd SO * 0<1 *d SO
datanunad by Znd-ardar intaipalatian * z
Smvea lacotiona ond
aniaaion otranglbo Vortlcol dlffuoion oolnod by
laplicit nuaarical aenaaa
• lha OISCOCF and OISOKP-Z andalo oro todmicolly only dioparaion. chonicol tronororaatian, and dry and «o)t aodulaa. In eoabinatian uith tho tranaport pho
Nat
Scovonging ratal
ara atatiat ically
paraaatariiad
Mat/dry durotion
porimfc utiliiad
Not traatad
Contiruad
oo of the Mt/ATM)
al. tha aodalint ayatoa 10 roforrod la by tho PC» aa tha CSECKO nodal.
-------
IM.C I-11 (Continued)
Itodel
Ml/
tcsonio
Kid>l Input
Objective determination
uf «md field*, mining
deplho. end itobltity
Hade I Output
Gridded
concentration*
far SOj end S0t
Ireneaort end
Oieperiiun AlgorithM
Point eaurre emieolono
•inetonteneoiMly' dlffuwd
•ithln grid cell
Physical end
Chemical Ironaformetion
linear SO^SO^
tranefermotioA, 2 SAi
MaMVaol rrUCMeWB
Dry
Oepoaition velocity
concept I
»d . 1 coVb, S01
Mat
No «et depoeition
preproceaoor
Source location* end
emiaaion etronotti*
Effective atech height datarainoe
layer receiving eaiMion*
CoarMWuiey tfethod of Ifaaente*
uaed for odvection
rbrizontol diffueion negligible
cuapared with trenapart
Vertical diffueion acheaet forverd
tiae>, centered differencing
< 0.1 cm/*,
CJ
I
VOrioblo Kf profile
CHI/ Q>j«cll.« determination
NCSOPUMC of «Kt field., uminf
depth*, ond otobility
f rooj piepcocooour
Source locotiono ond
eaiulon etrongtho
Cridded concontrotion O deterained by Tumor oetuoteo s°yS°o tr*M'"rMtion
fioldo for SQj, S04 ml diotonceo < 100 ho io linoor, I Vn
O fro* thrrter foroulo ot
diotoncee > 100 ta
Couoolon diet ribution ooouMd
Oioice of Couooion or Halted •t«ing
(for ooteraining depth of pollutant
dietribution)
Depoaitian velocity
concept i
»d « 1 cm/e. S02
« 0.1 cm/a. SO^
Ovpoeition of
pollutont dcpende on
choice of verticet '
diotribution
Not treated
O froa Tumor curveo
ol dietonceo < 100 ho
Of froa Mfaer fereulo
ot diotencoo > 100 Hi
Omtinued
-------
IMLC J-H (OmtuMod)
Iranaaart and
Itodel ntdel Input Ibdel Output Maparalon Algerlthaa
COI/ Objective determination Orldded cancantration Choice af Couaeian ar liaitod
ntSOTirr of .Md fialda. ailing fielde for SOJt SO^ mi 100. ta
MUM riaa catenated
0 detarainod by tinner (1970)
at diatancee < 100 ta
0 frea Im-fRer (194)) at
dletancee > 100 ta
Gauaaian diatribut ian aaaiaad
Ml/ ttMfl, ourface aetearo- Cancantratien tab lea Surface •inda advoct auffa uithi*
Htsartf r-ll logical data . tha aiied layer, upper level mtnda
Contour plate of con. ueed for puff advert ion above tha
Ifcpor atr re anaonde data caatration fielde boundary layer
Suurce lucatiana and Counian profile in tha vertical
eumun rata and haritantal
S«foco raudinan line-varying aining heigM
OMAOI Mrlicol taaaaratura and SOj and SO* cancan- Vertically varying diffuaivity
•ind euundinot trationa prof tie
Surface character let tea Dry depoait ion Vertical uind profile
Vertically averaged Hat dapaaitian
•cen Mnd opead
Pnyaical and IhanMi Prarenee
Otoaicel Iranafomation Dry act
SOj/SOj tranararaatian Oapoaition velocity Not treated
la linear, Z Mi concept i
e 0.1 cm/.. SO*
Oepooitlon of pollu-
tant depanda on
choice af vertical
diatribut ion
Space- and t tat-varying fcjietenc, aodel Linear SOj and
chemical tranafanetiona (epet telly and SO^ raaaval
and dry and Met dapoaition temporally vari-
able dapaaitian' Scavenging
firat-ordar reaction aachan- rataa) cuefficient
iaa, mith rata conatanta dependa on rain-
paraeeterited in term* of fall rata
environmental canditiona
lt> ta * apaciaa can be
aodeladl SOj. SO^. NO^,
laW,. »0,
Linear SO- to SO. dtd^tonc. approach, fir at arder
(0.1 ta O.OVh) diurnal ly. aaaaonally,
and meteorologically Orvertible for
varying SO-
Irreveraibla
for S0t
Continued
-------
IMU 1-11 (tbnllnuMl)
00
Model Hidel Input Model Output
IJeVlOl Spatially and toaaorally SOj concent rot lone
varying oind field* fro* at epeclfied tlte*
• M ab eurfecai hand for continental,
analysed beck trejecteriee regional, and local
treneport
Gridded eaieaion with
ettasunal variation
tion fielda
Suifaca und velocity
Surface preaoure
Surface teeperoture
l|iper-air «ind
Itoper-air teaperature
Cr idded aoiaeion*
UNL/AlhuS2 1-0 «ind field J-0 concentration
fie Ida
Su>«ce eeuaeion ratee
and releaae height a
Stability claoa
IU/WPIC Irefnirally end epet tally Gridded
varying oind field (nan- cuncantration*
divergent ) frua NMHM
I-O •ind aadel Particle pat it ion*
leaparelly end apot tally
varying diffualvity,
ai«ing depth field*
WicIOO • (local ecele)
Vie treneport equation ia ealved
nuMrically
Paeudaapectrat tcrnaa for hori-
•mtel trenaport
Vertical diffuaion ia treated Kith
e Crenk-Nlcolaon echeae
liae-veryina ailing heighta
(-theory baaed en Oarfey'e (IWt)
tecund order fora
Point eaurcee treated ee *cloul-
af particle*
Mvectlve velocity raabinad «llh •
diffueive velocity U^ far particle
•direction
Diffualve velocity defined by
UD'c-7'
ftiytictl and
tneoical Irene foroat ion
linear SOj/SO,
treneforaetlon
Different half-life of SO,
in hioli and lo> huaiditie*
»o> toe
tp,^ •.* h n h
I *.» h 1* h
1 lac *•' n • h
Linear SO » SO* cnamio-
try
Cbnatant oiidetion rate
Hi If -life decoy
Capability nleta for
redloectlv* dee** only)
particle eat* it reduced
according to half-life
MBOJOV*! PTiic*a*e*
Dry bet
rklf-life of SOj due rklf-life of SO,
to dry depaeitian due to wt depot t-
variee «ith acele of tion variea Kith
intereeti acele of intercept
Tcont * 14J h Tcont * "•* h
Treg • " h Tteg * 0>i h
llee e 20 h tloc < I.I h
for SO^ T > *» h
Ma SO, dry depueition
otaiatance oodel linear decay
(spatially and tea. cunatont propar-
porally variable tional to rote of
depaeitian retee) precipitation
obna None
User-epecified Iher-epecified
•article reauval •c*v*nging rat*
rota at laia>r
boundary
Qmtinued
-------
mac
(continued)
<*»
i
-P.
to
Ironopert ond
Model N«lel Input Model Mont Moperelen U get it hoe
fialde oitn a flrot-order upotreoo differ-
Saurce ooieeian rotea enclng technique
Initial and keuvJory
cimdiliono
Sorfec* «tnde
Inveraian kaaa height e
Salar lodiotion
taffueian carffuianta
Wr/IMHS UijcclMly onalyied Spatially and Geuaaieo plu«o apread
(audified geuatropMc) Hop orally varying
•imao fruo ripper air SOj, SO^ concentre- o frooj turner curvea for abort
aoMdinga tion fieldi diotonceo
•uint ouyrce eoieaian* Spatially and a froa Htfrtor foreulo foe
(eaoaunelly variabla) tooparatly varying large dietoncea
SOj, SO^ depaeition
StMimal 01. ing koigMa flelde Utiforo vortical pollutant dlotrl-
kution
f leldo Murnalty varying oiling depth
MII/NIIOMD SOj •oieeiano for apoc- Mnaol ovorage gridikjd Molytic oalution to SO, and SO, dio-
Ified eaurco ragiana SOj and Sft^ con- poraian oquot ion uaing canetent her-
eout ret ione inmtal dioporaion caefficiente
Anual total out fur
•at ond dry dopuoi-
tion
IHyalcal and Neoovol f
OvMicol Irenefaroetion Dry
Kinetic reaction aochanioa Dtpoaition velocity
for oultlple photachooical concept
apocioa
SMoanal ly and dlumol ly Suaonal ly and
varying linaar SO./SO • diurnal ly varying
tranoforootlan (avaraga dry dapoaitian rota
rote t Vh) nvorege volueat
¥ * tt.n CB/a. SO
NO, » NO, reaction
(linaar) « O.f> cm/*, SO^
(M0» HOj taotantaneaua)
Linear SOj/Sq, treneforoe- Dmotant
lion (U» «/»)
¥d < t.i co/m. SOj
i 3.1 tJt S/h
Omtinued
-------
IMLC MI (Continued)
Nodal Model Input
Nil Mjdel Spatially varying «nd
roee for eech eeiaelon
region
lot el eniaaion* froa
oelected region*
Precipitation at at tat lea
Dif fueion coefficient
baaed un trajectory data
M&fA/NDt Surface Mnd epeed ond
•ind direction
Mixing height
GO
1
*— * Stack characterietica
cn
CD
MUM/a.MM Surface end upper-air
aeteorolagicel data
Satellite data
Aircraft data
Modal Output
long-tent average
gridded concent ret lone
and depoeilion
2«-*-ur SO,. S0%.
and eat el (Ni. Pb, Fe)
concent rat lone, Ml
deposition, ond dry
depoaition at receptor
locationa
tonjontel >and coa-
ponenta
Verticel aotion
leepereture
Irenepart and
Oiopemon Ugorithae
Analytic aolution ta SO, and SO,
diaperaion equation uaing conetent
herlnntal diaparaian coefficient
Linear treneforaetion and dry/vat
deposition allow aupariapoaition
of concentration field* and dopo-
aition field* free each eource
region
Uee vind at ejidpaint
between eaurce end
receptor
Ceueeien diffuoioni
tl X O.S Cll 1 H
„ •••* yij
a « O.M n0'**
Minde and eddy diffueivitiee are der-
ived froa dvnaaic x»>ti*« eouotion
•adel
Mvectian and diffueion dona uaing
upetrea* finite differencing aathod
•hyaical and
Cneeiical Iramforeat ion
Linear SOySO^
tranaroraation, 2 Vti
ria«-avaragad in I fora
eonvaraion
VK
O.t to 1.0 X/h baaed on
diatanca froa enure*
Detailed MX. HC. and S0<
choaical kmetica
(Mkineon and Lloyd)
•eaoval Proceeeeo
Dry
Conetant and unifora
raaoval retee correa-
pond to depoeilion
velocitiee given
beloM. «eth eaaueed
1 Ha aiiilng depth
d ' 2
» O.I em/*, SO
Oapoaition velocity
concept
«d « 0.2 - O.I cm/*, SO
> O.I cm/*, SO^
Dry depoaition of pal-
lutanta ta bued on
the aua of the aaro-
dynaaic, eub lever,
and aurfece reeia-
taneaa
Wet
Cone tent and
unifora reeovel
rale depends on
regional avarege
Mahout coeffi-
cient and
precipitation rate
Scavenging
coefficient
-.,«O.°-"AH
for SO
• 1000 RAH
for 50-
OHailed cloud
•icrophyaica mth
aqueou* aquilibriu
for geee* end
me loot ion for
particle epeeia*
SO,. NO,, end RIC
point and oriddod area
eource oaiaaiona data
Miter vapor conetant
Sp9Citffl OBpMtteMK.
Diurnally and eee-
aanelly verying
Initial and boundary caw- frecipitetion
diltone for ell epeciaa
Cridded %-dieeneionel
concontrot ion ond dep-
oeitionef ell epeciee
Continued
-------
IMtC J-ll (Continued)
Iranapart and
Diapereien AlaarilhM
>hy*ic*l end
1 Iranefi
ROM**! froceeeee
im
Dry
Met
ftlM/Uf Spetielly end looperelly
veryieg Mod field* froe
•M ok oerf ece
Conetent oo>**iena
froo grid coll* MlliM
oodeling region
tolly e»m SOj. S^
concent rot wno et
•oocirtod location*
He KertreMal dieeercien
Iktiforoj dtotrtbutian
Hlil««d*oth**t looted froe
cooper lean*
LiMer
Irene reraet ton (rote
eotuotod free prediction
tocey rale concept
far SOj (rote
collected froe
prediction
vo. ooe
Cone tent eel
depoeUion ret*
unifarei ailing fteidM
conatant m 1 MM
•IrWrWf Surfeco «nd velocity
Standard depletion *f
Crtdda
tratian Couooion foro»datien
field*
0 (1) < o (•) »
— „ •
1 *i 1 VT) ** *
1 loyerai oe«fOLe, o
roaormur
riret-erder SO, » Mt
tranafaraotian
•niatanca aadol
(apatlolly and
teaporally varying
dapaaitian)
I
»-•
en
DllWaiO Surfm •net
Cr Hided i
tratian
field*
•wfoc* •ind direct ian
Mind Mlecttf and
direction itewiction
•I IM ->»*.
DM 2-diaeneionel continuity oota^ion
•o integrated ooporotery far eecn
leyer, ueing o poavlaapoctrol echooo
Comtoid. horuontel tyrnulont diffu-
eien t.
SO » SO firet-ord*r
react ion, dependent an
oiler radiation
•ealatence nodal
(apatially and
tonporolly verioble)
Hlll
Source MtMUM ret**
wu/nrfcm
tecopter c
trait**;
•ind direction
Couoeion faraulaluna
0Z(0)
Photadiaaical Oteeicel
1 inetic HMtunia*
laniataneo approach
(epetiolly end
dl molly varying)
None
~1 f "
1
-------
IMLC I-1) (Cblll IIMKMl)
»d.l
OME/
SIMMO
ORM/AIUMS
OML/
PHCMIX
PM./MUCP
PMAII
Mxfal Input M>del Output
Mr*! Hintl velocity and Spatially varying
diapereiim at et let Ice at average SO. end SO.
each aource caneant rat iona
Onatanl •iied-layer depth Spatially varying
total dry and wet
Aunt oource aaiaaiona SO. and SO.
depoaition
legrengian dry and ««t
tutu acalea
Source oaiaaion rata Ooenvdnd concon-
t rat iona
Ni>tng depth
Juinl frequency diatri-
bution HUM) raae
fielda of prUMry
Ictciamtal diffuaian and derived apeciea
coefficient l*-9~t SOj and SO,)
uj>uer-air oind data
NMthly aaieoien rata Ooumeind concen-
t rat iona
Joint frequency dietri-
bul IINI Mnd raaa
leanucel ly and apat ial ly Gridoed SOj and S04
vai>ing *iod field* concent rot ion fialde
arfijoctlvely detanMnad
fru» laJtueunde data
Suture l««c«llana and
l«*iwielly «M| apat let ly
• eiyuMi lAinfall ralvo
Iranaport and
Uaparaion Algorithee
Utlfora pollutant dlatrlbutian
beneath conetent Invereion baaa
height
Cauaaian diet r lout ion aodified by
reaoval function
OK ando are linear functiona af
Statiatical aind raae
Brigoe (1*7)) diepereion
Qmtinuity equation eolved by a
tfarlrantel turbulent diffueion
t^ ia tioe -dependent
«H(t) * •* I k^ft) eT
aatiatical «ind raae
Sector averaged
Pollutant diet rlbut Ian aaaueed
ta ba Ceuaatan
O fro* paver lav fit of the
Paaojuill-Clfferd-Kiroer curvea
Hariantal diaperaion optional
o frooi eapirlcal faraulae
dependent on (lability
Phyaical and •eeovel Pruceeeea
OMoilcel tranaforaatian Dry Met
Oanatant oildetion rata Cunetent depoaition Conatant SOj
of IS^tr velocity! reaoval rate,
V < 0.1 cm/*, SO X < 11 Vh
In^laud aildetion
enhanced by increaainq * 0.0} cm/*, SO Dmetant SO
uet SO. reaoval roauval rate,
X « >a «/h
tvlf-life pollutant decay Nona None
Linear SOj » SO* Oepnaltion velocity' Omatant acaveng-
tranaforaation concept ing rata
Linear decay far radio-
act iva apeciea
rklf-lire pollutent decay Mma Mma
Linear Mj/SO, aaiaaiona Otpoaitian velocity Sravanging rata
rate ia uaer epecified concept i deterained by
»d • 1 ca/e, SO collection
efficiency, aren
* O.I cm/e, SO rein drop Hiawter,
and rainfall rate
OrpnaiUon of aaaa
drponde on qmund.
level concent ret ion
Cbntinued
-------
IMLC 1-11 (GentUMM
Ibdel Detent
Tronaport and
Oiapereioft AlgaritlM
•nyalcal «n4
Oiaairel Iraaef
Dry
PNL/MLSM farrow
eaiealen rate*
table* *f concentre-
tieno et receptara
Iva-etep iterative precedJie'fa
puff traneport lech tlneatep
Norn tnttcd
(umttr
Wen and upper air
aeteoialagical dote
Vertical
Hli
*MlM
lly
Hindi very vertically, tart-
mtclly. ami
fnV/ajo| Objective wind field
detetoMed fren 100-
IDOn • average ainde
Crida^SO, aod S*^
tratla* field
ttortontal diaaanlan
«j>d varlatten
Linear SO-/S*.
canveraian
U>
j and Sb
dry and mt
dipiiittaa rtalda
Source lacetten* and
aaiaeian etrengtne
teoporolly end epatlolly
varying reinfel I ratee
tration of each
eir parcel ia mi far*
Omcontretien diotrtbutian farter
determined fro* Ca«*oian aaauiptian
•r unifar* eeeiaytlan
•epartad nee of I */h far
deytine rate canatant,
OJ> «/» for nltfrttlne
0( ia • rwctten af atafctlity idtk
diurnal ly varying aixad layer
Oepeeitien velocity *a*oval rate. «.
concept i praportianal to
»d « t cm/*, $0^ precipitation rate
• 0.1 cmA, SO^ • « AP
Depeaitian aeee A > V*. SO^
depend* an vertical
concent rat Ian farter A « 2XV*. SO^
(detarained froa
Coueeion ar uiifor*
diatributian)
MaperaiM above and •ttnin *i«ed
layer independently deteminad
rnysiMN Oijective Kind fields SOj concent rat lane
d>ler*ined fro* MBper for aalected
eir date receptora
tffective eteck heigHte. SOj and S0%
atebilitue. and aim ing depoeition for
depth* at rogular tlao aalected receptor*
inter vote
All input i* apat tally
"""""*
«y andOj determined by twner
eetiantian at dietancee < IOO be
0 andOt froo tkfftar foroyle
at dietancee > 100 to
Ceueeian diatributian laeueed
Onlca of Ceuoeien *r limited
al. ing depth (unifora pollutant)
diotributioo)
SOj/SO^ trenefornatlen Cepaeition velocity
rote* different for four concept i
eiluotione invelving time *d ' ' cm/m» M2
or doy. relative hueidity.
and hydrocarbon levele < O.I ce/a. SO
(range « 1-« Vh)
Orpoalt ion of meao
deponde on choice of
vert ice 1 pollutant
dietribut ion
Maahout coefficient
coefficient concept
*so •e-00**'
A e 0.01 A
S01 «
(antinued
-------
IMLC MI (Continued)
Modal Input
fed* I Output
Iranapart and
Diaperaion AlgorithM
Fhyaical and
Oteaical franeforaetion
•1
Dry
Wbt
•.I/NIDD1S Hourly aurfaca data and
12-hour upper Mind data
Object iva routine
genarataa hourly
atebitity, aiixing
haighta, end Hindi
24-hour and running Hourly updatad Hindi at affectiva
3-hour average
concentration at
uear-epectfied grid
pointa
atack haitftt (detarainad by po«r
law) uead to edvact pluaa aegaent
Ceueaian puff dietribution with
aultiple reflection*
rallutant hair-lifa
depletion (i.e., linear
il rata)
a < O.X
• e (2 K t)
Depoeition velocity Neehout ratio
applied to Mixed-layer dependent on
avarag* or naar-ground precipitation
concentration deter- ret* uaing
ainad by Gauaaian Scott'a (1978)
foraula pereaatrintioni
cloud faeae
height a u**d
1/l
Eatant of puff ia *»y fro* canter
to
I
SA1/AMSHCO lerrain heioHta
Source eaiaaian rataa
Griddad Mind field
feaperatura
Nixing heidjita
Saler radiation
Criddad concentretion Conoarvation equatione anlvad by
fieldi the SHASTA integration achaaa
Concent ret ione at
racaptora
Multiple photodHMBieal
apaeiaa, carbon-bond
reaction aechaniaa
Mraiatanca aodal
None
Spatially and taaporelly varying
•ixing heid4a
Multiple vertical layara, *z profilee
Initial and boundary
condition!
SAI/CCMM laaporally varying gridded Short-tana pal lutant
mnd ftelde concent ration* and
dapoaitiona along
Surfaca typaa apacifiad trajectory
Precipitation rataa
S0m. M0m. and
SHC eweeion rataa
Cloud typee, aaomta,
and aiwironaant/
-------
IMC »-IJ (QMlMUad)
CO
I
01
in
Tranaaort and
Nodal Madel Input IMal Output Otaperelon Mgeritha*
SAJ/CCMN SU%. •*,, and ONC
continued tone ant rot ion*
trajectory location
SAI/IIMO Surface uindo I*-hour aa>laa and Ti|intod pluae *ppi«ocl>
Miinrnl and annual
U/par «lr leind* average concentre- Cbncaat rattan* unifara uithin
lion* of SOj. NO, lirti iiiaanl
Stability NOj. MJ( SO,,
light arattering Vie vertical and heriBMtal
laeperaturo coefficlente, vi*u*l (cro**-«dnd) dinenelane of **ch
range | vat and dry eegoant are ».*> and 4. B
MuBtdlty dapoaitian rate*
Precipitation nitroaan O . nin O , V|-\ , (Ct)*%
*\ 0'
Sautca ea^aaian rate*
C « U m 10* «*/•
0 .0 (i-f .vft.r.1,
dapandi on atabltity
S*I/*IH-II objectively or Griddad average SOj, Ceuetian pliaw (kivariate noraal)
aubjcctively analynd SO, ceacontrotion aubaudal for grid initialiration
•wd field* and dry/vat
dapaaition field SN*SI« ichni far numerical
Otjactively or integration
aiaijactively determnad kgian*! SOj, SO,
ai.ina dapth field bufcpHo Z l/l-l*yer aadal (aiud
layer, invaroion layer, and
Sulace raualrn*** field naatad aurfaca layer)
Saatially varying hourly ttorlmntal diffuaivity varlee "ilh
precipitation roteo flMderoraatian
Spatially and temporally
.aryimi « Manure claa*
•wurce Inratiana and
-.,«»,.., .lr«qjha
Phyalcal and ftaminl Pracoaaaa
Oiaaical lran*fanat ion Dry **t
•ta chaaical aodula involve* Depaaitien velocity SO, dtaoaitian
raactiana aaang NO, MO^, aaatiatly and
MO,, 0,, Oj, 5",, SO,. Vegetatian^epandent taaporally varying
OH , MjO, and OCD) uaing Seott'a
(1«78) parawttri-
ationt vaahout
and rainout froa
three cloud typo*
SOj dopoaition
apatially and tea-
parally varying
uaing equilibria
paranateriiat ion
of Hale* and
Sitter
-------
IMLC 1-11 (Gmtinued)
Mxfel ' Model Input
UI/RIM-il Saee •• S«I/*IH-II
e«cept Z»-hr precipi-
tation ralcft «ey be uead
in place of hourly rates
Nu e>pasure claaa
required
SAI/niH-lll Objectively or
eu>jectively analyied
mod f lelde and ailing
iMloht fields
Surf ac* roufmesa
(•posurs class spatially
Hidsl Output
Seae as SAI/aiH-ll
Cridded SO,, S«^, NO,
NO,, 0,, HC concm.
t rat tons for ai»«d
layer and inversion/
cloud laj«r
Transport and
Oispsrsion MoorithM
SaM as SJkl/RIH-ll oicoot
Gaussian puff amtal ia not
uaod for grid initlslltatUn
Cauasian puff autaodal to
trsat aifcgrid dlaparsion
naar the sourcs
fuffs ars trsstad in •
nonrsact i«o asrawr
SWSIA advsction scnaa*
Physical and Rasoval •riKssasa
Chaaicsl Ir snaforsat ton Dry
Saas ss SAI/«IH-1I Saaa as SAI/RIH-1 1
Oatailad MO,, HC, and Dry dapoait ion baaad
SO< chaaical Mnatica on dtrfuoion through
(Carbon-Bund Ibchaniai) surface layar and
absorption at aurfaca
Spacivs osoandant
Diurnal variability
Net
Sulfata scavenging
siapllfltd to
scavenging coef-
ficient (linear
a»thod)
SOj scsvenging
saas as RIM-1 1
Haahout and
rain-out
cuefftcisnt
for SO. only
(aarae aethod as
RTM-II)
en
and teeporally variable
Suurca lucationa end
eojiaaion atrengtha
Satellite photoa for
estiaating cloud cover
precipitation
JJwur tsaporsl reeolution
Variable tiae step
Vertical diffueion
Variable K, profile
dale
SCMAIIH Spatially and teeparally
tarying uind fields free)
8>0 all aurfaca
Hi.tng depth estUHtea
Sink rate*
sslane
icaluulas over region
ul interest
Distribution, of SOj Unifora diatributian canfined by
concentrationa •iming depth
Cella traneported aver eaissions
are dsfonasd by tdiid.riald
•tffactive* diffusion coefficient
ueed (aannthlng of concentration
field)
•tot explicitly •
accounted fa* (ganerel
•ink coefflclant ueed)
Mit •uplieitly
accounted for
(general sink
coefficient mad)
Wit explicitly
accounted for
(general sink
coefficient used)
Continued
-------
IAB.C »-!» (Gontinjodt
Ul
I >
cn
rfedal
S*l/
DiMMP-i
(i«)
S«l/
CNAMP-2
(S) (M)
SHI/
CIMWP-1
SRI/
[MMP-2
tbdel wtput hadel Output
•ind rtaldi derived rroa tktddad or contourod
•Ml a* eurraro SOj. SO, cone an -
t ration and o»ai
O»ilr total rainfall ait ion pattern*
lac Mlocted atteo
(I* uM* >-kjr rauirall Pollutant tranarer
tale*) oatrU
SiMirca location and
aotaaiun etrengtha
Seaeuwlly varying
•lilito, death*
Sane input aa OMMnP-U Sana aa OMMP-M utth
additional nitrogan
SjMt tally end teaparally Grideed SOj, SO,
•aryuuj utnd riald rroa concentration and
adjuoted no ob uindo depooltten petterne
O>il, total rainfall Pollutant tranarer
for e.lec.ed atta. ortrl.
Dmatant oi«ing depth
•f urn.
Source location* and
cat Mian atrengtha
PoMir leu fit to current Griddud SOj, SO,
•id 1)0 aJb uinde to deter- concentration and
•ino tioiaport uind* In di* mitten
tuu la>u(ai IhMO uinda
are uiignted by vertical Pollutant t rant far
inillulenl dietrthuttan to oatrU
ulitaiii net I ranapart uind
Iranapart and
Maperaion Manrithaa
Initial aiie af circular puff ia
eouivalont to grid ai»
•adiua or purr euaande ao
2 l"2
ttiifaro concentretien utthin puff
•btetentonMua- vertical oixing,
t.*.. uniforo concent rot ion
dietribution
Seoe aa CHMUP-U uith 2 oddittmol
layara
Vartical dirTueian uaing I theory
and Q-enk-Micelaan echeoe
•Inotontoneeu*' vertical oiling,
i.e., miraro cancantratten
dietributien
Initial a tie or circular pufr ia
oguivalant to grid alia
Ihdiua or purr eujwnde aa
1 t -I172
' « l»o * "*'
Unifero concent rat Ian uithin purr
Dirrualan depend* on deferent ian
or uind field
lloa-dvpandmt vortical oiaing
Phyalcal and •aaevel Pr
Oiaoicel Iranaroroatian Dry
Linear SOj/Se, Dry depoaltlan
tranefereetlen, raefflctanta vary'
1 Vh (1* ia oadtriod to uith eaeeon only
penit eeeoMlly and (W paruita verie-
dtumally varying rate*) tione uith vegeta-
tian. atehllity, end
tio. or doy)
(S) or (N) rerer to linear Sane aa CNMMF-M uith
onlfur chaoiotry (SOj * SO,) addition or nitraoan
•1* laWaMiT H*tl*foiJMI CntWlVoKpy •patClVfl •WpoWlClvMI
(NOj » hflj) velocitioa
Linear SOysO, Dry depaaltion
traneforaatian, n/hr reaoval rote
conetant uith
tiaei
SOj, 2.* Vh
SO,. 0.7 Vh
Linear SOj/SO, . Dry dapoaitian
trenaforoetian depende on purr
hrioht above eurface
oceeeea
•M
a»t depaaitian
coefficient*
conetant uhen
precipitetton
aceura at puff
locatiant
SOj. 21.4 Vh
SO,, 7 Vh
M allow* depaoitian
to he a function or
rainfall rote and
elaud preceea typea
Saw aa CNMWr-1*
uith additional nit-
par aoetera
ait depaaitian
coefficient*
conatant unm
precipitation
occur* at purr
location!
Sflj, 21.4 Vh
SO,, 7 Vh
Qmatant uct
depeaition
co»rfieiente
if rain occur at
SOj, 21.4 Vh
SO,, 7 Vh
Continued
-------
IM.C I-1) (Continued)
•I Output
Otaparaian MgorlthM
Physical Md
Iranafaraatian
Procoaaaa
Dry
Hit
SRI/ Niiing haig>t
CUWMP-7 radioaand* and aurfaco
(canlinuail) dato (diurnal ly varying)
CJ
I
»—•
in
CO
Saurca lacatum and
uiuian rat* intonation
fat nartabla ouaaiano
aatiaata*
SM./STH
IRI-RCGHQO
UI/M3M-*
GruMad «nd fialda
Cnieaiona location and
strength
Mind velocities at «00 •
•are uaad (darivad froa
radiosonde data)
Constant inifora aiiung
height
Sourca lacationa and
emission atrangtha
Saurca eat so u» rataa
ninthly and seasonal
resultant «nda and
paraietance factor*
Cancantration (norasl Cauaaian disptrslon Radioactive decay rata
level) at eelected
y *
Unifora vortical distribution
after O, > O.W
Criddad svaraga Uiifora vertical pillutant Linear SOj/SO,
SOj and SO, distribution ' tranaforaation, 1 Vh
concent rat uma
Peaudaapactral aathad used ta
aalva 2-O canaarvation aquBtian
Spatially varying Adept et ion of HIT aodel SOj » SO, eanatant, 1 Vh
avaraga concent re-
tuna and dry and Cbnetent hariaMtal diepireum rata
•at depoaitiona of
aulfur (kiifera vartical aiting in boundary
Mat traatad (tot traatad
Dapaattion velocity Not traatad
concepts
»d a 1 ca/«, S02
> 0.1 c«/a, SO,
Oapaaition velocity Dapaaition rate
•eightad by percent (Vh) ia given by
of dry tiaa
• n i w a i» r dry "*'
Iran rawinaanda atatiana
layar
•t
Spatially averaged
eeeeonal dry and «et
period duration for
••th etote
Saaaonally varying •lung haio>t
> 0.0), SO
•hare
• « total precipitation
I f • • r •(!•
'dry' 'MI ' •>*r«o»
duration af dry and net
ptriode
Continued
-------
IMC »-1» tContiiuad)
ltod.1 Output
tranopart and
Diaparaian UaotlthM
Phyaieal and
OMnical Iranafamotion
al
Dry
aat
UK/SIM! Surface end upper-air
uind obMrvat tone
Cloud cover, ceiltng,
ourface raugfcneeo,
evauMot ion roteat ettaiaoj
height, huaidity, end
taooareturo or of I to
Criddad eniMieno of
sal' •*.. CT
CriddKd SOj.
rtolda
Crank-MidwUon Calaiiiin nuMflcal
IranapMt and diaparalan aathad
SOj » SO, rat* OaaoaUion valacity Not apacifiad
an ealeulatad ION] concept
and (MOj) froa
*d diractly
proportional to
u.i alao dapanda
on ovoporotian rata
and at ability
phil organic ol aoo*l|
canatant natarooMMKuo
rota apMifiad a* Mil
uc/sicN2
Kdiaaneionel ortddatf Mind
vectere. toaperetvree,
reletive huatdlty, cloud
typaa and aaaunto free
objective analyato or
trationa of all
•Meiaai df7 and Hit
dapooitian aaiinil a «f
all apaciaa
Iranaoort aquationa a>l*ad Ming •
Crant-Nlcholaon Calalktn flnlta
alaant nathod
CJ
I
cn
to
Oatoiled SO,, NO^t end MC Dry dapMition of pal-. Choice of detailed
aee-phoaa cheatcaI kinetic lulente ie booed an cloud •ierephyaice
aacnaniM (oaltn end Patera) the eua of the Mro- (AdOMuyi) or pere-
dynMic. boundary notarized i
Detailed clout «tor oxide- layer, end aurface
tion react iona roeieteneee
IMS ACID
It
fro. M.
•xtol (mfUr. I9M)
••cvplar lacction
Mont If •c«cipit«tian
Sourc* loot ion and .
Probability fiald af
pot *ntial Mlfat*
(•r S«j) loadlnq at
Mlactad racaplar
lawn tranoeart dua ta input** Ira-
>ctoriM (i^tour intanal)
Oiaporaian aatuwtM (an, ay) aaaut
an aonraaa diaotaeaaant aktainad
fran diatribution of tra>ctory
ano>ointa in boundary layer
Saaaonally varying ntunf naio>t
Oiumally and Manorial ty
vorying tranafamation
rata (Vh)i
Spatially and
tanjmrally varying
oapaaition valacity
•««-»>
ain
»8.»
lat-ordar rata
for SOj
K « O.OOS P(t) « 10 /M
-.so2
K a 0.2)2 P(t)B'*2* x to'/M
"•S0«
uhere H » nixing height (o)
P « no. dayliofM. houra
IfV
SIOMACIO trajectorioe end precipi-
tation enounte along too
Hacaptoc locationa
Sourca location* and
*«i»aioM alreno^lw
SOj ami S«4 cancantr*-
tiona, aulfur and aul-
fat* uat and dry dapa-
altiona at aalaetad
racaptora. lha frac-
tion of SOj aaiaaiona
eontributin} ta aabtant
SOj and SO, and «at and
dry daaoaitiona at aach
racaptar, a* a function
of upxind traxl tina
M IK/ACID
o* IM/ACIO
ao IM/MIO
Son* *a IK/ACID
Gontim.wf
-------
I Oat I-U I Concluded)
CJ
I
cr>
CD
Nodal
IM/
SfOOUCIO
(continued)
US/IODH
U5SI/IBI
UNAIM-SOX
MWCIl
Nadel Input
Spatially and temporally
varying wind field* froe
ISO eb winds or Minds el
other leva la
Cat iMt ion of durel ion
end frequency of dry end
•el panada
Sourca locetiona end
e*ii»ion atranjtho of
total sulfur
Polynomial and linear
interpolation of surface
end «M et> <*nd* (>-O)
from rudioeonJo dote
Gridded SO- eeieeion
Hourly gruMed
•eleorutogicel date
(Netionel Clleatic
Canter dale) ueed to
coapute hourly inifora)
uind epaed and direction
Point, line, area source
emtaaion*
Hind ap»ed
HUH! direct lun
Model Output
Source-receptor trene-
fer eatricae for SO-,
S04, end eulfur «ajt
deposit ten
Distribution of dry,
end «at anthropogenic
eulfur
Distribution of sulfur
concent ret ion
temporally end
spatially varying
so,, so.
Crowd-level concen-
l rot ion, dry. «al,
end total deposition
of SO, end eulfete
Variable avereging
tiasa
Sulfur bufcpte
Precipltetlon pH
estsMle
SO, end SO, aablent
concent rat lone
Dry dopoattien
transport end Ptiysieel end
Dispersion AlgorilhM Cnaaicel Irensforeet ion
Dispersion statistic* baaed en Mane aantioned
t reject ory endpolnta
tBiaaion sisiemt to enter el
elevet ion of li e
Vertical cencentret ion profile
obtained by analytic solution
of vertical diffusion equation
ties-eplitting nuaarical lolutian Linear SO, oxidation
eathod (eecond-erdar in liaa
end specs)
Gridded field of eulfur deposition
Fint-ordar fully iaplicit linear noaoopneoua SO, »
nuaaricel *rhaa* SO^ traneforaationi
"dey m * ""
Plues riee celculetiene for eejer "niohl * °"4 *"*
paint eourceo
Boundary layer cherectorind by
•ind end dlffuelvily profile a.
dependent on •icraeeteorelogicel
variable*
K-theory Overell conversion rslst
r « - O.I - 1.0 ce./.
laafieratur*
-------
An Important point to realize when discussing parameterization methods,
technical attributes, and limitations of models 1s that process treatment
within models 1s based on a "sub-model" algorithm for the particular pro-
cess and Its associated parameters. Models can differ 1n their conceptual
approach to a specific process (which determines the form of the
algorithm), or can exhibit Identical approaches but utilize different
parameters. A simple example of this 1s the dry deposition parameteriza-
tion. Two common approaches are the deposition velocity and the resist-
ance concepts. Several models may treat the process via the deposition
velocity concept yet differ 1n the values of the parameter (Vd) or 1n the
variability allowed. Whereas the algorithm and Us parameter's varia-
bility are usually "hard-wired" Into the model, the specific parameter
values are usually easily altered. Thus, there 1s no meaningful differ-
ence between a model 1n which constant Vd « 1.5 cm/s and one 1n which Vd -
0.5 cm/s. The models will certainly produce different results (all other
treatment methods being equal), but the difference cannot be ascribed to
different model attributes or limitations. With this comment 1n mind, the
following subsections discuss similarities and differences 1n various
model requirements and components.
3.3.2.1 Model Input Requirements
Input requirements for regional and mesoscale air quality models exhibit
a considerable range of variability, but essentially consist of the fol-
lowing components:
Transport wind specification
Mixing depth specification (or vertical diffusion profile)
Precipitation data
Emission data
Because of the sophisticated methods of process parameterization 1n some
models, temperature, humidity, solar radiation, stability class, terrain
height, and other ancillary data are often required as well. Parameter
values, such as chemical transformation rates and deposition velocities,
are usually hard-wired Into the computer codes, and thus are not con-
sidered 1n the strictest sense as model Input. In some cases, however,
these parameters are specified as Input on a monthly basis.
Transport wind specification depends primarily on the type of model, but
a surprising degree of variability exists among models of the the same
type. Models that use temporally varying transport winds generally derive
wind estimates from 12-hour (or sometimes 6-hour) upper-air data. Some of
the Eulerlan and Lagranglan models requiring two-dimensional winds utilize
3-161
-------
wind data averaged over a representative layer of fixed depth (e.g., ANL/
STRAP, ANL/ASTRAP, BNL/AIRSOX, CCIW/LDAPTM, ERT/ATM, PNL/RAPT, PNL/STRAM,
UMACID), whereas others utilize a time- and space-varying transport layer,
(e.g., ARL/MTDF, ARL/ATAD, ERT/ESEERCO model). Transport winds may be
computed from observed winds at a fixed elevation (e.g., TRI/REGMOD) or a
variable elevation (e.g., UWATM-SOX). Others utilize the relatively dense
surface-wind measurement network with modifications to account for frlc-
tlonal effects (e.g., CAPITA/MCARLO) or a combination of surface and
upper-air measurements (e.g., ERT/SURAO, ERT/MESOPUFF-II, RI/MTOOIS, SRI/
ENAMAP-1).
Still other models use winds observed on a constant-pressure surface, such
as 850 mb (e.g., DMI/LROPM, EURMAP-1), or calculate geostrophic winds on a
constant-pressure surface (e.g., IAP/LRT, NIAR/LRT, SCAR/LRTM). Modified
surface geostrophlc winds have been used by MEP/TRANS, whereas a combina-
tion of surface geostrophlc and upper-level geostrophlc winds have been
chosen as representative transport winds for US/ROM.
The different methods used to define transport winds suggest that no one
method 1s clearly superior to another, as confirmed 1n studies by Pack
et al. (1978) and Hoecker (1977).
For multiple-layer models, the methods used to specify winds are many and
varied. The SAI/RTM-II, RTM-LT, RTM-IINL, and RTM-III have utilized
radiosonde data to define average wind velocity below and above the local
mixing depth. Convergence and divergence of the wind fields results 1n
vertical motion between layers. Other multiple-layer grid models (e.g.,
EPA/ROM, USSR/LRT, NCAR/RADM, ERT/ADOM) use fully three-dimensional winds
derived by various objective processes discussed briefly 1n Section 2 of
this report and more comprehensively 1n NCAR (1983, Chapter IV). Two of
the models surveyed (NCAR/RADM and ERT/ADOM) require three-dimensional
wind fields and other meteorological parameters obtained as output from a
dynamic meteorological model; four other models (UK/STEM2, SAI/RTM-III,
SAI/RTM-IINL, and SAI/RIVAD) have also been run using a dynamic model out-
put. The Lagranglan puff model ERT/MESOPUFF-II allows two layer-averaged
transport winds to be defined as an option, although the puffs are advec-
ted by only one of these layer-averaged winds at a time.
For those long-term models that simulate advectlon 1h a statistical sense,
wind Input generally takes the form of average cl1matolog1cal velocities
(e.g., MIT Model, UI/RCDM-III) or wind and trajectory roses (e.g., CERL/
LTSDM). Use of mean climatologlcal data does not necessarily overcome the
problem of uncertainty 1n transport wind definitions because the climato-
loglcal statistics are derived from the same geostrophlc winds or wind
measurements used by nonstatlstlcal models and hence are subject to the
same uncertainties.
3-162
-------
Specification of a representative mixing-layer height within which pollu-
tants eventually become uniformly distributed depends on the particular
type of model. Long-term models generally make use of mean cl1matolog1cal
mixing depths, such as those given by Holzworth (1972), which vary with
season and geographical location (e.g., UI/RCDM-3, CCIW/LDAPTM, SRI/
ENAMAP-1). 'Other long-term models (and some episodic models) are not
designed to accommodate variable mixing depths, and therefore only define
a region top.* those models that simulate vertical pollutant distribution
using a one-dimensional diffusion equation often define mixing-layer
depths Implicitly through the selection of a vertical d1ffus1v1ty profile,
which generally:varies spatially and temporally (e.g., ANL/STRAP, ANL/
ASTRAP, BNL/AIRSOX, ARL/MTDF).
Finally, a large percentage of long-range transport models specify tem-
porally and spatially varying mixing depths from temperature soundings.
Hourly values are usually obtained through linear Interpolation between
morning and afternoon soundings, or through the use of more elaborate
methods using hourly surface temperatures (e.g., RI/MTDDIS), or through
the use of an atmospheric boundary layer model (e.g., ERT/ADOM and NCAR/
RADM).
For those models designed to simulate wet deposition, precipitation rate
Information 1s usually required. Episodic models generally use either
dally rainfall data (e.g.. SRI/EURMAP-1 and 2, SRI/ENAMAP-1), 6-hour rain-
fall totals (e.g., ANL/ASTRAP, ARL/MTDF), or 3-hourly totals (e.g., SAI/
RTM-LT, SRI/ENAMAP-1A, ENAMAP2(S)(N) from standard surface meteorological
stations, or use hourly rates from the more numerous rain gauge sites
(e.g., PNL/LRT, RI/MTODIS. SAI/RTM-II, RTM-III, UMACID, UWATM/SOX). The
ERT/MESOPUFF-II utH1z1es categorical surface observations of precipita-
tion type (I.e., rain, freezing rain, etc.) and Intensity (light,
moderate, heavy) 1n adddltlon to hourly rainfall rate for determining wet
deposition rates. Other episodic models estimate wet deposition using
constant decay rates whenever a parcel 1s subject to rain (e.g., IAP/LRT,
NIAR/LRT). The NCAR/RADM and ERT/ADOM models obtain the precipitation
amount and type from dynamic meteorological models.
* Models allowing temporally and spatially variable mixing depths must
Incorporate temporally and spatially varying volume changes Into the
mass conservation equation. These models also have to address the
possibility of pollutant mass aloft that could be entrained because of
volume changes.
3-163
-------
Many long-term Lagranglan models use cl1matolog1cal estimates of the dura-
tion of wet and dry periods with appropriate scavenging rates (e.g., CERL/
LTSDM, OME/STADMOO, and the ERT/ESEERCO model). Other long-term models
may assume that a constant decay rate Independent of the occurrence of
precipitation 1s sufficient to describe the cl1matolog1cal mean wet depo-
sition.
Emission requirements depend 1n part on the resolution of the model.
Single-layer models usually require only source locations and emission
rates. In these models emissions are assumed to Instantaneously diffuse
vertically throughout a grid or other reference volume. On the other
hand, models with multiple vertical layers, or those that specify Gaussian
pollutant distributions 1n the vertical direction, require either actual
stack heights and emission parameters or effective stack heights. These
models are generally capable of resolving near-source pollutant distribu-
tions and simulating near-source dry deposition 1n a more realistic man-
ner.
Although all regional models simulate pollutant distributions arising
from major point sources, only EuleHan grid models and receptor-oriented
trajectory box models can easily accommodate area sources. Some Lagran-
glan source-oriented models Incorporate large area sources or aggregate
sources by Initializing puffs of very large dimensions. Both grid models
and Lagranglan models suffer from lack of spatial resolution at local and
mesoscale distances from the area or aggregated point sources. In grid
models this can be overcome by reducing the horizontal grid dimensions.
This 1s not possible with Lagranglan models. A strong advantage of grid
models over Lagranglan models 1s the ability to nest a high-resolution
grid for a mesoscale domain within a coarse-resolution grid for a regional
domain. This allows adequate spatial resolution for area source and
multiple point sources, yet preserves the ability to accommodate emissions
over continental scales. This approach will be discussed further when
recommendations are made for a modeling methodology to meet the require-
ments of modeling add deposition 1n the Rocky Mountain region.
Long-term models usually require seasonally varying emission rates because
of their limited temporal resolution, whereas most short-term models allow
specification of diurnal and weekly varying emissions as well.
3.3.2.2 Model Output Format
Most episodic models display grldded distributions of S02 and sulfate over
appropriate averaging Intervals (typically, 3 and 24 hours). Recently,
simplified (linear) nitrogen chemistry has been Incorporated Into pre-
viously sulfur-only models. Typically, these models display N02 and/or
3-164
-------
nitric add concentrations. Some models are designed for one pollutant
only, either S02 or a radlolsotope, whereas others treat complex photo-
chemistry and hence are capable of generating grldded fields of N02, 03,
and nitrates as well (e.g., EPA/ROM, SAI/AIRSHED, SAI/RTM-III, CIT/UAPM,
LLL/LIRAQ, NCAR/RADM, ERT/ADOM, SAI/CCAOM, UK/STEM2).
With few exceptions, receptor-oriented models yield pollutant concentra-
tions only at selected sites. UM/STOCHACID and UMACID generate grldded
fields, Indicating the probability of contributions at selected receptors
and, after multiplication of this field with an emission density field,
produce grldded fields of probable source contributions. A majority of
those models that parameterize dry and wet deposition also display fields
of deposited S02, S04, or total sulfur. Other models, because of their
design, do not have this capability. Several models produce Interregional
pollutant transfer matrixes (e.g,, SRI/EURMAP-1 and 2, SRI/ENAMAP-1, ANL/
ASTRAP, UI/RCDM-3), although this generally requires multiple simulations
and postprocessing of model results and thus could be considered a capa-
bility of all models. The SAI/RTM-II, SAI/RTM-LT, and UWATM-SOX models
produce total sulfur budgets as well. Precipitation pH, though an
extremely difficult variable to predict, 1s predicted by AES/LRT and
UWATM-SOX 1n a simple manner from Its rather weak statistical correlation
with precipitation-borne sulfate.
Most source-oriented models require transforming output variables from the
puff or plume segment coordinate to a fixed (EuleHan) coordinate system
(e.g.. ERT/MESOPUFF-II, ERT/ESEERCO model, SRI/ENAMAP-II). This trans-
formation usually Involves spatial and temporal Interpolation and, 1n the
case of puff models, may suffer from Inaccuracies due to nonoverlapping
puffs near the sources.
The SAI/RIVAD, developed Initially for regional visibility modeling, pre-
dicts light scattering coefficient and visual range. Rough approximations
of visual range may be calculated by any model capable of simulating sul-
fate concentrations using simple empirical relationships.
3.3.2.3 Transport and Dispersion
Eulerlan episodic grid models simulate horizontal transport of pollutants
by solving the advect1on-d1ffus1on equation using different numerical
solution techniques. Most of these models solve only the two-dimensional
equation and do not treat vertical transport. The vertical flux of pollu-
tants Is treated only by those multiple-layer models that use either fully
three-dimensional wind fields or divergent two-dimensional wind fields
(e.g., SAI/RTM-II, SAI/RTM-LT, SAI/RTM-III, EPA/ROM, LLL/AOPIC, USSR/LRT,
NCAR/RAOM, ERT/ADOM, and UK/STEM2). Dispersion 1s simulated 1n Eulerlan
3-165
-------
grid models using gradient transfer theory, which requires specification
of horizontal d1ffus1v1t1es and, for the above multiple-layer models,
vertical d1ffus1v1t1es as well. With the exception of the SAI/RTM-II AND
SAI/RTM-III, which Incorporate a Lagranglan plume-segment submodel for
treating near-source dispersion, the episodic EuleMan models Included 1n
the survey require that emissions diffuse Instantaneously Into the grid
volume. Depending on the numerical advectlon scheme, Eulerlan models
exhibit varying degrees of numerical diffusion as well.
Eulerlan long-term models, such as CCIW/LDAPTM, CERL/LTSDM, MIT Model, and
UI/RCDM-3, specify transport and diffusion 1n analytic form after Invoking
assumptions of steady state and horizontal homogeneity.
Lagranglan episodic models simulate atmospheric transport from the refer-
ence frame of a moving column of air or particle. Except for the ILL/
ADPIC model, all of these models use two-dimensional wind fields for the
transport.* As discussed by Stewart and L1u (1982), transport depends not
only on the wind field specification, but also on the method of trajectory
determination. Roughly two-thirds of the Lagranglan episodic models pre-
sented In the survey (ARC/ATAD Included) calculate trajectories using the
constant velocity method; several models (e.g., PNL/STRAM, PNL/MELSAR-
POLUT, ERT/MESOPLUME, MEP/TRANS, ERT/MESOPUFF, ERT/MESOPUFF-II, NIAR/LRT,
ARL/BAT, and IAP/LRT) use the more accurate constant acceleration method;
and none use the most accurate variable acceleration method; suggested by
Daggupaty, M1sra, and Munn (1979). Models that use the same trajectory
calculation methods can nevertheless exhibit variations 1n transport sim-
ulation because of different parcel tracking Intervals. Most Lagranglan
episodic models track parcels at three-hour Intervals; some use Intervals
of one hour or less (e.g., PNL/LRT, ARL/ATAD, SRI/EURMAP-2, NIAR/LRT,
IAP/LRT, and LLL/ADPIC). Tracking frequency may Influence overall pollu-
tant transport when the constant velocity or constant acceleration assump-
tions fall, I.e., under situations 1n which the flow field exhibits high
spatial and temporal variability.
Recently, attention has focused on the effects of the vertical shear of
horizontal winds or pollutant transport. This has prompted the modifica-
tion of several models from single mixed-layer versions to at least two-
layer versions (e.g., SAI/RTM-II, SAI/RTM-LT, SAI/RTM-III, ERT/MESOPUFF-
II, ARL/BAT). Two or more transport layers, communicating via Interfadal
fluxes arising from mixing depth variations and convergence/divergence,
Although the AES/LRT model utilizes three-dimensional winds, each
vertical column 1s advected according to the horizontal projection of
the three-dimensional wind.
3-166
-------
are conceptually straightforward to treat 1n grid models but pose problems
within the Lagranglan framework. The continuity of puffs 1s disrupted
during day-to-night transition periods. The ARL/BAT models allow puffs
to split Into upper and lower portions that are Independently advected
by the appropriate transport winds. While this 1s the logical approach
to directional shear treatment 1n Lagranglan source-oriented models, the
method results 1n a geometrically Increasing number of puffs for a single-
source multlday transport, taxing both computer resources and the concept
of a contiguous puff remaining coherent at large downwind distances. The
ERT/MESOPUFF-II utilizes two-layer wind fields but, Instead of dividing
puffs Into upper and lower portions and performing separate advectlon,
the entire puff remains coherent and 1s advected by either the lower- or
upper-layer wind field. The SAI/CCADM model uses a multilayer wind field,
In which the Lagranglan box 1s advected according to a user-specified tra-
jectory, where wind shear results 1n Increased diffusion.
The manner 1n which horizontal dispersion 1s parameterized 1n Lagranglan
episodic models generally depends on whether the model 1s source oriented
or receptor oriented. Most source-oriented models (both puff and plume
segment) specify the parcel dimensions using dispersion parameters, such
as oy or oh (for puffs). These are generally taken from Pasqu1ll-G1fford
stability curves (Turner, 1970) or from the survey of Heffter (1965).
With the exception of SRI/ENAMAP-1, SRI/ENAMAP-1A, and SRI/EURMAP-2, which
specify a uniform horizontal distribution, the Lagranglan puff and plume
models surveyed here assume a Gaussian mass distribution across the puff
or segment. Of the receptor-oriented models surveyed, only UMACID treats
horizontal diffusion; the others assume that pollutant distributions with-
in the parcel are uniform and that the parcel dimensions remain constant.
Lagranglan episodic models that simulate vertical dispersion do so either
by assuming a vertical Gaussian distribution or by a separate Integration
of the vertical diffusion equation. The former method requires specifica-
tion of an effective stack height and oz values; the latter method
requires a multiple-layer model and the specification of time-varying K_
profiles. Except for the local-scale diffusion simulated 1n the IAP/LRT
model, none of the receptor-oriented Lagranglan episodic models surveyed
treat vertical diffusion.
Among the Lagranglan long-term models, AES/LRT. SRI/ENAMAP-1, the ESEERCO
model, and SRI/EURMAP-1 simulate transport and dispersion 1n the same man-
ner as their short-term counterparts. The PNL/RAPT model neglects hori-
zontal dispersion completely, while retaining the sequential trajec-
tory concept. US/ROOM. ANL/STRAP, ANL/ASTRAP, and OME/STADMOO are
designed to predict dispersion statistics from trajectory endpolnt distri-
butions, and use these statistics In a Gaussian puff framework.
3-167
-------
3.3.2.4 Physical and Chemical Transformations
The physical and chemical processes that act on primary sulfur and nitro-
gen emissions result 1n the formation of secondary pollutants that, 1n
many cases, are long-lived and hence transported far downwind from the
source regions. To date, most long-range transport models focus only on
transformations of sulfur species. A few treat nitrate formation 1n a
simple linear fashion (e.g., AES/LRT, CAPITA/MCARLO, MEP/TRANS, SRI/
ENAMAP-2(s)(N)), eight of the surveyed models use detailed photochemical
mechanisms to simulate oxIdant formation (EPA/ROM, SAI/RTM-III, SAI/
AIRSHED, LLNL/LIRAQ, NCAR/RADM, ERT/ADOM, UK/STEM2, SAI/CCAOM), and one
(UK/STEM) uses detailed photochemical mechanisms to simulate homogeneous
S02 oxidation while another model (SAI/RTM-IINL) obtains S02 and NOX oxi-
dation rates from a look-up table generated from detailed photochemical
and aqueous chemical mechanisms. The ERT/MESOPUFF-II model treats SOX and
NOX oxidation (as an option) 1n a plecewlse linear manner, relying on
results from photochemical box model simulations to determine the appro-
priate conversion rates. Only three of the surveyed models (NCAR/RADM,
UK/STEM2, and ERT/ADOM) are capable of treating physical transformations,
such as aerosol nucleatlon and coagulation.
For those models that treat sulfur species only, the most common method
of parameterizing oxidation 1s through a constant linear transformation
rate. This method has been extended to seasonal or diurnally varying rate
constants by the ANL/ASTRAP, BNL/AIRSOX, MEP/TRANS, SAI/RTM-II, SAI/RTM-
LT, and UMACID models, among others, and to a transformation rate depen-
dent on local photochemical activity (ERT/MESOPUFF-II). Diurnal rates are
usually specified by a sinusoidal function during the daytime (represent-
ing the homogeneous oxidation due to photochemlcally generated Intermedia-
tes) and a constant minimum value at night (representing heterogeneous
oxidation). Humidity effects are also taken Into account by the IAP/LRT,
CSU-EPA/MODEL B, and PNL/STRAM models.
The ERT/DISCDEP2 model (part of the ERT/ESEERCO model) uses constant
transformation rates, but, because of Its formulation, can Incorporate
both constant dry oxidation (I.e., gas-phase) rates and constant wet oxi-
dation (I.e., aqueous-phase) rates 1n a statistical manner.
The overall complexity of the S02 oxidation mechanism has resulted 1n a
wide variety of choices of a representative rate constant for regional
model use. Most models choose a dally average rate constant of the order
of 1 percent per hour. Those with seasonal variations generally Increase
summer average rate by a factor of 2 over the wintertime rate. (Rate con-
stants are Indicated 1n Table 3-13.)
3-168
-------
3.3.2.5 Removal Processes
From the review of dry deposition mechanisms presented 1n Section 3.1.3 1t
1s evident that a simplified parameterization of pollutant removal at the
surface 1s required for practical applications of regional and mesoscale
models. Generally, those models capable of simulating or parameterizing
the turbulent transfer processes 1n the surface layer have the most com-
plex dry deposition parameterlzatlons (e.g.. AMI/STRAP, ANL/ASTRAP, ANL/
MTDF, BNL/AIRSOX, EPA/ROM, ERT/SURAD, ERT/MESOGRID, ERT/MESOPUFF-II, ILL/
ADPIC, SAI/RTM-II, SAI/RTM-LT, SAI/RTM-III, UK/STEM, UK/STEM2, MCAR/RADM,
ERT/AOOM, and UWATM/SOX).
Those models that specify the vertical concentration distribution as
either uniform or Gaussian have a more limited choice of parameterlza-
tlons. The most commonly used method 1s to apply a constant deposition
velocity to the mixed-layer average concentration (e.g., AES/LRT, ARL/
ATAD, CCIW/LDAPTM, ERT/ATM, OME/STADMOD, SRI/EURMAP-1, SRI/ENAMAP-1, and
UMACID). If the model has a vertical Gaussian or other analytic distribu-
tion option, the ground-level concentration may be used or, alternatively,
the vertical extent of the pollutant mass subject to removal may be
rescaled (e.g., CERL/LTSOM, ERT/MESOPLUME. ERT/MESOPUFF, PNL/LRT, PNL/
RAPT, PNL/STRAM, SRI/EURMAP-2, and US/ROOM).
The simplest dry deposition parameterization 1s to subject the pollutant
mass to a uniform decay rate (e.g., CAPITA/MCARLO, DMI/LRDPM, ERT/ATM, the
ERT/ESEERCO (DISCEP2) model, IAP/LRT, NIAR/LRT, SCAR/LRTM, MIT model, MIT/
MITEMAO, UI/RCDM-3, USSR/LRT). The choice of this particular parameteri-
zation method 1s often determined by the basic model structure. For
example, 1n order to obtain analytic solutions to the mass balance equa-
tion, the MIT model and UI/RCDM-3 require uniform decay rates.
Met removal rates of pollutants are dependent on precipitation rate,
hydrometer history (for particle scavenging), and solubility characteris-
tics (for gaseous scavenging). Because the mechanisms Involved are com-
plex (see Section 3.1.4) simplifying assumptions must be Invoked for
economical treatment 1n existing long-range transport models. The models
surveyed here exhibit a broad range of parameterization methods, from a
constant linear decay rate to nonlinear expressions Incorporating effects
of droplet pH and temperature.
Long-term models generally treat wet deposition as an exponential decay
process (I.e., linear decay rate) that 1s often completely Independent
of the occurrence of precipitation (e.g., ERT/ESEERCO (OISCOEP2), OME/
STADMOO, MIT model, UI/RCDM-3, and US/RDDM). This linear decay concept 1s
even utilized 1n the short-term model CAPITA/MCARLO. The justification
3-169
-------
for this type of treatment, which 1s open to criticism, 1s that the occur-
rence of precipitation 1s temporally and spatially random, and hence the
total region 1s subject to the same removal efficiency. With this
parameterization method, the wet deposition pattern 1s proportional to the
airborne concentration pattern. An Improvement over this method 1s the
Unking of the linear decay rate with the spatially and temporally vari-
able probability of precipitation occurrence, as 1s done 1n the CERL/
LTSDM.
Models (both long-term and short-term) that utilize time-varying transport
and precipitation fields have the most realistic capability of specifying
wet deposition as a function of the occurrence of precipitation. The
IAP/LRT and NIAR/LRT models Invoke a constant linear decay rate only for
the fraction of time 1n which the parcel encounters precipitation. Other
models specify the decay rate (scavenging coefficient) as a linear func-
tion of precipitation rate (I.e., ERT/MESOPUFF II, PNL/RAPT, PNL/LRT, SRI/
EURMAP-1. SRI/EURMAP-2, SRI/ENAMAP-1, CECB/TD, KNMI/MAPM).
Another commonly used method 1s based on the scavenging ratio, I.e., the
concentration of pollutant 1n rainwater divided by airborne concentra-
tion. The scavenging coefficient A and scavenging ratio a (often called
washout ratio) are related by the expression
a
P
where H and P are the mixing depth (or rainfall layer depth) and the pre-
cipitation rate, respectively. Because the scavenging ratio 1s, 1n fact,
a function of precipitation rate and other variables^ many parameterlza-
tlons focus on detailed expressions for a. The ANL/ASTRAP model assumes
that for total sulfur the scavenging ratio 1s proportional to the square
root of the precipitation rate. Separate scavenging ratios are calculated
for S0? and sulfate by MEP/TRANS, SAI/RTM-II, RTM-LT, RTM-IINL, and UWATM-
SOX. The S02 removal 1s assumed to be reversible, with functional depen-
dencies on temperature and pH, whereas sulfate scavenging depends 1n a
nonlinear way on precipitation. The NCAR/RADM, ERT/ADOM, SAI/CCADM, and
UK/STEM2 contain detailed cloud mlcrophyslcs modules that calculate appro-
priate scavenging based on gaseous equilibrium and particle nucleatlon.
Because of the variety of wet deposition parameterization methods (see
Section 3.1.4), 1t 1s difficult to estimate the uncertainty associated '
with modeling wet deposition. However, because of the complexity of the
scavenging mechanism, the uncertainty 1s quite large. A survey of depo-
sition parameters published by McMahon et al. (1979) and listed 1n Section
2-of this report provides a more quantitative estimate of this uncer-
tainty.
3-170
-------
The joint treatment of aqueous-phase oxidation and wet gaseous deposition
has been treated 1n only five of the models surveyed (NCAR/RADM, ERT/ADOM,
UK/STEM2, SAI/CCADM, and SAI/RTM-IINL). Of these, the simplest scheme 1s
contained 1n SAI/RTM-IINL, which uses a lookup table for the gas- and
aqueous-phase chemistry and standard wet scavenging schemes (same as RTM-
II and RTM-III). The NCAR/RADM, ERT/ADOM, SAI/CCADM, and UK/STEM2 contain
detailed cloud mlcrophyslcs modules to calculate aqueous-phase chemistry
and wet gaseous and participate deposition. Preliminary results using the
UK/STEM2 model Indicate that simulations done using the detailed cloud
mlcrophyslcs module require almost 100 times more computer capacity than
those simulations exercised 1n a dry mode. Initial results of the NCAR/
RADM, ERT/ADOM, and SAI/CCADM also show that the cloud mlcrophyslcs and
aqueous chemistry modules also require extensive computational resources.
3-171
-------
4 SELECTION OF CANDIDATE MESOSCALE METEOROLOGICAL AND
ACID DEPOSITION MODELS
In this chapter we describe the candidate mesoscale meteorological and
add deposition models chosen for further evaluation. The mesoscale
meteorological and add deposition models are ranked based on a scoring
system 1n which objective scores are assigned to various modeling
approaches to the processes that lead to air flows over complex terrain
and add deposition. Finally, the last section 1n this chapter describes
the selection of the final candidate mesoscale meteorological and add
deposition models based on additional, nontechnical criteria that Include
the needs of the potential users.
4.1 SELECTION OF MESOSCALE METEOROLOGICAL MODELS
This section presents the criteria used to select mesoscale meteorological
models for application to a mesoscale region within the Rocky Mountains.
The first step Involved elimination of those models that do not satisfy
the minimal requirements for mesoscale meteorological modeling of complex
terrain, as discussed 1n Section 2.1. The remaining models were classi-
fied according to the scheme presented 1n Section 2.2. At this point a
technical merit analysis of the candidate models was performed to deter-
mine whether any model 1s clearly superior to the others. Finally, the
candidate models were ranked on the basis of other considerations, Inclu-
ding flexibility, adaptability, and use of computer resources. The selec-
tion of the final candidate mesoscale meteorological models was made after
consideration of several nontechnical criteria (Section 4.3).
4.1.1 Criteria for Selection
Complex-terrain meteorology has been discussed 1n Chapter 2. To model the
meteorology of a mesoscale region 1n the Rocky Mountains, the model must
be able to simulate three-dimensional wind patterns over complex ter-
rain. The model must also have a documented application to mesoscale com-
plex terrain and be available for this project. Thus, we Identified three
minimum requirements for the model for a Rocky Mountain Mesoscale add
deposition modeling system:
4-1
-------
(1) The model must be able to predict a mesoscale three-dimensional
(or multilayer, two-dimensional) wind field;
(2) The model must be able to accommodate complex terrain; and
(3) The model must be currently operational.
4.1.2 Technical Merit Analysis
Of the 65 mesoscale meteorological models listed 1n Table 2-2, 17 do not
predict a three-dimensional mesoscale wind field. These models are either
one-dimensional, slab-symetrlc models (x,z); single-layer models (x,y);
or, 1n one case (NCAR/QNGM), a Lagranglan model that would be unable to
predict three-dimensional wind fields for an entire mesoscale region. Of
the remaining 48 models, 11 do not explicitly account for complex ter-
rain. Many of these models were designed for predicting the formation of
hurricanes or tropical storms over the open ocean, for which the Inclusion
of complex terrain 1s not necessary.
The requirement that the model be operational eliminates 22 of the remain-
Ing 37 models, leaving 15 models. Of those models considered nonopera-
tlonal, 7 were classified as nonoperatlonal because they were developed
outside of the U.S. and the model code or simulation results were deemed
too difficult to obtain. The NCAR/NGM model Is considered nonoperatlonal
because the grid nesting procedure 1s accomplished by running several
models 1n parallel 1n an Interactive sense. The remaining nonoperatlonal
models are research-grade models, as reported 1n the literature cited. It
should be noted that the classification of a model as nonoperatlonal 1s
subjective. The elimination of a model should not Imply that 1t could not
be useful for meteorological modeling 1n the Rocky Mountains, only that
the effort required of an outside modeler would be beyond the resources
available 1n this modeling project.
Of the remaining 15 qualified mesoscale meteorological models, 9 are diag-
nostic and 6 are prognostic models. These models are classified according
to the scheme presented 1n Section 2. Table 4-1 classifies these 13
models according to the taxonomlc tree given 1n Figure 2-10. The names
given these models are shortened versions of the names given 1n column 1
of Table 2-2 and may differ from the names given the models by the origi-
nal developers. Except for the SAI/MWVM, the 15 remaining qualified
models are grouped together 1n their own category.
The 15 models were subjected to a technical merit analysis to determine
whether any of the models were better suited to predicting wind flows over
4-2
-------
TABLE 4-1. Classification of qualified meso-
scale meteorological models according to
taxonomlc tree given 1n Figure 2-10.
Diagnostic Model Based on Primitive Equations
SAI/MVWM
Diagnostic Models Based on Mass Continuity
CIT/WIMD
LLNL/MATHEW
LANL/ATMOS1
PNL/MELSAR-MET
SRI/Diagnostic
SAI/CTWM
USFS/KRISSY
CAMM/NUATMOS
Prognostic Models Based on Primitive Equations
CSU/MESO
DREXEL/NCAR
MESO/MASS
LANL/HOTMAC
NCAR/MM4
NOAA/ERL
4-3
-------
complex terrain. Due to the basic differences 1n prognostic and diag-
nostic model formulations, the two types were evaluated separately.
4.1.2.1 Methodology
Four major categories of technical merit were Identified, corresponding to
the modeling needs presented 1n Section 2.1 and the model formulation dis-
cussed 1n Section 2.2:
Equations
Parameterization
Initial and boundary conditions
Phenomena
Each category 1s divided Into subcategorles that focus on the physical
properties of the atmosphere or Individual model components. Each sub-
category 1s represented by a set of common modeling methods used to
address the process under consideration, and these methods are ranked from
least to most sophisticated. Scoring within each subcategory 1s thus
based on the level of sophistication or level of detail Incorporated Into
the candidate model. The Intervals between scores 1s not the same 1n all
subcategorles, since a weighting factor 1s applied to each subcategory.
This technique permits scores to be summed directly to obtain a total
level of merit. The scoring system 1s outlined 1n Table 4-2.
Equations. This category has two subcategorles—the coordinate system and
dimensionality, and the basic form. Since the dimensionality of the model
represents the model's degree of freedom, a three-dimensional model 1s the
most desirable measure of merit and 1s scored the highest. A terrain-
following coordinate system has an advantage over a Cartesian coordinate
system 1n modeling areas of complex terrain. The mathematical formula-
tions of the governing equations of the meteorological variables are
divided Into four groups: (1) the objective analysis method, which uses
observed data to Interpolate the desired grid values; (2) diagnostic
models that use the mass continuity equation; (3) diagnostic models that
use the primitive equation set and Iterate to steady-state; and (4) the
prognostic method. This last method can predict time-varying, dynamic
characteristics and thus scores the highest 1n Table 4-2.
Parameterization. The specification of subgrld-scale and source-sink
processes using experimental data and simplified fundamental concepts 1s
categorized as parameterization. According to Plelke (1984). three main
physical processes should be parameterized 1n mesoscale models.
4-4
-------
TABLE 4-2. Technical attribute categories, subcategorles, and modeling methods
for candidate prognostic models.
Category
Equations
Subcategory
Dimensionality and
coordinate system
Sample Modeling Method
(x) 1-D
(x,z) 2-D, slab symmetric
(x,y) 2-0, single layer
(x.y.z) 3-D, Cartesian
(x,y,o(p)) 3-D, slgma pressure
(x,y,o(z) 3-D, slgma terrain
Score
0
0
0
1
2
2
Basic formulation
Parameterization Turbulence
Radiation
Cumulus
Stable precipi-
tation
Objective analysis 0
Diagnostic with mass continuity 1
Diagnostic with primitive equations 2
Prognostic with primitive equations 3
Drag coefficient 0.5
First-order closure 1
Second-order closure 2.0
Simple 0.5
Complete shortwave and longwave 1
Dry corrective adjustment 0.5
Explicit presentation with
prognostic equations 1
One-dimensional cloud model 1
Instantaneous fallout 0
Moisture cycle with prognostic
equations 0.5
Parameterized mlcrophyslcs 0.5
Detailed mlcrophyslcs 1
Initial and
boundary
conditions
Domain, grid,
resolution
Lateral
boundary
conditions
Regional or mesoscale only
Fixed regional and mesoscale
Flexible regional and mesoscale
Constant
Zero gradient
Radiative
0
0.5
1
0
0.5
1
continued
4-5
-------
TABLE 4-2 (Concluded)
Category Subcategory Sample Modeling Method Score
Top boundary Without absorbing layer 0
condition With absorbing layer 0.5
Bottom boundary Prescribed surface temperature 0.5
condition Surface heat budget 1
With surface soil moisture budget 0.5
With vegetation/canopy effects 0.5
Initialization With objective analysis 0.5
With dynamic Initialization 0.5
With nonlinear normal mode
Initialization 1
With variations! asslmulatlon 0.5
Phenomena With kinematic terrain effects 1
With dynamic terrain effects 1
With upslope-downslope effects 1
With orograpMc precipitation
effects 1
4-6
-------
(1) Averaged subgHd-scale fluxes
(2) Averaged radiation flux divergence
(3) Averaged effects of a change 1n water phase, Including precipi-
tation
The averaged subgrId-scale fluxes (turbulent quantities) are parameterized
within the planetary boundary layer (PBL) using three different methods:
(1) bulk-aerodynamic parameterization (drag coefficient representations);
(2) first-order closure models; and (3) second-order closure models. The
drag coefficient representations are simple, but Inadequate, parameteriza-
tion schemes for most atmospheric situations (especially 1n complex ter-
rain). First-order closure models generally represent mean values well
enough, but for complex turbulent problems the second-order closure models
provide better resolution.
The forcing resulting from the averaged radiation flux divergence 1s the
primary forcing function for diurnal and seasonal variations. Depending
on the characteristics of the earth's surface and atmosphere, 1t can also
create significant mesoscale effects due to spatial thermal contrasts
(e.g., a land-sea breeze or urban heat-Island). The parameterization of
radiation can be divided Into longwave and shortwave radiation, and can
vary from quite simple forms to much more complicated forms for cloudy and
polluted air.
Parameterization of moist thermodynamlc processes Includes cumulus cloud
parameterization under convectlve stable conditions. The simplest repre-
sentation 1s dry convectlve adjustment, which forces the lapse rate to be
moist adlabatlc when saturation occurs. In some cases cumulus clouds can
be represented explicitly 1n the same fashion as stratiform clouds. To
better represent cumulus clouds, more sophisticated parameterization
schemes have been developed through the use of one-dimensional cloud
models (e.g., Frltsch and Chappell, 1981, or Kuo and Raymond, 1980). When
air flows over complex terrain are modeled, the cumulus effect 1s not as
pronounced as are the effects of radiation and turbulence. However, cumu-
lus cloud processes are very Important 1n oxidizing add precursors and
delivering the resultant acidic substances to the ground, and are thus
scored as high as the radiation subcategory.
On the other hand, a stable precipitation parameterization scheme 1s used
to represent stratiform clouds 1n mesoscale models. All of the final can-
didate meteorological models contain a parameterization scheme for stable
precipitation. The simplest approach 1s to assume Instantaneous fallout
as saturation occurs; a more complicated approach 1s to Include a moisture
4-7
-------
cycle for calculating the variation 1n water phases (vapor, liquid,
solid). Parameterized mlcrophyslcs 1s a bulk representation used as an
alternative to the detailed mlcrophyslcal representation. Because of the
Importance of precipitation 1n add deposition, this subcategory 1s scored
as high as are cumulus and radiation subcategorles.
Initial and Boundary Conditions. This category comprises features of the
modeling domain, lateral, top, and bottom boundary conditions, and
Initialization. Although the domain, grid, and resolution of the model
nay depend on the purpose of the simulation, 1t 1s more flexible for a
model to have a variable domain size and resolution and a flexible grid
structure. For the Rocky Mountain project, a model capable of simulating
mesoscale phenomena with a flexible grid structure 1s the most desirable
and 1s thus scored highest.
Boundary conditions are also Important for mesoscale meteorological model-
ing. Using a constant lateral boundary conditions, the model can easily
generate spurious energy within the domain through reflection. Gradient
boundary conditions can reduce some of this reflection of energy by allow-
ing energy to move out of the domain. Radiative boundary conditions mini-
mize the reflection of energy Into the domain by handling the propagation
of disturbances near the boundary 1n a more appropriate way.
Reflection of energy from the top boundary .can also occur, especially 1n
complex terrain, which can generate an upward-progagatlng mountain wave.
This problem can be reduced by use of an absorbing layer near the top
boundary to smooth the perturbation and to avoid the generation of
unresolvable shortwave phenomena.
The Interface between the atmosphere and the earth's surface 1s the boun-
dary that has the most physical significance. It 1s the differential
gradient of the dependent variables along or near this Interface that
drives the atmospheric motion. Thus, the bottom boundary conditions are
crucial to mesoscale atmospheric systems and must be represented as
accurately as possible. Therefore, a method of specifying the bottom
boundary condition has the highest score within this category. The most
Important variables for the bottom boundary Include surface temperature,
surface heat flux, and surface moisture flux. Surface temperature and
surface heat flux can either be prescribed from observed data using simple
formulas, or predicted by surface heat budget calculation. Surface mois-
ture flux 1s calculated 1n some models using a surface soil moisture bud-
get. Several models also treat more complicated surface properties such
as vegetation and canopy effects.
Providing the model with Initial field data has become a very Important
part of meteorological modeling. To transform observed data Into model
4-8
-------
grid point values generally requires the use of objective analysis. After
such an Initial field 1s supplied, then either a dynamic Initialization
scheme, which allows the model to adjust Itself, or a much more complica-
ted normal mode Initialization scheme 1s used. Some models have the
ability to Input updated observed or objective data during model simula-
tion periods through the use of varlatlonal assimilation. The basic
principle of varlatlonal assimilation 1s to minimize the adjustments to a
field 1n a least squares sense, while meeting a set of data constraints;
this 1s useful 1f the conditions are constantly changing and the distur-
bances from outside the domain are allowed to propagate Into the modeling
region.
Phenomena. This category Includes the kind of phenomena the models are
capable of simulating and 1s divided Into four subcategorles (Section
2.2), each with equal weight. Although the category 1s redundant because
the basic governing equations and parameterlzatlons determine the ultimate
capabilities of the model, 1t 1s Included to highlight the abilities of
the candidate models. The first subcategory, "Kinematic Terrain Effects,"
refers to modifications to the basic air flow due to the physical presence
of the topography. The second subcategory, "Dynamic Terrain Effects,"
Includes those effects that Involve the Interaction between the terrain
and other forces (e.g., gravity). The third subcategory 1s "Upslope-Down-
slope Flows," which are locally Induced, thermally driven flows. The
final subcategory, "Orographlc Effects on Precipitation," refers to the
ability of the model to reproduce the observed effects of Increases 1n
precipitation with elevation.
4.1.2.2 Analysis of Prognostic Models
Table 4-3 lists the technical attributes of the seven candidate prognostic
mesoscale meteorological models. The scoring system given 1n Table 4-2, -
when applied to the model attributes described 1n Table 4-3, gives the
scores presented 1n Table 4-4. It should be mentioned that these scores
reflect a degree of subjectivity. In addition, scores are based on the
state of the nodel as given 1n the publications cited 1n Table 3-12, or
existing 1n the operational versions of the Indicated model. As shown 1n
Table 4-4, all the models received similar final scores, with only a 2.0
point spread between the highest and lowest score. More Important, all
the models are capable of simulating the Important phenomena pertaining to
air flows over the complex terrain of the Rocky Mountains.
4.1.2.3 Analysis of Diagnostic Models
Although not as technically rigorous as the prognostic models, diagnostic
models have been used more frequently for air pollution studies because of
4-9
-------
TABLE 4-3. Technical attributes of candidate prognostic mesoscale meteorological models.
Category
Governing equetlone
Simplifying assumptions
Coordinate eystas)
turbulence
(PBL)
Radietlan
(free etaoephere)
Cuaulua
1
^> Precipitation
Ooaeln
Grid
Reeolutton (horizontal)
(vertical)
teterel boundary
condition
lop bounder? condition
Button botmdery condition
CSU/HCSO
Primitive equations
Hydroetetlc
Incoapreeaibls
Diabetic
Dry
Similarity theory,
profile K » local Kt
prognoatic aquation
for PBL height
Seeemorl echeaa
tone
Stable preclpltetloni
Inetentengous fellout
Verteble
Verlable, e.g.,
JJ x 36 x 14
Verlable 1 - 11 ta
Variable
Zero gradient bound-
ary condition
Kith abeorblng layer
Surface energy bud-
get | eurface molatun
bu>V)at| vegatatation
NESO/MASS
Primitive equation*
Hydroat*tic
Incoapreeeible
Diabetic
Dry
Generalized elailerlty
theory, prognoatic aquation
for POL height
None/epeciric
situations only
Convectlva sdjust*snt/
Frltech « Oteppell scheas
Stable precipitation (EOW)
North Africa
128 x 36 x IV
1J7 x 117 K 14
47.4. 24. 12, 6 ta
(neat ad)
Variable
tims-dapsndant boundary
condition/ radietlve boundary
condition
with absorbing layer
Surface energy budget |
eoll moleture budget
NOAA/Em.
Primitive equation*
Hydrostatic
Incompressible
Diabetic
wet/cloud
0 . (P - P )/(p . p )
top • top
Similarity theory, profile
K, constant PBL height/
separata THE aquation for
PBL height
Modified Saaeaorl acheaa
ParaaeterinKt
•Icrophyslc*
250 x 2SO ta/
1600 x 1600 ta
26 x 26 x 16/
25 x 25 x «
10 ta/40 ta
Variable
Von Neuaan/
radietlve boundary condition
(time dependant boundary
condition
Kith absorbing layer
Specified/
Surface energy budget
(vegetation and eoll)
LANL/Yaaada
Primitive equetlone
Hydroatatlc
Incoapreeeible
Diabetic
Dry
2nd-order cloeure
aquations
(complete or
eimpllfled)
Seeeaorl achsss
Gausalon cloud
•odel
10 x 7ta/
5 x 1.2 ta
250 x 300 ta
26 x 21 x 16/
29 x 40
10 ta/0.2 ta
Variable
Zero gradient
(inflow)
linear extrapo-
lation (outflow)
Rigid
Surface energy
budget! tree
canopy affects
NCAR/MHt
Primitive aquations
Hydrostatic
Incoapreeeible
Olabatlc
Dry/wet
tf* 'top)/(f Zc * 'top1
Bulk aerodynamic PRL
/high resolution PBL
None/elmpllfied
Dry convectlve sdjustssnt/
cuaulua convsction/Kuo's
scnsw/Hsia'a echeme
Instsntanaoua felloutf mois-
ture cycls with prognostic
equetlone
10 - SOOO ta
Variable/ SO x SO x 10
2.5 - 200 ta (nested)
Variable
Open boundary condition
(•pacified from lerqe-ecele
model)
dO/dt * 0 at 0 « 0
Surface energy
budget (Bleckadan'a)!
aurfaca molaturs budget
ORCXEL/NCA8
Primitive equetlone
Hydroetetlc
Incoapreesibls
Diabetic
Dry/wet
* ~ ZC
let-order cloeure
Longwave to stmosphera only
Ssquentlsl plume modal
Instantaneous fallout/
Kaaalsr's psrameterlzetlon
of cloud and rain water
Msso - a
Variable
37 x JV 49 x 49 x 15
1.2** (140 ta)
0,3125* (~ 35 ta)
Variable
Porous apongs ( time-
dependent | spselflsd from
large-scsls modal)
dK/dt . 0
Surface energy budget
-------
TABLE 4-3 (Concluded)
Category
Initialization
•CSU/NESO
Horizontal ha
eauoi dynaalr
itation
HESD/NASS
noon- Real data, UN output,
initial- Bamaa objective analyaia,
variation*! aoolallatlon
MMA/EM.
dynaaie initialization/
real data, UN output,
objective analyaia. dynamic
LANL/Yaaada
horizontal haao-
qanaoua potential
taaparatura field
NCAK/MH4
MC •Mlyaie and
aatallite data,
objective analyaia,
nonlinear noraal aoda
DMEXEL/NCM
Objective data
analyaia
initialization,
initialization
Solution tachnlojuai
Spaca dirratonea
lloa Intagration
•tudlad
UjMtr«aa cubic-opllno Fourth/aiicth ardar
Foniord-ln-tiaa Cular badoMrd tlaa
Ing
flow, Nautoin-vallay flow,
forced air flow aver farced air flow aver rough
rough terrain, dry tarraln, dry dapoaitlon
deposition
Anthea and Nemar centered- MI
in-tlaa with TASU-Mataurio/
leap frag with Aaaalln filter
Nountain-vallay flow, foread
air flow avar rough terrain,
dry dapoaltlon, wat
•itlon
Computer
2* aln/12 hr far
30 x M K 13.
t*. >90.S
on OUV-1
30 Bin/ M hrt
79 Bln/n hr
(HKSS. ).0)|
on CDC 203
Twlca real tiaa on
CDC 750/175.
Nountain-vallay
flow, ronad air
flaw evor rouqh
torrain, dry
dapoaltion
4 x 1lf*a/
grid, Iti
140 S/ 2.5 hr,
tt « 90S
on CDC 7600
fad ordor
Explicit (Brown end
HDuntain-vallay flow, foread
air flow over rouoh torrain,
dry dapoaitioni wat dapool-
tioni eonvaetlva ayataaa,
frontal ayataaa
•7 Bin/ 24 hr for
SO « 50 x 10, tt • 205
an OUT-1
4th ordar
explicit (Brown and
Hbuntain-vallay flow.
forcod air flow over rough
torrain, dry dopooitlon,
wat dapoaitlon
20 aln/ 24 hr
for CRAV-1
-------
TABLE 4-4. Technical attribute scores for candidate prognostic models,
Category
Equations
Parameterization
Initial and
boundary
condition
Phenomena
Subcategory CSU/MESO MESO/MASS
Dimensionality and
coordinate system
Basic formulation
Turbulence
Radiation
Cumulus
Stable precipitation
Domain resolution
Lateral boundary
condition
Top boundary condition
Bottom boundary
condition
Initialization
Kinematic effects
Dynamic effects
Ups lope-downs lope effects
Orographlc effects on
precipitation
2
3
1
1
1
0
1
1
0.5
2
0.5
1
1
1
1
2
3
1
0.5
1
1
1
1
0.5
1.5
1
1
1
1
1
NOAA/ERL LANL/HOTMAC
2
3
1
1
0
1
1
1
0.5
1.5
1
1
1
1
1
2
3
2.0
1
1
0.5
1
0.5
0.5
1.5
0.5
1
1
1
1
NCAR/MM4 DREXEL/NCAR
2
3
1
1
1
i
1
1
0.5
1.5
1
1
1
1
1
2
3
1
0.5
1
0.5
1
1
0.5
1
1
0.5
1
1
1
1
Total:
17.0
17.5
17.0
17.5
18.0
16.0
-------
their ability to produce nondlvergent, mass-consistent wind fields 1n com-
plex terrain without requiring excessive computational resources. Diag-
nostic meteorological models primarily use the conservation of mass equa-
tion to Interpolate between measurements, and then apply some sort of
divergence minimization scheme. Thus, any fr1ct1onalf dividing stream-
line, or thermal effects must be represented either through observations
or parameter1zat1ons within the model.
The scoring scheme used for the technical merit analysis of the diagnostic
meteorological models 1s given 1n Table 4-5. The scores within each cate-
gory and siibcategory are weighted according to the Importance of that
category for simulating wind flow over a mesoscale region containing the
complex terrain of the Rocky Mountains. The first category, "Equations,"
1s the same as that presented for the prognostic models. Again, a ter-
rain-following coordinate system 1s considered more desirable than a Car-
tesian coordinate system. Also, models that use several primitive equa-
tions score higher than those models that use only mass conservation equa-
tions.
The next two categories, "Parameterization" and "Phenomena," concern the
ability of the model to simulate wind flows over complex terrain. Since
all the diagnostic models considered here Include the conservation of mass
equation, they are well able to simulate most kinematic terrain effects.
Some of the models (e.g., LANL/ATMOS1. LLNL/MATHEW, and CAMM/NUATMOS) use
the varlatlonal technique of Sasaki (1970),- which minimizes the error
between model predictions and observations 1n a least squares sense, while
predicting a mass-consistent wind field. Thus, these types of models
require larger amounts of observed data to Infer the presence of such
phenomena as drainage flows. Other types of diagnostic models (e.g.,
SAI/CTWM) parameterize the effects of these phenomena. Since observed
surface and upper-wind data 1n the Rocky Mountains are very sparse, 1t 1s
more desirable to parameterize these phenomena than to rely on surface
measurements to Infer them.
The highest technical attribute scores were received by the SAI/MVWM and
the SAI/CTWM. The SAI/MVWM 1s the only primitive equation model 1n the
group, and thus received an additional point for "basic form." As a
primitive equation model, 1t 1s expected to simulate all listed
phenomena. Use of the SAI/MVWM as a "diagnostic" model Involves the Inte-
gration of the conservat1on-of-momentum equation to a steady state assum-
ing a constant three-dimensional temperature field. We note that diag-
nostic use of the SAI/MVWM would be at least an order of magnitude more
expensive on a given grid than would the other models 1n this group.
The SAI/CTWM scores high 1n the "Phenomena" category because 1t treats all
listed phenomena except mountain waves.
4-13
-------
TABLE 4-5. Technical attribute categories, subcategorles, and modeling methods
for candidate diagnostic models.
Category
Subcategory
Description
Score
Equations
Coordinate system
Basic form
Parameterization Fr1ct1onal effects
Blocking effects
Phenomena
Kinematic
Mountain waves
Upslope/downslope
Data Reliance
1-D (x) 0
2-0 (x,z), slab symmetric 0
2-0 (x,y), single layer 0
3-D (x,y,z). Cartesian 1
3-D (x,y,o(p)), slgma pressure 2
3-D (x,y,0(z)), slgma terrain 2
Objective analysis 0
Diagnostic with mass continuity 1
Diagnostic with primitive equation 2
Without drag coefficient 0
With drag coefficient 1
Without blocking effects 0
With blocking effects 1
Without kinematic effects 0
With kinematic effects 1
Without mountain wave effects 0
With mountain wave effects 1
Without upslope/downslope effects 0
With upslope/downslope effects 1
Reliance on surface and upper-
air meteorological data for
solution 0
Reliance on surface and upper
air meteorological data for
Initial/boundary conditions 1
Reliance on synoptic-scale wind 2
4-14
-------
The degree of reliance on observed data 1s the final category for the
diagnostic model technical merit analysis. One of the models, the SAI/
MVWM, requires complicated three-dimensional Interpolation techniques to
derive the temperature and wind fields for Initialization. This high
reliance on Input data 1s an undesirable model feature for predicting wind
fields 1n the data-sparse Rocky Mountain region. Models that rely on
observed data for their solution score lower than those models who only
require observations for Initialization. The highest scores for the Data
Reliance category are given to those models that only require synoptic-
scale wind data as Input.
Using the scoring system given 1n Table 4-5, we obtain the technical merit
scores for the seven diagnostic models for each category and subcategory
given 1n Table 4-6. This technical merit analysis 1s not without a degree
of subjective Interpretation regarding the models' ability to predict wind
fields 1n the Rocky Mountains region. Some of the models (e.g., LANL/
ATMOS1 and LLNL/MATHEW) have shown skill 1n calculating complex flows
based on meteorological observations during extensive measurement pro-
grams. However, these same models may not exhibit as much skill for
regions with sparse data networks.
The selection of the final candidate mesoscale meteorological models 1s
described 1n Section 4.3.
4.2 SELECTION OF ACID DEPOSITION MODELS
The models surveyed 1n Chapter 3 were analyzed according to their techni-
cal merits to select those models most appropriate for application to a
mesoscale region that Includes the Rocky Mountains. This analysis 1s
described here. In addition, we discuss evaluations of some of the most
suitable models as well as other selection factors, such as computational
efficiency, flexibility, adaptability, and Input requirements. The selec-
tion of the final candidate add deposition models 1s presented 1n Section
4.3. However, before the model selection criteria and results are pre-
sented, 1t 1s Important to look at the Intended uses and constraints of
the model.
4.2.1 Desirable Attributes of a Rocky Mountain Add Deposition Model
One of the primary purposes of the proposed model 1s the prediction of wet
and dry add deposition In the complex terrain of the Rocky Mountain
region. This Includes calculation of the contributions to add deposition
4-15
-------
TABLE 4-6. Technical attribute scores for diagnostic models.
Cft
Category
Equations
Parameterization
Phenomena
Data Reliance
Total
Subcategory CIT/UIND
Coordinate System
Basic fora
Friction
Blocking effects
Kinematic
Mountain waves
Upslope/downslope
2
1
0
0
1
0
0
1
5
USFS/KRISSY LANL/MHOS1
2
1
0
0
1
1
0
0
5
2
1
0
0
1
0
0
1
5
LINL/HATHEU
1
1
0
0
1
0
0
1
4
CAMM/KUATMOS
2
1
0
1
1
0
0
1
6
PNl/Melsar-MET SAI/CTWM SAI/MVWM SRI/01 agnostic
2
1
0
1
1
0
0
0
5
I
I
I
I
1
0
0.5
2
7.5
1
2
1
1
1
1
1
0
8
2
1
0
0
1
0
0
0
4
-------
1n the Rocky Mountains of sources such as those enumerated 1n Section
1.2. In the eastern U.S. and Canada the calculation of total and Indi-
vidual source contributions to add deposition has focused primarily on
annual average wet sulfate deposition. However, as noted by the World
Resource Institute (Roth et al., 1985) and other researchers (e.g., Lat1-
mer et al., 1985a,b), nitrate, as well as sulfate, 1s a significant and
Increasing component of wet and dry deposition 1n the western U.S.
Another desirable feature of the model Is the ability to calculate PSD
Increment values as mandated by the Clean A1r Act of 1977. The allowable
Increments, shown 1n Table 4-7, Include short-term concentrations, which
are generally the most limiting Increments. Current modeling techniques
generally use a steady-state Gaussian plume formulation, which restricts
source-to-receptor distances to well under 50 km. Depending on the choice
of modules, the Rocky Mountain Add Deposition Model could calculate PSD
Increments In Class I areas at large distances downwind from the proposed
new sources.
The PSD air quality requirements are not limited to atmospheric concentra-
tions of pollutants; the Impacts of new sources on the A1r Quality Related
Values (AQRVs) of a Class I area must also be Identified. The only AQRV
specifically Identified 1n the Clean A1r Act 1s visibility. However, the
Forest Service has Identified several other AQRVs as wilderness attributes
that can be affected by changes 1n air.quality (Table 4-8). The Forest
Land Managers (FLM) have the responsibility to ensure that no new sources
will adversely affect any AQRV 1n areas under their Jurisdiction.*
Clearly a mesoscale add deposition model cannot determine all the effects
to all the AQRV Indexes listed 1n Table 4-8. However, results of work
relating visibility to concentrations of sulfate, nitrate, and other fine
particles (e.g., Latlmer et al., 1985a,b) and work on forest canopy models
show that output from a correctly formulated add deposition model can
help the FLM evaluate the possible effects of new sources on AQRVs.
The Rocky Mountain Add Deposition Model must also be computationally
efficient. Because the model will be Installed on the EPA's mini-computer
and must be transferable to other federal and state agency mini-computer
systems, those models that run only on the most sophisticated computer
systems (Cray, Cyber 205) may not be appropriate.
4.2.2 Technical Merit Analysis
Because the Intended application primarily concerns pollutant deposition,
23 of the surveyed models 1n Chapter 3 were rejected outright as a result
* An adverse effect 1n this context 1s any measurable effect.
4-17
-------
TABLE 4-7. Allowable PSD Increments.
Species
Sulfur dioxide
TSP
Averaging
Time
Annual
24-hour
3-hour
Annual
24-hour
Concentrations (vg/m )
Class I
2
5*
25*
5
10*
Class II
20
91*
512*
19
37*
Class III
40
182*
700*
37
75*
* Not to be exceeded more than once a year.
4-18
-------
TABLE 4-8. Indexes of a1r-qua!1ty-related values (AQRV) and
possible effects of changes 1n air quality.
Indexes
Effects
Flora (plants)
Fauna (animals)
Soil
Water
Visibility
Cultural-archeologlcal
(structures and petroglyphs)
Geologic
Odor
Growth
Mortality
Reproduction
Diversity
Visible Injury
Succession
Productivity
Cation exchange capacity
Base saturation
PH
Structure
Metals concentration
PH
Total alkalinity
Metal concentrations
Anlon and cation concentrations
Contrast
Visual range
Coloration
Decomposition rate
Decomposition rate
Odor
4-19
-------
of their neglect of wet deposition and/or dry deposition processes and are
thus not Included 1n the following analysis. The eliminated models
Include the urban air quality models (CIT/UAPM, SAI/AIRSHED), several
mesoscale models developed primarily for dispersion applications (DM/RADM,
SRL/SPM), and models for which no wet deposition treatment was reported 1n
the surveyed literature. However, the PNL/MELSAR-POLUT model was retained
because 1t contains unique modules that describe transport and dispersion
specifically for the Rocky Mountain region.
Of the 52 models considered here, 16 have been classified as long-term
models only. These models generally utilize c11matolog1cal averages or
statistical techniques that are hard to compare with the modeling method-
ologies utilized by the deterministic models (capable of short-term
averaging). In a complex-terrain setting, cl1matolog1cal averages may not
be valid for all cases. In addition, several of the features proposed for
the Rocky Mountain model (e.g., PSO Increments and visibility calcula-
tions) are not obtainable with the calculation of long-term averages
only. However, since much of the recent work 1n the eastern U.S. has
focused on long-term averages, these models are Included 1n the technical
merit analysis.
For evaluating the technical merits of the models, three major categories
were Identified, corresponding to the modeling methods discussed In Chap-
ter 3 and listed in Table 3-13:
Transport and dispersion
Physical and chemical transformation
Removal processes
Each category 1s divided Into subcategoHes that Identify physical and
chemical processes or Individual model components. Each subcategory 1s
represented by a 11st of common modeling methods used to address the pro-
cess under consideration, and these methods are ranked from least to most
sophisticated (Table 4-9).
Scoring within each subcategory 1s based on the level of sophistication or
level of detail Incorporated Into the model. The Intervals between scores
are different for different subcategorles, reflecting weighting factors
applied to the subcategorles. This method permits scores to be summed
directly to obtain a total score. The weighting factors are equal for the
subcategorles of horizontal transport, mixing depth treatment, chemistry,
species treated, and dry and wet deposition, reflecting the Importance of
these processes or Issues 1n addressing add deposition over mesoscale
distances 1n complex terrain. Under these conditions, horizontal trans-
port 1s more Important than horizontal and vertical dispersion and thus
4-20
-------
TABLE 4-9. Technical attribute categories, subcategorles, and modeling methods
for candidate add deposition models.
Category
Subcategory
Sample Modeling Method
Score
TRANSPORT AND
DISPERSION
Horizontal transport
Horizontal dispersion
(near source)
Horizontal dispersion
(large travel time)
Constant transport across 0
region (may be seasonally
variable)
2-d1mens1onal transport 1
(spatially and temporally
variable, one layer or
level only)
2-d1mens1onal nondlvergent 2
(Including multiple layers)
2-d1mens1onal transport (ver- 3
tlcal flux Implicit through
divergence consideration)
or 3-d1mens1onal transport
Instantaneous near-source 0
dispersion (applicable to
puff and grid models)
Near-source dispersion time 0.5
or downwind distance varies
Independent of stability
Near-source dispersion time 1
or downwind distance dependent
on stability
Buoyancy Induced dispersion 2
effects
Enhanced dispersion due to 3
wind shear and complex
terrain effects
Large travel time dispersion 0
Indistinguishable from
near source*
Large travel time dispersion 1
distinguishable from near
source*
Continued
4-21
-------
TABLE 4-9 (Continued)
Category
Subcategory
Sample Modeling Method
Score
Vertical transport
Emission release
height
Trajectory-Independent**
Trajectory-dependent**
Additional complex terrain
effects
No vertical transport
Explicit w component of
wind vector
Cloud transport
Large-scale transport
(explicit 3-d1mens1onal) with
Interactive cloud transport
or complex terrain effects
Vertical dispersion Uniform mixing
Near-source dispersion
stability Independent
Near-source dispersion
stability dependent
Vertical diffusion equation
solved
No release-height emission
Emissions segregated by height
categorically (or separate
day/night treatment)
Emissions segregated by height
via plume rise calculation
0
1
2
0
1
1
2
0
0.5
1
1.5
1.0
Mixing depth
treatment*
Spatially and temporally
constant
0
Continued
4-22
-------
TABLE 4-9 (Continued)
Category
Subcategory
Sample Modeling Method
Score
PHYSICAL AND
CHEMICAL
TRANSFORMATION
Chemistry
Species treated
Spatially variable only 1
(seasonal variation Included)
Space-time variation 2
(not directly from dally data)
Space-time variation 3
(derived from observations
or boundary-layer model)
Radioactive decay only (no pro- 0
ducts)
Constant S02 or NOX 1
oxidation rate
Simple S02 or NOX oxidation 2
rate expression (sinusoidal
diurnal and/or seasonal
variation)
S02 or NOX oxidation rate 2.5
depends on solar zenith angle
Variable S02 or NOX oxidation 3
rate (rate constant depends on
urban Influence, cloud cover,
humidity, simple aqueous phase
effects
Gas phase photochemistry 5
Gas and aqueous phase 6
chemistry (sophisticated)
Inert species only 0
Linear decaying species only 0.5
SOg, sulfate only with 1
linear oxidation
Continued
4-23
-------
TABLE 4-9 (Continued)
Category
Subcategory
Sample Modeling Method
Score
REMOVAL
Dry deposition
Wet deposition
SO?, sulfate, and NOX to
nitrate (or nitric add) with
linear oxidation
sulfate, NOX, nitrate or
nitric add, or other nitro-
gen products with simplified
nonlinear oxidation
Multiple reactive species
Linear decay
Deposition velocity concept
(constant value)
Deposition velocity concept
(time varying— diurnal or
seasonal)
Deposition velocity concept
(land use dependent— space
and time variability)
Resistance approach (time- and
space-varying aerodynamic and
surface resistance)
Linear decay (constant
1n space and time)
Spatially constant,
seasonally dependent, or
stochastic (dry-wet period
duratlgij utilized or other
means)
4
0
1.5
2.5
Space-time variation (short-
term precipitation rate
dependent); linear (Irreversible)
decay for gas and aerosol)
Continued
4-24
-------
TABLE 4-9 (Concluded)
Category Subcategory Sample Modeling Method Score
Space-time variation (as above) 3
with reversible gaseous
scavenging
Detailed cloud mlcrophyslcs 4
^ Deterministic models
CUmatologlcal models
* Score 0 1f pollutant mass lost through fluctuating 1n mixing depth. I.e.,
1f mixing depth varies, pollutant should be kept track of.
4-25
-------
receives greater weight. Although vertical transport 1s very Important 1n
complex terrain and 1n cumulus clouds, relatively few studies have addres-
sed either of these Issues. Additionally, few models Incorporate a verti-
cal transport mechanism (except for Implicit vertical motion due to two-
dimensional divergence). Thus the vertical transport subcategory also
receives less weight than the horizontal transport subcategory. In com-
plex terrain there 1s frequent decoupling of the locally driven thermal
flows from the synoptic flows above an Inversion. Thus the correct
specification of the mixing height also receives a relatively high
weight. Since the primary goal of the model 1s the prediction of wet and
dry add deposition, these subcategorles also receive a relatively high
weight.
Transport and dispersion. Seven subcategorles have been Identified 1n
this category—three primarily address the horizontal dimension, three
address the vertical dimension, and one 1s concerned with both dimen-
sions. Sample methods by which models treat horizontal transport range
from the highly simplified uniform transport to single-layer and multiple-
layer transport. Single-layer wind fields (spatially and temporally vary-
ing) may be divergent or nondlvergent. They are weighed the same since a
single-layer model formulation will not conserve pollutant mass lost
through the model's Nl1d" due to convergence. Multiple-layer models are
scored differently depending on whether or not divergence (and the resul-
tant mass flux) 1s permitted.
Horizontal dispersion 1s subdivided Into "near source" and "large travel
time" primarily because of the need to address mesoscale as well as
regional transport 1n the Rocky Mountains.
The "vertical transport" subcategory was designated 1n order to assign
merit to the models that use fully three-dimensional wind fields or cloud
mass flux for pollutant transport. However, the absence of sophistication
among most surveyed models regarding vertical transport and the lack of
sufficient Information documenting Its Importance result 1n a lower
weighting factor than the horizontal transport category. Vertical disper-
sion and emission release height are considered Important for mesoscale
distances and hence 1s weighted as much as the vertical transport subcate-
gory. Treatment of the mixing depth 1s considered Important because of
Us close association with trapped pollutant concentration levels. Those
models utilizing radiosonde measurements or any real-time observation of
the mixing depth receive higher scores than those using the simple time-
averaged mixing depth specification or methods that specify mixing depth
variations using analytic or quasi-empirical expressions.
Physical and Chemical Transformation. In this category the subcategory of
"chemistry" receives a higher weight than horizontal transport to reflect
4-26
-------
the considerable Increase 1n sophistication of those models treating
detailed photochemistry and aqueous phase kinetics. The simplest chemis-
try modeling method (score • 0) treats radioactive or linear decay of a
primary pollutant only, I.e., no products are simulated. The next four
methods consider only linear reactions of SOX or NOX species. The level
of sophistication of these four modeling methods Increases as the varia-
bility of the transformation rate constant Increases. Simple temporal
variation (diurnal or seasonal), reflecting the availability of photo-
chemical reactive Intermediates, scores higher than a constant oxidation
rate specification. Latitudinal variation 1n oxidation rates (solar
zenith angle dependency), an Important consideration when simulating
transformations over regional and mesoscale scales, scores slightly higher
than a constant linear chemistry method. A variable linear transformation
rate that considers the effects of humidity, cloud cover, and urban emis-
sions scores still higher. Models that contain detailed gas-phase chemis-
try are scored 5; 1f, 1n addition, they contain aqueous-phase chemistry,
they receive a score of 6.
Within the "species treated" subcategory, models that treat only Inert or
linearly decaying species receive the lowest score. For add deposition
modeling 1n the West 1t 1s Important to treat conversion of N02 to nitric
add. Any model that uses linear oxidation of S02 to sulfate can be modi-
fled to linearly oxidize NOg to nitrate; however, evidence Indicates that
this oxidation rate 1s an order of magnitude faster than SOX oxidation.
Also, 1n view of the Importance of NOX 1n nonlinear photochemistry, the
linear simplification 1s of questionable merit, thus the Increase 1n score
for models that treat both SOX and NOX linear oxidation 1s restricted to
one unit. If the model treats the SOX and NOX oxidation with a simplified
nonlinear chemical mechanism, 1t receives a higher score, and 1f 1t treats
a full photochemical kinetic mechanism, 1t receives the highest score.
Removal. The most realistic method of parameterizing dry deposition 1s
through the resistance concept, which allows Incorporation of the effects
of lower atmospheric stability 1n addition to surface vegetation charac-
teristics (see Section 3.1). This modeling method therefore receives the
highest score. The more the methods listed 1n Table 4-9 resemble the
resistance method, the higher they are scored.
The manner 1n which a model treats wet deposition depends to some extent
on whether or not the model 1s of a cl1matolog1cal nature. Incorporation
of monthly or seasonal statistics of wet and dry periods provides more
realism than temporally constant removal rates. Utilizing precipitation
data for the temporal and spatial variability of wet deposition 1s even
more realistic. Models that recognize that gaseous scavenging tends to be
reversible, 1n contrast to aerosol scavenging (see Section 3.1), receive a
4-27
-------
slightly higher score than those that treat both processes as Irrever-
sible. Models that Include a detailed cloud mlcrophyslcs module are
ranked the highest.
Before presenting the results of the technical merits analysis, we empha-
size that the selection of representative modeling methods and their
scores are not without subjectivity. Several models use methods that He
between those delineated 1n Table 4-9, and thus received scores somewhere
between those listed. Some models use methods that are difficult to score
within the framework of Table 4-9 because the approach 1s substantially
different. An effort has been made to rank the sophistication of such
models as consistently as possible with that of the other models.
The scoring system reflects our desire to represent the average modeling
methods (from those models surveyed) by a mid-range score. A more objec-
tive scoring selection might assign a score of 10 to models utilizing
photochemical mechanisms because such mechanisms are far more sophistica-
ted relative to linear chemistry than linear chemistry 1s relative to no
chemistry. However, because few regional models possess photochemical
mechanisms, this degree of objectivity was not pursued.
All models are not equally familiar to us. While our knowledge of some of
the models 1s derived 1n part through our use of the models, most of the
model Information has been derived from the literature and from contact
with model developers. Because of these elements of subjectivity', the
technical merit analysis has been used to select a group of models for
further consideration. As will be shown within the next subsection, scor-
ing based on technical merit does not 1n and of Itself Indicate the
model's performance ability.
Table 4-10 lists the scores, by subcategory, of the 52 models under con-
sideration. Because the add deposition model selected for this project
must be able to use the wind fields generated by the mesoscale meteoro-
logical model, the cllmatological and statistical models have been elimi-
nated from further consideration. Based on technical merit alone, the 10
deterministic models with the highest scores are as follows (from highest
to lowest):
Eulerlan
NCAR/RADM
ERT/AOOM
UK/STEM-II
EPA/ROM
SAI/RTM-III
SAI/RTM-IINL
4-28
-------
TABLE 4-10. Technical attribute scores for short-term/long-term and long-term only models
considered for application to acid deposition In the Rocky Mountains.
ro
Transport and Dispersion
JL
Physical/chemical Removal
Transformation Processes
^ ^ -? ^ :
Model
AES/LRT
ANL/ STRAP
ANL/ASTRAP
ARL/ATAD
ARL/BAT
ARL/MTDF
BNL/AIRSOX
CAPITA/MCARLO
CCIW/LDAPTM
CEGB/TD
CERL/LTSDM
CSU-EPA/NODEL B
EI/EDRAB
EPA/ROM
ERT/ADOM
ERT/OISCDEP2
ERT/MESOPUFF
ERT/MESOPUFF-II
IAP/LRT
KNMI/MAPM
LLL/AOPIC
HEP/ TRANS
•v
*y *j
/£
^*?
^ 4
2.5
1
1
1
2
1
1
1
1
1
1
2
2
3
3
1
1
2
1
2
3
1
•*
'//
?g *
0
1
1
0.5
0.5
0.5
1
0
1
0.5
0.3
1
0.5
1
0
0.5
1
1
1
0
1
1
^^^
h. 4ft I
7/,
$g 4
0
1
1
0
0
0
1
0
0
0
0.3
1
0
1
1
0
1
1
0
0
0
1
fe
'/
**£
0
0
0
0
0
0
0
0
0
0
0
0
0
2
2
0
0
0
0
0
1
0
/**
£*?
tm V
£
^Sf
f^r
^^t
i
2
2
1
0
3
2
1
1
1
1
2
1
3
^
1.5
1.5
^
2
*
1
2
§ £
r «j
V*
^^
00 '
^^
A*
^**
2.5
2.2
2.5
2.5
0
2.5
2.5
1.5
2.5
1.5
1.5
2.5
2.5
2.5
^
1.5
0
2.5
1.5
2.5
0
2.5
Cont
«
,,
^
12.0
15.7
16.0
11.5
8.0
14.0
17.0
10.0
10.0
13.5
8.6
15.0
12.0
28.0
29.0
11.0
12.5
21.5
12.0
15.0
12.5
13.5
inued
-------
TABLE 4-10. Continued.
Physical/chemical
Transport and Dispersion Transformation
^
r
^
^£* ^ "
fii / / / *
VJ ^V
^^* ^^te
1
1
6
1
1
1
1
2
3
0
6
3.5
4
2.5
5
2.5
0
1
2
2
1
vli
1
1
4
1
1
1
1
1
1 .
0.5
4
2.5
3
1
4
1
0.5
1
1
1
1
«?
0
0
4
0
1
1
1.5
1.5
1.5
1.5
4
3
4
4
4
4
0
1
4
4
1
«u
*
0
0
4
0
1.5
0
2.5
2.5
2.5
2.5
4
3
3
3
3
3
0
2
2
2
2
^
4.0
4.0
29.0
3.0
6.0
5.0
14.0
13.0
17.0
11.5
25.0
21.5
25.0
21.5
27.0
17.5
3.5
7.0
11.0
17.5
6.0
-------
TABLE 4-10. Concluded.
Physical/chemical Removal
Transport and Dispersion Transformation Processes
SRI/EURMAP-2
UI/RCOM-3
UK/STEM2 30120236444 29.0
UHACIO 2 0 1 00 012 1 3 2.5 12.5
UH/STOCHAC1D 20000031102 9.0
US/RDDH 1 0 0.5 0 0.3 0.3 1 0 01 1.5 5.6
USSR/LRT 200000311029.0
UHATM-SOX 20 0 0 1.5 2 32 1 4 2.5 18.0
PNL/MELSPR-POLUT 3 3 2 21 2 30 0 0 016.0
-------
Lagranqlan
SAI/CCADM
ERT/MESOPUFF-II
SAI/RIVAO
PNL/MELSAR-POLUT
Selection of the most appropriate model should be based on considerations
other than technical merit, especially considering the subjectivity
Involved 1n this type of screening. The following subsections deal with
some nontechnical attributes of the 10 models with the highest technical
scores. These nontechnical attributes Include past application, past per-
formance evaluations and availability.
4.2.3 Past Applications
Due to the similarities between the SAI/RTM-IINL and Its predecessor, the
SAI/RTM-II, these two models will be considered as one model 1n evaluating
the nontechnical attributes of the 10 highest-scoring add deposition
models. This 1s also true for the UK/STEM and UK/STEM-II. Table 4-11
lists the past applications of the 10 candidate models as reported 1n the
published literature. Due to the recent development of the highest rank-
Ing models, NCAR/RADM and ERT/ADOM, the number of documented applications
1s very limited. The UK/STEM-II has had several 1-day to 3-day simula-
tions that focused on Individual processes. The EPA/ROM has been applied
only to a region containing the northeast corridor of the eastern U.S.,
and the SAI/CCADM and PNL/MELSAR-POLUT have only one documented applica-
tion. The SAI/RTM series of models has the largest number of applica-
tions, with a large range of conditions. The ERT/MESOPUFF-II has only one
documented application; however, the predecessor model, ERT/MESOPUFF, has
been applied to several different situations.
4.2.4 Model Performance Evaluation
Of the 10 models selected from the technical merit analysis, only the
SAI/CCADM and the PNL/MELSAR-POLUT documented performance evaluation. The
lack of a documented evaluation 1s not sufficient cause to eliminate a
model from further consideration. However, 1t 1s useful to examine the
performance of the other models that have been evaluated so that further
comparison 1s possible.
4-32
-------
TABLE 4-11. Past applications of the 10 highest scoring acid deposition Models.
CJ
Model
EPA/ROM
ERT/AOOM
ERT/MESOPUFF-II
Reference
Schere and Posslel, 1984
Schere, 1986
Venkatran, Sclre. and
Ple1». 1984
MPCA, 1985
Study
Application to a mil t1 day oxldant episode 1n the
northeast corridor of the eastern United States
Application to add deposition episode, July 1-
14. 1980, eastern United States and Canada
Application and evaluation— a nesoscale region
NCAR/RADM
NCAR, 1986
PNL/HELSAR-POLUT Allwlne and Whitman, 1985
SAI/RIVAO
SAI/CCADH
SAI/RTM-III
Latlner et al., 1985a,b
Latlaer et al., 1984
Systems Applications, Inc.,
1987
Stewart, Morris, and
Reynolds, 1987
Yocke, Morris, and Haney,
1985
containing the state of Minnesota for 1980
and 1981
Two 3-day applications to the eastern United
States during a period of oxldant scavenging
characteristic of April rains (OSCOR
experiments I and IV)
Application to a complex terrain region 1n the
western Rocky Mountain region
Application and evaluation--the western
United States for 1981
Applications and evaluation—the eastern
United States states for 1978.
Application for a single trajectory over the
eastern United States
Application and evaluation—a 5-day oxldant
episode 1n western Europe
Effect of urban hydrocarbon emissions on rural
ozone 1n a nesoscale region containing
St. Louis, Missouri
Continued
-------
TABLE 4-11 (continued)
Model
Reference
Study
CO
Uu, Morris, and Klllus.
1984
SAI/RTM-IINL Stewart et al., 1986b
and SAI/RTM
Stewart. Morris, and L1u,
1985
Stewart et al.. 1983a
Latlner et al., 1984
Systems Appl1catIons, Inc.
1982
Yocke, Ruchlls, and L1u,
1981
L1u and Ourran, 1977
UK/STEM-II
Camlchael, Peters, and
Kltada, 1986
Application and evaluation—an 8-day oxldant
episode 1n the eastern United States
Application and evaluation—a regional-scale
region of the midwest and eastern United
States and a mesoscale region containing the
State of Minnesota during the year 1980.
Application to four complex-terrain mesoscale
regions containing national parks 1n the
eastern United States for periods of 1-2 weeks
during 1978
Application and evaluation—a 9-day sulfate
episode In the eastern United States during
July 1978
Application and evaluation—the eastern
United States for 1978
Application and evaluation—source-receptor
relationships 1n the Ulhta Basin of the Rocky
Mountains
Application to a mesoscale region containing the
state of North Dakota
Development and application to the Great Plains
region
One-day, 2-dlmenslonal simulation of sulfate
formation within convectlve clouds
Continued
-------
WBLE 4-11 (concluded)
Model Reference Study
KHada, Canalchael, and Three-day, 2-dimensional simulation of sulfate,
Peters. 1984, 1986 nitrate, and oxldant formation In the presence
of a land/sea breeze
UK/STEM-I Balko and Peters, 1983 Transport of S02/NOx/hydrocarbon urban plumes
to the background troposphere
Canalchael and Peters, Regional-scale transport of SO? and sulfate for
1984b a 3-day period 1n the eastern United States
Dronamraju, 1986 Application to the eastern United States for a
6-day period 1n July 1974, and a 7-day period 1n
July 1978.
•£»
I
-------
Comparing the skill of regional models 1s not a straightforward task.
Many models are evaluated 1n a "hands-off" mode, which often does not do
justice to the model. Others are evaluated by the developers or experi-
enced modelers who are aware of the Importance of the quality and
Integrity of Input data. But even when a group of models 1s evaluated by
experienced modelers, the skill of the models 1s often not comparable
because the statistics chosen for evaluations of Individual models are not
consistent.
Various performance measures for judging model performance have been
recommended. One survey of statistical measures (Moore, Stoeckenlus, and
Stewart, 1982) presents a hierarchy of performance measures that have been
utilized for various types of models. Fox (1981) suggested a series of
performance measures and model evaluation strategies that have been
adopted by many regional modeling groups. A comprehensive set of perform-
ance measures has been recommended by the NCAR (1983) as well. Wlllmott
(1982) explained why several performance measures should not be Inter-
preted alone because of possible misinterpretation, and suggested appro-
priate combinations of statistics. Finally, the US-Canadian Transboundary
Air Pollution Modeling Workgroup developed a standard set of performance
measures for comparing eight regional-scale models (MOI, 1982).
Table 4-12 presents general Information on the evaluation studies of the
four models considered appropriate for application to the Rocky Mountain
region. Because of similarities between SAI/RTM-IINL and Its predecessor
SAI/RTM-II and between UK/STEM-II and UK/STEM, evaluation studies for both
versions of these models are presented.
Table 4-13 lists a set of commonly selected statistical performance mea-
sures for annual, monthly, and hourly ambient SO^j , S02, and 03 concentra-
tions and total wet sulfur deposition. Since the evaluation data bases
generally pertain to different time periods and spatial scales, no direct
comparisons of model performance can be made. The statistics do, however,
reveal certain characteristics of the models' performance abilities. For
example, It 1s evident that prediction of annual total wet sulfur deposi-
tion 1s more skillful than that of annual average ambient sulfate concen-
trations, which In turn 1s more skillful than that of annual average S02
concentrations. This 1s typical of a number of regional models (Stewart
et al., 1983b; Sch1erme1er and M1sra, 1984).
The data 1n Table 4-13 Indicate the difficulty encountered when attempting
to select the most skillful model on the basis of Independent evaluation
studies. The episodic and long-term evaluation studies by Stewart and co-
workers (Stewart and L1u, 1982; Stewart et al., 1983) and another study
sponsored by EPRI (Ruff et al., 1985) represent attempts to compare model
performance using consistent Input data from the same period of record,
4-36
-------
TABLE 4-12. General information on evaluation studies of final candidate add deposition models.
Model
SAI/MTM-III
Reference
(•) Uu. Morris ft
Mllus. 1984
Data Base
SURE. 8 class I
air quality sites
03. S02, SOJ.
NO, N02
Monitoring and
Modeling Period
16-23 July 1978
Averaging
TIM
Hourly
Spatial
Scale
2080 km x
1840 km
(80 km grid
cells)
Umber of
Sources
Grldded area
sources, f 400
•ajor point
sources
I
OJ
(b) Reynolds et al. Europe, up to 39
1986 air quality sites
(03, SOj. NO. N02.
CO)
22-26 July 1980
Hourly Approximately Grldded area
2232 km x sources, f 500
1388 km Major point
sources
SA1/RTH-II
SAI/RTH-IINL
UK/STEM- I
UK/STEM- II
SA1/CCAOH
NCAM/RAUH
Stewart et al.
1983a
. SURE, up to 54
S02 and SOJ sites
Stewart. Morris ft SAROAO. up to 1321
Vocke. 19866 SOj air quality
sites;
NAP3S. NADP. UAPSP.
up to 77 wet depo-
sition sites (SO?.
S02. NOj)
(a) Dronaaraju
(b) Oronawaju
(No evaluation
(No evaluation
NCAR. 1986
. 1986 SURE. 12 air quality
sites (S02. SOJ)
. 1986 SURE. 54 air quality
sites (SOg. SOJ)
documented)
doomented)
Eastern United
States and Canada.
8 MAP3S wet sulfate
deposition sites
16-23 July 1978
January 1980 to
December 1980
4-9 July 1974
15-21 July 1978
Two 3-day epi-
sodes In April
1981 (OSCAR I
and IV)
24 hours 2080 km x
1940 km
(40 km grid
cells)
Monthly Approximately
and 3183 km x
annual 3552 km (f 100
km grid cells)
2080 km x 2480
km (80 In grid
cells)
2080 km x 2480
km (80 km grid
cells)
Event by 2160 km x
event 2160 km
(80 km grid
cells)
Grldded area
sources, f 400
•ajor point
sources
Grldded area
sources, f 2000
•ajor point
sources
Grldded Missions
stratified by
release height
Grldded missions
stratified by
release height
Grldded emissions
with vertical
variations
Continued
-------
TABLE 4-12.. Concluded.
Model
ERT/ADOM
Reference
Venkatram, Sclre A
Plelm. 1984
Data Base
Eastern United
States and Canada,
(SURE) 8 MAP3S wet
sulfur deposition
sites
Monitoring and
Modeling Period
14-day episode
in July 1980
Averaging
Time
Event by
event
Spatial
Scale
2400 km x
2080 km
(80 km
grid cells)
Number of
Sources
Grldded emissions
with vertical
variations
SAI/RIVAU
(a) Latlmer et al..
1904
SURE, up to 54 S02
and SOj sites;
MAP3S up to 7 Met
deposition sites
(b) Latlmer et al..
1985a
OJ
CD
ERT/NESOPUFF-1I MPCA. 1985
NPS. 18 bext;
WAQS, 7 bscat.
10 bext + SOj;
«>CS. 24 §64;
NADP. 19 wet sulfur
deposition
NADP, up to 4 sites
In Minnesota,
wet sulfur depo-
sition
October 1977 to
September 1978
1980
1981
Monthly 2080 km x 66 grid cell
and 1840 km emissions within
annual (80 km region (every
emission 9th grid, results
grids) scaled for
remaining 8
emission grids)
60 emission source
areas: 43 point
sources (13 smelters
and 30 power plants)
and 17 area sources
(urban areas and
energy production
areas)
611 km x ~ 230 emission
628 km source regions
January to
December 1981
Monthly
and
annual
2200 km x
1600 km
(100 km
grid)
Monthly
and
annual
Abbreviations:
SURE: Sulfate Regional Experiment sponsored by Electric Power Research Institute (EPRI).
MAP3S: Multistate Atmospheric Power Production Pollution Study sponsored by the Department of Energy and Environmental
Protection Agency.
CANSAP: Canadian Network for Sampling Precipitation sponsored by Atmospheric Environment Service
and Environment Canada.
APN: Air and Precipitation Monitoring Network sponsored by Atmospheric Environment Service and
Environment Canada.
APiUS: Acid Precipitation In Ontario Study sponsored by the Ontario Ministry of the Environment.
NAOP: National Atmospheric Deposition Program sponsored by the U.S. Environmental Protection Agency.
UAPSP: Utility Acid Precipitation Study Program sponsored by EPRI.
-------
TABLE 4-13. Suomary of performance statistics for models ranked highest in the technical merit analysis
(UK/STEMI and SAI/CCADM have no documented evaluations). See Table 4-14 for explanation of "a" and "b."
SM/IBMlt
SAI/IRMINL/IM-II
SM/IIVIO
«*ruM* St*ti«tic
U»
Ck)
(•!
(k)
U)
(k)
UK/SfOM
tn/Msonrr-u MCM/MM CM/IDM '• u> <•)
*« sc<
r
•.} to • M> -«.•» to . 11) —
1.4 •.»! —
•.M •. 71 —
Ml total k lh|/1u)
', wlfw
»; s«
r
•.»?
X.C
-1.7 to • 7) -«.•• to • It) •.! (• . 4)
•.«» «^«
•.71 I
l.« '
•.II
i.n
•.M
1.4 to « 1»M) -».1 to • »H —
IkjKtfcly Ut*l
•^ MtfV
•.» to • 71) 4.B7 to • MIJ •.•! to • 71) —
r
k
•* Mlf«r >
CM* k,
-I.I (• • 4M) —
11.1 -4.i
— M.t
•.M e.»o
Iburly wr k
14-kM* SB, St
trctun
!.• (• * MM) —
-•.4
0.4»
•.*»
-a.o -».*
— 44.4
•.71 0.11
•.7. to > IMi) 1.4 to a mt) ~
II.1
•.M ».4J —
k • III*. «.f«A«l M tiw Mr«|* n«*Ml. •*•» tta n«i*Ml i« tk> i
otntion tor
mini* tta
xtratim tor da|>oaitian).
St « StMbwil itoviMiun .r Ik* rMtoyal*. rttich in4ieatM tlw tcoracy »r mt*l frcdictun*. SMllar Sc iMlicatca grMt>r •ecirwy Uwn Urg* Sj.
-------
evaluation data bases, and performance measures. All the models analyzed
by Stewart and co-workers (I.e., SAI/RTM-II, SAI/RTM-LT, ERT/MESOPUFF,
ARL/ATAD, SRI/ENAMAP-1A. and MIT) showed comparable skill. However, the
Eulerlan models (SAI/RTM-II and SAI/RTM-LT) showed slightly greater over-
all skill than the Lagranglan or analytical models. The EPRI study also
showed comparable skill for all models (SAI/RTM-II, SRI/ENAMAP-II, and
UMACID), though the overall skill of these models was much worse than that
of the models analyzed by Stewart and co-workers even though some of the
models and the period of simulation were the same as those studied by
Stewart. The UK/STEM-I was also run during the same simulation period
used 1n the above two comparative evaluation studies and demonstrated a
level of skill between that established by the two studies (Dronamarju,
1986). This discrepancy 1s most probably due to the methods used to pre-
pare the Input data. In the EPRI study an attempt was made to prepare In-
puts 1n a hands-off manner using existing preprocessor programs.
The technical merit analysis, the comparison of model performance, and the
documented skill levels of many regional models suggest that each of the
candidate models might be appropriate to application 1n the Rocky Moun-
tains. However, one of the most Important criteria for an add deposition
model 1n this project 1s that 1t be currently available. This Issue 1s
discussed 1n the next section.
4.2.5 Avatlability of the Models Reviewed
Several of the 10 highest-scoring models are considered research models
and are not currently available for routine applications. Five of the
models-the NCAR/RADM, ERT/ADOM, UK/STEM-II, SAI/CCADM. and EPA/ROM-are
newly developed models that have not been fully tested. Of these five,
only SAI/CCADM 1s currently available for this project. The UK/STEM-II
model has a new cloud m1crophys1cs, wet deposition, and aqueous-phase
chemistry modules and has not been released by the developer; however, the
UK/STEM-I model, which now Includes a detailed gas-phase photochemical
mechanism but no cloud wet deposition or aqueous-phase module, 1s cur-
rently available. The remaining models—the SAI/RTM-III, SAI/RTM-II (and
IINL), ERT/MESOPUFF-II, SAI/CCADM, PNL/MELSAR-POLUT, and SAI/RIVAD—are
all currently available.
Because of nonavailability and the lack of comprehensive past model appli-
cations and evaluations, the following models have been eliminated from
consideration for the Rocky Mountain add deposition model: NCAR/RADM,
ERT/AOOM, EPA/ROM, and UK/STEM-II. Note that these models have been
eliminated because of nontechnical model attributes. When these models do
4-40
-------
become available and tested, they should be considered for the calculation
of add deposition. The seven final candidate add deposition models are:
Eulerlan
SAI/RTM-III
UK/STEM-I
SAI/RTM-II (and IINL)
Lagranqlan
SAI/CCAOM
ERT/MESOPUFF-II
SAI/RIVAO
PNL/MELSAR-POLUT
The unavailable UK/STEM-II has been replaced by the available version,
UK/STEM-I, which does not contain any wet-phase processes.
4.2.6 Available Candidate Models
The seven available models consist of three Eulerlan grid models and four
Lagranglan source-oriented models. The computational demands of Lagran-
glan puff or segmented plume models generally Increase linearly with the
number of emission sources considered because of the practice of super-
positioning the pollutant mass of each puff or plume segment. To perform
Lagranglan model simulations over large spatial areas, either a large num-
ber of sources must be considered or sources must be aggregated. Many of
the Lagranglan models exercised by the U.S.-Canadian Transboundary model-
Ing subgroup aggregated emissions by state, each state's emissions being
treated as one source emitted from the geographical center of the state
(HOI, 1982). This method 1s likely to Increase the errors 1n concentra-
tion and deposition predictions as the distance from sources to receptor
decreases.
The 1978 EPRI/SURE emission Inventory Includes more than 3000 major point
sources In the eastern third of the U.S. (Klemm and Brennan, 1981). In
their evaluation of many models, Stewart and co-workers (1983a) reduced
this number to 190 so that the computational demands of the ERT/MESOPUFF
and ARL/ATAD models were not excessive.* Reduction of major point sources
Four hundred point sources aggregated within a 2.5 km radius were
utilized by the third model (SAI/RTM-II).
4-41
-------
to 190 required the lumping of emissions within a radius of 40 km. This
loss of resolution 1s not too critical at source-receptor distances beyond
500-1000 km, but represents an unacceptable compromise at mesoscale dis-
tances. Thus, to simulate pollutant behavior over both mesoscale and
regional distances using the Lagranglan approach, aggregation of emission
sources must be performed 1n a spatially nonunlform manner, which adds to
the complexity of Input preparation.
Of course, grid models lose spatial resolution of emission sources as
well, unless major point sources are treated with a subgrld-scale disper-
sion algorithm. By their hybrid approach, the SAI/RTM-II, SAI/RTM-IINL,
and SAI/RTM-III possess such a capability. The computational demands
Increase with the number of sources treated 1n this manner, but the model
allows the user to merely Identify the sources requiring subgrld-scale
treatment. Therefore, the SAI/RTM series of models 1s more flexible and
less demanding from the standpoint of Input preparation.
Lagranglan models possess several additional limitations relative to
Eulerlan models:
Linear superpos1t1on1ng of the pollutant mass requires linear chemis-
try and linear removal rates. It 1s Incorrect to superimpose puff
concentrations calculated from nonlinear chemistry that 1s Indepen-
dent of neighboring puff concentrations.
Long-range transport of Individual coherent puffs or plume segments
becomes unrealistic as greater distances are considered because of
Increased "puff" size. In most puff models the puff-specific
parameters (e.g., deposition parameters) and the transport wind are
chosen from values at the puff centrold. This may be a serious
shortcoming 1.n complex terrain situations where wind patterns tend
toward uniformity.
Background pollutant levels due to diffuse area sources must either
be added to concentration and deposition predictions or Incorporated
Into emission sources along with additional assumptions pertaining to
Initial puff size and dispersion rates.
Eulerlan grid models are not restricted to linear chemical and removal
processes. Nor do they suffer .from unrealistic specifications of deposi-
tion parameters or transport winds because the spatial resolution does not
vary as the pollutants travel away from the source regions. Deformation
of pollutant parcels by spatially variable winds 1s automatically calcu-
lated. Eulerlan grid models also require no additional assumptions 1n
order to Incorporate diffuse emissions.
4-42
-------
However, Lagranglan models also have several advantages relative to
Eulerlan models:
If Impacts from only a few sources are required, a Lagranglan model
1s more cost effective than Its Eulerlan counterpart.
For short to medium travel distances, a Lagranglan model can have
higher resolution of Individual plumes since the plume resolution 1s
not limited by the grid spacing.
For short-term averages 1n complex terrain, a Lagranglan model can
more easily depict the relationship between the plume centerline and
the terrain features.
Lagranglan models do not suffer from numerical diffusion, which
results from the finite differencing techniques used 1n Eulerlan
models.
The ERT/MESOPUFF-II model possesses the option of allowing nonlinear
chemical reaction rates (via plecewlse linear approximation). This Is
accomplished by performing puff superpos1t1on1ng at frequent Intervals so
that the accumulated concentration at a fixed point 1n space may be used
to determine the appropriate nonlinear oxidation rate constant for the
subsequent time step. While this 1s a technically sound method of Incor-
porating nonlinear effects Into a Lagranglan model, the method requires
that all emission sources be simulated simultaneously, thereby potentially
limiting the total number of sources that can be accommodated by the
model.*
The SAI/CCADM uses nonlinear gas-phase and aqueous-phase chemistry
mechanisms by moving the Lagranglan box through an existing field of con-
centrations and entraining at a rate dependent on the wind shear and dif-
fusion rates. Thus, an Eulerlan reference grid of concentrations 1s
required as Input.
One advantage often assigned to the Lagranglan source-oriented modeling
approach 1s the computational ease with which the Incremental Impacts of a
single source and source attribution can be calculated. This advantage 1s
valuable for assessing the Impact of Individual sources on concentration
* Strictly linear Lagranglan models may simulate smaller sets of point
sources and superimpose the results as a postprocessing activity,
thereby having no I1m1tat1on~bn the number of sources that can be
accommodated.
4-43
-------
and deposition Impacts at sensitive receptors because 1t reduces the simu-
lation costs considerably. However, the assumptions that allow these cal-
culations to be readily made are those of linear chemical transformation
and deposition. With these assumptions, pollutant mass transformation and
deposition within each puff or plume segment are computed Independent of
neighboring or overlapping puffs.* These assumptions make Incremental
Impact and source attribution studies with Lagranglan models appropri-
ate. Currently, we do not know how uncertain calculations based on
Lagranglan models are, and how Important nonlinear effects may be. The
trade-off between computational efficiency and accurate treatment of non-
linearity may determine whether Lagranglan or Eulerlan models are prefer-
red.
Strictly speaking, source-attribution calculations are possible using non-
linear chemistry only 1f all emissions that Interact with the source 1n
question are Included 1n the simulation. The Impact of a single source or
group of sources requires parallel simulations—one with all sources
treated and one with all sources except those 1n question. The Impact of
the sources 1n question 1s obtained by subtracting the results of the
"full emission" simulation from those of the "partial emission" simula-
tion. Although this approach 1s computationally demanding, 1t 1s the
necessary approach when nonlinear affects are Important.
The advantages of Eulerlan models over Lagranglan models 1n terms of rigor
and flexibility are probably best exemplified by the fact that an Eulerlan
framework was chosen for the second-generation add deposition models cur-
rently under development, e.g., the NCAR add deposition model (NCAR,
1983). Complex, nonlinear gas- and aqueous-phase chemistry and deposition
processes are readily accommodated by the Eulerlan framework.
Although model Input requirements are a potentially Important criterion
for selecting an appropriate model for the Rocky Mountains application,
all of the models considered 1n this chapter require similar Input.
Models that demand excessive Input requirements are more research-oriented
and are not yet available for applications.
For this project the requirement 1s to develop an add deposition model
that can be applied by western regulatory agencies to Isolated sources
within the complex terrain setting of the Rocky Mountains. In order to
make a final selection of potential candidate models, advice from the
* The nonlinear transformation option Incorporated Into ERT/MESOPUFF-II
prohibits puff superposition and hence removes the economical source-
attribution calculation.
4-44
-------
potential users 1s required. In the next section the needs and desires of
users are discussed and factored Into the selection criteria to produce
the final candidate add deposition models.
4.3 ADDITIONAL SELECTION CRITERIA AND SELECTION OF FINAL CANDIDATE
MESOSCALE METEOROLOGICAL AND ACID DEPOSITION MODELS
The objective of the Rocky Mountain Add Deposition Modeling Assessment
Project 1s to assemble an add deposition model based on currently avail-
able air quality and meteorological models and modules that properly
describe the processes of add deposition 1n the Rocky Mountain region.
The model would be used by federal and state agencies 1n the western
states to assist In the PSD permitting process. This effort differs from
other add deposition model development efforts, such as the NCAR/RADM and
SAI/CCADM, 1n that the model 1s to be designed specifically for use and
application by western regulatory agencies to estimate add deposition
Impacts from proposed new sources, such as those discussed 1n Section
1.2. Clearly there 1s a trade-off between the most technically rigorous
method of treating all the physical processes that lead to add deposition
1n the western U.S. and the computational resources available to the regu-
latory agencies. Thus the detail 1n which the add deposition processes
will be treated will be balanced by the Importance of these processes to
add deposition in the West and the computational requirements and needs
of the potential users. In August 1986 a questionnaire was sent to the
western regulatory agencies and other Interested parties requesting com-
ments and recommendations on the Issues leading to the design of the
hybrid add deposition model. This document was sent to assure that the
selected modeling approach would meet the needs of the potential users.
Written responses were received from seven groups, and verbal comments
were received from several others. There were many diverse and worthwhile
comments on the model design, but 1t 1s only possible to meet a limited
number of these goals. However, there was enough of a consensus to aid 1n
the design of the modeling system. These recommendations can be summari-
zed as follows:
The primary objective of the model 1s to provide an Improved modeling
tool to estimate add deposition Impacts at sensitive receptors from
specific sources for PSD permits. Along with the need to determine
long-term add deposition Impacts, there 1s also a need to estimate
short-term (3-hour, 24-hour) S02 and TSP Impacts for PSD Increment
consumption. Thus the model should be primarily concerned with a
mesoscale region within the Rocky Mountain region.
The modeling system will Include a mesoscale meteorological model
that can generate wind fields within the complex terrain of the Rocky
4-45
-------
Mountains where monitoring data are Inadequate. Since cost con-
siderations preclude the use of a prognostic (dynamic) wind model, a
diagnostic (mass-conserving) or 1nterpo1at1ve model win form the
basis for the wind field generator.
Transport, diffusion, chemical transformation and deposition would be
handled by a Lagranglan model that uses the wind fields generated by
the mesoscale meteorological model. Since this model will only be
concerned with the specific sources within a mesoscale region of com-
plex terrain, many of the processes that are believed to be non-
linear, such as chemical transformation and wet deposition, will have
to be highly parameterized.
There was almost complete consensus of all parties who responded that
the modeling system must be evaluated. The meteorological model
should be evaluated in applications to both flat terrain and complex
terrain and compared against measurements and other meteorological
models. The Lagranglan air quality and deposition model should be
evaluated with and without the meteorological driver. During Ideali-
zed conditions (flat-terrain, uniform flow fields, no chemical reac-
tions), the predictions from the Lagranglan model should produce
results similar to those of the steady-state Gaussian plume models
currently used for obtaining PSD permits.
4.3.1 The Final Candidate Mesoscale Meteorological Models
Although the prognostic meteorological models based on the primitive equa-
tions are more technically rigorous than the diagnostic models, the addi-
tional cost and complexity eliminate them from consideration for this pro-
ject.
Cost considerations and data Input requirements also eliminate the
SAI/MVWM from further consideration. The SAI/MVWM 1s essentially a prog-
nostic model 1n which the temperature field 1s Input and held constant and
the model runs to steady-state conditions. The remaining eight diagnostic
wind models all produce mass-consistent wind fields, and, based on the
technical merit analysis discussed 1n Section 4.2, are ranked as follows:
SAI/CTWM
CAMM/NUATMOS
PNL/MELSAR-MET
CIT/WIND
USFS/KRISSY
4-46
-------
LANL/ATMOS1
LLNL/MATHEU
SRI/DIAGNOSTIC
The SAI/CTWM received the highest ranking of the diagnostic models because
1t contains parameterlzatlons of the kinematic effects of topography,
boundary layer effects, critical Froude number effects, and thermally
forced upslope/downslope flows. It also requires the least amount of
meteorological Input data, requiring only a region-wide mean flow for
boundary conditions. However, the model 1s also formulated 1n a Cartesian
coordinate system, which, as discussed 1n Chapter 2, 1s not desirable for
application to add deposition studies. The SAI/CTWM also does not
explicitly use observations except through the specification of Internal
boundary conditions that must be made on a grid. In an application mode,
the differences between the predictions and observations are minimized
through repeated runs and adjustments to the boundary conditions. Because
of Its ranking and unique parameterlzatlons, the SAI/CTWM will be retained
as a candidate mesoscale meteorological model for evaluation 1n Chapter 5.
The CAMM/NUATMOS, LANL/ATMOS1, and LLNL/MATHEW are all similar diagnostic
wind models that use the varlatlonal technique (Sasaki, 1958) that pro-
duces mass-consistent wind fields while minimizing the differences between
the predictions and observations 1n the least squares sense. Since the
CCAM/NUATMOS 1s the most recent of this family of models, and contains an
additional parameterization of spatial variations 1n Froude numbers, 1t
would be the most appropriate candidate for further consideration. How-
ever, 1t has not yet been released by the developers (Us release 1s
scheduled for spring 1987). Until Its release, the predecessor model
LANL/ATMOS1 will serve as the candidate diagnostic model with a varla-
tlonal formulation.
The remaining models are PNL/MELSAR-MET, CIT/WIND, USFS/KRISSY, and
SRI/DIAGNOSTIC. These models are similar 1n the sense that they all
Interpolate from available observation to obtain an Initial wind field,
and then use some sort of mass adjustment procedure. The PNL/MELSAR-MET
model not only produces a wind field but also produces grldded fields of
mixing heights, stability classification, temperature, pressure, and other
meteorological variables that are required by an add deposition model.
Since the PNL/MELSAR-MET was designed for applications 1n the western
Rocky Mountain region, 1t seems prudent to Include 1t as a final candidate
mesoscale meteorological model, despite the fact that 1t has not been
extensively tested. It 1s also the only one of the four remaining diag-
nostic models that contain blocking effects.
4-47
-------
Of the remaining diagnostic wind models, the CIT/WIND model has been
applied the most often under a variety of complex-terrain conditions. The
CIT/WIND has also shown Us versatility 1n merging observations with out-
puts from a primitive-equation prognostic meteorological model to produce
more realistic mountain-valley wind fields (Moore, Morris, and Daly,
1987). The USFS/KRISSY has been applied only to regions where field data
are extensive, while the SRI/DIAGNOSTIC model has only four applications
documented 1n the literature, and these applications placed little empha-
sis on modeling In complex terrain.
As a result of these considerations, the following four mesoscale meteoro-
logical models were selected as final candidates and are evaluated 1n
Chapter 5:
SAI/CTWM
PNL/MELSAR-MET
LANL/ATMOS1 (1n place of the currently unavailable CAMM/NUATMOS)
CIT/WIND
4.3.2 The Final Candidate Add Deposition Models
Given the additional, nontechnical criteria for model selection, all of
the EuleHan transport and diffusion models can be eliminated.' It should
be emphasized that these EuleHan models have been eliminated for nontech-
nical reasons. The elimination of the Eulerlan models 1s based on the
needs of the potential users to calculate source-specific add deposition
Impacts 1n a cost-effective manner.
Of the add deposition models surveyed 1n Section 4.2, the highest-ranking
Lagranglan models based on technical merit alone are as follows:
SAI/CCADM
ERT/MESOPUFF-II
SAI/RIVAD
PNL/MELSAR-POLUT
These models represent distinctively different modeling approaches for
.predicting ambient concentrations and add deposition Impacts within a
Lagranglan formulation.
The CCADM contains a detailed treatment of gas and aqueous-phase chemistry
while having a more simplified treatment of transport and dispersion. It
also uses a cloud physics module and the dry deposition algorithm employed
4-48
-------
by the NCAR/RADM. As the CCADM 1s the most complicated, costly, and 1s
least like the existing regulatory PSD models, 1t would probably not be
the most suitable model to serve as a basis for a regulatory model of add
deposition 1n complex terrain. However, the CCADM contains several
modules, Including the chemistry cloud physics and dry and wet deposition
algorithms, which should be considered for Integration Into the final
model framework.
The MESOPUFF-II and the MELSAR-POLUT are Lagranglan puff models that simu-
late a continuous plume by releasing circular puffs of pollutants at regu-
lar Intervals. They retain a Gaussian distribution 1n the horizontal and
vertical. In the MESOPUFF-II dry deposition 1s handled through a resis-
tance approach; wet deposition uses the scavenging coefficient; and chemi-
cal transformations are limited to first-order or pseudo first-order reac-
tions based on empirical data.
The RIVAD 1s a rectangular plume-segment model 1n which a continuous plume
1s represented by Interpolating between the discrete plume segments. When
RIVAD 1s used for regional-scale Impacts, uniform concentrations are
assumed within the plume-segment. However, the RIVAD does contain
algorithms that allow the pollutants to be represented by a Gaussian dis-
tribution 1n the horizontal. It also uses first-order and pseudo first-
order chemical reactions for SOX and NOX species. Dry deposition 1s based
on the deposition velocity while wet deposition uses solubility assump-
tions.
In complex-terrain, maximum short-term concentrations from elevated
sources are generally associated with plume Impingement or nocturnal
drainage flows. Plume Impingement occurs when a concentrated plume of
pollutants Impacts elevated terrain. Air pollutants released 1n valleys
can be transported down-valley 1n nocturnal drainage flows. These pollu-
tants will tend to pool at the bottom of the valleys until at sunrise up-
slope winds carry them up the valley walls and out of the valley. If the
model 1s to be used to calculate the maximum short-term concentration, the
ability to simulate transport and dispersion during nocturnal drainage
flows may be an Important component of the add deposition models. This
may be particularly Important since the NAAQS short-term concentration
standards are based on health effects, and population densities are
highest on the valley floors.
The MESOPUFF-II and, to a lesser degree, RIVAD can simulate plume 1mp1ng-
ment 1n complex terrain; however, neither of these models contain explicit
treatment of drainage flow conditions. Of the models surveyed In Chapter
3, only two models—the PNL/MELSAR-POLUT and the SRL/2DFLOW-MCARLO—con-
tain explicit treatment of drainage flow conditions. The SRL/2DFLOW-
MCARLO 1s a specialized model that couples a two-dimensional dynamic wind
4-49
-------
model with a Monte-Carlo treatment of discrete Inert particles of mass for
drainage flow conditions only. The PNL/MELSAR-POLUT model 1s an extension
of the ERT/MESOPUFF, having been developed specifically to simulate trans-
port and dispersion within a complex-terrain region of the Rocky Mountains
(the Green River Ambient Model Assessment, or GRAMA, region). Although
the PNL/MELSAR-POLUT does not treat chemical transformation or dry and wet
deposition, 1t does contain several unique algorithms for simulating
transport and dispersion 1n complex terrain. The final candidate Lagran-
gian air quality and add deposition models to be examined 1n the next
chapter are
SAI/CCADM
ERT/MESOPUFF-II
SAI/RIVAD
PNL/MELSAR-POLUT
4-50
-------
5 EVALUATION OF CANDIDATE MESOSCALE METEOROLOGICAL AND
ACID DEPOSITION MODELS
5.1 EVALUATION OF CANDIDATE MESOSCALE METEOROLOGICAL MODELS
In this section we attempt a preliminary evaluation of the four candidate
diagnostic wind models Identified 1n the previous chapter: CIT/WINO,
PNL/MELSAR-MET; SAI/CTWM, and LANL/ATMOS1. The first part of this section
1s a comparative description of the models. This 1s followed by a display
and discussion of simulation results obtained when each model was sub-
jected to Initially uniform flow over Idealized terrain configurations.
In a future report the candidate models will be used to simulate actual
airflows over the Rocky Mountain region.
5.1.1 Comparative Description of the Candidate Models
Exercise of the diagnostic models In this group Involves two steps: (1)
Initialization of the wind field; and (2) adjustment of the Initial wind
field to produce a "final" wind field that satisfies a three-dimensional
conservat1on-of-mass equation. The discussion which follows 1s organized
In terms of the above two steps.
5.1.1.1 Initialization of the Wind Field
Data Requirements. The CIT, ATMOS1, and MELSAR can be expected to require
substantial Input of surface and upper-air wind data to produce realistic
wind fields 1n complex terrain. Use of these models Involves the Inherent
assumption that the Input data, rather than the model, account for most of
the mesoscale variability of the wind field. The CTWM, on the other hand,
requires Input of only a domain mean wind. The parameter1zatIons con-
tained 1n the model are expected to account for the mesoscale variability
of the model wind field.
The CIT, MELSAR, and CTWM require sufficient surface and upper-air tem-
perature or potential temperature data for crude determination of atmo-
spheric stability. In the CIT model atmospheric stability determines the
degree to which the Initial grldded wind field 1s smoothed before 1t 1s
subjected to mass adjustment.
5-1
-------
ATMOS1 does not require direct Input of stability data. However, domain-
scale stability should be taken Into account when specifying the para-
meters alphal and a1pha2 (discussed 1n Section 5.1.1.2).
All of the models require Input of grldded topographic data. For the CIT,
ATMOS1, and CTWM the horizontal resolution of the topographic data must
equal that of the desired wind field. MELSAR currently requires terrain
resolution of 1 km. We note that terrain data are available on tape from
the National Cartographic Center at horizontal resolutions of 200 feet and
30 seconds 1at1tute/long1tude. If model grldded terrain 1s to be extrac-
ted from these tapes a terrain preprocessor will be necessary.
Vertical Coordinate Formulation. The CIT, MELSAR, and ATMOS1 are formu-
lated 1n terrain-parallel vertical coordinates, I.e., each coordinate
surface 1s located at a constant height above the terrain surface. Ter-
rain-parallel coordinates are highly desirable for a variety of reasons.
First, they allow vertical resolution to be concentrated near the ground
surface 1f desired. Also, winds at a uniform height above ground are
necessary for computation of m1crometeorolog1cal and add deposition para-
meters.
The CTWM 1s formulated 1n Cartesian vertical coordinates, 1n which the co-
ordinate surfaces Intersect the terrain. It should be possible to trans-
form the CTWM Into terrain-parallel coordinates.
Initialization Procedures. In the CIT and ATMOS1 Input surface and upper-
air wind data (u,v) are Interpolated to a desired three-dimensional grid
utilizing variants of a standard l/rn weighting scheme, where r 1s the
distance of a given observation point from a given grid point. A "radius
of Influence" (maximum value of r at which observations are considered at
a given grid point) must be specified. In the CIT model the weighting
scheme 1s modified to account for topographic barriers. In both models
sparse upper-air wind data can be augmented via power-law upward extrapol-
ation of surface wind data.
The resulting three-dimensional grldded fields of u and v constitute the
Input to the mass-conservation portion of these models. These fields may
be very sensitive to the arbitrarily specified "radius of Influence." As
the radius of Influence 1s Increased the Interpolation scheme essentially
acts as a smoother; this can be undesirable 1n regions where closely
spaced observations are representative of mesoscale variability.
From the Input surface and upper-air wind data MELSAR computes analytic
expressions 1n three-dimensional co-ordinate space for the u and v compo-
nents. These expressions are linear combinations of a set of basis func-
5-2
-------
tlons which are polynomial functions of the horizontal co-ordinates. The
amplitude coefficients uniquely define the MELSAR wind field; they are
functions of the MELSAR vertical co-ordinate and are computed from the
data via a least-squares fitting procedure.
A 9
Sparse upper-air data (say, two sounding locations within a IQ knr
domain) may produce grossly unrealistic wind fields 1n regions far removed
from the sounding locations. To mitigate this problem upper-air "sound-
Ings" are generated via a l/rn Interpolation scheme (1) at each surface
station location, and (2) at a relatively large number of user-selected
points on the boundaries of the domain. These additional soundings are
Included 1n the least-squares fit.
A "Froude number adjustment" of the MELSAR wind field 1s carried out after
the first least-squares fit. The MELSAR terrain processor computes
"obstacle heights," "lower terrain heights," and "drainage directions" for
each cell of a coarse "Froude grid." The size of the Froude grid 1s spec-
ified by the user to represent the dominant horizontal scale of the ter-
rain features. The obstacle height, the lower terrain height, the atmo-
spheric stability, and the MELSAR wind define an obstacle Froude number,
which In turn defines a "critical dividing streamline height." At each
vertical level 1n each Froude grid cell, the wind 1s adjusted to the ter-
rain-tangent direction, with no change In speed. 1f the vertical level
lies below the critical streamline height and the flow has an upward com-
ponent. For all Froude grid cells 1n which this adjustment 1s made a
"sounding" 1s generated and added to the original Input data.
The least-squares fit 1s then repeated, yielding a new set of MELSAR
amplitude functions defining u and v. This 1s the final product of the
MELSAR wind Initialization. Note that the MELSAR wind field may be very
sensitive to the Froude grid spacing. The local obstacle heights are
determined by the relationship between the Froude grid spacing and the
horizontal scale of the terrain. As the Froude grid spacing 1s decreased
below the terrain scale, the local obstacle heights decrease and the local
Froude numbers thus Increase, decreasing the number of Froude grid points
at which the winds are deflected. It 1s unclear whether a theory based on
the concept of an Isolated obstacle can provide useful Information con-
cerning flow 1n extremely complex terrain with a broad range of horizontal
length scales.
The Initialization phase of the CTWM generates a grldded field of Car-
tesian vertical velocities. The vertical velocity at each grid point 1s
the sum of parameterized contributions from the following effects
Kinematic effects of topography
Boundary layer effects
5-3
-------
Urban heat Island effects
Thermally forced upslope/downslope flows
Contributions from each of the above effects are added to produce a sur-
face vertical velocity; the vertical velocity 1s assumed to decay exponen-
tially with height. The coefficient of exponential decay 1s a function of
stability. Under stable conditions the forcing due to kinematic topo-
graphic effects 1s reduced by a multiplicative factor 1f the Froude number
(based on obstacle height, the mean wind, and stability) 1s less than an
assumed critical value and the mean flow 1s uphill.
Each of the parameterized contributions to vertical velocity contains at
least one arbitrary coefficient There 1s little physical guidance for the
prescription of these coefficients; the model wind fields may be very
sensitive to the specification of the coefficients. In the case of slope
flows, a temperature difference between the slope surface and the "envi-
ronment" must be specified; diurnal variation of this parameter should be
specified.
Also, the parameterized contribution to vertical velocity from "boundary
layer effects" 1s of doubtful validity; 1t seems to be an artifact of the
formulation of this model 1n Cartesian vertical coordinates.
5.1.1.2 Mass-Consistent Wind Field Adjustment
A wind field 1s mass-consistent 1f 1t satisfies a form of the three-dimen-
sional conservat1on-of-mass equation. It 1s Important to note that mass-
consistency does not Imply uniqueness. If a wind field 1s not mass-con-
sistent any or all of Its components might be adjusted to satisfy mass
consistency, depending upon the assumptions Inherent 1n the specific
adjustment procedure. In diagnostic wind models the following considera-
tions are generally taken Into account to varying extents
The adjustment should produce a (u.v) field that deviates minimally
from observations at the observation locations.
The final adjusted field should contain realistic vertical veloci-
ties.
Blocking and deflection of airflow by topography should be properly
represented.
We discuss below the adjustment procedures 1n each of the candidate diag-
nostic wind models.
5-4
-------
CIT Wind Model. The adjustment procedure 1n the CIT wind model Involves
four steps. The first step 1s an adjustment of grldded surface winds to
account for topography. Following Anderson (1971) a surface Cartesian
vertical velocity 1s computed from the surface winds and terrain slope,
assuming free slip flow. The vertical velocity 1s assumed zero at an
arbitrarily defined height; the variation of w between the surface and
this height 1s assumed to be linear. Thus, a single value of dw/dz 1s
defined at each horizontal grid location. Assuming that the horizontal
flow 1s 1rrotat1ona1. manipulation of the Incompressible conservat1on-of-
mass equation yields a Polsson equation for a "velocity potential" which
can be uniquely solved. The solution for the velocity potential yields 1n
turn an adjusted surface velocity field.
The second step 1s the application of a standard five-point smoother to
the u and v fields above the surface. The objective of this step 1s to
•reduce as much of the anomalous divergence as possible." (Goodln et al.,
1980). It 1s apparently left to the user to determine exactly what frac-
tion of the horizontal variability 1n the grldded (u,v) fields constitutes
"anomalous divergence"; Goodln and co-workers Imply that this fraction
Increases with stability. In the version of the CIT wind model utilized
by Godden and Lurmann (1983) the number of applications of the smoother 1s
an Increasing function of Input thermal stability.
The third step Is the computation of a grldded field of vertical veloci-
ties. Given grldded three-dimensional (u,v) .fields a simple Integration
of the Incompressible continuity equation would yield a grid of w values;
the (u.v.w) field would then be mass-consistent. However, Goodln and co-
workers believe that w values obtained this way would be "unreal1st1cally
large" at the top of the domain. Therefore, they make the arbitrary
assumption that the vertical velocity at the top of a model layer 1s the
Integral of the convergence within that layer only. In later applications
of this model (e.g., Godden and Lurmann, 1983) this assumption 1s replaced
by a procedure suggested by O'Brien (1970). The O'Brien procedure modi-
fies the "Incompressible" w values so that the vertical velocity vanishes
at the top of the modelling domain.
With the w field obtained via either of the above assumptions the (u,v,w)
field 1s not mass-consistent. Thus a fourth step 1s required. In this
step, equal adjustments 1n the u and v fields at each grid point are made
In order that (u.v.w) satisfy the Incompressible conservat1on-of-mass
equation given the assumed w. This procedure 1s carried out 1terat1vely
until the residual three-dimensional divergence 1s reduced below a user-
specified level.
MELSAR. Given the analytic expressions for (u,v) generated 1n Its
Initialization phase, MELSAR simply Integrates the anelastlc continuity
5-5
-------
equation (density 1s allowed to vary with height only) to obtain a field
of w 1n terrain-following co-ordinates. Thus, (u.v.w) satisfy mass con-
sistency; no additional adjustment 1s made. The realism of the vertical
velocities thus calculated 1s not discussed by the model developers.
CTWM. The Initialization phase of the CTWM generates a grldded field of
vertical velocities, which are numerically differentiated to obtain a
grldded field of dw/dz. The values of u and v are unknowns. As 1n the CIT
surface wind adjustment, 1f the (u,v) fields are assumed to be Irrota-
tlonal, they can be expressed 1n terms of a "velocity potential." Fixing
dw/dz. and assuming 1ncompress1b1l1ty. a two-dimensional Polsson equation
can be uniquely solved for a velocity potential, from which (u,v) are
extracted. The resulting (u.v.w) field 1s mass-consistent. Note that 1n
a stable atmosphere the decay of the model-generated vertical velocity
with height will be rapid and the adjustments of u and v from the Input
mean flow will be maximized at low levels.
The adjustment of the Initial grldded wind fields by ATMOS1 1s constrained
by both the Incompressible continuity equation and by minimization of the
difference between Initial and final (u,v,w) fields. Based on the formal-
Ism developed by Sasaki (1958,1970ab) an Integral of the form
f [*\ (u - uzQ)2 + o^ (v - vQ)2 + a2 (w - w0)2 + x(du/dx + dv/dy +
dw/dz)1 dx dy dz (5-1)
1s minimized. The subscript 0 denotes Initial fields, and x denotes a
"Lagrange multiplier." The alphas are arbitrarily specified. As shown by
Sherman (1978) 1n her description of MATHEW, the predecessor of ATMOS1,
the final (u,v,w) required to minimize the Integral are functions of the
Initial fields and of x. Sherman derives a three-dimensional elliptic
equation for x; the solution for x 1s manipulated to yield the final mass-
consistent (u,v,w). Note that 1n ATMOS1 the above equations are trans-
formed Into terrain-following coordinates.
The relative degree to which (u,v) and w are adjusted to achieve mass
consistency 1s governed by the ratio alphalZ > alphal/alphaZ. Sherman
suggests an alphalZ of about 0.01, comparable to the ratio of w/u under
typical conditions. As this ratio decreases the adjustment of (u,v) pre-
•domlnates over that of w; thus, low values of alpha!2 should enhance
deflection around topographic obstacles. As demonstrated by Lewellen and
Sykes (1982), and also by Ross and Smith (1986), an alpha!2 of 1 1s equiv-
alent to three-d1mens1onally 1rrotat1onal "potential flow."
5-6
-------
In ATMOS1 a1pha!2 1s a constant 1n space; Its specification might depend
on an assessment of domain-scale stability. Ross and Smith Indicate that
the performance of ATMOS1 might be Improved If a1pha!2 were allowed to
vary In space as a function of local Froude number; this might result 1n
Improved simulation of blocking effects.
5.1.2 Evaluation Using a Hypothetical Terrain Obstacle
For the Rocky Mountain region a wind model might have to be exercised with
a single Input wind observation over a given domain. It 1s desirable to
know whether a model can generate useful Information about terrain-
generated mesoscale airflow patterns 1n the absence of significant Input
wind data.
To this end we carried out simulations with each model of Initially uni-
form flow over a three-dimensional bell-shaped mountain of the form h(x,y)
» hQaz/(x2 + y2 + a2). In this expression the mountain 1s centered at x «
y » 0; h0 1s the maximum height of the mountain, and a 1s the horizontal
distance from the mountain top to a point at which h(x,y) • h0/2. In our
experiments we set n0 - 2000 m and a • 25 km, representative of the larger
horizontal and vertical terrain scales 1n the Rocky Mountain region. The
Initially uniform flow was specified as 2 m/s from the west, sufficiently
weak so that the Froude number adjustments In MELSAR and the CTWM would be
activated. Winds were generated at heights of 50, 200, 500, 1000, and
2000 m above ground on a 21 x 21 grid with horizontal grid spacing of 5
km.
Results obtained for each of the candidate models are discussed below.
5.1.2.1 CIT Wind Model
The version of the CIT wind model utilized by Godden and Lurmann (1983),
dubbed "WINDMOD", was utilized 1n these experiments. With Initially uni-
form flow, the Anderson (1971) procedure generates horizontal variability
1n the "surface" (level 1) winds. Thus dw/dz 1s nonzero across layer 1
("levels" are at the midpoint of "layers"); above layer 1 dw/dz reverses
sign and 1s considerably decreased 1n magnitude, as the absolute value of
w decreases monotonlcally with, height to zero at the top of the model
domain (2700 m above ground). The CIT divergence reduction procedure
adjusts u and v so that they are mass-consistent with the vertical velo-
city field. The mass-adjustment procedure was executed until maximum
three-dimensional divergence was reduced to 10"6 s"1
5-7
-------
Figure 5-1 depicts CIT model wind fields over the bell-shaped mountain at
50, 200, and 500 m above ground (levels 1,2, and 3), respectively. Note
that the level 1 winds accelerate upwind of the mountain top and deceler-
ate downwind, with negligible horizontal deflection. Note also that the
level 2 and 3 winds are minimally perturbed by the terrain.
CIT model winds are expected to be very sensitive to the specified layer
thicknesses, especially to the thickness of layer 1. In these experiments
we found that the mass adjustment procedure would not converge 1f the
thickness of layer 1 was less than 50 m. To obtain near-ground winds with
this model 1t may be necessary to resort to power-law Interpolation/extra-
polation procedures, rather than attempt direct computation of anemometer-
level winds.
5.1.2.2 MELSAR
The only adjustment which MELSAR makes to an Initially uniform flow 1s for
blocking effects, as detailed 1n the previous section. In this experiment
the atmosphere 1s assumed to be uniformly Isothermal, so that the Froude
number 1s minimized.
Figure 5-2 depicts MELSAR wind fields at 50, 200, and 500 m above
ground. Note the deflection of the level 1 airflow upwind of the mountain
top, and the tendency of the MELSAR winds to follow the terrain con-
tours. The final MELSAR least-squares-fit procedure appears to smooth the
blocking effect somewhat. At levels 2 and 3 the blocking effects are
reduced, which 1s to be expected since the effective obstacle heights are
smaller.
Note that the level 1 MELSAR winds behave erratically near the south boun-
dary of the domain. As noted 1n the previous section, a sufficient number
of artificial "soundings" must be generated at the boundaries via a l/rn
Interpolation scheme 1f the MELSAR least-squares-fit procedure 1s to be
well-behaved. In this experiment such soundings are generated at 16
points on the boundaries. The reason for the upslope flow near the south
boundary 1s unclear. Possibly, this behavior Implies that MELSAR boun-
daries should be located relatively far from the domain of Interest. .
5.1.2.3 ATMOS1
ATMOS1 was exercised with the parameter a1pha!2 set equal to 0.02; an
attempted model run with alphalZ - 0.01, as suggested by Sherman (1978)
produced an unacceptably chaotic model wind field. Sherman, along with
Ross and Smith (1986) note that 100 to 400 Iterative passes through a
5-8
-------
0 5
NORTH
10 15 20 25 30 35 40 45 50 55 60 65 70 75 BO 85 90 95
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
WINDMOD WIND VECTORS AT LEVEL
05 10 15
WIND SPEED (M/S)
FIGURE 5-la. CIT model-generated wind fields at 50 m above ground.
Scaling of plotted vector winds is given at lower left. Topography
is contoured in meters. Distance along axes is given in kilometers.
5-9
-------
5 10 15 20 25 30 35
NORTH
40 45 50 55 60 65 70 75 SO 85 90 95
0 5 10 15 20 25 30 35
40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
WINDMOD WIND VECTORS AT LEVEL
1 ii IM 1 1
0 5 10 15
WIND SPEED (M/S)
FIGURE 5-lb. At 200 m.
5-10
-------
NORTH
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
WINDMOO WIND VECTORS AT LEVEL
I""!""!"11!
0 5 10 15
WIND SPEED (M/S)
FIGURE 5-lc. At 500 m.
5-11
-------
NORTH
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
10Po
\* \f LX IX [/ If \f It
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
MELSAR WIND VECTORS AT LEVEL - 1
05 10 15
WIND SPEED (M/S)
FIGURE 5-2a. MELSAR model-generated winds at 50 m above ground.
Scaling of plotted vector winds is given at lower left. Topography
is controued in meters. Distance along axes is given in kilometers,
5-12
-------
100
.0 5
NORTH
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
0 5
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
MELSAR WIND VECTORS AT LEVEL
tjjJi ii f i m yn
05 10 15
WIND SPEED (M/S)
- 2
FIGURE 5-2b. At 200 m.
5-13
-------
NORTH
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
109,
00
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
MELSAR WIND VECTORS AT LEVEL - 3
u^
0 5 10 15
WIND SPEED (M/S)
FIGURE 5-2c. At 500 m.
5-14
-------
successlve-overrelaxatlon solution of a three-dimensional elliptic equa-
tion are usually necessary to satisfy the criterion for three-dimensional
mass-consistency. However, we found that the horizontal wind components
changed negligibly after 10 Iterations. Thus, we believe that 1t may be
reasonable to obtain a mass-consistent wind field by running ATMOS1
through 10 or 20 Iterations, and then Integrating the conservat1on-of-mass
equation to produce vertical velocities that are exactly mass-consistent
with the ATMOS1 horizontal wind component fields. This hypothesis needs
further testing with complex terrain and spatially varying Input wind
data. In the experiment reported here the model run was halted after 20
Iterations.
Figure 5-3 Illustrates the ATMOS1 solutions at 50, 200, and 500 m above
ground. At 50 m flow 1s deflected around the mountain, accelerated over
the mountain top and to the north and south of the mountain top, and
decelerated upstream and downstream of the mountain top. Note the extreme
convergence and divergence zones near the mountain top In Figure 5-3a. At
200 m the effects of the terrain are sharply reduced, and at 500 m the
flow 1s essentially undisturbed.
ATMOS1 u-component fields upstream and downstream of the mountain top are
essentially symmetric, while v-component fields are mirror Images of each
other. This result 1s consistent with kinematic arguments. Given a con-
stant value of alpha!2, blocking effects, which would be expected to act
only on the upstream side of the mountain, cannot be simulated by ATMOS1.
The results obtained with alphal2 • 0.02 are apparently representative of
an atmosphere sufficiently stable to deflect low-level flow around the
mountain. A similar ATMOS1 run with a1phal2 • 1 (not shown) produced
acceleration to 2.7 m/s above the mountain top with minimal deflection at
all model levels; this 1s consistent with the "potential flow" results
obtained by Ross and Smith (1986) 1f the major difference 1n horizontal
obstacle scales 1s taken Into account. Application of ATMOS1 to the Rocky
Mountain region would require development of a methodology for relating
alphal2 to some measure of atmospheric stability. Ross and Smith suggest
a procedure for relating alpha!2 to local Froude number; this procedure
may eventually be Incorporated Into the NUATMOS model developed by these
Investigators.
5.1.2.4 CTWM
For this study several modifications were made to CTWM to facilitate com-
parison with the other candidate models. The vertical velocity formula-
tion was transformed to terrain-parallel coordinates, allowing computation
of winds at constant levels above ground. The "boundary-layer" parameter-
5-15
-------
tfl
100r-
4
95 -
t
90 -
85 -
•
80-
75-
70-"
65-"
60 -
»
55-
50 -
•
45-
»
40-
•
35-
30 -
a
25-
20-'
<
15-
10-1
NORTH
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 10
P
'00
95
90
85
80
75
70
65
60
55
50 §
u
45
40
35
30
25
20
15
10
5
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
10$
ATMOS1 WIND VECTORS AT LEVEL
inn Minimi i
- 5
0 5 10 15
WIND SPEED (M/S)
FIGURE 5-3a. ATMOS1 model-generated winds at 50 m above ground.
Scaling of plotted vector winds is given at lower left. Topography
is contoured in meters. Distance along axes is given in kilometers.
5-16
-------
10
NORTH
15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 10(
00
I I I ! I I I I I I I III
5 10
15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
ATMOS1 WIND VECTORS AT LEVEL = 4
0 5 10 15
WIND SPEED (M/S)
FIGURE 5-3b. At 200 m.
5-17
-------
100,
NORTH
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 10
?c
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
ATMOS1 WIND VECTORS AT LEVEL - 3
0 5 10 15
WIND SPEED (M/S)
FIGURE 5-3c. At 500 m.
5-18
-------
1zat1on was omitted because Its contribution appears to be absent 1n ter-
rain-parallel coordinates. Finally, the CIT divergence reduction proce-
dure, rather than the solution of the Polsson equation for velocity poten-
tial, was used to adjust the wind field to mass-consistency. We believe
that these modifications simplify the model without compromising Us basic
approach.
As noted 1n the previous section, the CTWM can be extremely sensitive to
the specification of several arbitrary constants. For this study these
constants were specified as detailed by Yocke (1981) In his simulations of
airflows 1n the Phoenix and Los Angeles areas.
So that the effects of the various parameterlzatlons can be displayed, we
carried out four CTWM simulations of Initially uniform airflow over the
bell-shaped mountain. In simulation 1 an Isothermal atmosphere 1s
assumed, and only the kinematic effects of terrain are parameterized.
Simulation 2 1s Identical to simulation 1 except that blocking effects are
Included. Simulation 3 1s Identical to simulation 2 except that slope
flow effects are Included, with the slope flow parameterization tuned to
produce downslope flow; thus simulation 3 1s Intended to represent noc-
turnal conditions. Simulation 4 1s Identical to simulation 3 except that
the atmosphere 1s assumed to be neutrally stratified and the slope flow
parameterization 1s tuned for upsiope flow; thus simulation 4 Is Intended
to represent conditions around noon.
Results of simulation 1 are displayed In Figure 5-4. The CTWM treatment
of terrain kinematic effects produces acceleration across the mountain
top, deceleration upstream and downstream of the mountain, and very slight
deflection of the flow around the mountain. We note that the arbitrary
coefficients 1n the CTWM could be tuned to Intensify the deflection. As
1n the ATMOS1 simulation, the u-components on the upstream and downstream
sides of the mountain are symmetric and the v-components are mirror
Images. The terrain kinematic effects are sharply reduced at 200 and 500
m above the ground, consistent with the rapid exponential decay of CTWM-
parameterlzed vertical velocities 1n a stable atmosphere.
Results of simulation 2 are displayed 1n Figure 5-5. The Froude number
adjustment 1n the CTWM Involves the multiplication of the parameterized
Cartesian vertical velocity by the fraction Fr/Frc, where Fr 1s the
"local" Froude number and Frc Is a "critical" Froude number, set equal to
1 1n this simulation. If Fr < 1 and flow 1s uphill, w, and therefore
dw/dz. 1s reduced 1n magnitude, 1n turn reducing the compensating diver-
gence/convergence of the horizontal flow. Thus the CTWM "blocking" para-
meterization reduces the kinematic effect of the terrain on the upstream
side of the mountain; 1n simulation 2 upstream deflection 1s virtually
absent and acceleration at the mountain top 1s sharply reduced. This
5-19
-------
NORTH
0 5 10 15 20 25 30 35 40 45 50 55
CO
95"
90"
85-
80 -
75
70
65
60
55
50
45
40
35
30
25
2C
1
1C
T 1 1 1 T
1—I 1 1 1
60 65
T T
70 75
T—r
80 85
I—
90 95
T—r-
5" 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
SOUTH
95
90
85
80
75
70
65
60
55
50 §
LJ
45
40
35
30
25
20
15
10
5
0
CTWM WIND VECTORS AT LEVEL
0 5 10 15
WIND SPEED (M/S)
.FIGURE 5-4a. CTWM model-generated winds for simulation 1 at 50 m
above ground. Scaling plotted vector winds is given at lower left
Topography is contoured in meters. Distance along axes is given
in kilometers.
5-20
-------
NORTH
10 15 20 25 30 35 AQ 45 50 55 60 65 70 75 80 65 90 95
i—i—i—i—i i i—i—i—i—i—i—r
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
CTWM WIND VECTORS AT LEVEL - 2
I""!""!"11!
0 5 10 15
WIND SPEED (M/S)
FIGURE 5-4b. At 200 m.
5-21
-------
NORTH
5 10 15 20 25 30 35 4Q 45 50 55 60 65 70 75 80 85 90 95
"05 10 15 20 25 -30 35 40 45 50 55 60 65 70 75 80 85
SOUTH
CTWM WIND VECTORS AT LEVEL - 3
0 5 10 15
WIND SPEED (M/S)
FIGURE 5-4c. At 500 m.
5-22
-------
NORTH
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
CTWM WIND VECTORS AT LEVEL = 1
05 10 15
WIND SPEED (M/S)
FIGURE 5-5a. CTWM model-generated winds for simulation 2 at 50 m
above ground. Scaling of plotted vector winds is given at lower
left. Topography is contoured in meters. Distance along axes is
given in kilometers.
5-23
-------
NORTH
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 B5 90 95
"0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
CTWM WIND VECTORS AT LEVEL
ijjjy M 111111111
0 5 10 15
WIND SPEED (M/S)
FIGURE 5-5b. At 200 m.
5-24
-------
NORTH
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
5 10 15 20 25 30 25 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
CTWM WIND VECTORS AT LEVEL - 3
0 5 10 15
WIND SPEED (M/S)
FIGURE 5-5c. At 500 m.
5-25
-------
result 1s very different from that generated by the MELSAR treatment of
blocking (compare Figures 5-2a and 5-5a), and does not seem Intuitively
realistic. Given an obstacle of this size we might expect to see Froude
number effects at 200 m above ground similar to those at 50 m; however 1n
simulation 2 Froude number effects are sharply reduced at 200 m and absent
at 500 m. As 1n simulation 1 this 1s probably the result of rapid expo-
nential decay with height of the CTWM vertical velocity field.
Results of simulation 3 are displayed 1n Figure 5-6. At the 50 m level
(Figure 5-6a) a strong downslope component appears. At 200 m the flow 1s
upslope, and at 500 m above ground the flow 1s virtually unperturbed. In
simulation 4 (Figure 5-7), tuned for upslope flow, a very strong upslope
component appears at 50 m; the winds at 50 m are stronger than those 1n
simulation 3 due to the absence of the Froude number effect under neutral
stratification. Downslope components are apparent at the 200 m and 500 m
levels. It should be noted that arbitrary constants 1n the CTWM slope
flow parameterization govern magnitude as well as direction of the slope
flow.
Although observed near-surface slope flows are frequently compensated by
return flow aloft (see, for example, Atkinson, 1981); the CTWM as cur-
rently formulated should not be able to reproduce return flow. The magni-
tudes of the Cartesian vertical velocities should decrease monotonlcally
with height; thus dw/dz should not change sign and the sense of the con-
vergence/divergence patterns should be unchanged.
It 1s probable that the reversal of the slope flow components between the
first two levels 1n simulations 3 and 4 are the result of our transforma-
tion of the parameterized CTWM vertical velocities Into terrain-parallel
co-ordinates. If w represents Cartesian vertical velocity, W represents
vertical velocity 1n terrain-parallel co-ordinates, and h represents
terrain height, then
W « w - u dh/dx - v dh/dy (5-2)
Consider a position due west of the mountain top, and assume the presence
of a thermal upslope component us 1n addition to the uniform west-to-east
flow U specified 1n these experiments. Assume no Froude number effects
(Fr > Frc). In the CTWM as currently formulated
w « (U + us) dh/dx exp(-kz) (no Froude number effects) (5-3)
where z 1s height above the surface. In transforming from w to W, we have
assumed u = U + us at the surface and u « U at all levels above the sur-
face. Thus 1t 1s possible for dW/dz to change sign between the first and
second layers even though dw/dz does not. Since a terrain-following ver-
5-26
-------
0 5
NORTH
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
', 't 'f 'I 'I 7 7 7 7 //y/»!
If///
\\\\\\
\ \ ,\ \ \ \ N N
90 95
CTWM WIND VECTORS AT LEVEL
H'"|tni|iiii|
05 10 15
WIND SPEED (M/S)
FIGURE 5-6a. CTWM model-generated winds for simulation 3 at 50 m
above ground. Scaling of plotted vector winds is given at lower
left. Topography is contoured in meters. Distance along axes is
given in kilometers.
5-27
-------
NORTH
0 5 10 15 20 25 30 35 40 45 50 55
60 65 70 75 80 85 90 95
"0 5 10 15 20 25 30 35 40 45 50 55
SOUTH
60 65 70 75 80 85 90 95
CTWM WIND VECTORS AT LEVEL
I mi mi i <
I""!""!""!
05 10 15
WIND SPEED (M/S)
FIGURE 5-6b. At 200 m.
5-28
-------
NORTH
5 10 15 20 25 30 35 4Q 45 50 55 60 65 70 75 80 85 90 95
i i i i i i i i i i i
0
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
CTWM WIND VECTORS AT LEVEL -
0 5 10 15
WIND SPEED (M/S)
FIGURE 5-6c. At 500 m.
5-29
-------
NORTH
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
CTWM WIND VECTORS AT LEVEL -
0 5 10 15
WIND SPEED (M/S)
FIGURE 5-7a. CTWM model-generated winds for simulation 4 at 50 m
above ground. JScaling of plotted vector winds is given at lower
left. Topography is contoured in meters. Distance along axes is
given in kilometers.
5-30
-------
5 10
NORTH
15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
"0 5" 10 ~ 3 20 23 38 35 4B 45 50 5S
SOUTH
60 &S 70 75 80 85 90 95
CTWM WIND VECTORS AT LEVEL
I IlllII II Illllll
0 5 10 15
WIND SPEED (M/S)
- 2
FIGURE 5-7b. At 200 m.
5-31
-------
NORTH
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
0 5
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
CTWM WIND VECTORS AT LEVEL
I ii n ii 111 [ n 111
05 10 15
WIND SPEED (M/S)
FIGURE 5-7c. At 500 m.
5-32
-------
slon of the CTWM 1s necessary 1f 1t 1s to be adopted for the Rocky Moun-
tain application, the mechanics of the coordinate transformation 1n the
case of slope flows needs further Investigation.
5.1.2.5 Remarks
The comparative simulations discussed 1n this section cannot by themselves
serve as a basis for model evaluation. A comprehensive model evaluation
would Involve tests of the ability of the models to simulate actual obser-
vations 1n complex terrain; we hope to Include such a study 1n a future
report. Another approach to model evaluation 1s comparison with analytic
theory, as discussed by Plelke (1984). As noted previously, Ross and
Smith (1986) demonstrate that ATMOS1 can reproduce analytic solutions for
unstratlfled potential flow over Idealized obstacles. However, based on
the analyses of mountain-generated airflows by Smith (1979) and others, 1t
1s unclear to us that the potential-flow solutions are relevant on the
terrain scales to be simulated 1n the Rocky Mountain region.
More relevant, perhaps, are two types of mountain wave disturbances:
trapped lee waves and vertically propagating hydrostatic waves. Durran
and Klemp (1982,1983) demonstrate the ability of a primitive-equation
nonhydrostatlc model to reproduce analytic solutions for each of these
types of mountain waves. Additionally. Clark and Gall (1982) have util-
ized a nonhydrostatlc primitive-equation model to simulate observed lee
waves near Elk Mountain, Wyoming, a location which lies within the domain
of current Interest. We note here that none of the candidate models 1s
capable of simulating either type of mountain wave disturbance unless the
disturbance 1s fully accounted for by Input wind data. There may be cir-
cumstances under which mountain waves play a significant role 1n horizon-
tal and vertical transport of pollutants; primitive-equation numerical
simulations would be necessary to delineate this role.
The blocking and deflection of airflow by terrain obstacles, especially
Important 1n the Rocky Mountain region under weak synoptic flow condi-
tions, 1s simulated to varying extents by each candidate model. MELSAR,
1n particular, 1s designed to simulate this effect alone. The CIT model
lacks a provision for Froude number flow adjustment and thus cannot simu-
late blocking effects 1f they are not defined by Input wind data. ATMOS1
similarly lacks a Froude number term, but can provide a gross simulation
of blocking depending on the magnitude of alpha!2. Ross and Smith (1986)
propose a scheme for calculation of a space-variable alpha!2 as a function
of local Froude number; such a treatment might Improve the ability of
ATMOS1 to simulate blocking effects. The CTWM appears to be capable of
treating kinematic deflection of airflow; while 1t attempts to parameter-
ize blocking effects, the treatment produces somewhat questionable
results.
5-33
-------
The CPU time on a Prime 750 minicomputer required for the Idealized simu-
lations with each of the candidate models 1s tabulated below:
CIT wind model 86 s
MELSAR 23 s
ATMOS1 (20 Iterations) 154 s
CTWM 140 s
MELSAR 1s Inexpensive because Its Initial grldded wind fields are assumed
mass-consistent without additional constrained adjustments. If the
details of the model vertical velocity field are unimportant, MELSAR may
be sufficient to represent blocking and deflection of the horizontal wind
components by terrain. .
If reasonable vertical velocities are desired, ATMOS1, which attempts
adjustment of the vertical velocity based on gross stability considera-
tions (I.e., the specification of alphal2), may be a better choice. The
CIT wind model 1s less desirable 1n sparse data situations due to Its lack
of ability to simulate blocking and deflection; however, 1f Input wind
data 1s plentiful and representative, the flexibility of the CIT model
Interpolation scheme may be of value.
The CTWM, alone among the candidate models, 1s designed to generate wind
fields with only domain-scale Input wind Information. It 1s the only
candidate model which explicitly attempts to treat thermally-generated
upsiope and downslope flows. The CTWM requires specification of several
arbitrary coefficients with little physical guidance. This would probably
require a specific "tuning" of the coefficients for the Rocky Mountain
region, based on available observed wind data. Also, the ability of the
CTWM to utilize more than one wind observation within the model domain 1s
unclear.
5.1.3 Conceptual Design of a Mesoscale Meteorological Model
for the Rocky Mountain Range
In the previous section no one diagnostic wind model appeared to have
superior performance or formulations than the other models. If there 1s a
total lack of observational data, the CTWM would be the best choice for a
wind field generator; however, 1t cannot take full advantage of any exist-
ing meteorological data. The MELSAR wind model 1s attractive because of
Its ability to represent blocking and deflection 1n a cost-effective man-
ner. However, the MELSAR, ATMOS1, and CIT wind models all require meteor-
ological measurements to Infer any dynamic properties 1n the wind field.
5-34
-------
Not mentioned 1n all the previous discussions 1s the need to generate
fields of other meteorological variables besides winds that are required
for an add deposition model. These other meteorological variables
Include boundary layer heights, temperatures, relative humidities, stabil-
ity, and m1crometeorolog1cal variables such as friction velocity and
Monln-Obukov length. The only candidate model that also generates fields
of meteorological variables besides wind fields 1s the MELSAR. The MELSAR
1s coded 1n a highly modular fashion, which allows for ease of addition,
replacement, or modification of any existing module.
Thus the conceptual design for a mesoscale meteorological model will make
use of the PNL/MELSAR-MET code as a basis for generating wind fields and
other meteorological variables needed for add deposition modeling 1n
complex terrain. However the wind field module 1s still to be chosen from
the four candidate diagnostic wind modules. Due to Its Cartesian coordi-
nate system the CTWM 1s not an appropriate choice; however, the para-
meter 1zat Ions of upslope/downslope and other dynamic effects contained
within the CTWM should be added to the wind module. The choice between
the MELSAR, CIT, and ATMOS1 (or NUATMOS If 1t becomes available) will be
made after the models are evaluated using actual complex terrain situa-
tions 1n the Rocky Mountain region.
5.2 EVALUATION OF THE CANDIDATE ACID DEPOSITION-MODELS
In Chapter 4, four candidate add deposition/air quality models were
chosen for further consideration for Incorporation Into an add deposition
model for application to the Rocky Mountain region. These four models
consist of a Lagranglan box model, the SAI/CCADM, two Lagranglan puff
models, the ERT/MESOPUFF-II and PNL/MELSAR-POLUT, and a Lagranglan plume
segment model, the SAI/RIVAD. These four models contain different model-
Ing approaches and parameterlzatlons of the processes that lead to add
deposition and pollutant transport 1n complex terrain. The choice of
these four candidate models was not based on the assumption that any one
of the models will serve as the final add deposition model, but that each
of the models contain, modules and parameterlzatlons that can be Incorpor-
ated Into the final add deposition model. In this section the methods
used by the four models to treat the major processes of transport, dis-
persion, chemical transformation, dry deposition and wet deposition are
evaluated 1n the context of predicting add deposition and air quality
Impacts from specific sources 1n complex terrain.
5-35
-------
5.2.1 Transport
The transport winds will be generated by a diagnostic wind model as
described 1n the previous sections. These winds will consist of several
vertical levels at terrain following fields of the three wind components
u, v, and w. These wind fields will represent the airflows over complex
terrain, Including blocking, channeling, dividing streamline concept, and
the decoupling between mountain-valley and synoptic winds. Estimating
plume trajectories 1n complex terrain 1s complicated by the large amounts
of wind shear and vertical velocities present. None of the four candidate
add deposition models allow for the explicit treatment of wind shear by
vertically spHtlng the plume, and allowing for the separation of the two
air parcels on separate trajectories. Three of the models MESOPUFF-II,
RIVAD, and POLUT use the wind at the plume centerline for horizontal
transport of the plume. The MESOPUFF-II 1s currently configured to oper-
ate with two vertical levels of wind fields (Sclre and Lurmann, 1983), the
RIVAO typically uses four vertical levels (Latlmer et al., 1985a,b) and
the POLUT can use up to 15 vertical levels of wind fields. The trajectory
used by the CCAOM 1s left as user Input. The number of vertical levels of
Input wind fields for the CCADM 1s also user defined. Wind shear 1n the
CCAOM Is handled by entraining ambient air from outside the Lagrang1an
box, and detraining air, Including source emissions, from Inside the
box. Although this method of handling.wind shear 1s technically accurate,
the loss of source emissions mass from the Lagranglan model would be an
undesirable feature for a regulatory source apportionment modeling study.
Thus the four candidate Lagranglan models all use a similar methodology
for horizontal transport; using a wind vector at the plume centerllne to
advect the plume. This may lead to erroneous results, for example, when
elevated source emissions become well-mixed during the day, and the night-
time transition period results 1n a decoupling between the synoptic flow
and the nocturnal drainage flow. Under these conditions the source emis-
sions will be advected according to the synoptic wind, even though some of
the emissions will be caught up 1n the drainage flows. The uncertainty
Introduced by using a single wind vector to transport a plume 1s unknown,
however, since the source-receptor distances will generally be on the
order of 50 to 200 km, the uncertainty Introduced by the boundary layer
variation during the diurnal cycle should be minimal.
A further consideration 1n the treatment of horizontal transport by the
candidate models 1s how they will represent a continuous plume 1n complex
terrain. Since the wind fields generated by the diagnostic wind model
will conta1n~the Influences of channeling, blocking and the dividing
stream-line concept, 1t 1s Important that the modeling approach taken
5-36
-------
takes full advantage of the complex terrain wind fields. As discussed 1n
Section 3.3.1, a Lagranglan puff model, such as MESOPUFF-II and POLUT,
represents a continuous plume by releasing a series of overlapping puffs
(see Figure 3-38). A continuous plume 1s accurately represented by a
series of overlapping puffs as'long as the separation distance between
adjacent puffs, Ad, 1s not too large. Although there 1s not a consensus
1n the literature as to what the maximum separation distance should be to
assure the representation of a continuous plume, Zannettl (1980) states
that this separation distance should not exceed the horizontal standard
deviation of the puff (Ad s o ), while Ludwlg et al. (1977) recommends
twice the distance (Ad s 2o )y. The MESOPUFF-II and the POLUT models use
two different techniques 1n order to assure the accurate representation of
a continuous plume.
In a Lagranglan puff model a concentration at a receptor 1s obtained by
summing the contributions of each nearby puff, generally by taking a
"snapshot" of each puff at particular time Intervals (sampling steps)
specified as program Input. The concentration at a receptor due to a
single puff can be given by:
C(s)
Q(s
exp
• •
-r2(s)
(5-4)
9(s)
?- 1 «p
(He + 2nz1)
I~of(0
where
C(s)
s
Q(s)
°y(s)
r(s)
21
the ground-level concentration,
the distance traveled by the puff,
the mass of pollutant 1n the puff,
the standard deviation of the Gaussian distribution 1n the
horizontal,
the standard deviation of the Gaussian distribution 1n the
vertical,
the radial distance from the puff center,
the mixed-layer height, and
5-37
-------
He - the effective height of the puff center.
The vertical term, g(s), reduces to the uniformly mixed layer of 1/z^ for
/z, * 1.6. In general, puffs within the daytime mixed layer satisfy
criterion about an hour or two after release.
In the MESOPUFF-II a more accurate representation of a continuous plume 1s
obtained by Integrating Equation 5-4 over the distance the puff travels,
AS, during the sampling step:
exp
ds (5-5)
This Integral can be evaluated by assuming that the most significance
downwind distance dependence during the sampling step 1s 1n the r(s) and
Q(s) terms and evaluating the horizontal o (s) and vertical, g(s), terms
at the midpoint of the sampling step.
The POLUT model uses a puff splitting technique that assures that the
separation distance between the adjacent puffs, Ad, Is always less than
the horizontal standard deviation of the puffs (Ad s o ). Under these
conditions sampling puffs using a "snapshot" approach tEquatlon 5-4) will
provide an accurate representation of the.continuous plume.
A Lagrang1 an plume-segment model, such as the RIVAD, simulates a continu-
ous plume by Interpolating between the plume segments (see Figure 3-38).
If, as expected 1n complex terrain, the plume trajectory 1s curvaceous and
the plume segment release Interval 1s not frequent enough, the use of
linear Interpolation between the adjacent plume-segments may lead to
erroneous results. The RIVAD model does not contain any provisions for
Increasing the number of plume segments 1f the separation distance 1s too
large. For mesoscale modeling 1n complex terrain the plume segment
approach may not be the most appropriate modeling methodology.
The CCADM model also relies on the frequent release of the Lagranglan box
1n order to simulate a continuous plume. As currently configured, the
CCADM does not Interpolate between adjacent boxes, thus there may be gaps
1n the representations of a continuous source.
The MESOPUFF-II and RIVAD models do not consider variations of the plume
centerline due to vertical velocities. The CCADM does consider the ver-
tical transport of mass within its vertical layers due to horizontal div-
ergence of the wind fields, and updraft/downdraft velocities caused by
cumulus and stratus clouds. The POLUT model contains a user option of
computing puff movement with or without the vertical velocities. If the
3-38
-------
vertical velocities generated by the diagnostic wind model represent the
vertical velocities 1n complex terrain, then using them for puff movement
would provide a more accurate description of the puff movement relative to
the terrain than empirical correction techniques. However, 1t 1s uncer-
tain how well the diagnostic wind models represent the actual vertical
velocities. If the user exercises the option of not using the vertical
velocity fields for plume transport 1n the POLUT model, then the heights
of the plume above ground 1s determined using an empirical correction
adjustment (Schulman and Sdre, 1980):
He(t + At) - HQ - (1 -Cj) minimum [H0, T(t + At) - T(t)] (5-6)
where
H0 - hs + Ah, the Initial effective stack height (m A6L)
T(t + At) - terrain height at time t + At (m MSL)
T(t) - terrain height at time t (m MSL)
Cj • terrain adjustment coefficient
Values of Cj recommended by Shulman and Sdre (1980) for stabilities A
through F are 0.5, 0.5, 0.5, 0.3, and 0.3, respectively. This technique
1s known as the half-height correction, since for neutral and unstable
conditions, the puff 1s lifted one-half of the difference between the
elevation of the terrain at the puff location and the elevation of the
terrain at the stack base, with the additional restriction that the height
of the plume always be at least half the height above ground as the Ini-
tial effective stack height.
5.2.2 Dispersion
The MESOPUFF-II, RIVAO, and POLUT models all contain algorithms for pre-
dicting Initial stack heights for the source plumes. The RIVAO uses
BHggs's (1972) formula for final plume rise and the POLUT and MESOPUFF-II
use plume rise equations as described by BHggs (1975). The CCADM assumes
all emissions are Instantaneously mixed within the mixed-layer of the
Lagranglan box. As described below, both the RIVAO and the POLUT models
also contain algorithms that account for buoyancy Induced dispersion for
strongly buoyant plumes.
5-39
-------
The MESOPUFF-II, RIVAD, and POLUT all represent dispersion by expanding
the puffs or plume-segments 1n terms of the puff dispersion parameters oy
and oz. The MESOPUFF-II calculates o and DZ for distances out to 100 km
using formulas fitted to the curves of Turner (1970). For distances
greater than 100 km the plume growth rates given by Heffter (1965) are
used. The Implementation of the plume expansion at each time step 1s 1n
the differential form:
AS) » ou(s) + AS
do
s
AS
(5-7)
*r
so that the puffs always grow with time. The Integral formulas for o and
oz for travel distances less than 100 km are as follows:
oy(s) » a s°'9
oz(s) - c Sd (5-8)
where a, c, and d are stability-dependent constants (Benkley and Bass,
1979b). For distance greater than 100 km, dispersion 1s based on time, t
(seconds), Instead of downwind distance using the following formulas
(Heffter. 1965):
oy(t + At) - oy(t) + 0.5 At
o2(t + At) - oz(t) + d-At//t (5-9)
where d 1s a stability-dependent parameter. The vertical extent of the
plume defined by oz 1s constrained by the mixing depth.
Horizontal dispersion 1n the RIVAD 1s treated using an approach suggested
by Randerson (1972) that accounts for the effect of vertical wind shear.
On the basis of field measurements, Randerson found that d1ffus1v1t1es
Increase rapidly during a transition phase typically lasting about 10
hours. During this phase
where
-------
°y
where KH- « 7 x 108 cm2/s.
Vertical dispersion 1n the RIVAD 1s handled 1n a somewhat similar way 1n
that dispersion 1s keyed to transport time:
where k 1s determined from Pasqu111-G1fford curves to be 2.10, 1.09, 0.53,
0.36, and 0.30 for stabilities A, B, C, D, E, and F, respectively.
Vertical downward dispersion 1n RIVAD 1s ultimately limited by the ground,
and vertical upward dispersion by the height of the mixed layer Hm.
The horizontal dispersion 1n the PQLUT model assumes that the square of
the total horizontal diffusion, c , 1s the sum of the squares of three
components: an 1nt1a1 buoyancy Induced dispersion (Ay), diffusion result-
Ing from atmospheric turbulence (By), and diffusion resulting from hori-
zontal wind shear (Cy).
°y " (Ay * By + cy) • (
Similarly, the vertical diffusion coefficient 1s
°z " Az * Bz
where AZ 1s the Initial buoyancy Induced dispersion and BZ 1s the vertical
diffusion due to atmospheric turbulence. The formulas used 1n calculating
the diffusion coefficient 1n the POLUT are complex and Involve the use of
downwind distance, travel time, standard deviation of the horizontal and
vertical components of the wind, terrain roughness, Monln-Obukov length,
and friction velocity. These formulas are presented 1n Appendix A.
The CCAOM uses the simplest description for horizontal and vertical dif-
fusion. In the CCADM the user has an option of either specifying a hori-
zontal and/or vertical diffusion coefficient along the edges of the Lag-
rang 1 an box, and/or specifying a box expansion rate.
5-41
-------
5.2.3 Chemical Transformation
The POLUT model 1s currently configured for Inert species only and thus
contains no chemistry module. The MESOPUFF-II and the RIVAD both contain
pseudo first-order chemical reactions that basically contain the following
reactions:
S02 * SO'
NOX * HN03
The MESOPUFF-II has several parameter 1zat Ions of the rate constants for
these reactions. For the ERT method the rate constants for these reac-
tions have been parameterized 1n terms of the environmental conditions
such as solar radiation, relative humidity, temperature, and background
ozone concentrations. These pseudo first-order reaction rate constants
were obtained by exercising a Lagranglan photochemical box model (Atkinson
et al., 1982) over a wide range of environmental conditions. Stepwlse
linear regression on the logarithms of the output variables was performed
to determine the controlling variables and the best regression equa-
tions. The aqueous-phase conversion of S02 to SO* was tied to relative
humidity and limited to a maximum of 3 %/h. Other parameterlzatlons of
rate constants are based on the work of 611lan1 and co-workers (1981)
after analyzing data from the St. Louis plume and results obtained by
Henry and H1dy (1981, 1982) who employed principal component analysis of
urban aerometrlc data.
The RIVAD pseudo first-order rate constants are determined by estimating
the steady-state OH* concentration 1n the plume as a function of sunlight
NOX and Og. This calculation 1s tuned for a rural hydrocarbon concen-
tration and there 1s an urban adjustment factor for Industrialized
areas. The heterogenous (aqueous) oxidation of S02 to SOj 1s simulated by
adding a constant 0.2 %/h oxidation rate to the homogeneous oxidation
rate.
The CCAOM model contains the most sophisticated gas-phase and aqueous-
phase chemical mechanism of the four candidate models. The gas-phase
mechanism uses the Carbon Bond Mechanism, CBM-X (WMtten et al., 1985a,b),
while the aqueous-phase mechanism 1s based on extensive reviews of work by
Jacob (1986), Walcek and Stockwell (1986), Hoffman and Calvert (1985), and
others. This mechanism represents the current state-of-the-art knowledge
5-42
-------
of chemistry that leads to add deposition. The CCAOM chemistry mechanism
1s described 1n detail 1n Appendix B.
5.2.4 Dry Deposition
Two of the candidate models, the MESOPUFF-II and the CCADM, use the pre-
ferred resistance approach for the parameterization of dry deposition.
The RIVAO uses the dry deposition velocity concept, while the POLUT does
not consider pollutant loss due to dry deposition.
The flux of pollutants to the ground due to dry deposition can be
expressed as:
Fd « Vd c (5-15)
where Vd 1s the deposition velocity, and c 1s concentration at some refer-
ence height. In the RIVAO the deposition velocity 1s a function of land-
use type and 1s set to zero at night to account for the shielding effect
of the stable nocturnal boundary layer. During the day however, the depo-
sition velocity 1s applied to the whole mixed layer concentration which
effectively enhances the rate of vertical diffusion of pollutants because
mass removed at the surface 1s Immediately replaced with material from
above.
\
In the resistance approach to dry deposition, the deposition velocity 1s
expressed as the Inverse sum of the atmosphere, surface, and canopy resis-
tances (see Section 3.1.3):
vd ' <5
In the MESOPUFF-II an option exists to treat puffs that have become ver-
tically well-mixed with a three-layer model. This parameterization
essentially removes the enhanced rate of vertical diffusion by only con-
sidering the loss of pollutants out of the surface layer.
The CCADM uses the dry deposition algorithm 1n the NCAR/RADM for gaseous
species (Walcek et al., 1986) and the algorithm 1n the ERT/AOOM for par-
.tlculate species (Plelm, Venkatram, and Yamertlno, 1984). The CCADM
parameterization of dry deposition also uses a surface layer to minimize
the effect of enhanced depletion due to Instantaneous vertical mixing.
5-43
-------
5.2.5 Wet Deposition
Wet removal of pollutants Includes both 1n-c1oud scavenging (ralnout) and
below cloud scavenging (washout). The precipitation scavenging of par-
tlculates, such as sulfates and nitrates, 1s function of the cloud type
and history. For example, 1n an 1ce cloud 1t 1s very difficult for the
particles to become embedded 1n the 1ce crystals except through the pro-
cess of riming.
The scavenging efficiency of gaseous species depends on their solubility
and reactivity. Precipitation scavenging of gaseous 1s also reversible,
I.e., gases caught 1n cloud raindrops may :be released back Into the atmo-
sphere before the drop Impacts the ground.
The MESOPUFF-II uses the scavenging coefficient approach to wet deposi-
tion, while the CCADM and RIVAO use wet deposition algorithms based on
solubility (Henry's Law) for gases, and Scott's (1978) parameterization
for participates. The POLUT does not treat wet scavenging.
In the MESOPUFF-II, the loss of pollutant mass over a time step At(s) due
to a precipitation rate R (mm/h) 1s expressed as follows:
Q(t + At) - Q(t) exp(-AAt) (5-17)
where A (s ) 1s the scavenging ratio expressed as:
A « x(R/Rx) (5-18)
Here Rj 1s a reference rainfall rate (1 mm/h) and x 1s a scavenging coef-
ficient (s'1) whose value depends on the species and whether R 1s liquid
or frozen precipitation. The rainfall rate, R, 1s that observed at the
surface station closest to the center of the puff.
Although both the RIVAO and CCADM assume solubility for calculating
precipitation scavenging of gaseous species, the Implementation of these
calculations 1n the two models 1s quite different. Both follow the method
suggested by Hales and Sutter (1973), which assumes that gaseous
scavenging 1s a reversible process.
The scavenging of pollutant mass by precipitation can generally be
expressed by the following formula:
o R
W - -g- x (5-19)
where ou 1s the density of water, R the precipitation rate, H the layer
thickness, and x the ratio of mass of the pollutant 1n the raindrop to the
5-44
-------
mass of rainwater. For gaseous species x will be a function of the
Henry's law constant and the pH of the rainwater. The RIVAD assumes a
constant pH (4.5 for the eastern U.S.) while CCADM explicitly calculates
the pH based on an 1on/an1on balance (see Appendix B) and the rainwater
content Input from the cloud physics module. Both RIVAO and CCADM assume
that part1culate ralnout and washout 1s Irreversible and use the
algorithms of Scott (1978) and the ERT/ADOM model respectively.
5-45
-------
CONCEPTUAL DESIGN OF AN ACID DEPOSITION MODEL
FOR THE ROCKY MOUNTAIN REGION
The evaluation of the four candidate add deposition models (Section
5.2.1) has Indicated that no single candidate 1s the best choice for cal-
culating source specific add deposition Impacts 1n the Rocky Mountain
region. However, the evaluation has Indicated that the most flexible
modeling approach would be the Gaussian puff model formulation. However
neither of the candidate Gaussian puff models (the ERT/MESOPUFF-II or
PNL/MELSAR-POLUT) appear to be superior 1n all processes that lead to add
deposition. The POLUT model appears to describe transport and dispersion
In complex terrain better than the MESOPUFF-II; however, the POLUT does
not treat chemical transformation or scavenging. Thus for developing an
add deposition model for the Rocky Mountain region, the concept of a host
model will be used. The host model will use the Gaussian puff model for-
mulation, and the treatment of add deposition processes will be based on
modules taken from the four candidate models.
6.1 TRANSPORT
The transport of the Gaussian puff will be based on the horizontal wind
field defined by the puff centre1d. This 1s believed to produce more
realistic trajectories than 1s obtained by Interpolating between vertical
levels of wind fields, which may produce Impossible air parcel trajec-
tories 1f decoupled wind field conditions exist. Since there 1s great
uncertainty 1n the vertical velocities generated by the diagnostic wind
models, 1t would be Inappropriate to use them for puff transport. Instead
an empirical technique of plume height, as used 1n the POLUT model, would
be used.
Provisions will be made 1n the model coding to allow for future develop-
ment of vertical separation of a single puff when decoupled flow field
conditions exist. However the model development effort required to Imple-
ment the puff separation 1s beyond the scope of this project.
At this time 1t 1s unclear whether the puff sampling method used by the
MESOPUFF-II, or the puff splitting technique employed by the POLUT will
produce the most realistic representation of a continuous plume 1n complex
6-1
-------
terrain. The choice between these two techniques, or using a combination
of the two algorithms, will be made after the models are tested 1n actual
complex terrain conditions.
6.2 DISPERSION
The horizontal and vertical diffusion coefficients will be based on the
parameterization used by the POLUT model (see Appendix A). Of the four
candidate models this parameterization exhibits the most complete descrip-
tion of diffusion over complex terrain.
6.3 CHEMICAL TRANSFORMATION
The CCAOM contains the most comprehensive chemistry module of the candi-
date models. However the computational requirements and the need for an
ambient field of background concentrations may preclude Its use. The
parameterized pseudo-first-order chemistry mechanisms 1n the MESOPUFF-II
were generated based on simulations relevant to the eastern U.S. and aero-
metric data from St. Louis and other urban areas. Thus these parameter-
ized mechanisms 1n the MESOPUFF-II may not be appropriate for the Rocky
Mountain west. The simplified chemical mechanism used 1n the RIVAD con-
tains no explicit account of the aqueous oxidation of S02 to sulfate.
There have been a few limited studies of the oxidation of SO? and NOX 1n
the Rocky Mountains (Roberts et al., 1983, 1984, 1985; Countess, Wolff,
and Cadle, 1980; Eldred et al., 1983; Henml and Bresch, 1985; Cahlll et
al., 1981; FloccMnl et al., 1981). These studies have focused mainly on
measurement studies on the eastern side of the front range. Only one
extensive western regional modeling study has been carried out to date
(Latlmer et al., 1985a,b). However, this study, using the RIVAO model,
used a simplified chemical kinetic mechanism, and thus did not predict the
concentrations of background oxidizing agents, and oxldant precursors. In
order to correctly specify the oxidation of S02 to sulfate, and NOX to
nitrates and nitric add, the characteristics of the background reactivity
of the atmosphere 1n the western Rocky Mountains 1s required. Thus we
propose the following approach for developing a cost effective, yet com-
prehensive chemistry module to use with an add deposition model for the
Rocky Mountain region.
The heart of the proposed chemistry module would Involve multiple simula-
tions of the SAI/CCAOM model over portions of the western Rocky Mountains
using wide variations of meteorological and aerometrlc conditions. The
CCAOM chemistry module Incorporates all current knowledge of gas and aque-
ous-phase chemistry that leads to add deposition (see Appendix B). The
6-2
-------
CCAOH would be run 1n a nested box mode, where a large Lagranglan box
trajectory would define the ambient conditions for a source specific CCAOM
simulation. These many simulations of the CCADM would form the basis for
a multidimensional lookup table that would contain the oxidation rates of
SOX and NOX as a function of the primary components of oxidation. These
primary components would Include, but not be limited to, the following
factors: solar Intensity, cloud cover, cloud type, relative humidity and
temperature, NOX, hydrocarbon, ozone, and hydrogen peroxide concentra-
tions. The determination of the principal components that describe the
oxidation, decay and formation of add species and their precursors would
be obtained through use of the statistical techniques of Factor Analysis
(FA) or the more recent extension of FA to nonlinear systems known as
Exploratory Projection Pursuit (Friedman, 1985). Once the cost effective
chemistry module based on the multidimensional lookup table 1s developed,
1t will be Inserted Into the CCADM and verified by doing s1de-by-s1de
simulations of the full CCAOM and the reduced CCAOM over the western Rocky
Mountain region.
It should be noted that a simplified version of this approach to simplify-
ing chemistry 1n regional models was used for the SAI/RTM-IINL model 1n
the development of Its nonlinear rate table. In the SAI/RTM-IINL rate
table,the principal components consisted of NOX, RHC, t^O? concentra-
tions, and the N02 photolysis rate. For this application It 1s expected
to Include more principal components, however the statistical techniques
described above will allow for the Identification, of the principal compo-
nents as well as an estimation of how much variance 1n the chemical rates
the principal components describe.
6.4 DRY DEPOSITION
The preferred approach for the modeling of dry deposition Involves the use
of the resistance concept. Both the MESOPUFF-II and CCADM employ the
resistance approach 1n their dry deposition modules. They also both
Incorporate a surface layer that reduces the enhanced dispersion associ-
ated with depositing pollutants out of the entire mixed layer. The choice
of one of these dry deposition modules cannot be made at this time.
Instead the two modules will be exercised using a range of conditions that
characterize the atmosphere over the Rocky Mountains. Since there are
currently no reliable measurement data for dry deposition, the choice of
one of these dry deposition modules will ultimately be based on the
reasonableness of the results, the Input requirements, and the flexibility
for adaptation to the Rocky Mountain region.
6-3
-------
6.5 WET DEPOSITION
The three wet deposition modules employed by the MESOPUFF-II, RIVAD, and
CCADM make use of a scavenging coefficient approach, a simple solubility
calculation, and a more complicated solubility calculation. To use the
solubility approach, Information concerning the total load of pollutants
1n the atmosphere 1s needed to characterize the cloud pH, and the amount
of pollutants that are 1n the liquid phase within the cloud. Since the
conceptual design of the hybrid model will be based on the Gaussian puff
model formulation, the model will only have Information pertaining to the
specific source concentrations. The use of the solubility approach may
not be valid 1n a puff model.
Thus the conceptual design of the Rocky Mountain add deposition model
will make use of a modified version of the MESOPUFF-II scavenging coef-
ficient approach for the wet deposition module. Modifications to this
module will Include use of grid cell specific storm characteristics,
rather than relying on Information from the nearest surface stations.
6-4
-------
REFERENCES
AeroVlronment, Inc. 1986. Letter to Systems Applications, Inc., cri-
tiquing Latlmer et al. (1986) study of visibility 1n the western
states.
Allwlne, K. J., and C. D. WMteman. 1985. "Green River A1r Quality Model
Development: MELSAR--A Mesoscale A1r Quality Model for Complex Ter-
rain. Volume 1—Overview, Technical Description and User's Guide."
Prepared by Pacific Northwest Laboratory, Rlchland, Washington, for
U.S. Environmental Protection Agency (EPA 600/8-85-0176; NTIS PB 85-
247211).
Alpert, P. 1981. Implicit filtering 1n conjunction with explicit filter-
ing. J. Comput. Phy.. 44:212-219.
Alpert, P., and J. Neumann. 1983. A simulation of a Lake Michigan
winter land breeze on the 7th of November 1978. Hon. Weather .Rev..
111:1873-1881.
Ames, J., T. C. Myers, L. E. Reid, 0. C. Whitney, S. H. Goldlng, S. R.
Hayes, and S. D. Reynolds. 1978. "User's Manual for the SAI Airshed
Model." Systems Applications, Inc., San Rafael, California (SYSAPP
78-89).
Anderson, G. E. 1971. Mesoscale Influences on wind fields. J. Appl.
Meteor.. 10:377-386.
Anthes, R. A. 1983. A review of regional models of the atmosphere 1n
middle latitudes. Mon. Weather Rev.. 111:1306-1335.
Anthes, R. A., Y.-H. Kuo, S. G. Benjamin, and Y.-F. L1. 1982. The evolu-
tion of the mesoscale environment of severe local storms: Prelimi-
nary modeling results. Mon. Weather Rev.. 110:1187-1213.
Anthes, R. A., and T. T. Warner. 1978. Development of hydrodynamlc
models suitable for air pollution and other mesometeorological
studies. Mon. Weather Rev.. 106:1045-1078.
R-l
-------
Ashbaugh, L.t E. Fujlta, M. Ahaja, J. Sanders, K. Tonnessen, and 0. Wes-
terdahl. 1985. "The Third Annual Report to the Governor and the
Legislature on the A1r Resources Board's Add Deposition Research and
Monitoring Program." California A1r Resources Board, Sacramento,
California.
Atkinson, B. W." 1981. Meso-scale Atmospheric Circulations. Academic
Press.
Atkinson, R., A. C. Lloyd, and L. Wlnges. 1982. An updated chemical
mechanism for hydrocarbon/NOx/SOx photooxldatlons suitable for Inclu-
sion 1n atmospheric simulation models. Atmos. Environ.. 16:1341.
Austin, D. I., A. W. Bealer, and W. R. Goodln. 1981. "Random-Walk Advec-
tlon and Dispersion Model (RADM),n Dames & Moore.
Balko, J. A., and L. K. Peters. 1983. A modeling study of SOx-NOx-hydro-
carbon plumes and their transport to the background troposphere.
Atmos. Environ.. 17(10):1965-1978.
Ballentlne, R. J. 1980. A numerical Investigation of New England coastal
frontogenesls. Mon. Weather Rev.. 108:1479-1497.
Ballentlne, R. J. 1981. Numerical simulation of land-breeze-Induced
snowbands along the western shore of Lake Michigan.' Mon. Weather
Rev.. 110:1544-1553.
Baron, J. 1986. Personal communication.
Barry, R. 6. 1973. Climatologies! transect on the east slope of the
Front Range, Colorado. Arctic and Alpine Research. 5(2):89-110.
Belike, S. 1970. "Laboratory Investigations on Washout of Trace
Gases." U.S. Atomic Energy Commission, AEC Symposium Series 22, pp.
261-269.
Bender, M. A., R. E. Tuleya, and Y. Kurlhara. 1985. A numerical study of
the effect of a mountain range on a landfalUng tropical cyclone.
Mon. Weather Rev.. 113:567-582.
Bengtsson, L., M. GM1, and E. Kail en. 1981. Dynamic meteorology: data
assimilation methods. Applied Mathematical Sciences. Vol 36.
Sprlnger-Verlag, New York.
R-2
-------
Benjamin, S. G. 1986. Some effects of surface heating and topography on
the regional severe storm environment. Part II: Two-dimensional
Idealized experiments. Mon. Weather Rev.. 114:330-343.
Benjamin, S. G., and T. N. Carlson. 1986. Some effects of surface heat-
Ing and topography on the regional severe storm environment. Part
I: Three-dimensional simulations. Mon. Weather Rev.. 114:307-329.
Benjamin, S. G., and N. L. Seaman. 1985. A simple scheme for objective
analysis 1n curved flow. Mon. Heather Rev.. 113:1184-1198.
Benkley, C. W., and A. Bass. 1979a. Development of Mesoscale A1r Quality
Simulation Models. Vol. II: User's Guide to MESOPLUME (Mesoscale
Plume Segment) Model. U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina (EPA-600/7-80-057).
Benkley, C. W., and A. Bass. 1979b. "Development of Mesoscale A1r
Quality Simulation Models, Vol. Ill: User's Guide to MESOPUFF (Meso-
scale Puff) Model." Environmental Research and Technology, Inc.,
Concord, Massachusetts.
Benkovltz, C. M. 1980. MAP3S/Ra1ne Emissions Inventory Progress
Report. Brookhaven National Laboratory, Long Island, New York (BNL
51378).
Bhumralkar, C. M. 1973. An observational and theoretical study of atmo-
spheric flow over a heated Island. Parts I and II. Mon. Weather
Rev.. 101:719-745.
Bhumralkar, C. M. 1974. Effect of moisture on hydrostatic pressure.
J. Appl. Meteorol.. 13:174-175.
Bhumralkar, C. M., W. B. Johnson, R. L. Mancuso, R. A. ThullHer, 0. E.
Wolf, and K. C. N1tz. 1980. "Interregional Exchanges of Airborne
Sulfur Pollution and Deposition 1n Eastern North America." Second
Joint Conference on Application of A1r Pollution Meteorology and
Second Conference on Industrial Meteorology (24-27 March 1980, New
Orleans, Louisiana).
« »
Blondln, C. 1979. "Simulation numerlque de 1'atmosphere a mesoechelle.
In Proceedings of the First Symposium on Physico-chemical Behaviour
of Atmospheric Pollutants. ISPRA, October, pp. 479-491.
Bolln, B., and C. Persson. 1975. Regional dispersion and deposition of
atmospheric pollutants with particular application to sulfur pollu-
tion over western Europe. Tellus. 27(3):281-309.
R-3
-------
Bornsteln, R. D. 1972. "Two Dimensional Non-steady Numerical Simulations
of Nighttime Flow of a Stable Planetary Boundary Layer over a Rough,
Warm City." Ph.D. thesis, New York University.
Bornsteln, R. D. 1975. The two dimensional URBMET urban boundary layer
model. J. Appl. Meteorol.. 14:1459-1477.
Boudra, D. B. 1981. A study of the early winter effects of the Great
Lakes. I: Comparison of very fine scale numerical simulations with
observed data. Mon. Weather Rev.. 109:2507-2526.
Brlggs, G. A. 1972. Chimney plumes 1n neutral and stable surroundings.
Atmos. Environ.. 6(7):507-510.
Brlggs, G. A. 1975. "Plume Rise Predictions." In Lectures on A1r Pollu-
tion and Environmental Impact Analyses. American Meteorological
Society, Boston, Massachusetts, pp. 59-111.
Brown, P. S., Jr., J. P. Pandolfo, and S. J. Thoren. 1982a. GATE air-sea
Interactions. I: Numerical model calculation of local sea surface
temperatures on diurnal time scales using the GATE version III grld-
ded global data set. J. Phys. Oceanogr.. 12:483-494.
Brown, P. S., Jr., J. P. Pandolfo, S. J. Thoren, and G. D. Robinson.
19825. GATE air-sea Interactions. II: Numerical model calculation
of regional sea-surface temperature fields using the GATE version III
grldded global data set. J. Phys. Oceanoqr.. 12:1051-1070.
BurHdge, D. M., and J. Haseler. 1977. A model for medium range weather
forecasts—adlabatlc formulation. ECMWF Tech. Rep.. No. 4. 46 pp.
Butzer, K. W. 1976. Geomorpholoqy from the Earth. Harper and Row, New
York.
Cahlll, T. A., B. H. Kusko, L. L. Ashbaugh, J. B. Barone, R. A. Eldred,
and F. G. Walther. 1981. Regional and local determinations of
part1culate matter and visibility 1n the southwestern United States
during June and July, 1979. Atmos. Environ.. 15(10/11):2011-2016.
CahUl, T. A., R. G. Flocch1n1, R. A. Eldred, P. J. Feerney. 1983.
"Western Partlculate Characterization Study." U.S. Environmental
Protection Agency, Environmental Systems Laboratory, Las Vegas, NV.
R-4
-------
Calvert, J. G.( and W. R. Stockwell. 1983. "The Mechanism and Rates of
the Gas-Phase Oxidation of Sulfur D1xo1de and the Nitrogen Oxides 1n
the Atmosphere." In Add Precipitation; S02, NO, and N02 Oxidation
Mechanisms: Atmospheric Considerations, Ann Arbor Science
Publishers, Ann Arbor, Michigan.
Carlson, J. D., and N. R. Foster. 1986a. Numerical study of some
unstably stratified boundary-layer flows over a valley at moderate
Richardson number. J. CUm. Appl. Meteorol.. 25:203-213.
Carlson, J. D., and M. R. Foster. 1986b. Numerical study of some neu-
trally and unstably stratified boundary-layer flows over a valley at
small Richardson number. Tellus. 38A(l):49-59.
Carmlchael, G. R., and L. K. Peters. 1979. "Numerical Simulation of
the Regional Transport of S02 and Sulfate 1n the Eastern United
States." Proc. Fourth Symposium of Turbulence. Diffusion, and A1r
Pollution (15-18 January 1979, Reno, Nevada), American Meteorological
Society, Boston.
Carmlchael, G. R., and L. K. Peters. 1984a. An Eulerlan transport/
transformation/removal model for SO? and sulfate—I. Model develop-
ment. Atmos. Environ.. 18(5):937-95l.
Carmlchael, G. R., and L. K. Peters. 1984b. An Eulerlan transport/
transformation/removal model for S02 and sulfate—II. Model calcula-
tion of SOX transport 1n the eastern United States. Atmos. Environ..
18(5):953-967.
Carmlchael, G. R., L. K. Peters, and T. Kltada. 1986. A second genera-
tion model for regional-scale transport/chemistry/deposition.
Atmos. Environ.. 20(1):173-188.
Ceselskl, B. F., and L. L. Sapp. 1975. Objective wind field analysis
using line Integrals. Mon. Weather Rev.. 103:89-100.
Chamberlain, A. C. 1967. Transport of Lycopodlum spores and other small
particles to rough surfaces. Proc. Roy. Soc. London (A). 296:45-70.
Chan, S. T., H. C. Rodean, and 0. L. Ermak. 1982. "Numerical Simulations
of Atmospheric Releases of Heavy Gases over Variable Terrain." Proc.
NATO/CCMS 13th International Technical Meeting on A1r Pollution
Modeling and Its Applications. September 1982, France.
R-5
-------
Chang, C. B., 0. J. Perkey, and C. W. Kreltzberg. 1981. A numerical case
study of the squall line of 6 May 1975. J. Atmos. Sc1.. 38:1601-
1615.
Chang, L. P., E. S. Takle, and R. L. San1. 1982. Development of a two-
dimensional finite-element PBL model and two preliminary model appli-
cations. Hon. Weather Rev.. 110:2025-2037.
Chlen, H.t and P. J. Smith. 1973. On the estimation of kinematic
parameters 1n the atmosphere from radiosonde wind data. Hon. Weather
Rev.. 101:252-261.
Cho, S.-Y. 1986. "Mathematical Modeling and Sensitivity Analysis of Add
Deposition." Ph.D. thesis, University of Iowa.
Clark, T. L. 1977. A small-scale numerical model using a terrain-follow-
ing coordinate system. J. Comput. Phys.. 24:186-215.
Clark, T. L. 1979. Numerical simulations with a three-dimensional cloud
model: Lateral boundary condition experiments and multi-cellular
severe storm simulations. J. Atmos. Sc1.. 34:2191-2215.
Clark, T. L. 1982. Cloud modeling 1n three spatial dimensions. Hail-
storms of the Central High Plains. Vol. 1. National Hall Research
Experiment, Charles A. Knight and Patrick Squires, Eds. Colorado
Associated University Press, 282 pp.
Clark, T. L., and R. D. Farley. 1984. Severe downslope windstorm calcu-
lations 1n two and three spatial dimensions using anelastlc Interac-
tive grid nesting: A possible mechanism for gustlness. J. Atmos.
Sc1.. 41:329-350.
Clark, T. L., and R. Gall. 1982. Three-dimensional numerical model simu-
lations of airflow over mountainous terrain: A comparison with
observations. Mon. Weather Rev.. 110:766-791.
Countess, R. J., 6. T. Wolff, and S. H. Cadle. 1980. The Denver winter
aerosol: A comprehensive chemical characterization. J. A1r Pollut.
Control Assoc.. 30(11):1194ff.
Cox, R. A., and S. A. Penkett. 1972. Aerosol formation from sulfur
dioxide in the presence of ozone and olefinlc hydrocarbons. J. Chem.
Soc.. Faraday Trans. I. 68:1735-1753.
R-6
-------
Cram, J. M. 1985. "VaHatlonal Assimilation of Asynoptlc Data on 10
April 1979." 7th Conference on Numerical Weather Prediction, Mon-
treal, Canada (17-20 June).
Cram, J. M., and M. L. Kaplan. 1985. Var1at1onal assimilation of VAS
data Into a mesoscale model; assimilation method and sensitivity
experiments. Mon. Weather Rev.. 113(4).
Daggupaty, S. M., P. K. M1sra, and R. E. Munn. 1979. "A Numerical
Experiment 1n Trajectory Modeling for the Transport of A1r Pollutants
over the Great Lakes Region." WMO Symposium on the Lonq-Ranqe Trans-
port of Pollutants and Its Relation to General Circulation Including
StratospheHc/Tropospherlc Exchange Processes (Sofia, October 1-5,
1979), WMO No. 538.
Oana, M. T., D. R. Drewes, D. W. Glover, and J. M. Hales. 1976. Precipi-
tation Scavenging of Fossil Fuel Effluents. U.S. Environmental Pro-
tection Agency (EPA-600/4-76-031).
Dana, M. T., J. M. Hales, and M. A. Wolf. 1975. Rain scavenging of S02
and sulphate from power plant plumes. J. Geophys. Res.. 80:4119-
4129.
Danielson, E. F. 1974. Review of trajectory methods. Adv. 1n Geophys..
18(B):73-94.
Dasch, J. M. 1982. A comparison of surrogate surfaces for dry deposition
collection. Proc. Fourth International Conference on Precipitation
Scavenging. Dry Deposition, and Resuspenslon. Santa Monica, Califor-
nia, 29 November - 3 December, 1982.
Davis, C. G., S. S. Bunker, and J. P. Mutschlecner. 1984. Atmospheric
transport models for complex terrain. J. CUm. Appl. Meteorol..
23:235-238.
Davis, W. E., and L. L. Wendell. 1976. "Some Effects of Isentroplc
Vertical Motion Simulation 1n a Regional Scale Quasi-Lagranglan A1r
Quality Model." Proceedings of the Third Symposium on Atmospheric
Turbulence. Diffusion, and A1r Quality (October 19-22, 1976, Raleigh,
North Carolina), American Meteorological Society, Boston.
Deardorff, J. W., K. Ueyoshi, and Y. J. Han. 1984. Numerical study of
terrain-Induced mesoscale motions and hydrostatic form drag in a
heated, growing mixed layer. J. Atmos. Sc1.. 41:1420-1441.
R-7
-------
Defant, A. 1951. "Local Winds." In Compendium of Meteorology, American
Meteorological Society, Boston, Massachusetts, pp. 655-672.
DelTOsso, L. 1984. High-resolution experiments with the ECMWF model: A
case study. Mon. Weather Rev.. 112:1853-1883.
Dlckerson, M. H. 1978. MASCON—A mass consistent atmospheric flux model
for regions with complex terrain. J. Appl. Meteorol.. 17:241-253.
Dlckerson, M. H. 1980. "A Collection of Papers on Drainage Wind Studies
1n the Geysers Area of Northern California Part I." U.S. Department
of Energy (DOE ASCOT-80-7; NTIS UCD-18884).
Dleterle, D. A., and A. 6. Tingle. 1976. "A Numerical Study of Mesoscale
Transport of A1r Pollutants 1n Sea-Breeze Circulations." Proceedings
of the Third Symposium on Atmospheric Turbulence, Diffusion, and A1r
Quality (October 19-22, 1976, Raleigh, North Carolina), American
Meteorological Society, Boston.
Dlffenhoefer and Ferullo. 1983.
Draxler, R. R. 1977. A Mesoscale Transport and Diffusion Model. A1r
Resources Laboratory, Silver Spring, Maryland (NOAA Technical
Memorandum ERL ARL-64).
Draxler, R. R. 1979. Modeling the results of two recent mesoscale dis-
persion experiments. Atmos. Environ.. 13:1523-1533.
Draxler, R. R., and J. L. Heffter. 1981. "Workbook for Estimating the
Climatology of Regional-Continental Scale Atmospheric Dispersion and
Deposition over the United States." A1r Resources Laboratories, Sil-
ver Spring, Maryland (NOAA Technical Memorandum ERL ARL-96).
Dronamraju, M. 1986. "Evaluation of Regional A1r Pollution Models."
Ph.D. dissertation, University of Kentucky, Lexington.
Duewer, W. H., M. C. MacCracken, and J. J. Walton. 1978. "The Llvermore
regional air quality model: II. Verification and sample application
1n the San Francisco Bay Area." J. Appl. Meteorol.. 17:273ff.
Durran, D. R., and J. B. Klemp. 1982. The effects of moisture on trapped
mountain lee waves. J. Atmos. Sc1.. 39:2490-2506.
Durran, 0. R., and J. B. Klemp. 1983. A compressible model for the simu-
lation of moist mountain waves. Mon. Weather Rev.. 11:2341-2361.
R-8
-------
Durran, D., M. J. Meldgln, M. K. L1u, T. Thoem, and D. Henderson. 1979.
A study of long-range air pollution problems related to coal develop-
ment 1n the northern Great Plains. Atmos. Environ.. 13:1021-1037.
Eldred, R. A., L. L. Ashbaugh, T. A. CaMll, R. G. Flocch1n1, and M. L.
Pltchford. 1983. Sulfate levels 1n the Southwest during the 1980
copper smelter strike. J. A1r Pollut.. Contr. Assoc.. 33(2):110ff.
EUassen, A., 0. Hov, I.S.A. Isaksen, J. Saltbones, and F. Stordal.
1982. A Lagranglan long-range transport model with atmospheric boun-
dary layer chemistry. J. Appl. Meteorol.. 21:1645ff.
EUassen, A., and J. Saltbones. 1975. Decay and transformation rates of
S0£, as estimated from emission data, trajectories, and measured air
concentrations. Atmos. Environ.. 9:425-429.
Ellenton, G., B. Ley, and P. K. M1sra. 1985. A trajectory puff model of
sulfur transport for eastern North America. Atmos. Environ..
19(5):727-737.
Endllch, R. M., F. L. Ludwlg, and C. M. Bhumralkar. 1982. A diagnostic
model for estimating winds at potential sites for wind turbines. CL
C11m. Appl. Meteorol.. 21:1441-1454.
EPA. 1982. "Addendum/Supplemental Information for the MESOPUFF Model."
U.S. Environmental Protection Agency, Research Triangle Park, North
Carolina (EPA-DF-83/007d).
Epstein, I. R., K. Kustln, and R. H. S1moy1. 1982. Nitrous add decompo-
sition catalyzed by an Iron (III) complex: trls (3, 4, 7, 8-tetra-
methyl-1, I0-phenanthral1ne) Iron (III). J. Am. Chem. Soc.. 104:712-
717.
Fandry, C. B., and L. M. Leslie. 1984. A two-layer quas1-geostroph1c
model of summer trough formation 1n the Australian subtropical
easterlies. J. Atmos. Sc1.. 41:807-818.
Fawcett, E. B. 1977. Current capabilities 1n prediction at the National
Weather Service's National Meteorological Center. Bull. Am.
Meteorol. Soc.. 58:143-149.
Fay, J. A., D. GoTomb, and S. Kumar. 1985. Source apportionment of wet
sulfate deposition 1n eastern North America. Atmos. Environ.,
19(11):1773-1782.
R-9
-------
Fay, J. A., and J. J. Rosenzwelg. 1980. An analytical diffusion model
for long-distance transport of air pollutants. Atmos. Environ.,
14:359-365.
Fett, R. W., and P. M. Tag. 1984. The sea-breeze-Induced coastal calm
zone as revealed by satellite data and simulated by a numerical
model. Mon. Weather Rev.. 112:1226-1233.
Fisher, B.E.A. 1975. The long-range transport of sulfur dioxide. Atmos.
Environ.. 9:1063-1070.
Fisher, B.E.A. 1978. The calculation of long-term sulfur deposition 1n
Europe. Atmos. Environ.. 12:489-501.
Fltzjarrald, 0. R. 1984. Katabatic wind 1n opposing flow. J. Atmos.
Sc1.. 41:1143-1158.
Fleagle, R. G. 1947. The fields of temperature, pressure, and three-
dimensional motion 1n selected weather situations. J. Meteorol..
4:165-185.
Flocch1n1, R. 6., T. A. CahUl, M. L. Pltchford, R. A. Eldred, P. J.
Feeney, and L. L. Ashbaugh. 1981. Characterization of particles 1n
the arid West. Atmos. Environ.. 15(10/11):2017-2030.
Forchtgott, J. 1949. Wave currents on the leeward side of mountain
crests. Meteoroloqlcke Zpravy. 3:49-51. (Translation 1n British
Meteorological Office library).
Fosberg, M. A. 1984. "A Diagnostic Three-Dimensional Mass and Momentum
Conserving Wind Model for Complex Terrain." Third Conference on
Mountain Metetorology. American Meteorological Society, Boston,
Massachusetts.
Foster, M. R. 1985. Numerical study of some neutrally and unstably
stratified boundary-layer flows over a valley at small Richardson
number. Tellus. In press.
Fowler, D. 1978. Dry deposition of S02 on agricultural crops. Atmos.
Environ.. 12:369-373.
Fox, D. G. 1981. Judging air quality model performance. Bull. Am.
Meteorol. Soc.. 62:599-609.
R-10
-------
Friedman, J. H. 1985. "Exploratory Projection Pursuit." Department of
Statistics, Laboratory for Computational Statistics, Stanford Univer-
sity (Technical Report No. 18).
Fritsch, J. M., and C. F. Chappell. 1981. Preliminary numerical tests of
the modification of mesoscale convectlve systems. J. Appl.
Meteorol.. 20:910-921.
Fuchs, N. A. 1964. Mechanics of Aerosols. Pergamon Press. (Translated
from Russian).
Garrett, A. J., and F. G. Smith III. 1984. Two-dimensional simulations
of drainage winds and diffusion compared to observations. J. CUm.
Appl. Meteorol.. 23:597-610.
G11lan1, N. V. 1978. Project MISTT: Mesoscale plume modeling of the
dispersion, transformation and ground removal of S02. Atmos.
Environ.. 12:569-588.
G1llan1, N. V., S. KohH, and W. E. Wilson. 1981. Gas-to-part 1cle con-
version of sulfur 1n power plant plumes: I. Parameterization of the
gas phase conversion rates for dry, moderately polluted ambient con-
ditions. Atmos. Environ.. 15:2293-2313.
Godden, D., and F. Lurmann. 1983. "Development of the PLMSTAR Model and
Its Application to Ozone Episode Conditions 1n the South Coast A1r
Basin.' Environmental Research & Technology, Inc., Westlake Village,
California (ERT P-A702-200).
Godske, C. L., T. Bergeron, J. Bjerkness, and R. E. Bundgaard. 1957.
Dynamic Meteorology and Weather Forecasting. American Meteorological
Society, Boston, Massachusetts.
Goldberg, E. D. 1982. Atmospheric Chemistry; Report of the Dahlem Work-
shop on Atmospheric Chemistry, Berlin, 1982. Sprlnger-Verlag, New
York.
Goodln, W. R., et al. 1980. "Evaluation and Application of RADM, a
Lagranglan, Random-Walk Model for Advectlon and Dispersion." Dames
& Moore, Los Angeles, California.
Goodln, W. R., G. J. McRae, and J. H. Seinfeld. 1979. A comparison of
Interpolation methods for sparse data: Application to wind concen-
tration fields. J. Appl. Meteorol.. 18:761-771.
R-ll
-------
Goodln, W. R., G. J. McRae, and J. H. Seinfeld. 1980. An objective
analysis technique for constructing three-dimensional urban scale
wind fields. J. Appl. Meteorol.. 19:98-108.
Goodln, W. R., G. J. McRae, and J. H. Seinfeld. 1981. Reply. J. Appl.
Meteorol.. 20:92-94.
Greenfield, S. M. 1957. Rain scavenging of radioactive partlculate
matter from the atmosphere. J. Meteorol.. 14(2):115-125.
Greenland, D. E., and H. J. de BUj. 1977. Earth 1n Profile: A Physical
Geography. Canfleld Press, San Francisco.
Haagenson, P. L., M. A. Shapiro, and P. Mlddleton. 1981. A case study
relating high ground level ozone to enhanced photochemistry and 1 sen-
tropic transport from the stratosphere. J. Geophys. Res., 86:5231-
5237.
Hales, J. M. 1978. Wet removal of sulfur compounds from the atmo-
sphere. Atmos. Environ.. 12:389-399.
Hales, J. M., and M. T. Dana. 1979. Precipitation scavenging of urban
pollutants by convectlve storm systems. J. Appl. Meteorol., 18:294-
316.
Hales, J. M., D. C. Powell, and T. D. Fox. 1977. STRAM—An A1r Pollution
Model Incorporating Non-Linear Chemistry. Variable Trajectories, and
Plume Segment Diffusion. U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina (EPA-450-3-77-012).
Hales, J. M., and S. L. Sutter. 1973. Solubility of sulfur dioxide 1n
water at low concentrations. Atmos. Environ.. 7:997-1001.
Hales, J. M., J. M. Thorpe, and M. A. Wolf. 1970. "Washout of S02 from
the Plume of a Coal-Fired Power Plant." Battelle Memorial Institute,
Pacific Northwest Laboratory, Richland, Washington.
Hane, C. E. 1973. The squall line thunderstorm: Numerical experimenta-
tion. J. Amos. Sc1.. 30:1672-1690.
Heffter, J. L. 1965. The variations of horizontal diffusion parameters
with time for travel periods of one hour or longer. J. Appl.
Heteorol.. 4:153-156.
R-12
-------
Heffter, J. L. 1980. A1r Resources Laboratories Atmospheric Transport
and Dispersion Model, U.S. Department of Commerce, Silver Spring,
Maryland (NOAA Technical Memorandum ERL ARL-81).
Heffter, J. L. 1983. Branching Atmospheric Trajectory (BAT) Model. U.S.
Department of Commerce, Silver Spring, Maryland (NOAA Technical
Memorandum ERL ARL-121).
He11man, U.( and R. Dobosy. 1985. A nocturnal atmospheric drainage flow
simulation Investigating the application of one-dimensional modeling
and current turbulence schemes. J. CHrn. Appl. Meteorol.. 24:924-
936.
Henml, T., and J. F. Bresch. 1985. Meteorological case studies of
regional high sulfur episodes 1n the western United States. Atmos.
Environ.. 19(11):1783-1796.
Henml, T., and E. R. Relter. 1979. Long-Range Transport and Transforma-
tion of SO? and Sulfate. U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina (EPA-600/4-79-068).
Henry, R. C., and G. M. H1dy. 1981. Discussion of "MultlvaHate analysis
of particulate sulfate and other air quality variables. Part I.
Annual data from Los Angeles and New York." Atmos. Environ.. 15:424.
Henry, R. C., and 6. M. H1dy. 1982. Mult1var1ate analysis of participate
sulfate and other air quality variables by principle components.
Part II. Salt Lake City, Utah and St. Louis, Missouri. Atmos.
Environ.. 16:929-943.
Hess, S. L. 1959. Introduction to Theoretical Meteorology. Holt, R1ne-
hart and Winston, New York.
H111, A. C. 1971. Vegetation: a sink for atmospheric pollutants.
3. A1r Pollut. Control Assoc.. 21:341-346.
Hodur, R. M. 1982. Description and evaluation of NORAPS: The Navy
operational regional atmospheric prediction system. Mon. Weather
Rev.. 110:1591-1602.
Hoecker, W. H. 1977. Accuracy of various techniques for estimating
boundary-layer trajectories. J. Appl. Meteorol.. 16:374-383.
R-13
-------
Hoffman, M. R., and J. G. Calvert. 1985. "Chemical Transformation
Modules for Eulerlan Add Deposition Models. Vol. 2—The Aqueous-
Phase Chemistry. U.S. Environmental Protection Agency, Research Tri-
angle Park, North Carolina (EPA/600/3-85/017).
Hogstrom, U. 1974. Wet fallout of sulfurous pollutants emitted from a
city during rain or snow. Atmos. Environ.. 8:1291-1303.
Holnka, K. P. 1985. A comparison of numerical simulations of hydrostatic
flow over mountains with observations. Mon Weather Rev;. 113:719-
735.
Holton, J. R. 1979. Introduction to Dynamic Meteorology, second
edition. Academic Press.
Holzworth, G. C. 1972. Mixing Heights. Wind Speeds, and Potential for
Urban A1r Pollution throughout the Contiguous United States. U.S.
Environmental Protection Agency, Research Triangle Park, North Caro-
lina (AP-101).
Hong, M.-S., and G. R. Carmlchael. 1986. An Investigation of sulfate
production 1n an orographlc storm using a detailed transport/chemis-
try model coupled with a detailed cloud scavenging model. Atmos.
Environ.. 20(10]1:1989-1997.
Hov, 0., J. Schjoldager, and B. M. Wathne. 1982. Measurement and model-
Ing of the concentrations of terpenes 1n coniferous forest air. Sub-
mitted to J. Geophys. Res.
Huber et al., 1976. "Stack Placement 1n the Lee of a Mountain Ridge: A
Wind Tunnel Study." U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina (EPA-600/4-76-047).
Husar, R. B., and 0. E. Patterson. 1981. "Documentation of the Capita
Monte Carlo Model MCARLO." Atmospheric Sciences and Analysis Work
Group, U.S./Canada Memorandum of Intent on Transboundary A1r Pollu-
tion, 5 August 1980.
Jacob, D. J. 1986. Chemistry of OH 1n remote clouds and Us role 1n the
production of formic add and peroxymonosulfate. J. Geophys. Res..
91(D9):9807-9826.
Johannes, A. R., J. N. Galloway, and 0. E. Troutman. 1980. "Snowpack
storage and 1on release." Ecological Impact of Add Precipitation;
Proceedings of an International Conference. Sandefjord. Norway. 11-14
March 1980. D. Drablos and A. To!Ian, eds.
R-14
-------
Johannessen, H.t and A. Henrlksen. 1978. Chemistry of snow meltwater:
Changes 1n concentration during melting. Hater Resources Research,
14:615-619.
Johannessen, M., A. Skartvelt, and R. F. Wright. 1980. "Streamwater
Chemistry Before, During, and After Snowmelt. In Ecological Impact
of Add Precipitation: Proceedings of an International Conference.
Sandefjord, Norway, 11-14 March 1980. D. Drablos and A. To!Ian, eds.
Johnson, VI. B. 1983. Interregional exchanges of air pollution: Model
types and applications. J. A1r Pollut. Control Assoc.. 33(6):563-
574.
Johnson, W. B., C. M. Bhumralkar, R. L. Mancuso, and D. E. Wolf. 1978.
"The European Regional Model of A1r Pollution (EURMAP) and Its
Application to Short-Term Episodes." Proc. Ninth International
NATO/CCMS Technical Meeting on A1r Pollution Modeling and Its Appli-
cation. (28-31 August 1978, Toronto, Canada), North Atlantic Treaty
Organization.
Jones, R. W. 1980. A three-dimensional tropical cyclone model with
release of latent heat by the resolvable scales. J. Atmos. Sc1.«
37:930-938.
Jude1k1s, H. S., and A. G. Wren. 1978. Laboratory measurements of S02
and N02 depositions onto soil and cement surfaces. Atmos. Environ..
12:2315-2319.
Junge, C. E. 1958. The distribution of ammonia and nitrate 1n rain water
over the United States. Trans. Am. Geophys. Union. 39:241.
Junge, C. E. 1964. "Modification of Aerosol Size Distribution 1n the
Atmosphere." Meteorological and Geophysical Institute, Johannes
Gutenberg Un1vers1taet.
Junge, C. E., and P. E. Gustafson. 1956. Precipitation sampling for
chemical analysis. Bull. Am. Meteorol. Soc.. 37:244.
Junge, C. E., and P. E. Gustafson. 1957. On the distribution of sea salt
over the United States and Its removal by precipitation. Tellus.
9:164.
Junge, C. E., and R. T. Werby. 1958. The concentrations of chloride,
sodium, potassium, calcium, and sulfate 1n rain water over the United
States. J. Meteorol.. 25:427.
R-15
-------
Kaplan, M. L., J. W. Zack, V. C. Wong, and G. D. Coats. 1984. The Inter-
active role of subsynoptlc scale jet streak and planetary boundary
layer processes in organizing an Isolated convectlve complex. Mon.
Weather Rev.. 112(11).
Kaplan, M. L., J. W. Zack, V. C. Wong, and J. J. Tucdllo. 1982a.
Initial results from a mesoscale atmospheric simulation system and
comparisons with the AVE-SEMAME I data set. Mon. Weather Rev..
110(11).
Kaplan, M. L., 0. W. Zack, V. C. Wong, and J. J. Tucdllo. 1982b. "A
Sixth-Order Mesoscale Atmospheric Simulation System Applicable to
Research and Real-Time Forecasting Problems." Proceedings of the
Symposium on Mesoscale Modeling. Cooperative Institute for Mesoscale
Meteorological Studies, University of Oklahoma.
Kern, C. 0. 1975. "A Weather Information and Display System for Research
and Emergency Response, Savannah River Laboratory Environmental
Transport and Effects Research, Annual Report FY-1975." E. I. du
Pont Nemours and Company, Savannah River Laboratory, A1ken, South
Carolina (U.S. ERDA Report DP-1412).
Kessler, 1986. Personal communication.
Keyser, D., and R. A. Anthes. 1982. The Influence of planetary boundary
layer physics on frontal structure 1n the Hosklns-Bretherton hori-
zontal shear model. J. Atmos. Sc1.. 39:1783-1802.
King, 0. S., and S. S. Bunker. 1983. Application of atmospheric trans-
port models for complex terrain. J. CUm. Appl. Meteorol.. 23:239-
246.
KHada, T., G. R. Carmlchael, and L. K. Peters. 1984. Numerical simula-
tion of the transport of chemically reactive species under land- and
sea-breeze circulations. J. Climate Appl. Meteorol.. 23(8):1153ff.
Kltada, T., G. R. Carmlchael, and L. K. Peters. 1986. Effects of dry
deposition on the concentration-distribution of atmospheric pollu-
tants within land- and sea-breeze circulations. Atmos. Environ..
20(10): 1999-20.10.
R-16
-------
Klelnman, L. I. 1983. A regional scale modeling study of the sulfur
oxides with a comparison to ambient and wet deposition monitoring
data. Atmos. Environ.. 17:1107-1121.
Klemm, H. A., and R. J. Brennan. 1981. "Emissions Inventory for the SURE
Region." Electric Power Research Institute, Palo Alto, California
(EPRI-EA1913).
Koch, S. E. 1985. Ability of a regional-scale model to predict the gene-
sis of Intense mesoscale convectlve systems. Mon. Weather Rev..
113(10).
Koch, S. E., W. C. Sklllman, P. J. Kodn, P. J. Wetzel, K. F. Brill, D. A.
Keyser, and M. C. MeCumber. 1985. Synoptic scale forecast skill and
systematic errors 1n the MASS 2.0 model. Mon. Weather Rev.. 112(10).
Kodn, P. J., L. W. Uccell1n1, J. W. Zack, and M. L. Kaplan. 1984.
"Recent Examples of Mesoscale Numerical Forecasts of Severe Weather
Events Along the East Coast." National Aeronautics and Space Admini-
stration, Goddard Space Flight Center (Tech. Memo 86172).
Kodn, P. J., L. W. UccelUnl, J. W. Zack, and M. L. Kaplan. 1985. A
mesoscale numerical forecast of an Intense convectlve snowburst along
the east coast. Bull. Am. Meteorol. Soc.. 66(11).
Kreltzberg, C. W., and M. J. Leach. 1979. "Diagnosis and Prediction of
Tropospherlc Trajectories and Cleansing." Proceedings; Advisory
Workshop to Identify Research Needs on the Formation of Add Precipi-
tation. Electric Power Research Institute, Palo Alto, California
(EPRI EA-1074 WA-78-98).
Kr1shnamurt1, T. N. 1968. A diagnostic balance model for studies of
weather systems of low and high latitudes, Rossby Number less than
1." Mon. Weather Rev.. 96:197-207.
KMshnamurtl, T. N., M. Kananrftsu, R. Godbole, F. Carr, C. B. Chang, and
J. H. Chow. 1976. Study of a monsoon depression (II), dynamical
structure. J. Meteorol. Soc. Jpn.. 54:208-224.
Kr1shnamurt1, T. N., S. L. Low-Nam, and R. Pasch. 1983. Cumulus parame-
terization and rainfall rates. II. Mon. Weather Rev.. 111:815-828.
Kr1shnamurt1, T. N., H. L. Pan, C. B. Chang, J. Ploshay, 0. Walker, and
A. W. Oodally. 1979. Numerical weather prediction for GATE. Q. J.
Roy. Meteorol. Soc., 105:979-1010.
R-17
-------
Kuo, H. L., and W. H. Raymond. 1980. A quasi-one-dimensional cumulus
cloud model and parameterization of cumulus heating and mixing
effects. Mon. Weather. Rev.. 108:991-1009.
Kurlhara, Y.t and M. A. Bender. 1980. Use of a movable nested-mesh model
for tracking a small vortex. Mon. Weather Rev.. 108:1792-1809.
Kurlhara, Y.t and R. E. Tuleya. 1981. A numerical simulation study on
the genesis of a tropical storm. Mon. Weather Rev.. 109:1629-1653.
Lamb, R. G. 1983. Regional Scale (1000 km) Model of Photochemical Air
Pollution — I. Theoretical Formulation. U.S. Environmental Protec-
tion Agency, Research Triangle Park, North Carolina (EPA-600/3-83-
035).
Lange, R. 1978. ADPIC—a three-dimensional part1cle-1n-cell model for
the dispersion of atmospheric pollutants and Its comparison with
regional tracer studies. J. Appl. Meteorol.. 17:320-329.
Lange, R., and J. B. Knox. 1974. Adaptation of a Three-Dimensional Atmo-
spheric Transport-Diffusion Model to Ralnout Assessments. Lawrence
Llvermore Laboratory, University of California, Llvermore, Califor-
nia.
Latlmer, D. A., G. E. Anderson, H. Hogo, R. G. Ireson, R. E. Morris, .and
P. Saxena. 1984. "Visibility and Other A1r Quality Benefits of Sul-
fur Dioxide Emission Controls 1n the Eastern United States, Volume
1." Systems Applications, Inc., San Rafael, California (SYSAPP-83-
248).
Latlmer, D. A., et al. 1985a. "Modeling Regional Haze 1n the South-
west: A Preliminary Assessment of Source Contributions. Volume
1." Systems Applications, Inc., San Rafael, California (SYSAPP/85-
038).
Latlmer, D. A., L. R. Ch1nk1n, M. C. Dud1k, H. Hogo, and R. G. Ireson.
1985b. "Uncertainties Associated with Modeling Regional Haze 1n the
Southwest." Systems Applications, Inc., San Rafael, California
(SYSAPP-85/108).
Latlmer, D. A., and H. Hogo. 1983. "Effect of Class I Area Visibility
Protection on Power Plant Siting In the West." Volume I. Systems
Applications, Inc. San Rafael, California (SYSAPP-83/269).
Latlmer, D. A., et al. 1983. "Final A1r Quality Technical Report for the
Ulntah Basin Synfuels Development Environmental Impact Statement."
Systems Applications, Inc., San Rafael, California.
R-18
-------
Lavery, T. F., J. W. Thrasher, D. A. Godden, A. C. Lloyd, and G. M.
H1dy. 1978. "Regional Transport and Photochemical Model of Atmo-
spheric Sulfates." Proc. Ninth International NATO/CCMS Technical
Meeting on A1r Pollution Modeling and Its Application (28-31 August
1978, Toronto, Canada), North Atlantic Treaty Organization.
Lazrus, A. L., P. L. Haagenson, G. L. Kok, B. J. Huebert, C. W. Kreltz-
berg, G. E. Likens, V. A. Mohnen, W. E. Wilson, and J. W. Winches-
ter. 1983. Acidity 1n air and water 1n a case of warm frontal pre-
cipitation. Atmos Environ.. 17:581-591.
Lee, R. L., J. M. Leone, Jr., P. M. Gresho, and S. T. Chan. 1982. A
finite element model for environmental problems Involving complex
terrain. In Finite Element Flow Analysis. T. Kawal, ed., University
of Tokyo Press.
Lee, Y. N., and S. E. Schwartz. 1981. Reaction kinetics of nitrogen
dioxide with water at low partial pressure. J. Phys. Chem.. 85:840-
848.
Leslie, L. M., G. A. Mills, and D. J. Gauntlett. 1981. The Impact of
FGGE data coverage and Improved numerical techniques 1n numerical
weather prediction In.the Australian region. 0. J. Roy. Meteorol.
Soc.. 107:629-642.
Levlne, S. Z., and S. E. Schwartz. 1982. In-cloud and below-cloud
scavenging of nitric add vapor. Atmos. Environ. 16(7): 1725-1734.
Lewellen, W. S., R. I. Sykes, and D. Oliver. 1981. "The Evaluation of
MATHEW/AOPIC as a Real-Time Dispersion Model." Aeronautical Research
Associates of Princeton, Inc., Princeton, New Jersey (Report No. 442).
Lewellyn, W. S., R. I. Sykes, and D. Oliver. 1982. "The Evaluation of
MATHEW/ADPIC as a Real Time Dispersion Model." (NUREG/CR 2199; ARAP
Rep. No. 442).
Ley, B. 1981. "Documentation of the Ontario Ministry of the Environment
Statistical LRT Model OME-LRT." Atmospheric Sciences and Analysis
Work Group, U.S./Canada Memorandum of Intent on Transboundary A1r
Pollution, 5 August 1980.
L1u, M. K., and 0. R. Durran. "Development of a Regional A1r Pollution
Model and Its Application to the Northern Great Plains." U.S.
Envlronmenal Protection Agency, Denver, Colorado (EPA-908/1-77-001).
R-19
-------
L1u, M. K., R. E. Morris, and J. P. Klllus. 1984. Development of a
regional oxldant model and application to the northeastern United
States. Atmos. Environ.. 18:1145-1161.
L1u, M. K., T. C. Myers, and J. L. McElroy. 1979. Numerical modeling of
land and sea breeze circulation along a complex coastline. Math.
Computers S1mu1.. 21(4):359-367.
Liu, M. K., 0. A. Stewart, and D. Henderson. 1982. A mathematical model
for the analysis of add deposition. J. Appl. Meteorol.. 21:859-873.
Ludwlg, F. L., L. S. Gasldrek, and R. E. Ruff. 1977. Simplification of a
Gaussian puff model for real-time minicomputer use. Atmos. Environ..
11:431-436.
MacCracken, M. C., D. J. Wuebbles, J. J. Walton, W. H. Duewer, and K. E.
Grant. 1978. The Llvermore Regional A1r Quality Model: I. Concept
and development. J. Appl. Meteorol.. 17:254ff.
Maddox, R. A., and J. M. Frltsch. 1984. A new understanding of thunder-
storms: the mesoscale convectlve complex. Heatherwlse. 37:136.
Maddox, R. A., 0. J. Perkey, and J. M. Frltsch. 1981. Evolution of upper
tropospherlc features during the development of a mesoscale convec-
tlve complex. J. Atmos. Sc1.. 38:1664-1674.
Mahrer, Y., and R. A. Plelke. 1978. A test of an upstream spline Inter-
polation technique for the advectlve terms 1n a numerical mesoscale
model. Mon. Weather Rev.. 106:1758.
Makhon'ko, K. P. 1967. "Diurnal Changes of the Dust Content of the Sur-
face Layer of the Atmosphere." Army Foreign Science and Technical
Center, Washington, DC (translated from unidentified Russian source).
Mancuso, R. L., et al. 1979. "The Exchange of Sulfur Pollution between
the Various Countries of Europe Based on the EURMAP Model." Fourth
Symposium on Turbulence. Diffusion, and Air Pollution (15-18 January,
1979, Reno, Nevada), American Meteorological Society, Boston.
Martell, E. A. 1974. Radioactivity of tobacco trlchomes and Insoluble
cigarette smoke particles. Nature. 249:215-217.
Martin, L. R. 1983. "Kinetic Studies of Sulflte Oxidation 1n Aqueous
Solutions." In Add Precipitation; S02, NO, and N02 Oxidation
Mechanisms; Atmospheric Considerations. Ann Arbor Scientific Publi-
cations, Ann Arbor, Michigan.
R-20
-------
Martin, L. R., and D. E. Damschen. 1981. Aqueous oxidation of sulfur
dioxide by hydrogen peroxide at low pH. Atmos. Environ., 15:1615-
1621.
Martin, L. R., 0. E. Damschen, and H. S. Judelkos. 1981. The reactions
of nitrogen oxides with SO? 1n aqueous aerosols. Atmos. Environ.,
15:191-195.
Mason, B. J. 1971. The Physics of Clouds. Oxford: Clarendon Press.
Mass, C. F., and 0. P. Dempsey. 1985. A one-level, mesoscale model for
diagnosing surface winds 1n mountainous and coastal regions. Mon.
Weather Rev.. 113:1211-1227.
Mathur, M. B. 1983. A quasi-Lagranglan regional model designed for
operational weather prediction. Mon. Weather Rev.. 111:2087-2098.
Maul, P. R. 1977. The mathematical modeling of the mesoscale transport
of gaseous pollutants. Atmos. Environ.. 11:1191-1195.
Maul, P. R. 1978. "A Prototype Model for the Long Range Transport of
Sulfur Dioxide and Associated Pollutants." C.E.G.B., Nottingham,
U.K. Internal Report (SSD/MID/R17/78).
Maul, P. R., F. R. Barber, and A. Martin. 1980. Some observations of the
meso-scale transport of sulphur compounds 1n the rural East Mid-
lands. Atmos. Environ.. 14:339-354.
McGregor, J. L., L. M. Leslie, and D. J. Gauntlett. 1978. The ANMRC
limited area model: Consolidated formulation and operational
results. Mon. Weather Rev.. 106:427-438.
McMahon, T. A., and P. J. Denlson. 1979. Empirical atmospheric depo-
sition parameters—a survey. Atmos. Environ.. 13:571-585.
McMahon, T. A., P. J. Denlson, and R. Fleming. 1976. A long-distance air
pollution transportation model Incorporating washout and dry depo-
sition components. Atmos. Environ.. 10:751-761.
R-21
-------
McNaughton, D. J. 1980. Initial comparison of SURE/MAP3S sulfur oxide
observations with long-term regional model predictions. Atmos.
Environ.. 14:55-63.
McNIder, R. T. 1981. "Investigation of the Impact of Topographic Circu-
lations on the Transport and Dispersion of A1r Pollutants." Ph.D.
dissertation, University of Virginia, Department of Environmental
Sciences, Charlottesvllle, Virginia.
McNIder, R. T., S. R. Hanna, and R. A. Plelke. 1981. "Numerical Simula-
tion of the Transport and Dispersion of Pollutants 1n Slope and Moun-
tain Valley Flows." Extended Abstracts. Fifth Symposium on Turbu-
lence, Diffusion, and A1r Pollution (9-13 March, 1981, Atlanta,
Georgia).
McNIder, R. T., and R. A. Plelke. 1981. Diurnal boundary-layer develop-
ment over sloping terrain. J. Atmos. Sc1.. 38:2198-2212.
McNIder, R. T., and R. A. Plelke. 1984. Numerical simulation of slope
and mountain flows. J. Climate Appl. Meteorol.. 23(10):1441-1453.
McRae, G. J., W. R. Goodln, and J. H. Seinfeld. 1982. "Mathematical
Modeling of Photochemical A1r Pollution." Environmental Quality
Laboratory, California Institute of Technology, Pasadena, California
(EQL Report No. 18).
Mellor, G. L., and T. Yamada. 1974. A hierarchy of turbulence closure
models for planetary boundary layers. J. Atmos. Sc1.. 31:1791-1806.
Meslnger, F. 1981. Horizontal advectlon schemes of a staggered grid—an
enstrophy and energy-conserving model. Mon. Weather Rev.. 109:467-
478.
Meslnger, F., and R. F. Strlckler. 1982. Effects of mountains on Genoa
cyclogenesls. J. Meteorol. Soc. Jpn.. 60:326-338.
Meyerhofer, P. M., R. M. EndUch, B. E. Cantrell, R. Brodzlnsky, and C. M.
Bhumralkar. 1982. "Long-Term SO? and Sulfate A1r Pollution Model:
Refinement of Transformation and Deposition Mechanisms." U.S.
Environmental Protection Agency (EPA-600/3-82-063).
Meyers, R. E., et al. 1979. A Lonq-Ranqe Transport Model for Calculating
the. Atmospheric Impacts of Residual Sulfur Oxides. Brookhaven
National Laboratory, Upton, New York (BNL 26185-R1).
R-22
-------
Mills, 6. A. 1981. An objective limited area analysis/prognosis experi-
ment using FGGE data 1n the Australian region. Mon. Weather Rev.,
109:1898-1913.
Mills, G. A., and C. M. Hayden. 1983. The use of high horizontal resolu-
tion satellite temperature and moisture profiles to Initialize a
mesoscale numerical weather prediction model—a severe weather event
case study. J. CUm. Appl. Meteorol.. 22:649-663.
Mills, M. T., and A. A. Hlrata. 1978. "A Multi-Scale Model for Local and
Regional Sulfur D1ox1de/Sulfate Concentrations: Formulation and
Initial Evaluation." Ninth International Technical Meeting on A1r
Pollution Modeling and Its Application (28-31 August 1978, Toronto,
Canada), North Atlantic Treaty Organization.
Mohnen, V. A., A. Hogan, and P. Coffey. 1977. Ozone measurements 1n
rural areas. J. Geophys. Res.. 82:5889-5895.
MOI. 1982. "Atmospheric Sciences and Analysis: Final Report, Technical
Basis." Report of the Regional Modeling Subgroup, Work Group 2,
U.S./Canada Memorandum of Intent on Transboundary A1r Pollution,
5 August 1980.
Moore, G. E., C. Daly, M. K. L1u, and S.-J. Huang. 1986. Modeling of
mountain-valley wind fields 1n the southern San Joaquln Valley, Cali-
fornia. Submitted to J. CUm. Appl. Meteorol..
Moore, G. E., R. E. Morris, and C. Daly. 1987. "Application of a
Regional A1r Quality Model to the California San Joaquln Valley."
Systems Applications, Inc., San Rafael, California (SYSAPP-87/025).
Moore, G. E., T. E. Stoeckenlus, and D. A. Stewart. 1982. "A Survey of
Statistical Measures of Model Performance and Accuracy for Several
A1r Quality Models." Systems Applications, Inc., San Rafael, Cali-
fornia (SYSAPP-82/122).
Moore, R. E., C. F. Baes III, L. M. McDowell-Boyer, A. P. Watson, F. D.
• Holtman, J. C. Pleasant, and C. W. MUler. 1979. "A1r DOS-EPA: A
Computerized Methodology for Estimating Environmental Concentration
and Dose to Man from Airborne Releases of Rad1onucl1de." Oak Ridge
National Laboratory, Oak Ridge, Tennessee (ORNL-532).
Morris, C. S., C. W. Benkley, and A. Bass. 1979. Development of Meso-
scale A1r Quality Simulation Models. Volume IV; User's Guide to
MESOGRID (Mesoscale Grid) Model. U.S. Environmental Protection
Agency, Washington, DC (EPA-66/7-80-059).
R-23
-------
MPCA. 1985. "Statement of Needs and Reasonableness." Minnesota Pollu-
tion Control Agency.
Munn, R. E., and B. Bolln. 1971. Global air pollution—meteorological
aspects. Atmos. Environ., 5:363-402.
Murphy, B. D. 1976. Influence of Ground Cover on the Dry Deposition Rate
of Gaseous Materials. Oak Ridge National Laboratory, Tennessee
(Report UCCCND/CSD-19).
Murphy, B. D. 1981. "PHENIX--A. Pseudo-Spectral Model of Long-Range Atmo-
spheric Transport." Oak Ridge National Laboratory, Tennessee (ORNL-
5761).
NCAR. 1983. "Regional Add Deposition: Models and Physical Pro-
cesses." National Center for Atmospheric Research, Boulder, Colorado
(Technical Note NCAR/TN-214+STR).
NCAR. 1985. "NCAR EuleHan Regional Add Deposition Model." National
Center for Atmospheric Research, Boulder, Colorado (Technical Note
NCAR/TN-256+STR).
NCAR. 1986. "Preliminary Evaluation Studies with the Regional Add Depo-
sition Model (RADM)." National Center for Atmospheric Research,
Boulder, Colorado (Technical Note NCAR/TN-265+STR).
NRC. 1983. Add Deposition: Atmospheric Processes 1n Eastern North
America. National Research Council. National Academy Press, Wash-
ington, DC.
Nappo, C. J., Jr. 1978. Workshop on Lonq-Ranqe Trajectory-Puff and Plume
Modeling of Continuous Point Source Emissions. A1r Resources Labora-
tory, Silver Spring, Maryland (NOAA Technical Memorandum ERL ARL-72).
Nlckerson, E. C. 1979. On the numerical simulation of airflow and clouds
over mountainous terrain. Contrlb. Atmos. Phys., 52:161-177.
Nlckerson, E. C., E. Richard, R. Rosset, and D. R. Smith. 1986. The
numerical simulation of clouds, rain, and airflow over the Vosges and
Black Forest Mountains: A meso-a model with parameterized micro-
physics. Mon. Weather Rev.. 114:398-414.
R-24
-------
Nlemann, B. L.t and E. H. Pechan. 1981. "Model Profile: Documentation
of the Regional CUmatologlcal Dispersion Model (RCDM)." Work Group
2, Atmospheric Sciences and Analysis Work Group, established under
the Memorandum of Intent on Transboundary Air Pollution signed by the
United States and Canada on 5 August, 1981 (Report 2-9).
Nordlund, G. G. 1975. A quasi-Lagranglan cell method for calculating
long-distance transport of airborne pollutants. J. Appl. Meteorol..
14:1095-1104.
Olson, M. P., and E. C. Voldner. 1981. "Model Profile. Documentation of
the Atmospheric Environment Service Long-Range Transport of A1r Pol-
lutants Model AES-LRT." Work Group 2, Atmospheric Sciences and
Analysis Work Group, established under the Memorandum of Intent on
Transboundary A1r Pollution signed by the United States and Canada on
5 August, 1980 (Report 2-5).
OokoucM, Y., M. Uryu, and R. Sawada. 1978. A numerical study of the
effects of a mountain on the land and sea breezes. J. Meteorol. Soc.
Jpn.. 56:368-385.
OokoucM, Y., and Y. Wakata. 1982. "Three-Dimensional Numerical Model of
the Land and Sea Breezes with a Mountain (1n Japanese). (Rep.
Y.N.C.T., No. 4, pp. 9-18).
Oppenhelmer, M. 1984. "An Analysis of the Sulfur Budget and Interstate
Sulfur Transport for Colorado." Environmental Defense Fund, Inc.,
New York, New York.
Orglll, M. M., J. E. Cermak, and L. 0. Grant. 1971. Research and
Development Technique for Estimating Airflow and Diffusion Parameters
Related to the Atmospheric Water Resources Program. Colorado State
University Fluid Dynamics and Diffusion Laboratory. Fort Collins,
Colorado (Final report CER71-72-MMO-JEC-LOG-20).
Orlanskl, I. 1981. The quasi-hydrostatic approximation. J. Atmos. Sc1..
38:572-582.
Orlanskl, I., and B. B. Ross. 1973. Numerical simulation of the genera-
tion and breaking of Internal gravity waves. J. Geophys. Res..
78(36):8808-8826.
Orlanskl, I., and B. B. Ross. 1977. The circulation associated with a
cold front. Part I: Dry case. J. Atmos. Scl.. 34:1619-1633.
R-25
-------
Orlanskl, I., and B. B. Ross. 1984. The evolution of an observed cold
front. Part II: Mesoscale dynamics. J. Atmos. Sc1., 41:1669-1703.
Pack, D. H.f G. J. Ferber, J. L. Heffter, K. Telegadas, J. K. Angell,
W. H. Hoecker, and L. Machta. 1978. Meteorology of long-range
transport. Atmos. Environ., 12:425-444.
Paegle, J., and D. W. McLawhorn. 1983. Numerical modeling of .diurnal
convergence oscillations. Hon. Weather Rev., lll:January Issue.
Patterson, D. E., R. B. Husar, W. E. Wilson, and L. F. Smith. 1981.
Monte Carlo simulation of dally regional sulfur distribution: com-
parison with SURE sulfate data and visual range observation during
August 1977. J. ADD!. Meteorol.. 20:404-420.
Peltier, W. R., and T. L. Clark. 1983. Nonlinear mountain waves 1n two
and three spatial dimensions. J. Roy. Meteorol. Soc.. 109:527-548.
Perkins, R. W., C. W. Thomas, J. A. Young, and B. C. Scott. 1970. "In-
Cloud Scavenging Analysis from Cosmogenlc Rad1onucl1de Measure-
ments." In Precipitation Scavenging. AEC Symposium Series (NTIS
CONF-700601).
Perla, R. I., and M. MartlnelH, Jr. 1976. "Avalanche Handbook." U.S.
Department of Agriculture, Forest Service, Fort Collins, Colorado.
Agriculture Handbook 489.
Pfluger, U. 1983. "FITNAH Ilia (2D-Vers1on)." Abwarmeprojekt
Oberrhe1ngeb1et. Rep. Bundesumweltaint. Berlin, Germany, 51 pp.
Phillips, D. S. 1984. Analytical surface pressure and drag for linear
hydrostatic flow over three-dimensional elliptical mountains. J..
Atmos. Sc1.. 41:1073-1084.
Plelke, R. A. 1974. A three-dimensional numerical model of the sea
breezes over southern Florida. Mon. Weather Rev.. 102:115-139.
Plelke, R. A. 1984. Mesoscale Meteorological Modeling. Academic Press.
Plelke, R. A., and Y. Mahrer. 1978. Verification analysis of the Univer-
sity of Virginia three-dimensional mesoscale model prediction over
south Florida for July 1, 1973. Mon. Weather Rev.. 106:1568-1589.
R-26
-------
Plelm, J., A. Venkatram, and R. Yamartlno. 1984. "ADOM/TADAP Model
Development Program, Volume 4: The Dry Deposition Module." Environ-
mental Research and Technology, Inc., Concord, Massachusetts (P-B980-
520).
Powell, D. C., D. J. McNaughton, L. L. Wendell, and R. L. Drake. 1979.
"A Variable Trajectory Model for Regional Assessments of A1r Pollu-
tion from Sulfur Compounds." Battelle Pacific Northwest Laboratory,
Richland, Washington (PNL-2734).
Prahm, L. P., and 0. Christensen. 1977. Long-range transmission of pol-
lutants simulated by a two-dimensional pseudospectral dispersion
model. J. Appl. Meteorol., 16:896-910.
Proctor, F. H. 1985. "Three-Dimensional Simulation of the 2 August CCOPE
Hailstorm with the Terminal Area Simulation System." International
Conference on Cloud Modeling, Irsee, Federal Republic of Germany.
Pruppacher, H. R.f and J. D. Klett. 1978. M1crophvs1cs of Clouds and
Precipitation. D. Reldel, Dordrecht, The Netherlands.
Radke, L. F., E. E. Hlndman II, and P. V. Hobbs. 1977. A case study of
rain scavenging of aerosol particles from an Industrial plume. Proc.
Symposium on Precipitation Scavenging (1974). ERDA Symposium Series
NTIS (pp. 425-436).
Ragland, K. W., and K. E. Wllkenlng. 1980. Intermediate Range Grid Model
for Atmospheric Sulfur Dioxide and Sulfate Concentration and Depo-
sition. U.S. Environmental Protection Agency, Duluth, Minnesota.
Ragland, K. W., and K. E. Wllkenlng. 1981. "Relationship Between Meso-
scale Add Precipitation and Meteorological Factors." Department of
Mechanical Engineering, University of Wisconsin, Madison; for the
Conference on Effects of Add Precipitation on Ecological Systems 1n
the Great Lakes of the United States (1-3 April 1981).
Randerson, D. 1972. Temporal changes In horizontal diffusion parameters
of a single nuclear debris cloud. 3. Appl. Meteorol.. 11:670-673.
Rao, K. S., J. S. Lague, and B. A. Egan. 1976. "An A1r Trajectory Model
for Regional Transport of Atmospheric Sulfates." Third Symposium on
Atmospheric Turbulence, Diffusion, and A1r Quality (19-22 October
1976, Raleigh, North Carolina), American Meteorological Society, Bos-
ton.
R-27
-------
Raynor, G. S., and J. V. Hayes. 1982. Variation 1n chemical wet depo-
sition with meteorological conditions. Atmos. Environ.. 16:1647-
1656.
Renner, E., U. Ratzlaff, and W. Rolle. 1985. A Lagranglan multi-level
model of transport, transformation and deposition of atmospheric sul-
fur dioxide and suflate. Atmos. Environ.. 19(8):1351-1359.
Reynolds, S. 0. 1977. "SAI Urban Airshed Model: An Introduction to Its
Usage." Systems Applications, Inc., San Rafael, California (SYSAPP-
77/106R).
Richards, L. W., J. A. Anderson, D. L. Blumenthal, J. A. McDonald, G. L.
Kok, and A. L. Lazrus. 1983. Hydrogen peroxide and sulfur (IV) 1n
Los Angeles cloud water. Atmos. Environ.. 17:911-914.
Ritchie, I. M., J. D. Bowman, and G. B. Burnett. 1983. A mesoscale atmo-
spheric dispersion model for predicting ambient air concentration and
deposition patterns for single and multiple sources. Atmos.
Environ.. 17(7):1215-1223.
Roberts, J. M., F. C. Fehsenfeld, 0. L. Albrltton, and R. E. S levers.
1983. Measurement of monoterpene hydrocarbons at N1wot Ridge,
Colorado. J. Geophy. Res.. 88(C15):10,667-10,678.
Roberts, J. M., F. C. Fehsenfeld, S. C. L1u, M. J. Bollinger, C. Hahn,
D. L. Albrltton, and R. E. Slevers. 1984. Measurements of aromatic
hydrocarbon ratios and NOX concentrations 1n the rural troposphere:
Observation of air mass photochemical aging and NO- removal. Atmos.
Environ.. 18(11):2421-2432.
Roberts, J. M., R. S. Hutte, F. C. Fehsenfeld, D. L. Albrltton, and R. E.
Slevers. 1985. Measurements of anthropogenic hydrocarbon concentra-
tion ratios 1n the rural troposphere: Discrimination between back-
ground and urban sources. Atmos. Environ.. 19(11):1945-1950.
Rodhe, H., and J. Grandell. 1972. On the removal time of aerosol
particles from the atmosphere by precipitation scavenging. Tellus.
24:442-454.
Ross, B. B., and I. Orlanskl. 1982. The evolution of an observed cold
front. Part I: Numerical simulations. J. Atmos. Sc1.. 39:296-327.
R-28
-------
Ross, D. G.t and I. Smith. 1986. "Diagnostic Wind Field Modelling for
Complex Terrain— Testing and Evaluation." Centre for Applied Mathe-
matical Modelling, Chlsholm Institute of Technology (CAMM Report No.
5/86).
Roth, P. M., C. Blanchard, J. Harte, H. Michaels, and M. T. El-Ashry.
1985. The American West's Add Rain Test. World Resources Institute
(Research Report II).
Ruff, R. E., K. C. N1tz, F. L. Ludwlg, C. M. Bhumralkar, J. D. Shannon,
C. M. Shelh, I. Y. Lee, R. Kumar, and D. J. McNaughton. 1985.
Evaluation of three regional air quality models. Atmos. Environ.,
Runchal, A. K. 1980. "A Random Walk Atmospheric Dispersion Model for
Complex Terrain and Meteorological Conditions." Second Joint Con-
ference on Application of A1r Pollution Meteorology (24-27 March
1980, New Orleans, Louisiana), American Meteorological Society,
Boston.
Russell, A. G., and G. R. Cass. 1986. Verification of a mathematical
model for aerosol nitrate and nitric add formation and Its use for
control measure evaluation. Atmos. Environ.. 20(10) :2011-2025.
Samson, P. J. 1980. "Diagnostic Modeling of Summertime Sulfate Concen-
trations 1n the Northeastern United States." Second Joint Conference
on Application of A1r Pollution Meteorology (24-27 March 1980, New
Orleans, Louisiana), American Meteorological Society, Boston.
Samson, P. J. 1981. "Interim Model Profile: Documentation of the Uni-
versity of Michigan Atmospheric Contribution to Inter-Regional Depo-
sition Model— UMAC ID." Work Group 2, Atmospheric Sciences and Analy-
sis Work Group, established under the Memorandum of Intent on Trans-
boundary A1r Pollution signed by the United States and Canada on 5
August, 1980 (Report 2-10).
Sasaki, Y. 1958. An objective analysis based on the varlatlonal
method. J. Meteor. Soc. Japan, 36:77-88.
•
Sasaki, Y. 1970a. Some basic formalisms 1n numerical varlatlonal analy-
sis. Mon. Wea. Rev.. 98:875-883.
Sasaki, Y. 1970b. Numerical varlatlonal analysis formulated under the
constraints determined by longwave equations and lowpass filter.
Mon. Wea. Rev., 98:884-898.
R-29
-------
Saxena, P., A. B. Hudlschewskyj, and C. Seigneur. 1986. "The Effect of
Meteorology on the SCWSulfate Relationship." Systems Applications,
Inc., San Rafael, California. Presented at the APCA Second Inter-
national Specialty Conference on the Meteorology of Acidic Depo-
sition, Albany, New York, 17-20 March 1986.
Saxena, P., and C. Seigneur. 1986. The extent of nonllnearlty 1n the
atmospheric chemistry of sulfate formation. Submitted to Science.
Schere, K. L. 1986. "EPA Regional Oxldant Model: ROM1 Evaluation for 3-
4 August 1979 " U.S. Environmental Protection Agency, Research Tri-
angle Park, North Carolina (EPA/600/3-86/032).
Schere, K. L., and N. C. Posslel. 1984. "U.S. EPA Regional Oxldant
Model-Background and Overview." Presented at the 77th Annual
A.P.C.A. Meeting, San Francisco, California (paper 84-47.2).
Schulman, L. L., and J. S. Sdre. 1980. "Buoyant Line and Point Source
Dispersion Model Users Guide." National Technical Information Ser-
vice, Springfield, Virginia (PB81-164642).
Schwartz, S. E., and W. H. White. 1982. Kinetics of reactive dissolution
of nitrogen oxides Into aqueous solution. Adv. Environ. Sc1. Tech-
no! ., vol. 12.
Sdre, J. S., and F. W. Lurmann. 1983. "Development of the MESOPUFF II
Dispersion Model." Presented at the 6th Symposium on Turbulence and
Diffusion, American Meteorological Society (paper PI.11).
Sdre, J. S., F. W. Lurmann, A. Bass, and S. R. Hanna. 1983. Development
of the Mesopuff II Dispersion Model. U. S. Environmental Protection
Agency, Research Triangle Park, North Carolina.
Scott, B. C. 1978. Parameterization of sulfate removal by precipita-
tion. J. Appl. Meteorol.. 17:1375.
Scott, B. C. 1982. Predictions of in-cloud conversion rates of S02 to
SO* based upon a simple chemical and kinematic storm model. Atmos.
Environ.. 16:1735-1752.
Scott, W. D., and P. V. Hobbs. 1967. The formation of sulfate In water
droplets. J. Atmos. Sc1.. 24:54-57.
R-30
-------
Segal, M., R. T. McNIder, and R. A. Plelke. 1985. "Use of a Mesoscale
Primitive Equation Model 1n the Design of the 1985 SCCCAMP." Atmo-
spheric Simulation, Testing and Research, Inc., Fort Collins,
Colorado.
Sehmel, G. A. 1970. Particle deposition from turbulent airflow.
J. Geophys. Res.. 75:1966-1781.
Sehmel, 6. A. 1980. Particle and gas dry deposition: a review. Atmos.
Environ.. 14:983-1011.
Sehmel, G. A., S. L. Sutter, and M. T. Dana. 1973. "Dry Deposition Pro-
cesses." BatteHe Pacific Northwest Laboratories, Rich!and, Washing-
ton (BNWL-1751).
Seigneur, C., P. Saxena, M. Oudlk, G. Z. Whltten, and P. M. Roth. 1984.
"Modeling Studies of Sulfate and Nitrate Chemistry. The Effect of
Changes 1n Sulfur Dioxide, Nitrogen Oxide, and Reactive Hydrocarbon
Levels." Systems Applications, Inc., San Rafael, California (SYSAPP-
84/013).
Seigneur, C., and P. Saxena. 1986. "The S02/Su1fate Relationship 1n the
Northeastern United States." Systems Applications, Inc., San Rafael,
California (SYSAPP-86/038).
Se1p, H. M. 1980. "Add Snow—Snowpack Chemistry and Snowmelt." In
Effects of Add Precipitation on Terrestrial Ecosystems. Plenum
Press, New York, pp. 77-94.
Seliars, F. M. 1985. "Development of an Emissions Inventory to Support
Testing of the Eulerlan Regional Add Deposition Model." GCA/Tech-
nology Division, for the Second Annual Add Deposition Emission
Inventory Symposium, Charleston, South Carolina (12-14 November).
Shannon, J. D. 1979. Advanced Statistical Trajectory Regional A1r Pollu-
tion Model. Argonne National Laboratory, Argonne, Illinois (ANL/RER-
79-1).
Shannon, J. D. 1981. A model of regional long-term average sulfur atmo-
spheric pollution, surface removal and net horizontal flux. Atmos.
Environ.. 15:689-701.
Shelh, C. M. 1977. "Application of a Lagranglan Statistical Trajectory
Model to the Simulation of Sulfur Pollution over the Northeastern
United States." Third Symposium on Atmospheric Turbulence. Diffu-
sion, and Air Quality (19-22 October 1976, Raleigh, North Carolina),
American Meteorological Society, Boston.
R-31
-------
Sheih, C. M., M. L. Wesely, and B. B. Hicks. 1979. Estimated dry depo-
sition velocities of sulfur over the eastern United States and sur-
rounding regions. Atmos. Environ.. 13:1361-1368.
Shepherd, J. G. 1974. Measurements of the direct deposition of sulphur
dioxide onto grass and water by the profile method. Atmos. Environ..
8:69-74.
Sherman, C. A. 1978. A mass-consistent model for wind fields over com-
plex terrain. J. Appl. Meteorol.. 17:312-319.
Shim, S.-G., S.-Y. Cho, and G. R. Carmlchael. 1986. The sensitivity of a
3-d1mens1onal Eulerlan model to uncertainties 1n emissions and dry
deposition velocities. Atmos. Environ.. 20(10):1979-1987.
Shuman, F. G. 1978. Numerical weather prediction. Bull. Am. Meteorol.
Soc.. 59:5-17.
Singh, H. B., W. Vlezee, VI. B. Johnson, and F. L. Ludwlg. 1980. The
Impact of stratospheric ozone on tropospherlc air quality. J. A1r
Pollut. Control Assoc.. 30:1009-1017.
Small, H. J. 1982. "Stochastic Model of Atmospheric Regional Trans-
port." Ph.D. thesis, University of Michigan.
Smith, R. B. 1979. The Influences of mountains on the atmosphere. Adv.
Geophys.. 21:87-230.
Som1esk1, F. 1984. Numerical simulation of mesoscale flow over ramp-
shaped orography with application of a special lateral boundary
scheme. Mon. Weather Rev.. 112:2292-2302.
Sorbjan, Z. 1984. A model study of the stably stratified steady-state .
atmospheric boundary layer over a slightly Inclined terrain. J..
Atmos. Sc1.. 41:1863-1874.
Stewart, 0. A., and M. K. Uu. 1982. "A Review of Episodic and Long-Term
Regional Transport Models for Long-Range and Intermediate Applica-
tions." Systems Applications, Inc., San Rafael, California (SAI
82137).
Stewart, 0. A., R. E. Morris, A. B. Hudlschewskyj, M. K. Uu, and
D. Henderson. 1983a. "Evaluation of Episodic Regional Transport
Models of Interest to the National Park Service." Systems Applica-
tions, Inc., San Rafael, California (SYSAPP-83/020).
R-32
-------
Stewart, D. A., M. A. Yocke, and J. E. Langstaff. 1986a. "Selection of
an Appropriate Model for Minnesota Add Deposition. Part 1: Review
of Models and Recommendations." Systems Applications, Inc., San
Rafael, California (SYSAPP-86/011).
Stewart, D. A., R. E. Morris, and M. K. L1u. 1983b. "Evaluation of Long-
Term Regional Transport Models of Interest to the National Park Ser-
vice." Systems Applications, Inc., San Rafael, California (SYSAPP
83/215).
Stewart, 0. A., R. E. Morris, and M. K. L1u. 1985. "Applications of
Regional A1r Quality Models to National Parks. Part II. Detailed
Modeling Methodology and Simulation Results for Four Parks.w Systems
Applications, Inc., San Rafael, California (SYSAPP-85/088).
Stewart, D. A., R. E. Morris, M. K. L1u, and D. Henderson. 1983c.
Evaluation of an episodic regional transport model for a multi-day
sulfate episode. Atmos. Environ.. 17:1225-1252.
Stewart, D. A., R. E. Morris, and S. D. Reynolds. 1987. "Application of
the Regional Transport Model (RTM-III) to Western Europe." Systems
Applications, Inc., San Rafael, California (SYSAPP-87/030).
Stewart, D. A., R. E. Morris, and M. A. Yocke. 1986b. "Selection of an
Appropriate Model for Minnesota Add Deposition. Part 2: The NSP
Modeling System." Systems Applications, Inc., San Rafael, California
(SYSAPP-86/034).
Sun, W. 1984. Numerical analysis for hydrostatic and nonhydrostatlc
equations of Inertia!-Internal gravity waves. Mon. Weather Rev..
112:259-268.
Systems Applications. 1982. "A1r Quality Impact Analysis of Synthetic
Fuel Development 1n the U1nta Basin." Systems Applications, Inc.,
San Rafael, California (SYSAPP-82/185).
Systems Applications. 1987. "Development of a Comprehensive Chemistry
Add Deposition Model (CCADM). Technical Document." Systems Appli-
cations, Inc., San Rafael, California (SYSAPP-87/002).
Szepesl, 0. J. 1978. Transmission of sulfur dioxide on local, regional
and continental scales. Atmos. Environ.. 12:529-535.
Tag, P. M. 1983. "Marine Fog/Stratus Forecasting with a 3-0 Boundary
Layer Model." Preprints. Sixth Conference on Numerical Heather Pre-
diction. American Meteorological Society, Boston.
R-33
-------
Tag, P. M., and T. E. Rosmond. 1980. Accuracy and energy conservation 1n
a three-dimensional anelastlc model. J. Atmos. Sc1.. 37:2150-2168.
Thorp, J. M., and B. C. Scott. 1982. Preliminary calculations of average
storm duration and seasonal precipitation rates for the northeast
sector of the United States. Atmos. Environ.. 16:1763-1774.
Tledtke, M., J-F. Geleyn, A. HolUngsworth, and J-F. Louis. 1979. ECMWF
model parameterization of subgrld scale processes. ECMWF Tech. Rep.t
No. 10. 46 pp.
THjonls, J. C., and D. Shapland. 1979. "Existing Visibility Levels 1n
the United States." U.S. Environmental Protection Agency (EPA-450/5-
79-010).
Tripoli, 6. J., and W. R. Cotton. 1982. The Colorado State University
three-dimensional cloud/mesoscale model—1982. Part I: General
theoretical framework and sensitivity experiments. J. Rech. Atmos..
16:185-219.
Tuleya, R. E., M. A. Bender, and Y. Kurlhara. 1983. A simulation study
of the landfill of tropical cyclones using a movable nested-mesh
model. Hon. Weather Rev.. 112:124-136.
Turk, J. T., and 0. B. Adams. 1982. "Sensitivity of Acidification of
Lakes 1n the Flat Tops Wilderness Area, Colorado." Prepared for sub-
mission to Water Resources Research, U.S. Geological Survey, Lake-
wood, Colorado.
Turner, D. B. 1970. Workbook of Atmospheric Dispersion Estimates. U.S.
Environmental Protection Agency, Research Triangle Park, North Caro-
lina (AP-26).
UccelUnl, L. W., R. A. Petersen, P. Kodn, M. L. Kaplan, J. W. Zack, and
V.-C. Wong. 1983. "Mesoscale Numerical Simulations of the Presi-
dents' Day Cyclone: Impact of Sensible and Latent Heating on the
Pre-cyclogenet1c Environment." Sixth Conference on Numerical Weather
Prediction (Preprint volume of extended abstracts), American Meteoro-
logical Society, Boston, Massachusetts.
Ueda, H. 1983. Effects of external parameters on the flow field 1n the
costal region—a linear model. J. dim. Appl. Meteorol.. 22:312-321.
Van Oop, H., and B. J. de Haan. 1983. Mesoscale air pollution dispersion
modelling. Atmos. Environ.. 17:1449-1456.
R-34
-------
Van Oop, H., B. J. de Haan, and C. Engeldal. 1982. "The KNMI Mesoscale
A1r Pollution Model." Konlnkluk Nederlands Meteorologlsch Instltuut
(W.R. 82-6).
Van Egmond, N. 0., and H. Kesseboom. 1983a. Mesoscale air pollution dis-
persion models, II. Lagranglan puff model and comparison with
EuleHan grid model. Atmos. Environ.. 17:267-274.
Van Egmond, N. 0., and H. Kesseboom. 1983b. Mesoscale air pollution dis-
persion models, I. EuleHan grid model. Atmos. Environ.. 17:257-265.
Veltlscheva, N. S. 1979. "A Model for Evaluation of Transboundary Sulfur
Dioxide Flux with the Inclusion of Vertical Motions." HMO Symposium
on the Long-Range Transport of Pollutants and Its Relation to General
Circulation Including Stratospher1c/Tropospher1c Exchange Processes
.(Sofia, 1-5 October, 1979) (WMO No. 538).
Venkatram, A,, B. E. Ley, and S. Y. Wong. 1982. A statistical model to
estimate long-term concentrations of pollutants associated with long-
range transport. Atmos. Environ.. 16:249-257.
Venkatram, A., and J. Plelm. 1985. Analysis of observations relevant to
long-range transport and deposition of pollutants. Atmos. Environ..
19(4):659-667.
Venkatram, A., J. Sdre, and J. Plelm. 1984. "ADOM/TADAP Model Develop-
ment Program. Volume 1. Approach to the Evaluation of the
ADOM/TADAP Model." Environmental Research & Technology, Inc., Con-
cord, Massachusetts (P-B866-100).
Vogel, B., G. Grob, and F. Wlppermann. 1986. Mesok11p (first special
observation period): observations and numerical simulation—a com-
parison. Boundary-Layer Met.. 35:83-102.
Voldner, E. C., M. P. Olson, K. Olkawa, and M. Lolselle. 1981. Compari-
son between measured and computed concentrations of sulfur compounds
1n eastern North America. Proceedings of CACGP Symposium on Trace
Gases and Aerosols (August 1979). J. Geophys. Res.. 86(C6):
5339-5346.
Walcek, C. J., R. A. Brost, and J. S. Chang. 1986. S02, sulfate, and
HNOj deposition velocities computed using regional landuse and
meteorological data. Atmos. Environ.. 20(5):949-964.
R-35
-------
Walcek, C. J., and W. R. Stockwell. 1986. The contribution of aqueous-
phase radical chemistry to sulfur and ^2 Cnem1str>y ^n cloudy
environments. Submitted for publication.
Wallbaum, T. 1982. "Numerlsche Simulation Atmospharlscher Stromungen 1m
Meso-scale Gamma" (Numerical Simulation of Atmospheric Flows 1n the
Meso-scale Gamma), Ph.D. dissertation, Technical University Darm-
stadt, Germany.
Walmsley, J. L., J. R. Salmon, and P. A. Taylor. 1982. On the applica-
tion of a model of boundary-layer flow over low hills to real ter-
rain. Bound. Layer Meteorol.. 23:17-46.
Wang, I. T., and T. L. Waldron. 1980. "User's Guide for MTDDIS: Meso-
scale Transport, Diffusion, and Deposition Model for Industrial
Sources." Rockwell International, Creve Coeur, Missouri
(EMSC6062.1UG(R2)).
Wang, I. T., T. L. Waldron, and R. R. Bushey. 1980. "A Mesoscale Trans-
port and Dispersion Model for Industrial Plumes." Second Joint Con-
ference on Application of A1r Pollution Meteorology (24-27 March
1980, New Orleans, Louisiana), American Meteorological Society, Bos-
ton.
Warner, T. T., R. R. F1zz, and N. L. Seaman. 1983. A comparison of two
types of atmospheric transport models—use of observed winds versus
dynamically predicted winds. J. C11m. Appl. Meteorol.. 22:394-406.
Washington State University. 1985. "Development of a National Inventory
for Natural Hydrocarbon Emissions." Washington State University,
Pullman, Washington (Technical proposal to U.S. Department of Com-
merce, Mountain Administrative Support Center).
Weber, A. H., M. R. Buckner, and J. H. Weber. 1982. Statistical perform-
ance of several mesoscale atmospheric dispersion models. J. Appl.
Meteorol.. 21:1633ff.
Welsman, B. 1980. "Long-Range Transport Model for Sulfur." Presented
at the 73rd Annual Meeting of the A1r Pollution Control Association,
22-27 June 1980, Montreal, Quebec (Paper 80-54.6).
Wendell, L. L., 0. C. Powell, and R. L. Drake. 1976. "A Regional Scale
Model for Computing Deposition and Ground Level A1r Concentration of
S02 and Sulfate from Elevated and Ground Sources," Third Symposium on
Atmospheric Turbulence. Diffusion, and A1r Quality (19-22 October
1976, Raleigh, North Carolina), Boston: American Meteorological
Society.
R-36
-------
Wesely, M. L., J. A. Eastman, D. R. Cook, and B. B. Hicks. 1978. Daytime
variation of ozone eddy fluxes to maize. Boundary-Layer Meteorol..
15:361-373.
Wesely, M. L., and B. B. Hicks. 1977. Some factors that affect the depo-
sition rates of sulfur dioxide and similar gases on vegetation.
J. A1r Pollut. Control Assoc.. 27:1110-1116.
Whlteman, C. D., and T. B. McKee. 1979. "Temperature Inversion Destruc-
tion 1n Mountain Valley—Implications for A1r Pollution Disper-
sion." NATO/CCMS A1r Pollution Pilot Study, 10th International Tech-
nical Meeting on A1r Pollution Modeling and Its Application, Rome,
Italy (23-26 October).
Whltten, G. Z. 1983. "The Chemistry of Smog Formation: A Review of Cur-
rent Knowledge." Systems Applications, Inc., San Rafael, California
(SYSAPP-83/246).
Whltten, 6. Z., and M. W. Gery. 1985. "Development of CBM-X Mechanisms
for Urban and Regional AQSMs." Systems Applications, Inc., San
Rafael, California (SYSAPP-85/162).
Whltten, G. Z., J. P. Klllus, and R. G. Johnson. 1985a. "Modeling of
Auto Exhaust Smog Chamber Data for EKMA Development. U.S. Environ-
mental Protection Agency, Research Triangle Park, North Carolina'
(EPA-600/3/85-025).
Whltten, G. Z.. J. P. Klllus, and R. G. Johnson. 1985b. "Development of
a Chemical Kinetic Mechanism for the U.S. EPA Regional OxIdant
Model." U.S. Environmental Protection Agency, Research Triangle
Park, North Carolina (EPA-600/3-85-026).
W1lken1ng, K. E., and K. W. Ragland. 1980. "A User's Guide to the Uni-
versity of Wisconsin Atmospheric S02/S04 A1r Pollution Computer Model
(UWATM-SOX)." Department of MechanTcal Engineering, University of
Wisconsin, Madison.
Wlllmott, C. J. 1982. Some comments on the evaluation of model perform-
ance. Bull. Am. Meteorol. Soc.. 63:1309-1313.
Wnioughby, H. E., H.-L. J1n, S. J. Lord, and J. M. P1otrow1cz. 1984.
Hurricane structure and evolution as simulated by an ax1symmetr1c
nonhydrostatlc numerical model. J. Atmos. Sc1.. 41:1169-1186.
R-37
-------
Wojdk, M., T. Myers, J. KUlus, 0. Serang, and M. K. L1u. 1978.
"Development and Evaluation of a Mesoscale Photochemical A1r Quality
Simulation Model." Systems Applications, Inc., San Rafael, Califor-
nia (EI78-118).
Wong, V. C., J. W. Zack, M. L. Kaplan, and G. D. Coats. 1983. "A Nested-
Grid Limited-Area Model for Short Term Weather Forecasting." Sixth
Conference on Numerical Weather Prediction (Preprint volume of
extended abstracts), American Meteorological Society, Boston.
Yamada, T. 1978. "A Three-Dimensional Numerical Study of Complex Atmo-
spheric Circulations Produced by Terrain." Proc. American Meteoro-
logical Society Conference on Sierra Nevada Meteorology, 1978. pp.
61-67.
Yamada, T. 1979. An application of a three-dimensional, simplified
second-moment closure numerical model to study atmospheric effects of
a large cooling-pond. J. Atmos. Environ., 13:693-704.
Yamada, T. 1980. "Numerical Simulation of Mesoscale Atmospheric Circula-
tions over Lake Michigan." Winter Annual Meeting, American Society
of Mechanical Engineers, 80-WA/HT-65 (November).
Yamada, T. 1981. A numerical simulation of nocturnal drainage flow.
J. Meteorol. Soc. Jpn.. 59:108-122.
Yamada, T. 1982. A numerical model study of turbulent airflow 1n and
above a forest canopy. J. Meteorol. Soc. Jpn.. 60:439-454.
Yamada, T. 1983. Simulations of nocturnal drainage flows by a q^ - 1
turbulence closure model. J. Atmos. Sc1.. 40:91-106.
Yamada, T., and G. Mellor. 1975. A simulation of the Wangara atmospheric
boundary layer data. J. Atmos. Sc1.. 32:2309-2329.
Yamada, T., and G. L. Mellor. 1979. A numerical simulation of the BOMEX
data using a turbulence closure model coupled with ensemble cloud
relations. Q. J. Roy. Meteorol. Soc.. 105:915-944.
Yamartlno, R. J., J. E. Plelm, and W.-S. Lung. 1983. "Development of an
Add Rain Impact Assessment Model." Environmental Research & Tech-
nology, Inc., Concord, Massachusetts (ERT Document No. PB770-1).
Yocke, M. A. 1981. "A Three-Dimensional Wind Model for Complex Ter-
rain." Ph.D. dissertation, University of California, Berkeley.
R-38
-------
Yocke, M. A., M. K. L1u, and J. L. McElroy. 1977. "The Development of a
Three-Dimensional Wind Model for Complex Terrain." Joint Conference
on Applications on A1r Pollution Meteorology, American Meteorological
Society and A1r Pollution Control Association, pp. 209-214.
Yocke, M. A., R. E. Morris, and J. L. Haney. 1985. "Regional Ozone
Modeling Study Using RTM-III." Systems Applications, Inc., San
Rafael, California (SYSAPP-85/068).
Yocke, M. A., M. L. Ruchlls, and M. K. Liu. 1981. "Application of a
Regional A1r Pollution Model to North Dakota." Systems Applications,
Inc., San Rafael, California (SYSAPP-80/200).
Yu, Chang-Han, and R. A. Plelke. 1985. "Mesoscale A1r Quality Under
Stagnant Synoptic Cold Season Conditions 1n the Lake Powell Area."
Department of Atmospheric Science, Colorado State University, Fort
Collins, Colorado.
Zack, J. W., V. C. Wong, M. L. Kaplan, and G. D. Coats. 1983. "A Nested-
Grid Mesoscale Numerical Simulation of an Isolated Tornadlc Convec-
tlve Complex." 13th Conference on Severe Local Storms (Preprint
volume), American Meteorological Society, Boston.
Zannettl, P. 1981. An Improved puff algorithm for plume dispersion simu-
lation. J. Appl. Meteorol.. 20:1203-1211.
Zlshka, K. M., and P. J. Smith. 1980. The climatology of cyclones and
anticyclones over North America and surrounding ocean environs for
January and July, 1950-77. Mon. Weather Rev.. 108:387-401.
R-39
-------
Appendix A
DESCRIPTION OF THE DIFFUSION COEFFICIENTS
USED IN THE PNL/MELSAR-POLUT MODEL
(From "Green River A1r Quality Model Development:
MELSAR—A Mesoscale A1r Quality Model for Complex
Terrain. Volume 1—Overview, Technical Description
and User's Guide." K. J. Allwlne and C. D. Whlteman,
Prepared by Pacific Northwest Laboratory, Richland, Washington
for U.S. Environmental Protection Agency
(EPA 600/8-85-0176; NTIS PB 85-247211)
-------
2'.5.10 Diffusion Coefficients
The horizontal and vertical diffusion coefficients are based on an
approach presented by Pasqulll (1976) and further discussed by Irwln (1979)
and Ramsdell, Hanna and Cramer (1982). The square of the total horizontal
diffusion 1s the sum of the squares of three components: 1) an Initial
diffusion resulting from nonatmospheric processes (e.g., buoyant plume
rise), 2) diffusion resulting from atmospheric turbulence, and 3) diffusion
resulting from horizontal wind-direction shear. The square of the total
vertical diffusion Is the sum of the squares of two components: 1) an
Initial diffusion resulting from nonatmospheric processes, and 2) diffusion
resulting from atmospheric turbulence. In mathematical form, the horizontal
diffusion coefficient Is
29 91/9
. M£ . A£\ i/t in
°y * l*y y y
where
cy • the horizontal diffusion coefficient (m)
Ay • Initial diffusion resulting from buoyant plume rise (m)
BT, « horizontal diffusion resulting from atmospheric turbulence (m)
C* • horizontal diffusion resulting from horizontal wind shear (m).
The vertical diffusion coefficient 1s
- _ /»' , n2\ 1/2 If.
«z • (AZ + BZ) (2
where
o- « vertical diffusion coefficient (m)
At • Initial diffusion due to buoyant plume rise (m)
Bz * vertical diffusion due to atmospheric turbulence (m).
In the rest of this section the determinations of Ay and Az, Cy, By,
and B2 are given.
2.5.10.1 Diffusion Resulting from Buoyant Plume Rise - Ay and A2
As discussed by Pasqulll (1976) and recommended by the AMS Workshop
(Hanna et al. 1977), for strongly buoyant plumes the vertical motion of the
plume relative to the surrounding air contributes to both vertical and
horizontal plume spread. Numerous observations of plumes near the stack by
BMggs (1972) show that the plume radius 1s approximately equal to half the
plume rise. Based on this and assuming the horizontal and vertical
A-l
-------
diffusion to be affected equally by plume rise, Bj oriel and and Bowers (1982)
compute Initial diffusion as 0.5 Ah/2.15. Therefore, Ay 1n Equation (2-111)
and AZ 1n Equation (2-112) are evaluated by y
Ay - Az ' TTIT <2
2.5.10.2 Diffusion Resulting from Wind Direction Shear -
As pointed out by Pasqulll (1976), horizontal wind shear adds a compo-
nent to total horizontal diffusion. As a rough rule, Pasqulll suggests add-
ing a component of variance to o~ (Cy 1n Equation (2-111)), which Is equal
to 0.03 A62s2 where A6 1s the turning of wind direction over the total depth
of the plume, 1n radians. This correction 1s for a steady-state plume. In
MELSAR the diffusion resulting from wind shear must be numerically
accumulated along the path of the puff. This 1s done using the equation
Cy(s+A$) • Cy(s)
where
.1
* 0.173 A6
S+AS/2
(2-114)
A8 « wind direction shear through depth of puff evaluated at s+As/2
(radians)
As « distance puff center of mass traveled 1n one time step (m)
s » puff travel distance (m).
A8 1s determined by computing the difference 1n wind direction between the
winds at three o. above the puff center of mass and three a- below the puff
center of mass. If three az above the puff center of mass is above the top
of the mixed layer, then the height of the mixed layer 1s used as the upper
limit. If three a, below the puff center of mass 1s below the terrain, then
the terrain 1s used as the lower limit. For a uniformly distributed puff 1n
the vertical direction, the upper boundary 1s the mixing height and the
lower boundary 1s the ground surface.
2.5.10.3 Horizontal Diffusion Resulting from Atmospheric
Turbulence - By
The modeling of the horizontal diffusion process 1n puff trajectory
models requires that both plume diffusion and puff diffusion be treated.
A-2
-------
Hanna, Brlggs and Hosker (1982) and Gifford (1982) both point out the need
for caution when using plume diffusion coefficients extrapolated to long
travel distances 1n puff trajectory models.
The transition from plume diffusion to puff diffusion 1s dependent on
the frequency of the Input meteorology. As discussed by Hanna, Brlggs and
Hosker-(1982), plume horizontal diffusion coefficients are measured experi-
mentally when the sampling time, ts, 1s much greater than the travel time,
t, from the release point to the receptor arc. When t. 1s much less than t,
then Instantaneous plume diffusion or puff (relative) diffusion 1s
measured. The same principle applies to puff trajectory modeling. The
sampling time 1s specified by the period of Input meteorology, t_ (e.g.,
1 hour). For any pollutant travel times 1n the model less than tm, plume
diffusion should be used. For any travel times much greater than tm, puff
diffusion should be used. A mix of plume and puff diffusion treatment 1s
required when tffl 1s approximately equal to travel time.
The use of plume diffusion coefficients In puff trajectory models out
to long travel distances (>50 to 100 km) could theoretically lead to the
underpredlctlon of short-term pollutant concentrations at these distances.
This would result from the overpredlctlon of horizontal diffusion caused by
the double-counting of plume meander, by using plume diffusion coefficients,
and by the Integration of Individual puffs In a time- and space-varying wind
field. 'True1 plume diffusion should already account for this Integration
of puffs In a varying wind field.
Gifford (1982) points out that If typical short-range plume diffusion
coefficients are extrapolated to travel times corresponding to distances on
the order of 100 to 200 km, the results will fall short of both observed and
theoretical values by amounts ranging up to nearly an order of magnitude.
Gifford attributes this to short-period turbulence sampling 1n determining
horizontal plume diffusion. In other words, the sampling time was too short
1n comparison to the Lagranglan time scale and, therefore, not all the
turbulent fluctuations were accounted for. Consequently, the particle
spread more nearly resembles relative diffusion than time-averaged
diffusion.
It can be deduced from 61fford's observations that, 1n light of the
current state of knowledge on this subject, the use of extrapolated
empirical 'plume' diffusion coefficients 1n puff trajectory models may not
be a bad estimate of near-field plume diffusion as well as later 'puff dif-
fusion. However, the air pollution modeler should be aware that approaches
for bounding the problem do exist (Gifford 1982), and the extrapolation of
short-range plume diffusion coefficients 1s certainly not the final answer.
The approach used 1n MELSAR for treating horizontal dlffuslo'n combines
a method proposed by Pasqulll (1976) for plume diffusion and Gifford1s
approach for treating both plume and puff diffusion.
A-3
-------
Near-Field Plume Diffusion— The horizontal diffusion coefficient for plumes
used in MELSAR is of the form recommended by Hanna et al. (1977) after
Pasquill and described by Irvrin (1979)
By - ov t fy (2-115)
where
ov » the standard deviation of the horizontal cross-wind component of the
wind (m/s)
f « a nondimensional function of travel time
t « puff travel time (s).
For nonstationary, nonhomogeneous conditions, By 1s evaluated using a
'virtual time1 , tv, in Equation (2-115) for a puff traveling from time t to
time t+At, where At 1s the model time step. The virtual time is the total
time a puff would have had to travel to be the size of the puff at time t
for atmospheric conditions at time t+At. The equation for By then becomes
By (t+At) - oy [ty+At] fy (ty+At) (2-116)
The value of ty 1s determined from Equation (2-116) using Newton's method.
Rewriting Equation (2-116) in the form
t-.O (2-117)
and then finding the first derivative
g'(tv) - Cf'(t¥)] t + [fv(t n
V y V t+At V y y t+At
we can set up an iterative scheme to converge on ty where
In Equation (2-119) [tv]n at n«l 1s set equal to the actual puff travel time
t. Then [tv]ntl 1s evaluated. If the ratio g/g' 1s <1 the iteration stops,
otherwise the iterative procedure continues until g/g' <1 or n « 6.
This general form of By as given in Equation (2-115) or (2-116) allows
the effects Of terrain roughness on diffusion to be incorporated directly
through specification of ov as a function of terrain roughness.
A-4
-------
Universal function f..—Irwln (1983) reviewed several schemes for
estimating diffusion coefficients. He described the horizontal diffusion
after the form given 1n Equation (2-115) and compared the results of several
field experiments with five models of the universal function fy. His
results showed that a scheme by Cramer (1976) performed best fir elevated
releases and Pasqulll's (1976) best for surface releases. However, Irwln
also concluded that a scheme by Draxler (1976) performed almost as well as
PasqullTs for the near-surface releases and almost as well as Cramer's for
the elevated releases.
Bv 1n MELSAR [Equation (2-116)] Is written as a function of travel
time, and Draxler's forms of fy are also written as functions of travel
time. Irwln (1983) recommends^that a simplified set of Draxler's equations
be used to overcome some bias In Draxler's equations. Therefore, the
'Model 4* equation proposed by Irwln Is used 1n MELSAR:
l/fy -1+0.9 (t/1000)°«5 (2-120)
Standard deviation of horizontal component of winds a.."Current1y the
user has the option of specifying ov In one of two ways: I7~an Interim
scheme proposed by Irwln (1979), or 2) empirical relationships developed by
MacCready, Baboolal, and Ussaman (1974) which account for terrain rough-
ness. In the future It will be desirable to set up MELSAR to use measured
values of oy.
The first method for computing ov In MELSAR 1s a scheme proposed by
Irwln (1979) 1n the absence of onslte measurements of ov. He proposes that
for unstable conditions a relationship developed by Panofsky et al. (1977)
should be used
ov/u* - (12 - 0.5 Z^L)1'3 (2-121)
where
u » surface friction velocity (m/s)
It * mixing height (m)
L « Mpnln-Obukov length (m).
For neutral and stable conditions, Irwln recommends the relationship of
B1nkowsk1 (1978)
oy/u* -1.78 (2-122)
The determination of uw, Zt. and L 1s given 1n Section 2.4.
A-5
-------
The second option for ay that the user can specify Is based on an
approach given by MacCready et al. (1974) that accounts for the Influence of
terrain roughness on ov. Using the LO-LOCAT (Lo-Low Altitude Clear Air
Turbulence Project) turbulence data, MacCready et al. developed empirical
equations of ov and ow as a function of a measure of terrain roughness, wind
speed, and height aboveground.
The turbulence data used by MacCready et al. (1974) were from a major
observational program (LO-LOCAT) conducted for the A1r Force Flight Dynamics
Laboratory, by Boeing Aircraft Company. LO-LOCAT used Jet aircraft with
excellent Instrumentation to assess complete spectra 1n three dimensions;
along 32-mile path lengths; at 250- and 750-ft altitudes; covering high
mountains, low mountains, desert, plains, and water; for all seasons; for
three times of day Including dawn; for four areas near Edwards Air Force
Base (AFB) (California), Griffins AFB (New York), Peterson Field (Colorado)'
and McConnell AFB (Kansas). The LO-LOCAT goal was to provide statistical
data for work on aircraft gust load and control, but the reports do not
present the meteorological data stratified In a manner suitable for
diffusion studies In complex terrain. However, MacCready et al. examined
the LO-LOCAT data from the Edwards AFB flights from the standpoint of
diffusion 1n the mountain-desert regions of the southwest United States.
The empirical equations of oy developed by MacCready et al. are
for very stable conditions
av - 0.186 R°-35U°-18Z-°-13 (2-123)
for stable conditions
., - 0.231 RO.28,,0.06,-0.04 ;
V I
where
• a measure of terrain roughness (m)
• average wind speed (m/s)
Z » height aboveground (m).
MacCready et al. (1974) did not present results for neutral and
unstable conditions. If these relationships are available, they will be
Included in MELSAR at a later date. Irwln (1983) pointed out that Hinds and
Nlckola (1967) concluded from their analysis of the Dry Gulch, Prairie
Grass, Green Glow, and Mountain Iron data that the influence of local
climate and terrain on diffusion 1s increased during stable atmospheric
conditions and decreased during unstable conditions; that is, the
A-6
-------
atmospheric mixing 1s sufficient during moderately unstable atmospheric
conditions to outweigh local terrain effects. Therefore, the Interim scheme
for determining cv for option 2 1n MELSAR 1s as follows: Equation (2-123)
will be used for very stable conditions, Equation (2-124) will be used for
stable and neutral conditions, and option 1 will be used for unstable condi-
tions. That 1s, Irwln's (1979) scheme described earlier will be used for
determining oy during unstable conditions. A plot of Equations (2-123) and
(2-124) are given In Figure 2-22 for a height of 50 m and a wind speed of
5 m/s'for the neutral case and 2.5 m/s for the very stable case.
RT Is a measure of terrain roughness derived by MacCready et al. (1974)
that best correlated with the turbulence data they analyzed. The method for
computing Rj Is given In Section 2.3.
FIGURE 2-22
PLOT OF ov VERSUS TERRAIN ROUGHNESS (RT) FROM
MACCREADY et al.'s (1974) EMPIRICAL RELATIONSHIPS ASSUMING A HEIGHT
ABOVEGROUND OF 50 m AND A WIND SPEED OF 5 m/s FOR NEUTRAL CONDITIONS
AND 2.5 m/s FOR VERY STABLE CONDITIONS
600
Far-Field Puff Diffusion and the Transition from Plume to Puff--G1 fford
(1982) solved a form of Lagevln's equation applicable to atmospheric dif-
fusion. The resulting equation describes both puff and plume horizontal
diffusion and combinations of the two, for small and large diffusion
times. Currently, Glfford's equation 1s used In MELSAR to describe puff
diffusion. However, after more testing of his plume diffusion approach 1n
the future, Glfford's full scheme may be used In MELSAR.
A-7
-------
The horizontal diffusion 1s described using the following equation
- 2 Kt -
1 - e
CK
1 - e
(2-125)
where
B
horizontal diffusion coefficient resulting from atmospheric
turbulence (m) _
large-scale eddy d1ffus1v1ty (nr/s)
the standard deviation of the horizontal component of the wind
;ant travel time (s)
pollute
1-V§/
source.
(m/s)
where V« 1s the Instantaneous turbulence level at the
The parameter C In Eguatlon (2-125) varies between zero and unity depending
on the value of v£/0y, which, In effect, Is the ratio of the Instantaneous
turbulence level at the source to that of the entire flow. Therefore, when
C » 0, plume diffusion 1s described by Equation (2-125) and when C « 1, puff
diffusion Is described. A typical value of the large-scale eddy
dlffuslvlty, K, Is 1.5 x 104 m2/s. which Is the default value specified 1n
MELSAR.
Figure 2-23 gives a plot of By versus travel time for the two extremes
of Equation (2-125): plume diffusion (C - 0) and puff diffusion (C » 1).
The K value 1s 1.5 x ICr nr/s and ov 1s 0.6 m/s. The oy Is representative
of neutral atmospheric stability determined from a relationship by B1nkowsk1
(1978) where ov/u • 1.78. The surface friction velocity, u , for this
example case was determined from the logarithmic wind profile,
U • u /0.4 In (Z/Z0); assuming the surface roughness,
height, Z » 10 m; and the wind speed, U » 5 m/s.
0.03 m; the
Also plotted 1n Figure 2-23 are curves of horizontal plume diffusion
coefficients based on universal functions proposed by various researchers,
determined from diffusion data. Shown are curves using Cramer (1976),
Doran, Horst and Nlckola (1978), Draxler (1976), and Pasqu1ll-G1fford
(Turner 1970) relationships. These are all assumed to be representative of
neutral atmospheric stability. Doran, Horst and Nlckola's and Cramer's
curves are plotted from the relationship By » ov t fy(x/U) assuming oy •
0.6 m/s and U * 5 m/s, Draxler's curve Is plotted from By - oy t fy(t), and
Pasqu11l-G1fford's curve from the well-known curves 1n Turners Workbook
(1970) assuming U « 5 m/s to convert from downwind distance to travel time.
The fy functions for Ooran, Horst, and Nlckola (1978), Cramer (1976), and
Draxler (1976) are given 1n Table 2-12.
The plume curves are shown extrapolated beyond the limits of the data
used to derive them. All the empirical plume curves are below the theo-
retical plume curve of G1fford's. It 1s not clear why G1fford's curve 1s
A-8
-------
TABLE 2-12
f FUNCTIONS USED TO COMPUTE HORIZONTAL PLUME DIFFUSION
Doran, Horst X (km) 0.1 0.2 0.4 0.8 1.6 3.2 lo(a)
and Nlckola (1978) f* 1.04 0.98 0.92 0.85 0.77 0.67 0.54
Cramer (1976) fy • (50/X)[(X-5)/45]°*9
Draxler (1976) fw - [1 + C
(a) Extrapolated.
(b) Sampling time « 3600 s.
above all the empirical curves, especially for travel times less than 100 to
200 s. Since the experimental sampling times were on the order of 10 m1n to
1 hr, one would think that most of the turbulent fluctuations contributing
to the diffusion at these short travel times would have been accounted for
through the relatively long sampling times. This point will require further
Investigation.
At travel times near or longer than the experimental sampling times one
would expect the empirical plume curves to depart from Glfford's theoretical
plume curve. That 1s, the measured diffusion 1s more and more representa-
tive of puff diffusion for longer travel times. This 1s shown 1n
Figure 2-23, especially for Draxler's curve. Based on the hypothesis that
Draxler's 'plume' curve Is actually more and more representative of puff
diffusion for longer and longer travel times, and considering that there are
no data on transition from plume to puff, then the extrapolated Draxler
curves will be used to represent plume to puff transition 1n the Melsar
model. Consequently, Draxler's curves will be followed by MELSAR from
po1nt-of-re1ease to their Intersection with Glfford's theoretical puff
curve. Beyond that time, the MELSAR model will specify the puff growth rate
by Equation (2-125) with C » 1 (puff). Figure 2-24 shows plots of Draxler's
curve followed by Glfford's 'puff curve for four values of ov.
As 1n Equation (2-115), Equation (2-125) 1s evaluated using the concept
of a virtual time, tv, when-the puff travels from time t to time t + At.
The solution of Equation (2-125) uses the same procedure as that described
for Equation (2-115). To determine when puff growth 1s transferred from
Draxler's curve to Gifford's curve, MELSAR evaluates Equations (2-115) and
(2-125) for all time steps. As soon as By from Equation (2-125) 1s greater
than that from Equation (2-115), the results from Equation (2-125) are used
to grow the puff and Equation (2-115) is no longer evaluated.
A-9
-------
FIGURE 2-23
HORIZONTAL DIFFUSION VERSUS TRAVEL TIME FOR NEUTRAL CONDITIONS
(oy « 0.6 m/s)
10V , , , |.| MI|
10*
E
of
10*
101
10
r x
GIFFORD
DRAXLER
DORAN, HORST
AND NICKOLA
CRAMER
—— PASQUILL-GIFFORD-
...I
101
10* 103 10*
(0.5 km) (5km) (50km)
TRAVEL TIME t (sec)
10a
(500km) [U=5m/s]
A-10
-------
FIGURE 2-24
HORIZONTAL DIFFUSION VERSUS TRAVEL TIME USING ORAXLER'S (1976)
EQUATION FOR THE NEAR FIELD AND GIFFORD'S (1982) EQUATION FOR
THE FAR FIELD
I i ml t i i I i til
• I 11 nl i i i I 11 ill
10
101
103
(5km)
TRAVEL TIME t (s«c)
104
(50 km)
10*
(500km) [U=5m/s]
A-ll
-------
2.5.10.4 Vertical Diffusion Resulting from Atmospheric
Turbulence - B,
Few data exist for characterization of vertical diffusion, especially
at long travel times. In addition, characterization for long travel times
of vertical diffusion 1s not as Important as that for horizontal diffusion
because the pollutant becomes uniformly mixed through the mixing depth rela-
tively quickly. Therefore, properly characterizing the mixing depth and the
puff height relative to the mixing depth 1s more Important than characteriz-
ing the vertical diffusion coefficient for long travel times. Based on
these considerations, plume versus puff diffusion will not be treated 1n
detail as was done for horizontal diffusion. The vertical diffusion 1n
MELSAR will be determl ned.f rom empirical plume coefficients extrapolated out
to travel times of 2 x 107 s (100 km at 5 m/s winds) and by a method
recommended by Heffter (1965) for 'travel times greater than 2 x 10 s.
The vertical diffusion resulting from atmospheric turbulence used 1n
MELSAR for travel times less than 2 x 10 s Is of the form described by
Irwln (1979)
B2 • ow t fz (2-126)
where
ow » the standard deviation of the vertical component of the wind (m/s)
t « travel time (s)
f 2 * a nondlmenslonal function, primarily a function of travel time.
Equation (2-125) 1s evaluated using the concept of a virtual time, ty,
for the puff traveling from time t to time t+At. Equation (2-126) becomes
Bz(t+At) - ow (ty+At) fz(ty+At) (2-127)
The solution of Equation (2-127) uses the same procedure as that described
for Equation (2-116).
This general form of B- as given by Equation (2-126) or (2-127) allows
the effects of terrain rougnness on diffusion to be Incorporated directly
through specification of ow.
For travel times greater than 2 x 104 s, B. 1s evaluated using the
relationship given by Heffter (1965):
0.5 (2 KJ1/2
2t +
B(t) - - - (2.128)
A-12
-------
where
t * puff travel time (s)
At * model time step (s) 9
K2 » the vertical eddy d1ffus1v1ty (nr/s).
The default values of K_ are given 1n Table 2-13 and are equal to those used
1n MESOPUFF II (Sdre e! al. 1984).
TABLE 2-13
VERTICAL DIFFUSIVITIES BY PASQUILL-GIFFORD STABILITY CLASS
P-G A B C D E F
Kz (m2/s) 50 30 15 7 3 1
Universal Function f-—Irwln (1983) reviewed several schemes for estimating
diffusion coefficients. He described the vertical diffusion after the form
given 1n Equation (2-126) and compared the results of several field experi-
ments with five models of the universal function fz. His results showed
that a scheme by Draxler (1976) performed best 1n characterizing the verti-
cal diffusion. However, Irwln (1983) proposed a simplified set of Oraxler's
equations which perform as well as the full set and eliminate some bias at
longer travel times. Irwln1s 'Model 4*-equations for fz are used In MELSAR:
for unstable conditions
1/f, -1 + 0.9 (t/500)0'5
z m (2-129)
for stable conditions
l/fz -1 + 0.9 (t/50)0'5
Standard Deviation of the Vertical Wind Component of cy—Currently the user
has the option of specifying ow m one of two ways: 1) an Interim scheme
proposed by Irwln (1979) or 2) empirical relationships developed by
MacCready, Baboo!a1 and Llssaman (1974) which account for terrain roughness.
In the future 1t will be desirable to set up MELSAR to use measured values
of v
The first method for computing ow 1n MELSAR 1s a scheme proposed by
Irwln (1979) to be used 1n the absence of onsite measurements of ay He
proposes that the following relationships be used for unstable conditions:
A-13
-------
ww » 1.342 (Z/Zt)0>33 l/lt < 0.03
°w/w* * °'763 (Z/Z*)* 0-03 < Z/Z£ < 0.40 (2-130)
°w/w* " °'722 (1-Z/Z£)0*207 0.40 < Z/Zt < 0.96
» 0.37 Z/Zt > 0.96
where
w » convectlve velocity (m/s)
z » height aboveground (m)
Zt « depth of mixed layer.
The relationships for aw/w given 1n Equation (2-129) were determined by
Irwln from plotted data of Willis and Deardorff (1974) and Kalmal et al.
(1976). For neutral and stable conditions Irwln recommends the relation-
ships of B1nkowsk1 (1978) be used
(2-131)
where
u » surface friction velocity (m/s)
L » Monln-Obukov length (m)
Z • height aboveground (m)
2.0
The determination of u , Z,, L, and w Is given 1n Section 2.4 of this
report.
The second option for ow 1s based on an approach given by MacCready,
Baboo!a 1 and Ussaman (1974), which accounts for the Influence of terrain
roughness on crw. A background discussion on MacCready, Baboolal and
Ussaman's approach 1s given In Section 2.5.10.3 of this report. The
empirical equations of ow developed by MacCready, Baboolal and Ussaman are
for very stable conditions
* - 0.123 R0.*2u0.22z-0.17 (2_132)
A-14
-------
for stable conditions
0.251 RO-26uO.052-0.03
(2-133)
where
a measure of terrain roughness (m)
average wind speed (m/s)
Z • height aboveground (m).
MacCready, Baboolal and Ussaman did not present results for neutral
and unstable conditions. If these relationships are available, they will be
Included 1n MELSAR at a later date. The Interim scheme for determining ow
for option 2 1n MELSAR Is: Equation (2-132) Is used for very stable condi-
tions, Equation (2-133) Is used for stable and neutral conditions, and
option 1 1s used for unstable conditions. That 1s, Irwln's (1979) scheme
described earlier 1s used for determining ow during unstable conditions. A
plot of Equations (2-132) and (2-133) are given 1n Figure 2-25 for a height
of 50 m and a wind speed of 5 m/s for the neutral case and 2.5 m/s for the
very stable case.
FIGURE 2-25
PLOT OF ow VERSUS TERRAIN ROUGHNESS (RT) FROM MACCREADY,
BABOOLAL AND IISSAMAN'S (1974) EMPIRICAL RELATIONSHIPS ASSUMING
A HEIGHT ABOVEGROUND OF 50 m AND A WIND SPEED OF 5 m/s FOR THE
NEUTRAL CASE AND 2.5 m/s FOR THE VERY STABLE CASE
600
A-15
-------
2.6 Averaging Process
POLUT creates a disk file of pollutant concentrations for one or two
pollutants for all time steps of a simulation. The concentration at each
receptor from each source 1s written out. These concentration values can
then be operated on by the post-processor, POLPRC, to compute concentrations
at various averaging times that can be compared with standards.
POLPRC computes moving average concentrations of pollutants at each
receptor for up to three user-specified averaging times. The averaging
times can range from 1 to 24 hr. The outputs are tables of highest and
second highest pollutant concentrations for the duration of the simulation,
at each receptor for each pollutant for each averaging time. The highest
and second highest moving averages for the sum of all sources arid the
contribution of each source to the highest and second highest sum are
computed. In addition, highest and second highest moving averages can be
computed for each source Individually. Tables of the time of occurrence of
the highest and second highest values are also listed.
A moving average 1s simply an averaging process that progresses through
the data by first computing an average using the number of hourly values
required for the average and then computing a new average for each new
hourly value encountered. For example, moving 3-hr averages would be com-
puted by averaging the first three hourly values for the first average, the
second through fourth values for the second average, the third through fifth
values for the third average, and so on. Moving averages give a better
estimate of peak average values, encountered over a duration of time, than
block averaging.
Because all of the hourly concentrations resulting from a run of MELSAR
are written to a disk file, any form of post-processing can be done by the
user provided the software Is available.
A-16
-------
REFERENCES
Anderson, G. E., et al. A1r Quality Impacts of Anticipated Development 1n
011 Shale Operations 1n Western Colorado and Eastern Utah. SAI No. 155-
EF80-147, Systems Applications, Inc., San Rafael, California, August 1980.
Bader, D., and T. B. McKee. Dynamic Model Simulation of the Morning
Boundary Layer Development 1n Deep Mountain Valleys. J. Climate and Appl.
Meteor., 22(3):341-351, 1983.
Benkley, C. VI., and A. Bass. Development of Mesoscale A1r Quality Simu-
lation Models. Volume 3. User's Guide to MESOPUFF Model. Environmental
Research and Technology, Inc., Concord, Massachusetts, 1979.
Benkley, C. U., and L. L. Schulman. Estimating Hourly Mixing Depths from
Historical Meteorological Data. J. Applied Meteor., 18:772-780, 1979.
B1nkowsk1, F. S. A Simple Semi-Empirical Theory for Turbulence 1n the
Atmospheric Surface Layer. Atmos. Evlron., 13(2):247-253, 1978.
Bjorklund, J. R., and J. F. Bowers. User's Instructions for the SHORTZ and
LONGZ Computer Programs. EPA-903/9-82-004a, U.S. Environmental Protection
Agency Region III, Philadelphia, Pennsylvania, 1982.
Brehm, M., and C. Freytag. Erosion of the N1ght-T1me Thermal Circulation 1n
an Alpine Valley. Arch. Met. Geophys. B1okl., Ser. B., 31:331-352, 1982.
BHggs, G. A. Chimney Plumes 1n Neutral and Stable Surroundings.
Atmospheric Environment, 6(7):507-510, 1972.
BMggs, G. A. Plume Rise Predictions. In: Lectures on A1r Pollution and
Environmental Impact Analyses, American Meteorological Society, Boston,
Massachusetts, 1975, pp. 59-111.
Cramer, H. E. Improved Techniques for Modeling the Dispersion of Tall Stack
Plumes. In: Proc. 7th Int. Technical Meeting on A1r Pollution Modeling
and Its Application. N.51, NATO/CCMS, 731-780. (NTIS PB 270 799), 1976.
Doran, J. C., T. W. Horst, and P. W. Nlckola. Variations 1n Measured Values
of Lateral Diffusion Parameters. J. Applied Meteor., 17(6):825-831,
1978.
A-17
-------
Drake, R. L., and C. H. Huang. Mass-Consistent, Interpolated Wind Fields
for Complex Terrain. Presented at Symposium on Intermediate Range
Atmospheric Transport Processes and Technology Assessment, GatHnburg,
Tennessee, October 1-3, 1980.
Drake, R. L., C. H. Huang, and W. E. Davis. Green River Ambient Model
Assessment Program - Progress Report for the Regional and Mesoscale Flow
Modeling Components. PNL-3988, Pacific Northwest Laboratory, Rlchland,
Washington, 1981.
Draxler, R. R, Determination of Atmospheric Diffusion Parameters.
Atmospheric Environment, 10:99-105, 1976.
Duncan, D. C., and V. E. Swanson. Organic-Rich Shale of the United States
and World Land Areas. U.S.G.S. C1r. 523, U.S. Government Printing
Office, Washington, D.C., 1965.
Egan, B. A., R. D'Errlco, and C. Vaudo. Estimating Air Quality Levels In
Regions of High Terrain Under Stable Atmospheric Conditions. In:
Proceedings of Fourth Symposium on Turbulence, Diffusion, and A1r
Pollution, American Meteorological Society, January 15-18, 1979, Reno,
Nevada, pp. 177-181.
Glfford, F. A. Horizontal Diffusion In the Atmosphere: A Lagranglan-
Dynamlcal Theory. Atmospheric Environment, 16(3):505-512, 1982.
Hales, J.- M., D. C. Powell, and T. D. Fox. STRAM- An Air Pollution Model
Incorporating Non-linear Chemistry, Variable Trajectories, and Plume
Segment Diffusion. EPA450/3-77-012, U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina, 1977.
Hanna, S. R., G. A. Brlggs, J. Oeardorff, B. A. Egan, F. A. Glfford, and
F. Pasqulll. AMS Workshop on Stability Classification Schemes and Sigma
Curves-Summary of Recommendations. Bull. Am Meteor. Soc., 58{12):305-
1309, 1977.
Hanna, S. R., G. A. Brlggs, and R. P. Hosker, Jr. Handbook on Atmospheric
Diffusion. DOE-TIC-11223, Technical Information Center, U.S. Department
of Energy, Washington, D.C.,.1982.
Heffter, J. L. The Variation of Horizontal Diffusion Parameters with Time
for Travel Periods of One Hour or Longer. Journal of Applied Meteorology,
4:153-156, 1965.
Hinds, W. T., and P. W. Nlckola. The Mountain Iron Diffusion Program:
Phase I, South Vandenberg: Vol. I, BNWL-572, Pacific Northwest
Laboratories, Rlchland, Washington, 1967, p 220.
Irwln, J. S. Estimating Plume Dispersion- A Recommended Generalized
Scheme. In: Proceedings of Fourth Symoosium on Turbulence, Diffusion,
and A1r Pollution, American Meteorological Society, January 15-18, 1979,
Reno, Nevada, 1979.
A-18
-------
Irwin, J. S. Estimating Plume D1spers1on-A Comparison of Several Sigma
Schemes. Journal of Climate and Applied Meteorology, 22:92-114, 1983.
Kalmal, J. C., J. C. Uyngaard, D. A. Haugen, 0. R. Cote, and Y. Izuml.
Turbulence Structure 1n the Convectlve Boundary Layer. J. Atmos. Sc1.,
33:2152-2169, 1976.
Kronenberger, L., et al. The Impact of the Clean A1r Act on 011 Shale
Development: API Oil Shale Case Study Group Report. American Petroleum
Institute, Washington, D.C., 1981.
Latlmer, D. A., and J. R. Doyle. Prevention of Significant Deterioration
Policy Implications for Projected Oil Shale Development. Draft Final
Report. U.S. Environmental Protection Agency, Washington, D.C., 1981.
MacCready, P. B., Jr., L. B. Baboolal, and P. B. S. Ussaman. Diffusion and
Turbulence Aloft Over Complex Terrain. Presented at AMS Symposium on
Atmospheric Diffusion and A1r Pollution, Santa Barbara, California,
September 9-13,. 1974.
Maul, P. R. Atmospheric Transport of Sulfur Compound Pollutants, Central
Electricity Generating Bureau MID/SSD/80/0026/R, Nottingham, England,
1980.
Panofsky, H. A., H. Tennekes, D. H. Lenschow, and J. C. Wyngaard. The
characteristics of Turbulent Velocity Components 1n the Surface Layer
Under Convectlve Conditions. Boundary-Layer Meteor., 11:355-361, 1977.
Pasqulll, F. Atmospheric Dispersion Parameters In Gaussian Plume Modeling,
Part II. Possible Requirements for Change 1n the Turner Workbook
Values. EPA-600/4-76-030b. U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina, 1976, 53 pp.
PEDCO. Colorado's Climate, Meteorology and A1r Quality. PEDCO
Environmental, Inc., Cincinnati, Ohio. Prepared for U.S. Dept. of
Interior, Bureau of Land Management, Denver, CO, 1981.
Ramsdell, J. V., S. R. Hanna, and H. E. Cramer. Turbulent Diffusion
Coefficients Within and Above the Urban Domain. Prepared for the U.S.
Army, White Sands Missile Range, New Mexico, Pacific Northwest Laboratory,
Rlchland, Washington, 1982.
Ramsdell, J. V., G. F. Athey, and C. S. Glantz. MESOI Version 2.0: An
Interactive Mesoscale Lagranglan Puff Dispersion Model with Deposition and
Decay. NUREG/CR-3344, U.S. Nuclear Regulatory Commission, Washington,
D.C., 1983.
Rlchardus, P., and R. K. Adler. Map Projections. North-Holland Publishing
Company, Amsterdam, 1972.
A-19
-------
Schulman, L. L., and J. S. Sdre. Buoyant Line and Point Source Dispersion
Model Users Guide. PB81-164642, National Technical Information Service,
Springfield, Virginia, 1980.
Sdre, J. S., F. U. Lurmann, A. Bass, and S. R. Hanna. Development of the
MESOPUFF II Dispersion Model, Environmental Research and Technology, Inc.,
Concord, Massachusetts, 1984.
Shelh, C. M., M. L. Wesely, and B. B. Hicks. Estimated Dry Deposition
Velocities of Sulfur Over the Eastern United States and Surrounding
Regions. Atmos. Environ. 13(10).-1361-1368, 1979.
Slade, D. H., ed. Meteorology and Atomic Energy. National Technical
Information Service, Springfield, Virginia, 1968.
Snyder, W. H. Towing Tank Studies 1n Support of Field Experiments at Cinder
Cone Butte, Idaho. Phase III: Verification of Formula for Prediction of
Dividing Streamline Height. U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina, 1980.
Turner, D. B. Workbook of Atmospheric Dispersion Estimates. AP-26, U.S.
Environmental Protection Agency, Research Triangle Park, North Carolina,
1970.
U.S. Army. Universal Transverse Mercator Grid. TM5-241-8, Department of
the Army, Washington, D.C., 1973.
Venkatram, A. Estimation of Turbulence Velocity Scales 1n the Stable and
the Unstable Boundary Layer for Dispersion Applications. In: Eleventh
NATO-CCSM International Technical Meeting on A1r Pollution Modeling and
Its Application, pp. 54-56, 1980a.
Venkatram, A. Estimating the Monln-Obukhov Length In the Stable Boundary
Layer for Dispersion Calculations, Boundary-Layer Meteorology 19:481-485,
1980b.
Vergelner, I. Elne Energetlsche Theorle der Hangwlnde." Ann. Meteor.,
NF19:189-191, 1982.
Wang, I. T., and P. C. Chen. Estimations of Heat and Momentum Fluxes Near
the Ground. Proc. 2nd Joint Conf. on Applications of A1r Pollu.
Meteorology, New Orleans, LA, pp. 764-769, March 24-27, 1980.
Whlteman, C. D. Breakup of Temperature Inversions 1n Colorado Mountain
Valleys. Ph.D. Dissertation, Colorado State University, pp. 250.
Available from University Microfilms, Ann Arbor, Michigan, or as
Atmospheric Science Paper No. 328 from Department of Atmospheric Science
Colorado State University, Fort Collins, Colorado, 1980.
Whlteman, C. D. Breakup of Temperature Inversions 1n Deep Mountain
Valleys: Part I. Observations. J. Applied Meteorology, 21:270-289, 1982.
A-20
-------
Uhlteman, C. D., and T. B. McKee. Breakup of Temperature Inversion 1n Deep
Mountain Valleys: Part II, Thermodynamic Model. J. Applied Meteor.,
21:290-302. 1982.
WMteman, C. D., and K. J. Allwlne. Green River Air Quality Model Development:
VALMET - A Valley A1r Pollution Model. Prepared by Pacific Northwest Labor-
atory, Rlchland, Washington, 1985, for U.S. Environmental Protection Agency
(EPA 600/3-85-064; NTIS PB 86-104106).
W1111S, 6. E., and J. W. Deardorff. A Laboratory Model of the Unstable
. Planetary Boundary Layer. J. Atmos. Sc1., 31(5):1297-1307, 1974.
Zannettl, P. An Improved Puff AlgoMtm for Plume Dispersion Simulation. J.
Applied Meteor., 20:1203-1211, 1981.
A-21
-------
Appendix B
DESCRIPTION OF THE SAI/CCAOM CHEMISTRY MODULE
-------
Appendix B
DESCRIPTION OF THE SAI/CCADM CHEMISTRY MODULE
The CCADM chemistry module consists of four submodules that treat daytime-
cloudy, daytime-dry, n1ght1me-cloudy, and nightIme-dry conditions.
Depending on the physical conditions 1n the cell, one of these four sub-
modules 1s selected to describe the chemical changes occurring 1n that
cell. Cloud cells are multiphase systems comprising complex Interactions
between gas- and aqueous-phase chemical reactions linked by mass-transfer
and equilibrium processes. Dry (noncloud) cells consider gas-phase reac-
tions only. In addition, because there are significant differences 1n
these chemical systems during the day and at night, the cloud and dry
(noncloud) representations are each further divided depending on the radi-
ation characteristics (day and night), resulting 1n four distinct submod-
ules. In this section, we first discuss the principle components of the
chemistry module—the gas- and aqueous-phase mechanisms, chemical equili-
brium relationships, and mass-transfer processes. This discussion 1s
followed by a description of the module structure and a demonstration of
chemical mass balance.
THE GAS-PHASE CHEMISTRY MECHANISM
The gas-phase photochemical kinetics mechanism represents both the chemi-
cal reactions that occur 1n the noncloud portion of a cell, and the reac-
tions that occur 1n the Interstitial air of the cloud. The CCADM gas-
phase mechanism 1s based on an expanded version of the Carbon-Bond
Mechanism—CBM-X (WhUteri et al., 1985a,b). Results obtained with the
CBM-X have been compared with smog chamber data from the University of
North Carolina, the University of California, and Battene smog cham-
bers. The CBM-X has also been tested for urban and regional multlday ox1-
dant scenarios. Three general types of changes were made to the CBM-X to
effectively couple the gas-phase oxidation chemistry with droplet chemis-
try and to represent the gas-phase oxidation of SO? and NOX to sulfurlc
and nitric acids under long-range transport conditions of Interest 1n this
study: (1) the chemistry of additional species needed for coordination
with the aqueous mechanism were Included, (2) the kinetic, product, and
Sto1ch1ometr1c representations 1n the basic CBM were updated, and (3), the
gas-phase oxidation of S02 was reviewed and Included.
B-2
-------
Additional chemical species and reactions were Included 1n the gas-phase
mechanism by reformulating this modified version of the CBM-X and conden-
sing the new reactions Into a format similar to that of CBM-IV (Whltten
and Gery, 1985). The new mechanism (Table B-l) Includes the chemistry of
gaseous sulfur, organic nitrates, methylhydroperoxlde, peroxynltrlc add,
peroxyacetlc add, and formic add. Peroxyacetic add and organic nitrate
were Included 1n the original CBM-X as products only, and were not con-
sidered Important enough to track (1n light of the additional computing
requirements) 1n the urban oxldant representation of the CBM-IV. However,
the gas-phase chemistry of these species was Included 1n the CBM/CCADM
mechanism (reaction 52 and reactions 60, 65, 69, 74 and 75). The chemical
reactions that represent the gas-phase equilibrium of peroxynltrlc add
(reactions 24 and 25) were also Included and the rate expressions were
reformulated. k24 was modified slightly to reflect the values given 1n
NASA (Jet Propulsion Laboratory, 1985); the recommendations of Atkinson
and Lloyd (1984) were used for k^. Because the decomposition of
methylhydroperoxlde 1s poorly understood, and XC^ 1s not always a methyl-
peroxy radical, reaction 77 was used to estimate the upper limit of pos-
sible gas-phase methylhydroperoxlde formation.
Three reaction groups 1n the CBM form gas-phase formic add. The reac-
tions of ozone with ETH and OLE result 1n similar formation pathways that
depend on the secondary reactions of b1radical products. In the gas-phase
component of the CCAOM chemistry module, we have utilized the reaction
Sto1ch1ometr1es proposed by Dodge and Arnts (1979), and have assumed that
1n the atmosphere the thermal1zed blradlcal product reacts exclusively
with water to produce formic add, which leads to the formic add yields
shown 1n reactions 59 and 63. A third possible source of formic add 1s
presented by Su et al. (1979), who demonstrate that the reaction of H02
and formaldehyde leads to an Intermediate product that may (1) decompose
to the original reactants, (2) oxidize NO to N02 to form products Includ-
ing formic add, or (3) react with a second HO? molecule to form products
Including formic add (reactions 39 through 42). These reactions are also
Included 1n the CBM/CCADM. Finally, the reaction of OH with formic add,
reaction 43, has been Included as a gas-phase sink. The resulting gas-
phase chemical mechanism treats 35 species (Table B-2), though nine of
these species are assumed to be at steady state 1n the CCADM.
Several reaction rate constants and activation energy values were modified
slightly (changes of less than 10 percent) 1n the gas-phase mechanism to
reflect values recently reported 1n the literature. However, more dra-
matic differences are seen 1n the new rate constants and activation
energies used 1n the blmolecular reactions,*
* R-nn refers to reaction numbers given 1n Tables B-l, B-3, and B-4.
B-3
-------
TABLE B-l. Gas-phase reaction mechanism.
Rate Expression2
1
2
3
4
5
6
7
8
9
10
H
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
N02
0 (402 * M)
03 + NO
0 4 N02
0 * N02 (4 M)
0 4 MO (4 M)
N02 * 03
0, (4 • H20)
w *
°3
03 * OH
03 * H02
N03 * MO
N0a 4 N02
N03 * N02 (4M)
N205 4 H20
«205 (4 M)
NO 4 N02 4 H20
HN02 4 HN02
HN02
N02 + OH (4 N)
NO * OH (4 N)
H02 * NO
NO * NO (4 02)
H024N02
PNA
¥>2
OH 4 H202
OH 4 HN02
OH 4 HN03
"°3
H02 4 H02 (4 M)
H02 * H02 * H^O
OH * CO
FORM 4 OH
FORM
FORM
FORM * 0
FORM » N03
FORM + H0?
Reaction1
* NO * 0
» 03 * M
» N02 •» 02
• NO * 02
* N03 * M
* N02 + M
» »o3 * o2
» (2a) OH * (1 - a) 0 * 02
* 0 + 02
* H02 + 02
•» OH * 2 02
* 2 H02
• NO 4 N02 4 Q2
* N205 4 M
* 2 HN03
4 N03 4 N02 4 M
4 2 HN02
4 NO 4 N02 •» H20
4 NO 4 OH
4 HN03 * N
4 HN02 4 M
4 OH 4 N02
4 2 N02
4 PHA
4 H02 4 N02
4 2 OH
4 H02 4 H20
4 N02 4 H20
4 N03 4 H20
4 0.85 N02 4 0.85 0 4 0.15 NO 4 0.15
4 H202 4 02 4 M
4 H202 4 H20 4 02
4 H02 4 C02
4 H02 4 CO 4 H20
4 2 H02 4 CO
4 CO 4 H2
4 OH 4 H02 4 CO
» HN03 4 H02 4 CO
* FO,
K298
radiation dependent
4.24 x 10*6
26.6
13750.
2320.
2460.
0.0473
rid1«t1on dependent
tpeclil function
0.042 x kj
100.
3.0
35400.
0.59
1870.
1.9 x 10'6
3.12
1.6 x ID'11
1.5 x W5
0.18 x k}
17000.
9890.
12300.
1.52 x W4
2050.
4.35
0.0014 x kj
2520.
9770.
218.
02 30.6 x kj
4300.
0.218
355.
15000.
rid fit Ion dependent
ndlitlon dependent
237.
0.93
14.9
E/R
•650.
1370.
450.
•350.
2450.
940.
580.
•122.
1230.
•230.
10840.
•560.
•610.
•240.
•530.
•635.
10015.
187.
•1000.
•800.
•5800.
1550.
References/Notes
3
4,5
6
6
<.7
4.8
6.9
3.10
3
6.9
6.9
11
9
4,12
9
13
9
9
3
4,14
4.15
6,9
9
6
9
3
6.9
9
6
3,i6
17
9
6
6.9
3
3
6,9
18
19 ,
(Continue0
B-4
-------
TABLE B-l. Continued.
Rate Expression2
40
41
42
43
44
45
46
47
48
49
SO
51
52
S3
54
55
56
57
58
59
(0
•1
62
63
64
65
66
67
68
69
70
roz
F02 * W
F02 * H02
FACD 4 OH
AL02 * 0
ALD2 4 OH
ALD2 .4 N03
ALD2
C203 4 NO
C203.4 N02
PAN
C203 * C203
C203 4 N02
UGLY
M6LY 4 OH
OH (4 CM4)
PAR 4 OH
0 * OLE
OH 4 OLE
0,4 OLE
m
N03 •» OLE
0 * ETH
OH * ETH
03 4 ETH
TOL 4 OH
N03 4 TOL
WEN 4 NQ3
PHO 4 N02
XYL 4 OH
N03 4 XYL
TLA * OH
Reaction1
* FORM * H02
• N02 * H02 * FACfl
» HjO + 02 * FACO
• C02 * H02 4 H20
* C203 4 OH
• C203 4 HjO
* C203 4 HN03
» X02 4 2 H02 4 CO 4 FORM
• N02 4 X02 4 FORM 4 H02
« PAN
* C203 4 N02
* 2 X02 4 2 FORM 4 2 H02
« PACO
• C^ 4 H02 4 CO
* xo2 4 c2o3
• X02 4 FORM 4 H02 4 HjO
« 1.49 XO, 4 0.067 X02N 4
0.93 H02 4 0.45 ALD2
• 0.75 PAR
» 0.9S AL02 4 0.35 HO, 4
0.20 X02 4 0.15 CO 4
O.OS FORM 4 0.05 C203 4
-0.35 PAR
« FORM 4 AL02 4 X0» 4 HO,
-PAR Z
* O.SALD2 4 0.66FORM 4 0.23 FACO 4
0.212CO 4 0.28H02 4
0.144X02 * 0.08 OH -PAR
• 0.91X0, 4 0.91H02 4
0.09X02N 4 NTR
* FORM 4 X02 4 CO 4 2 H02
» X02 4 H02 4 1.S6FORM 4 0.22AL02
» FORM 4 0.37CO 4 0.13H02 4 0.41FACO
• HO, 4 0.64X0, > 1.13FORM 4
0.56MCLY 4 0.36PHEN 4
0.36PAR 4 1.13CO
• NTR
* PHO 4 HN03
*
• HO, 4 0.72X0, 4 0.67CO 4
1.33MGLV 4 0.28PHEN 4
0.56PAR 4 0.06TLA 4
0.67FORM
* NTR
* XO, * PHO 4 2. PAR
*298
90.
12000.
3000.
473.
636.
24000.
3.7
radiation dependent
16500.
9000.
0.0222
3700.
9600.
0.02 x kj
26000.
21.
1150.0
5920.
42000.
0.018
11.4
1080.
12000.
0.0027
9750.
0.03
14000.
20000.
36000.
0.12
20000.
E/R
986.
•250
-250.
-250.
13500.
1710.
324.
-537.
1897.
800.
-411.
2840.
References/Notes
19
19
19,20
20
9
9
21
3
21
21
9
9
21
3
22
6
21
21
21
21
21
9
22,23
24
21
9
21
21
21
9
21
B-5
-------
TABLE B-l. Continued.
71
72
73
74
75
76
77
78
79
ISOP
ISOP
ISOP
ISOP
X02N
X02 *
X02*
X02*
SOj*
* 0
* OH
4°3
*N03
+ NO
NO
H02
C203
OH
Reaction1
* HO, * 0.8ALD2 + 0.550LE «•
0.5X0, + 0.5CO *
0.45ETH - 0.1PAR
• 0.87X0, «• 0.87FORM «•
0.87H02 * 0.29PAR
0.290LE «• 0.58ETH
0.29ALD2
• FORM * 0.45AL02 *
0.55ETH + 0.98HO, +
0.100LE + 0.06CO
-0.25PAR
• XO^ * MTR
* NTR
* N02
* UMHP
» X02 * H02 •» FORM
« H0? * SV1
Rate Expression'
K298
28000.
126000.
0.018
165.
1000.
12000.
5000.
2400.
1100.
E/R References/Notes
21
21.22
21.23
21
21
21
-1300. 21
21
-180. 4,25
1. Reactants 1n parentheses (M, 02. and CM4) art Included 1n (C298 valuts. CH4 was assumtd to bt 1.85 ppm, M MIS
1 x 10* ppm and (^ was 209000 ppm.
2. Ratt constants art 1n ppm and mln units at on* atmosphtr* prtssur*.
«n*rt K298 1s the reaction rat* at 298 K and T Is th* t*np*ritur* In K*lv1n d*gr**s.
3. Variable photolytlc ratts depend on light Intensity and zenith angle (MMtten, Klllus, and Johnson, 1985).
Rates designated as "x kj" are ratios to the N02 photolysis rate. Sources of quantum yield and absorption cross-
section data are:
Quantum Yield
Cross-Section
NO, , JPL (1985)
°3 * °;° Atkinson and Lloyd (1984)
'03 » 03P JPL (1985)
HNO, Baulch et al. (1984)
Assumed equal to unity
Calvert (1980)
AL02 Baulch et al. (1984)
N6LY Plim et al. (1983)
N03 Atkinson and Lloyd (1984)
JPL (1983)
Bass (1985)
Bass (1985)
JPL (1985)
Baulch et al. (1984)
Bass (1985)
Baulch et al. (1984)
Plum et al. (1983)
Atkinson and Lloyd (1984)
4. Pressure and temperature dependent. The falloff relationship of JLP (1985):
2"1
Ko(T)CM] ) „ l1 * tlog10(H0{T)[M]/K_(T))]
0(T)L«J/K.(T)j °'6
r .
was used to describe an Arrhtnlus expression.
S. K0(T) • 6.0 i 10'34 (T/300)'2'3
K. • 2.8 x
12
cm6 molecule'2 sec'1
cm3 molecule'1 sec"1
JPL (1985)
Baulch et al. (1984)
B-6
-------
TABLE B-l. Concluded.
6.
7.
JPL (1985)
K_(T) - 9.0 x 10"32(T/300)"2'°
K. - 2.2 x 10
rll
a. K0(T) » 9.0 x W32(T/300r1>5
9.
10.
K." 3.0 x 10"11
Atkinson and Lloyd (1984)
cm6 molecule"2 sec"1
cm3 molecule"1 sec"1
cm6 molecule"2 sec"1
cm3 molecule"1 sec"1
• [H20] Ka/([H20] Ka + [M] Kb)
Ka » 326000 ppm"1 mln'1
Kb - 42900 ,<-WO(l/T - 1/298)) ppn-l m1n-l
from Uhltten and Gery (1986)
11. Sander and Klrcher (1986)
12. K.(T) • 2.2 x 10"30 (T/300)'4'3
K.(T) • 1.5 x ID*12' (T/300)'0'5
Perner et al. (1985) and references therein.
cm6 molecule"2 sec"1
cm3 molecule"1 sec'1
13.
14. K0(T) - 2.6 x 10'30 (T/300)'3'2
KJT) - 2.4 x 10"11 (T/300)'1'3
15. K0(T) • 6.7 x lO'31 (T/300)'3'3
KJT) • 3.0 x 10'11 (T/300)'1'0
16. Combination of
N03 * N02 + 0
N03 •» NO + 02
cm6 molecule"2 sec"1
cm3 molecule'1 sec"1
cm6 molecule"2 sec"1
cm3 molecule"1 sec"1
K • 26.0 x K! mln"1
K » 4.6 x K mln"1
17. Combination of
H02 + H02 * H202 + 02 K29tt « 2512 ppm"1 mln"1
H02 + H02 + M * H202 * 02 + M K298 • 1783 ppm"1 nrin"1
18.
19.
20.
21.
22.
23.
24.
2b.
JPL (1985)
JPL (1985)
JPL (1985)
JPL (1985)
Atkinson and Lloyd (1984)
UMtten and Gery (1985)
JPL (1985)
Cantrell et al. (1985)
Su et al. (1979)
Jacob (1986)
Mitten et al. (19tt5a)
Atkinson (1985)
Stolchlometry adjusted to 22 percent yield of glycolaldehyde (represented by AL02).
Atkinson and Carter (1984)
K0(T) - 3.0 x lO'31 (T/300)-3'3
K. « 1.5 x 10
-12
cm6 molecule"2 sec'1
cm3 molecule"1 sec"1
JPL (1985)
B-7
-------
TABLE B-2. Chemical species 1n the CBM/CCADM gas-phase mechanism.
_ Species _ Abbreviation1
Nitric oxide NO
Nitrogen dioxide N02
Nitrogen trloxlde N03(*)z
Dlnltrogen pentoxlde N205(*)z
Nitrous add HN02
NUHc add HN03
Peroxy nitric add (H02N02) PNA
0(3P) atom 0(*)
Hydroxyl radical OH(*)
Water H20
Ozone
Hydroperoxy radical o
Hydrogen peroxide H202
Carbon monoxide CO
Formaldehyde (CH20) FORM
Hydroxymethylperoxy radical (HOCH202) F02
Higher aldehydes (RCHO, R>H) ALD2
Peroxyacyl radical (CH3C(0)02) C203(*)
Peroxyacyl nitrate (CH3C(0)02N02) PAN
Peroxyacetlc add (CH3C(0)OOH) PACD
Methyl hydroperoxlde (CH3OOH) upper limit UMHP
Methyl glyoxal (CH3C(0)CHO) MGLY
Paraffin carbon bond (C-C) PAR
Olef1n1c carbon bond (C-C) OLE
Ethene (CH2-H2C) ETH
Toluene (C6H5CH3) TOL
Xylene (C6H4(CH3)2) XYL
Phenol (Cresol) surrogate (aromat1c-OH) PHEN
Phenoxy radical surrogate (aromat1c-0) PHO(*)
Isoprene (CH2*C(CH3)CH>CH2) ISOP
Organic nitrate NTR
Sulfur dioxide S02
Sulfate (H2S04) SULF
NO-to-N02 operation X02(*)
N0-to-n1trate operation X02N(*)
Total » 35
(*) denotes steady-state species.
N03 and N20g mass accounted for 1n nocturnal steady-state
B-8
-------
NO + N03 > 2 NO (R-12)
OH + HN03 > H20 + N03 (R-29)
OH + CO > H02 + C02 (R-33)
which were replaced with the data given 1n Table B-l for these reac-
tions. In the case of reaction 12, recent evaluations have put this rate
significantly higher than that reported 1n Atkinson and Lloyd (1984). The
rates of Sander and Klrcher (1986) are used here, though even higher rates
have been recommended by NASA (Jet Propulsion Laboratory, 1985). The rate
expressions used for reactions 29 and 33 are pressure- and temperature-
dependent functions from NASA and replace earlier NASA and COOATA (Baulch
et al., 1982) formulations. The activation energies of reactions 3 and 55
were also adjusted to the values recommended by NASA.
Reaction rate constants and activation energies for three-body reactions
were modified to more recent values by fitting the results of falloff cal-
culations to Arrhenlus expressions over the temperature range of atmo-
spheric Interest. More recent, slightly different values (less than 10
percent) were determined for the rate constants and activation energies 1n
reactions 2 and 20, and a new activation energy was used for reaction 5.
All these changes were made at the recommendation of NASA (Jet Propulsion
Laboratory; 1985). We have used the new NASA fall-off expressions to
describe kg and have combined the NASA two- and three-body rate expres-
sions for the self reaction of H02 at atmospheric pressure to derive a
formulation for k^. Finally, though the kinetics of the reaction, NO +
OH + M > HONO + M (R-21), were recently reviewed and remain unchanged
1n the new NASA evaluation, Atkinson and Lloyd (1984) previously recom-
mended that they not be used because they were Inconsistent with the
reported data. We, therefore, use the recommendations of Atkinson and
Lloyd for the falloff data.
The gas-phase equilibrium of N20g was also studied and updated. The
formation rate of N205 was obtained from new NASA values. The activation
energy, though small, 1s an opposite sign from the previous value obtained
from Kerr and Calvert (1985). We have not altered k16 from the previous
value, which follows the recommendations of Perner et al. (1985) that the
Keq value of Graham and Johnston (1978) be used until discrepancies
between experimental determinations are resolved.
The oxidation of sulfur dioxide 1n the noncloudy atmosphere 1s mainly
through reaction with OH:
02, H20
OH + S02 + M > H2S04 + M (R-79)
B-9
-------
The new kinetic data of NASA (Jet Propulsion Laboratory, 1985) was used to
describe the Arrhenlus expression 1n k.
THE AQUEOUS-PHASE CHEMICAL MECHANISM
The aqueous-phase chemical mechanism 1s based on extensive reviews of work
by Jacob (1986), Walcek and Stockwell (1986), Hoffman and Calvert (1985),
Chameldes (1984), Schwartz (1984a,b), Slgneur and Saxena (1984), and Grae-
del and Wechsler (1981), and Graedel and Goldbert (1983). This Informa-
tion was condensed Into a mechanistic representation of 100 reactions,
with 41 Individual forms of aqueous-phase species, 10 of which are assumed
to be at steady state. Various Input parameters such as NHo, Mn , Fe
and Na"*" are also represented. New kinetic and mechanistic information up
to 1 November 1986 was Included 1n the CCADM aqueous mechanism. These
reaction sets can be conveniently divided Into nonsulfur and sulfur reac-
tions. These subsets are provided 1n Tables B-3 and B-4. A 11st of aque-
ous chemical species 1s shown 1n Table B-5.
Nonsulfur Aqueous Reactions
The aqueous-phase, non-sulfur-conta1n1ng reactions are presented 1n Table
'B-3. These reactions were drawn from the recent works of Jacob (1986),
Walcek and Stockwell (1986), Schwartz (1984b) and Chameldes (1984). When
no data were available, estimates of activation energies were performed
using the method of Jacob (1986). Reactions too slow to be of Importance
1n the time scale of Interest 1n this model were excluded from Table B-3;
however, some reactions of Intermediate significance, or which required
elucidation of their possible Importance, were Included. Some of these
reactions can probably be removed after sensitivity testing.
The Initial nonsulfur portion of the mechanism consists of simple Irrever-
sible oxygen-hydrogen reactions between equilibrium forms of the odd
hydrogen species, and the reactions of ozone. These reactions describe
the OH and H02 aqueous radical concentrations used to determine their
reversible mass transport rates. The Important reactions of Individual
equilibrium forms are usually presented explicitly. For Instance, the
reaction of OH(aq) with total dissolved H02 (H02 ) 1s represented 1n reac-
tions 103 and 104:
OH + H02 > H20 + 02 (R-103)
OH + 02~ > OH' + 02 (R-104)
since
8-10
-------
TABLE B-3. Aqueous-phase nonsulfur reactions,
Rate Expression2
102
10J
104
104
1U6
1U7
,0*1
109
110
111
112
1U«
114
IU
116
117
IU*
ll»
fMt
Reaction1
H202i^20H
°3 (*TijO) "z°2 * °2
OH * HOj* H.,0 » l)2
OH * 02 « OH" « 02
OH * HjOj • H20 * H02
HU2 * H02 » H^2 * 02
. C* «2°)
HUj » 0? =-+ H202 » 02 » OH
(2H20)
02 » 0", =-» H202 » 02 * 20H
H02 * H^ . OH * 02 » HjO
°2 * f2°2 • °" » 02 • «~
OH » 03 » H02 » 02
Ml2 * Oj . OH * 202
(H?0)
°2 * °3 — — * * * M2 * OM~
OH" + Oj — ^— . 02 * H02
H02 * Oj » OH » 0" * 02
H.,02 * Oj • H-jO * 202
HCOj + OH • H20 * COj
HCOj * 02 » H02 • COj
(H,0)
CO) + O« '— •- * HCO * 0 + OH
m" * H it * wi * MTrt~
(M* sec'1)
(• | CnQf 9cn*i^f a roQMi Cnmt stfy
9.5 x 10"7 kt
2.2 x 10'5 kt
7.0 x 10*
1.0 x 1010
2.7 x 107
8.6 x 10s
1.0 x 10*
< 0.3
0.5
0.13
2 x 109
< 1 x 10*
1.5 x 10*
70
2.8 x 10s
7.8 x 10--t03]-°-S
(b) Carbonate Cherts try
1.5 x 107
1.5 x 10s
4.0 x 10*
H n » in*
-E/R
-1500
-1500
-1700
-2365
-1500
0
0
0
0
0
-1500
0
-2520
0
-1910
0
-1500
-mm
Reference
Graedel and tfcschler (1981)
firaedel and tfcschler (1981)
Sehested et al. (1968); Jacob (1986)
Sehested et at. (1968); Jacob (1986)
Chrlstensen et al. (1982)
Bielskl (1978)
Blelski (1978); Jacob (1986)
Bielskl (1978)
Ualnsteln and Blelski (1979)
tfclnstetn and Bielskl (1979)
Stachell* et al. (1984)
Sehested et al. (1984)
Sehested et al. (1983); Jacob (1986)
Stacheltn and Hotgne (1982)
Stacheltn and Hotgne (1982)
Martin et al. (1986)
feeks and Rabant (1966); Jacob (1986)
Schrtdt (1972)
Behar (1970); Jacob (1986)
B*h«r (19701: Jacob (1986)
-------
TABLE B-3. Continued.
Rate Expression2
121
122
124
125
126
127
128
129
130
CO
i
t— •
IVJ
131
132
133
134
135
136
137
138
139
140
141
142
Reaction1
Ct" * OH » CtOH"
ClOH" • Ct" * OH
4
CtOH' J!-» U » H-O
(M20) . ,
Ct — *-* CtOH » H
H02 * Ctj » 2Ct" » 02 » H*
Ol » Ctt » 2Ct" + 02
H02 * Cl » Ct" » 02 » H*
H^j * C*2 » 2Ct" • H02 » H*
H202 * Ct » Ct" * H02 » H*
OH" » Ct2 » 2Ct" + OH
(H20)
NO * N02 e . 2ND" » 2H
-------
TABLE B-3. Continued.
Rate Expression2
143
144
145
146
147
148
149
ISO6
CO
^151
CJ
152
153
154
155
156
157
158
159
160
161
162
Reaction1
NOJ 4 M>3 » M>2 4 NO"
mi Tfipn Mz * m * m
N0j^» NO » 02
NO] * H02» NO" 4 H* » 02
NOj 4 02 » NOj 4 02
NOj 4 HgOg* NOj * H* » H02
NOj * tt~ » NO" » U
HCH(OH)2 4 OH-?* HCOOH * H02 * HgO
HCH(OH)2 * 03*
°2
HCOOH 4 OH-=» C02 * H02 » M20
HCOOH * NjOj* product «• H20
HCOOH » NOj-£» NO" + H* » C0? * H0?
HCOOH » Oj* CO. * HO- 4 OH
-°2 - *
HCOOH » Ct- -£» C02 » H02 » 2Ct * H
6,
HCOO" « OH-£» C0? * H02 + OH"
HCOO" * 03» C02 * OH * 02
0_
HCOO" » NOj-i* NO" * CO- * HO.
- °2 -
MCW" * CO, |u ni» CO, » HCO, 4 HO, 4 OH
3 QH20) 2 3 Z
HCOO" 4 C»2 -?» C02 » H02 * 2Ct"
PAN. NOZ 4 products
(M" sec'1)
1.2 x 109
3.2 x W7 k|
1.5 x W5 kt
4.5 x 109
1.0 x 109
1.0 x 106
1.0 x 10*
(e) Formaldehyde and Foralc Acid Chemistry
2.0 x 109
0.1
1.6 x 108
4.6 x W6
2.1 x 10s
5.0
6.7 x 103
2.5 x 109
100.
6.0 x 107
1.1 x 105
1.9 x 106
4.0 x 10"4
-E/R
-1500
-1500
-1500
-2800
-1500
-1500
0
-1500
-5180
-3200
0
-4300
-1500
0
-1500
-3400
-2600
0
Reference
Ross and Neta (1979); Jacob (1986)
Graedel and Weschler (1981)
Grade) and Meschler (1981)
Jacob (1986)
Jacob (1986)
Chaneldes (1984); Jacob (1984)
Ross and Neta (1979); Jacob (1986)
Narkovlc and Sehested (1972); Jacob (1986)
Holgne and Bader (1983a)
Scholes and Wilson (1967); Jacob (1986)
Shapllov and KostyckovskH (1974)
Oogllottt and Kayon (1967); Jacob (1986)
Holgne and Bader (19836)
Hagesawa and Neta (1978); Jacob (1986)
Anbar and Neta (1967); Jacob (1986)
Holgne and Bader (1983b)
Jacob (1986)
Chen et al. (1973); Jacob (1986)
Hagesawa and Neta (1978); Jacob (1986)
Lee (19845)
-------
TABLE B-3. Concluded.
NOTES;
1. All species and concentrations pertain to the aqueous phase.
2. The rate expression Is the product of all reactant concentrations and the ((299 9lven. Changes from unity dependence on each species are reflected
In the rate expression given by
where Kggg Is the reaction rate of 298 K.
3. Reaction rate assumed to equal zero.
4. 0~ * 03 * 0~ * 02 followed by Oj * H20 » OH + 02 * OH"; Buhler et al. (1984).
S. Reaction My not occur (Schwartz. 1984). Rate assumed to equal zero.
6. HCH(OH)2 * OH * •CH(OH)2 * ^0 followed by •CH(OH)2 * Oj » HCOOH * H02; Bothe and Schulte-Frohllnde (1980).
DO
-------
TABLE B-4. Aqueous-phase sulfur reactions.
Reaction1
201 S(IV) * 03 * S(VI) « 02
202 S(IV) * MjOj » S(VI) » HjO
i •«.*• rJ»
201 S(l») *f 0. "• ' * » S(»l)
i
R.t.E^.I«,
ktSOjl^jtHSO-^k^SO^W,]
k.tM^XSO,]
l. * i«N*3
fM < 4. IS(M3 > 10'V
_ j / 1* \
k0D* r AitFe 3CS(l»nA
01*3 * DO '
/. , 1.7 , lo3 nm**i'-5\
V 6.3 « 10"« » [F.^/
OH < 4. tS(l»)3 < WS K:
N .LfMn^3TJISO: 3 * k.[F.3*][S(l»)]\
Rate Cons!
(N-WC-)
f .4 x IO4
3.7 x IO5
1.5 * 10*
1.3 x 10*
4.7
0.82
5.0 x IO9
1.0 x IO7
tant^
-€/* Deference
HoffnaiM and divert (1985); Crlckton *t at
-5.530 (1977)
-5.280
-4.430 NcArdle and Moffnann (1983)
Martin (1984)
-13.700
-11.000
-13.700
-11.000
* on
H • 3. If botb [Fe3*] and P«n2*] are
otkendM N • 1
N:
DM > 4. [S(I»n >
•M > 4. r$(W)l < 10'5 M:
204
O?"
kacso|-]loii]
205 HSO3 * OH -1* SOj * HjO
I - °*
206a' SOj * HSOj -*• M»5 * SOj
2-
206b SOj * S0| -*. MS05 * S05
OH" *
* HCOOM -£* HSOj * COj * M02
201 SOj « 02 —•*-• MS05 » OH * 02
0.
208
209 SOJ * HCOO" -£» HSOj * COj * 0' k^SOpOWMn
4.6 x 10*
4.2 x 10*
3.0 x 10S
1.3 x IO7
1.0 x IO8
200
1.4 x IO4
-1500
-1500
-3100
-2000
-1500
-5300
-4000
Hule and Neta (1906); Jacob (1986)
Hule and Neta (1986); Jacob (1986)
Hule and Neta (1986); Jicoh(19B6)
Hule and Neta (1986); Jacob (1986)
Jacob (1986)
Jacob (1986)
Jacob (1986)
-------
TABLE B-4. Continued.
™"~ Rate Constant*
210
211
212
213
214
215
216
21?
218
219
220
co 221
£ '»
223
224
225
226
22?
228
229
2M
Reaction' *ate expression
SO* « SO" » 2SO* » Oj "0^*5^
* , . .
NSO* * HSOj lU 2SOJ* » 3H* ^[HSO'HHSOjJDT J
HSOj • OH » SO" * HjO ko[HSO*KOH]
NSO* » SO* » SO* » »|* • H* kc[MSOjK504]
NSO* * NOl • HSO* » NO* kttMSO*][NO"]
HSOJ * Cl" • SO** » product ka[NSOjXCt*]
0
SO* * HSOl -*» SOJ* » N » SOI k CSO'XKSO**!
2- °t t - 2-
SO" » SO j -*• SO* •» »5 . k^CSOflCSO] ]
SO* » H02 » S04* * H* » Oj k^CSO'XHOj]
»; • o* • soj* * o2 k.c»*ico;]
sol » OH" » so?" * OH k rso'Xw"]
•J •• O *
SO* » HjOj • SO** » M* » M02 kjSO'XNjOj]
so* » NO" . so** * NO, k0tso*M»-l
SO* » HCO" » SO?* * H* » CO* k^SOlXHCO*]
a J* J o**
SO* » HCOO — • SO^* » CO. * HOj k FSO^XHCOO'J
sol * ci" * so?" • ct k [so~xct~]
•» *v . o **
o
SO. « HCOOH -A so!" « H k [SOrXNCOM]
f!4 BaTi
k.twaxwo:]
»
«SO* » CNjOOM *-» SO*,* * 2H* kot"SO*J[CM^»HKH*]
« product
KSO" » CHjCfOKM H • SO** » N* (kjH*] » k1)[KSO^XCH]C(0)OOH]
* product
S,,W)»HO,.S«,,)»OH kol«fKHSO*l
S(l») « 0* -?-. S(«l) » OH » OH' I0t0*)[so|*]
(«" see*1)
2.0 x 10*
7.5 x 107
1.7 • 107
< 1.0 x 10S
0.31
1.1 • W3
1.3 x 10*
5.3 x 10*
5.0 x 10*
5.0 x 10*
8.0 a 107
1.2 x 107
8.8 x 10*
9.1 x 10*
1.7 x 10*
2.0 x 10*
1.4 x 10*
8.7 x 10'3
1.9 x 107
5.0 x lO7
6.0 x NT
1.0 x 10*
1.0 x 10S
-•/•
-1500
-4755
-1900
0
-4650
-7050
-1500
-1500
-1500
-1500
-1500
-2000
-1500
-2100
-1500
-1500
-2700
0
-3800
-4000
0
0
0
•eference
Jacob (1986)
Jacob (1986)
Jacob (1986)
Jacob (1986)
Jacob (1986)
Jacob (1986)
Jacob (1986)
Jacob (1986)
Jacob (1986)
Jacob (1986)
Jacob (1986)
•oss and Neta (1979); Jacob (1986)
Jacob (1986)
•oss wid Neta (1979); Jacob (1986)
Jacob (1986)
•oss and Neta (1979); Jacob (1986)
Jacob (1986)
Lee (1984)
HoffMM and Calvert (1985)
HorrMim and Calvert (1985)
Hnffnan and Calvert (I*8S)
-------
TABLE B-4. Concluded.
Rate Constant*"
Reaction
"0
1
tat* Expression
see')
Reference
231
232 2N02 «• HSOj
* NO, —*-» NO; * 2H*»so*' * so: k rw.][Hso:i
fttso")
3 *
,-M2°
* 3M » 2N0
tr ra > s
233a S(IV) » N(lll) » S(VI) » product
233b 2HSOJ * M>2 » OH" » product
Notes;
I. All species and concentrations pertain to the aqueous phase.
2. The rate Is In N
k ' k298 •"
.-1
Temperature dependence Is given by
1.0 x 10s 0
2.0 x 10S
1.42 x 10* 0
4.77 x 103 -6100
Chanel** (1984)
tee and Schwartz (1983); Lee (198«a)
Martin et al. (1981); Martin (1984)
Oblath et al. (1981)
234
23S
^236
1" *
^231
HCHO * HSO, « HOCtt-SO. k.DCNOKNSO^
•* C •» D J
HCHO * so*." » HOCH.SO; » OH" k rwHOHsof ']
J £ J O J
HOCNjSOj * OH" » SO*" » HCHO * NjO k .tHOCHjSO'HOH"]
o
HOCH^j » OH -1* SO" » HCHO » HjO k .[HOCHjSOjlfOH]
0
HSO; * Ctl -*-» SO" « 2C1" » H* k.CHSOl][Ctl]
J 4t _ 3 0 J C
,.0
SOj » Ctj -^ SOj « 2CI~ k0tSOj KCljp
2,t4 x 101
2.S x 107
3.6 x 103
1.4 x 10*
3.4 x 10*
1.6 x 108
-4900
-1800
-4500
-1500
-1500
-1500
Royce and Hoffiwnn (1984); Jacob (1986)
Noyce and Hoffmann (1984); Jacob (1986)
Monger et at. (1986); Jacob (1986)
Jacob (1986)
Hule and Neta (1986); Jacob (1986)
Hule and Neta (1986); Jacob (1986)
where kjvg '* reaction rate at 298 K. For each reaction, the values of ko, kj. kj. etc. are presented In the si
3. Assuring SOj » 0? » SOj rapidly (Mile and Neta. 1984). Protect stolchloMetry unknown (Mule and Neta. 1986).
-------
TABLE B-5. Comparison of the chemical species 1n the
CCADM aqueous-phase mechanism and equilibrium expressions
with their gas-phase adjuncts.
Aqueous-Phase Species
Gas-Phase Adjunct
S02-H20, HS03~, S032'
H2S04, HS04~, S042'
HN02, N02~
HN03, N03~
C02-H20, HC03', C032'
22* 2
HCHO, H2C(OH)2
HCOOH, HOXT
NO
N02
°3
PAN
CH3C(0)OOH
CH3OOH(g)
HC1, CT
OH
H02, Of
N03
NH4OHf NH4
S02(g)
H2S04
HN02(g)
HN03(g)
C02(g)
H202(g)
HCHO(g)
HCOOH (g)
M0(g)
N02(g)
03(g)
PAN(g)
CH3C(0)OOH(g)
CH3OOH(g)
HCl(g)
OH(g)
H02(g)
N03(g)
NH3(g)
Species only 1n the Aqueous Phase
C12~, C1, C10H-
S04", S05",
HOCH2S03-t 'OCH2S03"
C03", OH",
B-18
-------
H02 <«**»> 02~ + H+, (B-l)
with
Keq - [02-][H+]/[H02|, (B-2)
then the total dissolved H02 radical 1s represented by
IH02T] » [H02l + [Of], (B-3)
where
{H021 - [H02T]/(1 + Keq/[H+]) (B-4)
[Of] - IH02TJ/(1 + [H+]/Keq). (B-5)
Similar descriptions of H202T (H202 and H02~) are presented 1n the oxygen-
hydrogen portion of the mechanism.
Carbonate chemistry was taken from the mechanism presented by Chameldes
(1984). However, Schwartz (1984) points out that the reaction of 02~ with
either the bicarbonate or carbonate 1on (reaction 118) suggested by
Schmidt (1972) 1s unlikely. Accordingly, we have set the rate of that
reaction to zero 1n all simulations.
Because the chloride 1on may be an Important anlon 1n coastal environ-
ments, Us Inclusion enhances the applicability of this cloud chemistry
mechanism. As with the Ionic dissociation equilibria of H02 and H202, an
equH1bru1m exists between two forms of radical chlorine:
C12' <—»-> Cl + Cl" (B-6)
Hence, the chemical reactions for all forms of these species are given 1n
the aqueous-phase mechanism. The reactions were accumulated mainly from
the work of Chameldes (1984) and Jacob (1986). The aqueous chlorine reac-
tion set may be reduced after future sensitivity tests, but again, we feel
that the Inclusion of all reactions that appear even marginally Important
ensures the most comprehensive description of that chemistry.
The aqueous-phase nitrogen chemistry 1s presented 1n Table B-3 as reac-
tions 131 through 149. Because of the low Henry's Law solubility con-
stants for NO and N02, compared to the extremely high value for nitric
add, sensitivity testing will probably reveal that the main source of
B-19
-------
aqueous M(V) 1s gas-phase oxidation, followed by absorption of nitric add
by water. Some of the reactions 1n Table B-3, especially those of NO, may
later be deemed unimportant. On the other hand, the uncertainty 1n the
solubility constant for N03, which has been estimated to be as large as
that of nitric add 1n some studies (Chameldes, 1984), combined with the
Intermediate solubility of nitrous add, led to the Inclusion of all reac-
tions at this time.
The aqueous chemistry of formaldehyde and formic add 1s based mainly on
the work of Jacob (1986). Formaldehyde exists 1n equilibrium with Us
hydrated form, methylene glycol (H2C(OH)2); formic add undergoes 1on1c
dissociation. :Jacob presents evidence that formic add production through
the methylene glycol reaction with OH (reaction 150) 1s three times faster
that total gas-phase formic add production. However, the aqueous-phase
sinking of formic add Is orders of magnitude faster than the gas-phase
sink (reaction 43 of Table B-l). Hence, formic add produced 1n the aque-
ous phase can be stabilized through volatilization to the gas phase, 1n
which Us lifetime 1s much longer.
Sulfur Aqueous Reactions
The aqueous-phase, sulfur-containing reactions are given 1n Table B-4. In
this listing, the chemical reactions of the three S(IV) species (sulfur
with an oxidation number of 4) often have the generalized S(IV) repre-
sented as a reactant. The actual contribution of each Individual S(IV)
equilibrium form 1s shown 1n the rate expression. Aqueous S(VI) (sulfate)
formation mechanisms Include the reactions of S(IV) with H202, 03, 02
(catalyzed by Mnz+ and Fe3+), OH, CH302H, CH3C002H, H02, PAN, N03, N02,
nitrite and C12". Reactions for S(IV) oxidation by PAN, CH302H, CH3C002H,
H02, N03, N02 and C12~ were added to the aqueous-phase mechanism because
of the comprehensive nature of the chemical module. As noted, after sen-
sitivity testing, 1t may be possible to delete some less critical reac-
tions.
The oxidation of S(IV) by hydrogen peroxide 1s based on the work of
McArdle and Hoffmann (1983); the ozone reaction kinetics were obtained
from the work of Hoffmann and Calvert (1985). Catalytic (Mnz+ and Fe3*)
oxidation by oxygen 1s represented by the complex expressions of Martin
(1984); the OH oxidation scheme (reactions 204 through 226) 1s taken from
recent work by Jacob (1986) and Huie and Neta (1986). Huie and Neta have
recently measured the rates of S(IV) oxidation by C12" (reaction 237) and
Hoffmann and Calvert have reviewed the reactions of 5(IV) with methyl-
hydroperoxlde, peroxyacetlc add, and the H02 radical. Other reactions of
S(IV) with nitrogen species were obtained for PAN (Lee, 1984a), N03
(Chameldes, 1984), N02 (Lee and Schwartz, 1983) and N(III) (Martin et al.,
B-20
-------
1981; Oblath et al., 1981). Although the formation of hydroxymethane sul-
fonlc add (HMSA) from the complexation of S(IV) and formaldehyde has
often been Implemented as a chemical equilibrium, we have chosen to repre-
sent this chemistry as a set of reactions using the formation rate
described by Boyce and Hoffmann (1984) and the decomposition rate given by
Hunger et al. (1986). Also Included Is the reaction with OH proposed by
Jacob (1986).
A major source of uncertainty In the aqueous oxidation of S(IV) concerns
the contribution of the OH reaction (reactions 204 and 205), specifically
regarding the aqueous OH concentration and the ensuing reactions. The OH
concentration 1s uncertain for a number of reasons: (1) the concentration
1n a droplet probably varies radially, (2) the magnitude of the OH Henry's
Law solubility constant (and, therefore, the gas-phase contribution) 1s
far from certain, and (3) OH reacts rapidly with almost all dissolved
species. The mechanism of OH oxidation, particularly the chemistry of
secondary reaction products, 1s not yet fully explained and could be a
major source of chemical nonllnearlty. The sulfur-bearing product of OH
oxidation, $03", rapidly adds oxygen to form the radical SOe" under atmo-
spheric conditions. As shown 1n Table B-4, many of the S05 sinks produce
HSOs" (peroxymonosulfate) or 504". Further reactions of these species
result 1n additional oxidations of S(IV) to S(VI) (see reactions 211, 216
and 217) providing a greater-than-unlty yield of sulfate formation for
each OH plus S(IV) reaction. The OH mechanism 1s taken from the recent
work of Jacob (1986) and Hu1e and Neta (1986). Various aspects have yet
to be confirmed and some reactions may still be undiscovered. The compre-
hensive nature of this model, 1n particular, this chemical mechanism, pro-
vides an excellent tool for testing the Impact of chemical uncertainties
on ambient predictions.
MASS TRANSFER PROCESSES AND CHEMICAL EQUILIBRIA
Many of the chemical species treated 1n this module (Tables B-2 and B-5)
must diffuse across the droplet surface to achieve equilibrium at the gas-
I1qu1d Interface. If 1t can be determined that a given gas-aqueous reac-
tion 1s "phase-mixed," then the aqueous reaction rates can be evaluated 1n
terms of gas-phase partial pressures through the application of Henry's
Law (Schwartz, 1983). For cases 1n which the mass transport 1s limited by
gas-phase diffusion, 1nterfac1al (kinetic collision) resistance, or aque-
ous-phase diffusion coupled with chemical reaction, this application may
be restricted. This equilibrium limitation and the calculations needed to
describe the mass transfer process are described next. The solubility
equilibrium constants used 1n either set of calculations are given 1n
Table B-6.
B-21
-------
TABLE B-6. Equilibrium reactions with temperature dependence.*
Equilibrium Reaction
S02(g) * S02-H20
S0£ .H20 * HSOj + H*
HS03 * SO?,' + H*
H2S04(aq) * HSO; + H*
HSO; * SO2' + H*
H202(g) I H202(aq)
H202(aq) * HO' + H*
HN03(g) * HH03(aq)
HH03(aq) * HOj + H*
HN02(g) * MN02(aq)
HH02(aq) * MO' + H*
03(g) * 03(aq)
N02(g) * M02(aq)
fc T fc
N0(g) *MO(aq)
K29B
1.23
1.23 x 10'2
6.61 x 10'B
1 ,000.
1.02 x 10'2
7.45 x 104
2.2 x ID'12
2.1 x 105
15.4
49.
5.1 x 10'4
1.13 x 10'2
1.00 x 10'2
1.9 x 10'3
"4"
3.120
1.960
1.500
—
2.720
6.620
•3,730
8,700
4,780
-1,260
2,300
2,500
1,480
Reference
Smith and Martell (1976)
Smith and Martell (1976)
Smith and Martell (1976)
PerMn (1982)
Smith and Martell (1976)
Llnd and Kok (1986)
Smith and Martell (1976)
Schwartz (1984b)
Schwartz (1984b)
Schwartz and White (1981)
Schwartz and White (1981)
Kosak-Channlng and Helz (1983)
Schwartz (1984b); NBS (1965)
Schwartz and White (1981)
continued
B-22
-------
TABLE B-6. Continued.
Equilibrium Reaction
Reference
C02(g) * C02-H20
C02«H20 * HCOj + H*
HCOj * Cof " + H*
NH3(9) * NH4OH
NH4OH * NH4* + OH"
H20 * H* * OH"
PAN(g) * PAN(aq)
HCHO(g) * CH2(OH)2(aq)
(H.O)
HCHO(aq) * CH2(OH)2(aq)
HCOOH(g) * HCOOH(aq)
HCOOH(aq) * HCOO" + H4
HCi(g) * HCt(aq)
HCt(aq) * H* + Ct"
C*2 * Cl + C*~
CH302H(g) * CH302H(aq)
3.40 x 10'2 2.420
4.46 x 10"7 -1,000
4.68 x 10'11 -1,760
Smith and Kartell (1976);
NBS (1965)
Smith and Marten (1976)
Smith and Mart ell (1976)
75. 3,400 Hales and Drewes (1979)
1.75 x 10"5 -450 Smith and Kartell (1976)
1.0 x 10~14 -6,710 Smith and Kartell (1976)
2.9 5.910 Lt* (1984D)
6.30 x 103 6,460 Ltdbury and Blair (1925)
1.82 x 103 4.020 LeHinaff (1968)
3.5 x 103 5,740 Latlmer (1953)
1.78 x 10"* -20 Kartell and Smith (1977)
7.27 x 102 2.020 Harsh and HcElroy (1985)
1.74 x 106 6,900 Harsh and HcElroy (1985)
5.26 x 10'6 — Jayson et al. (1973)
2.27 x 102 5.610 Und and Kok (1986)
continued
B-23
-------
TABLE B-6. Concluded.
Equilibrium Reaction
*298
Reference
CH3C002H(g) * CH3C002H(aq)
N03(g) * N03(aq)
OH(g) * OH(aq)
H02(g) * H02(aq)
H02(aq) * H* * 02
HOCHjSOj * 'OCHgSOj* + H*
4.73 x 102 6,170
2.1 x 105 8.700
25 5,280
2.00 x 103 6,640
3.50 x ID*5 -
2.00 x W12 -
L1nd and Kok (1986)
Jacob (1986)
Jacob (1986)
Jacob (1986)
Perrln (1982).
Sorensen and Anderson (1970)
* The temperature dependence 1s represented by
K . r ..nl"-*" A
^QB "I P IT ™
where K Is equilibrium constant at temperature T (1n Kelvin).
f Units of K and K298 are M or M atnT1.
5 "4^ 1s 1n Kelvin.
B-24
-------
Various droplet constituents form acidic or 1on1c species through aqueous
equilibria, which exist 1n solution. The equilibrium expressions that
define the various forms of these species are also given 1n Table 3-6.
Aqueous equilibrium conditions and electroneutrallty are assumed to be
continuously maintained. Thus, for all dissociating species, 1t 1s pos-
sible to determine the distribution of 1on1c forms. As an example of this
distribution, consider the phase-mixed absorption of S02 1n water to yield
a distribution of S(IV) species,
S02(g) <».»«-> S02-H20 (B-7)
S02'H20 <««««»> H* + HS03" (B-8)
HS03" <««««»> H+ + S032' (B-9)
H20 <-»«««> H+ + OH" (B-10)
where
KH - [S02'H20]/ps02 (B-ll)
K! - [H+][HS03-]/lS02'H20] (B-12)
K2 - lH+HS033-)/lHS03-] (B-13)
^' [H+HOH-] (B-14)
The concentrations of the dissolved species are given by
IS02*H20] - KHpSQ2 (B-15)
[HSOf] - KHKlPso2/[H+] (B-16)
[S032"l - KHKiK2pso,/[H+]2 (B-17)
Z
and S(IV) 1s defined as the sum of these species. The electroneutrallty
relationship Is
[H+] - [OH'l + [HSOfl + IS032-] (B-18)
or, In terms of [H*]t
" WsoJl^ - 2 KHKiK2Pso2 • °' ^B-19)
B-25
-------
with pH defined as -log [H*]. In a more complex system, the equilibria
and electroneutrallty calculations are 1terat1vely performed until equili-
brium exists for all species at a pH that represents electroneutral con-
ditions.
Direct application of the phase-mixed assumption to determine the Instan-
taneous phase distribution of a species may be 1ncons1stant with the dyn-
amic nature of chemical reactions 1n the droplet. Thus, 1f mixing within
a droplet 1s less than 1s needed to counteract rapid aqueous reactions
that produce radially varying distributions of some species, averaging
across the droplet may be misleading. This 1s because Interfaclal equili-
brium depends on the concentrations at the droplet surface, which are not
the same as those predicted by chemical kinetics calculations that deter-
mine concentrations averaged across the droplet. Jacob (1986) has studied
this problem for the average droplet size used 1n CCADM (10 ym) and typi-
cal aqueous-phase diffusion coefficients. He found that the lifetime of
chemical removal and the characteristic time for aqueous phase production
were much longer than the characteristic time for diffusion within a drop-
let for all species except 03(aq), N03(aq), OH(aq), S04~, €03", Cl(aq),
and Clo". Fortunately, this dilemma can be solved by consideration of
only first-order concentration effects because the reactions between these
species are usually not significant. Therefore, we have used the formu-
lations of Jacob (1986) and Schwartz (1983) to describe this process.
This Involves the determination of droplet surface concentrations for some
species so that a kinetic description of the mass flux across the gas-
11qu1d Interface can be determined.
Molecular diffusion has been shown to dominate the mass transport process
1n the vicinity of cloud drops (Schwartz, 1983); thus, the absorption rate
(kgl) and volatilization rate (klq) can be defined for each species
(Jacob, 1986; Schwartz 1983):
kgl » 3nLDg/r2 , (B-20)
klg « 3nLDg/r2KHRT , (B-21)
where r 1s the droplet radius (cm), L Is the liquid water content (vol/
vol), Dg 1s the d1ffus1v1ty 1n air (cm2/s), KH 1s the Henry's Law constant
(M/atm), R 1s the gas constant (0.082058 1 atm/mole K), T 1s temperature
(K) and n is a coefficient correcting for free molecular effects. This
can be defined as (Fuchs and Sutugin, 1971):
B-26
-------
where Kn 1s the Knudsen number and a represents the accommodation coef-
fedent. There 1s very little Information concerning the values of accom-
modation coefficients for water surfaces other than that provided by sen-
sitivity studies and model recommendations. We have chosen a value of 0.1
1n the basic formulation of CCADM, but sensitivity to this value should be
tested, especially with respect to OH.
The 1nterfac1al mass fluxes can be determined by
•gl - kgl cg (B-23>
•ig-klgCi, (B-24)
where Cg 1s the bulk gas-phase concentration and Cls 1s the liquid concen-
tration at the surface of a drop.
Because 03(aq) and N03(aq) are rapidly removed by well-mixed species, the
solution to the aqueous-phase diffusion equation of Schwartz and Freiberg
(1981) can be used to determine surface concentrations from bulk
concentrations for either 03(aq) or N03(aq):
/ • • • i M-l
(B-25)
where q 1s defined for a species as
iv \l/2
ft •
(B-26)
and 1s the bulk liquid concentration, a 1s the droplet radius, kL 1s
the first-order loss rate and Dt 1s the aqueous diffusion coefficient for
the species of Interest.
The behavior of OH(aq) 1s more complicated than that of'N03*(aq) or
03(aq). This is because the aqueous formation rate of OH(aq) 1s radially
variable due to Its dependence on the radial distribution of 03(aq).
Jacob (1986) formulated an algebraic variation of the Schwartz and
Freiberg (1981) expression to account for the radially varying source term
and we have utilized this form.
B-27
-------
Finally, three radical species (OH, HC^, and NOo) undergo reactions 1n
both gas and aqueous phases. Therefore, reversible mass transfer must be
considered. Because these species are all treated as being 1n steady
state, the Interfadal flux calculations just described are used to link
the gas-phase and aqueous-phase steady-state calculation schemes. The
solubility equilibrium constants assumed are given 1n Table B-6. It
should be noted that these constants, along with the accommodation
coefficients are highly uncertain.
METHODOLOGY USED IN THE CHEMISTRY MODULE
The chemistry module provides the host model with a determination of the
changes 1n concentration for state species over a time step (At) used 1n
the host model, and for the physical conditions (temperature, light
Intensity, water vapor concentration, and cloud water content) provided by
the host model. The chemistry module can also provide concentration
estimates for the steady-state species and specific mass throughput rates
for each chemical reaction. These values are useful In determining the
significance of various processes for the conditions of the time step.
Numerical Solution Method .
Computational efficiency combined with numerical accuracy was the princi-
pal requirement 1n the design of the chemistry solution scheme. The
amount of computational time required to simultaneously solve the ordinary
differential equations that represent a chemical kinetics mechanism 1s
primarily determined by two factors: the number of state variables (non-
steady-state chemical species) and the stiffness of the equations repre-
senting them. The number of state variables can be reduced by eliminating
unimportant species and by performing steady-state calculations for some
species. The utility of these methods 1s necessarily limited, however, by
the Inherently large number of variables required to accurately describe
the complexity of real chemical systems. Elimination of too many species
renders the mechanism Insensitive to Important chemical changes;
therefore, 1t 1s used very selectively 1n a comprehensive model such as
the CCADM. Steady-state approximations (which reduce stiffness) can.only
be performed for the few species that satisfy the steady-state definition
over all possible conditions.
Stiffness 1s most easily described for linear systems. Consider the sys-
tem dy/dt » My, where M 1s a constant matrix whose eigenvalues are all
negative but of widely different magnitudes. Classical solutions to ordi-
nary differential equations require a time step small enough to avoid
Instability 1n the solution for the most rapid time constant. Thus, the
number of time steps required may be so large that completion of the cal-
culation becomes impractical. The situation is analogous, though more
complex, for nonlinear systems. _
-------
For the solution of chemical kinetics equations that do not Involve atmo-
spheric transport, a numerical algorithm based on the method of Gear
(1971) can be used. However, though the Gear algorithms can be used for
kinetic calculations 1n large multlcell models, they have an Inherent
shortcoming—I.e., they are designed for a system 1n which the solution
asymptotically decays to a smooth function. The Introduction of discon-
tinuities (such as boundary conditions and hourly changes 1n Input varia-
bles) Into the forcing function of a Gear solution results 1n major compu-
tational Inefficiencies. These Inefficiencies, together with the addi-
tional storage requirements of a high-order predictor-corrector method,
make such a scheme unattractive for use 1n complex, Input-driven models.
For CCADM, we chose a widely used lower-order numerical method known as
the Crank-Nicholson method.. This method 1s a universally stable technique
that 1s computationally efficient when coupled with the utilization of
steady-state approximations. That 1s, since the steady-state approxlma- .
tlon Is valid only for species with the smallest time constants of
asymptotic decay, elimination of those species as state variables reduces
the stiffness of the remaining set of ordinary differential equations and
results 1n a significant reduction 1n computer time.
A schematic description of our Implementation of the Crank-Nicholson tech-
nique 1s provided 1n Figure B-l. As the figure Indicates/each time the
chemistry module 1s called from the host model, tests are performed to
determine 1f a coupled gas-phase/aqueous-phase chemical solution 1s needed
(only for a cloud-containing cell) and 1f 1t 1s day or night (to determine
1f photolysis reactions occur). These two tests specify the bounds of
four separate chemistry submodules that have been optimized for the
specific regimes defined. The characteristics of the submodules are dis-
cussed 1n detail 1n the following paragraphs. After the Initial deter-
minations have been made, the chemical reaction rates and equilibrium con-
stant values are calculated for current conditions. The Crank-Nicholson
algorithm 1s then utilized to determine the changes 1n the concentrations
of state species caused by chemical reaction.
On the first entry Into the predictor-corrector loop, convergence (of all
roots) over the entire time step 1s attempted. That 1s, the working time-
step (DT), which 1s successively halved until a convergence 1s achieved,
1s Initially set to the time step requested by the host model (At). At
this time, and at all other times when a local convergence occurs but the
sum of all converged working time steps does not yet equal At, a test 1s
made to determine if a coupled gas-phase/aqueous-phase solution 1s needed;
1f so, the following special calculations are performed: (1) determina-
tion of current gas-phase/aqueous-phase equilibrium conditions, (2) deter-
mination of any reaction rates that are rapid enough to be decoupled from
the Crank-Nicholson solution and solved prior to 1t, and (3) tltratlon of
B-29
-------
HOST MODEL —calculation of physical conditions for which the
chemistry will be solved over At
for each cell
for At
CHEMISTRY MODULE
I. Determine 1f day or night.
Determine If aqueous chemistry to be solved.
II. Calculate reaction rate constants and equilibrium constants as
determined by conditions 1n I.
III. a. If aqueous chemistry: *
Calculate aqueous equilibrium
Determine aqueous species to titrate
b. If night:
Titrate gas-phase NO and Oj
IV. Crank-H1cholson Predictor-*
a. Calls appropriate CHEM SUB-MODULE as defined 1n I.
b. In that call—sends current concentrations and receives
for future concentration predictions.
V. Crank-Nicholson Corrector •«
a. Estimates new concentrations at working tlmestep DT (Initially
DT • At) with a higher order method than IV.b. and a new call to
appropriate CHEM sub-module for new
a2C
b. Test for convergence:
If the change 1n concentration of species 1s too large
compared to the concentration.
If maximum number of corrections attempted, set new DT
DT/2 : =
If a new concentration < 0, set new DT • r
-> Go to IV 1f all tests succeed
VI. If At accounted for after last convergence—EXIT TO HOST with new
concentrations
If not'
FIGURE B-l. Description of chemistry module solution technique.
B-30
-------
the decoupled species. Also at this point, if 1t 1s nighttime, gaseous NO
and Oj are titrated since their reaction rate 1s far more rapid than that
of any other gas-phase nocturnal reactions.
The predictor-corrector scheme 1s then used to simultaneously determine
the concentrations of the remaining state species. This 1s done by first
predicting concentrations and then successively correcting them using a
higher-order scheme. The Information needed 1s provided by calls to the
appropriate chemistry submodules, which use the current concentrations to
estimate the rates of change of each species (aC^/at) and the partial
derivatives of those rates with respect to each species (a C./aC^at).
After a correction step 1s taken, all concentrations are tested to guaran-
tee that (1) they are greater than zero (1f not, DT 1s halved and a new
prediction 1s made), (2) the change 1n concentration 1s not too large (If
so, a new correction step 1s taken), or (3) the maximum number of correc-
tions has been attempted (1f so, DT 1s halved and a new prediction 1s
made). If all these tests are successful, the current working time step
converges and a new prediction 1s made. When the sum of all OTs 1n the
current call to the chemistry module equals At, the calculations are com-
plete for this cell over At, and the new concentrations are returned to
the host model. Otherwise, a new DT 1s determined, and new tests are
Initiated prior to a new prediction, as described.
Chemistry Submodules
The core of the solution technique resides 1n the four chemistry sub-
modules noted earlier. Both the predictor and corrector sections require
calculation of the rates of change of each species (aC^/at) and-the
partial derivative of each rate with respect to each species (a Cj/aCat).
These values are determined 1n the chemistry submodules from the current
estimates of concentrations of state species and other physical
parameters. The submodules are highly optimized for the four regimes
noted; namely, coupled gas- and aqueous-phase chemistry for both night and
day, and gas-phase-only chemistry for both night and day. These distinc-
tions allow regime-specific consideration of Important processes while
minimizing Irrelevant code. The different chemical regimes require
significantly fewer calculations 1f only gaseous chemistry 1s considered
because there are fewer chemical kinetic expressions and there 1s no need
to perform phase equilibrium calculations. The separation of nocturnal
and daylight regimes provides simpler nighttime chemistry because photoly-
sis reactions do not occur at night. This separation also allows more
rigorous mass-balance calculations, which are needed for the nocturnal,
gaseous (^-NOg-NOj-NgOs system. The following subsections describe these
different regimes and examine our methods for handling stiffness through
steady-state calculations or numerical tltratlons.
B-31
-------
Steady-State Approximations
The elimination of state variables through valid steady-state approxima-
tions serves the dual purpose of reducing system stiffness and minimizing
the size of the system (number of ordinary differential equations). The
validity of the steady-state approximations for the concentration of
species y 1s related to the terms 1n the ordinary differential equation
for that species:
dy/dt « P - yD , (B-27)
where dy/dt 1s the time rate of change 1n the concentration of y, P 1s the
production rate of y, and yO 1s the rate of destruction. A steady-state
approximation 1s valid 1f P « yO » dy/dt and the concentration of y 1s
small compared to state variables producing y or resulting from a reaction
Involving y. The steady-state concentration of y 1s
yss • (P - dy/dt)/D . (B-28)
Since P » dy/dt. Equation 3-28 1s usually written as
yss - P/D . (B-29)
Note that dy/dt need not be zero, but Us absolute value must be Insigni-
ficant compared with either P or yO. There 1s a special case, however,
where y does not react with Itself or another species, and 1n which the
steady-state concentration 1s dependent upon the concentration of y. For
example, 1f y reacts with Itself, then
rss
P/f(y) (B-30)
and the solution of ys- becomes quadratic. Analysis of the mechanisms
presented 1n Tables B-I, B-3, and B-4 Indicates that even higher-order
solutions must be considered for some steady-state species. Since the
numerical solution of higher-order algebraic equations can be time-consum-
ing, both mechanism design and numerical solution techniques must be
examined 1n an attempt to minimize computational effort. That 1s, to the
degree realistically possible, reactions between steady-state species
should be minimized during the design of the chemical mechanism, and a
general technique for handling higher-order steady-state relationships
must be utilized.
B-32
-------
Finally, 1t should be noted that Invoking the steady-state relationship
for a species, 1n effect, also removes that species from mass balance con-
siderations, I.e., the steady-state expression assumes that dy/dt • 0. If
J|dy/dt| 1s very small, then removal of this species does not matter. If,
however, J|dy/dt| 1s sufficiently different from zero that 1t affects the
overall mass balance, then the steady-state assumption produces Invalid
results.
Gas-Phase Steady-State Approximations
The gas-phase species approximated via steady-state relationships (listed
1n Table B-2) are 0, OH, H02, C203, N03, N205, X02, PHO, and X02N. The
last two species do not react to form steady-state products. Thus, once
and [OH]SS are determined, 1t 1s possible to calculate [PHO]SS and
from
i"w3'SS «
IX02N]SS
[PHO]ss[N02]k67 - [N03]ss[PHEN]k66 (
and
lX02M]sslNO]k75 - 0.067[OH]ss[PAR]k56
+ 0.09[N03]sslOLE]k60MN03lsslISOP]k74 . (B-32)
The remaining steady-state species are related 1n a complex manner and
their concentrations must be determined simultaneously. We use a modified
Newton-Raphson method to determine the roots of these coupled steady-state
equations. This approach converges extremely rapidly on the proper roots
(steady-state concentrations) 1f good first approximations of those roots
can be made.
The first approximations for the concentrations of steady-state species
can be greatly simplified 1f some mechanistic assumptions are made. For
example, the steady-state equations for 0, N03, and N205 are
[0]SS00 - [NOglk! + (l-«)[03]kg + [03]k9
+ 0.85[N03]ssk3Q (B-33)
B-33
-------
+ lN205]ssk16 + [OH]sslHN03]k29 (B-34)
[N03]ss[N02]k14 . (B-35)
where « 1s a ratio described 1n Table B-l and Ox 1s the sum of partial
derivatives (with respect to steady-state species x) of all rate expres-
sions for first-order reactions of x or higher-order reactions with state
variables. For Instance,
D0 - k2 + [N02](k4 + ks) + [N0]k6 + [FORM]k37
+ [ALD2jk44 + [OLElk57 * [ETHlkgl + [ISOP]k71 . (B-36)
Except for the [OH][HN03JK2g term on the right side of Equation B-34,
these steady-state expressions are Independent of those of the remaining
four species. Under most conditions, Reaction 29,
OH + HN03 - - N03 + H20 , (R-29)
Is of minor Importance. If the effect of this reaction 1s temporarily
Ignored, the fourth unknown, the OH concentration, can be eliminated and
the remaining Independent equations can be used to calculate Initial esti-
mates of 0, N03, and N205 steady-state concentrations. Of course, after
the first estimates of roots are made, the entire steady-state expressions
must be used during subsequent calculations.
The chemical relationships between the remaining steady-state species are
more complex and Involve higher-order terms. An Initial estimate of
steady-state concentrations was made by assuming that all H02 and C203
reactions, Including self reactions, were Insignificant. Then, with
[Olss, (N031SS and (N205]ss previously determined, the functional depen-
dence of the simplified steady-state expressions Is
s = f(OH) <8
[H02]ss = f(OH, C203) (B-38)
B-34
-------
[OH]SS - f(H02) (B-39)
[X02]ss » f(OH. H02, C203) . (B-40)
Consecutive substitutions of (C203]ss Into the (H02)ss expression, and
[H02]ss Into the [OH]SS expression, provide an initial estimate of
[OHJSS. It 1s then assumed that the H02 self reactions represent half of
the total H02 radical reactions. This assumption, combined with the esti-
mated OH concentration, provides a somewhat constrained second-order equa-
tion for [H02)ss, which 1s solved using a modified quadratic equation.
Given these estimates of [OH]SS and [H02]ss, the complete X02 expression
can be substituted Into the complete C203 expression to yield a third-
order equation with respect to [C203]ss. Assuming that [C203]ss « 1, the
third-order term can be dropped and the quadratic equation solved to yield
an Initial estimate of (C203)ss.
Once these first approximations of the steady-state terms are performed,
the complete steady-state equations for all seven species are successively
Iterated to provide Initial estimates of the seven roots to be used 1n the
Newton-Raphson calculations. Tests of several single time step calcula-
tions for different chemical conditions have shown these approximations,
led by the Newton-Raphson method, not only converge to the proper solution
(defined here as the Gear solution for the same conditions), but to do so
with great efficiency. Once concentrations for the nine steady-state
species have been calculated, the Jacoblan elements needed 1n the Crank-
Nicholson calculations for state species can be determined, and the new
state concentrations can be calculated for the current time step.
Aqueous-Phase Steady-State Approximations
The concentrations of the aqueous-phase species OH, H02, N03, S04", S05~,
C03" and C10H" are estimated through steady-state calculations. Three of
these species (OH, H02, N03) also occur 1n the gas phase and can be
absorbed or volatilized by water droplets. This necessitates the linkage
of steady-state calculations between the two phases so that reversible
mass transfer can be considered for these species. Therefore, additional
steady-state production and destruction terms from Interactions with the
other phase must be Included 1f the chemistry under consideration 1s that
of a cloud cell. These terms are simply Included 1n the gas-phase steady-
state expressions Just presented and the aqueous-phase expressions
developed next. The formulation of these flux terms was described
earlier.
B-35
-------
Also unique to the aqueous-phase steady-state calculations 1s the Issue of
aqueous equilibria. Prior to each set of aqueous steady-state calcula-
tions, the equilibrium for H02(aq) must be reestablished so that the vary-
ing contributions of H02 and 02~ can be determined. This equilibrium was
discussed previously.
In the present mechanism, the Initial estimate of N03 can be decoupled
from the other aqueous-phase steady-state species (though not from the
gas-phase) because there are no aqueous sources of N03. Hence, it 1s pos-
sible to Independently estimate the concentration using only rate con-
stants, absorption rate, and concentrations of state species. The remain-
Ing terms are all interrelated and their functionality 1s described by
[S04-]ss - f(H02, N03, S05-) (
[OH]SS - f(H02, S04-, C10H-) (B-42)
[S05-)ss - f(OH, H02, S04-) (B-43)
[C10h"]ss - f(OH) (B-44)
[C03-]ss - f(OH. H02. S04') (B-45)
[H02]ss - f(OH, N03, S04", S05-, C03') . (B-46)
As with the gas-phase calculations, Initial approximations of [S04~]ss and
[H021SS can be estimated from a knowledge of the system and typical
values. This allows simultaneous estimates of [OH]SS and [C10H~]$S, and
finally, of the remaining species. These estimates are then refined by
successive Iterations through the complete set of steady-state expressions
to provide good Initial estimates of these concentrations using these
estimates, the Newton-Raphson algorithm 1s used to determine all steady-
state concentrations, ultimately leading to the determination of the terms
needed for the Crank-Nicholson solution of the state species concentra-
tions.
Nighttime Chemistry
Simulation of chemical reactions occurring as the sun sets and at night
presents computational problems that are not immediately apparent. As the
sun sets, the production rates of many radical species approach zero since
photolytlc rates are zero at night. Thus, the only nocturnal sources of
radicals in the gas-phase (Table B-l) are 03 reactions with N02 and ole-
f1n1c species (ETH, OLE, and ISOP), and PAN unimolecular decomposition.
B-36
-------
All Oj production and almost all NO production 1n the CBM-IV 1s due to
photolytlc reactions, whereas the relatively rapid rate of the second-
order reaction,
NO + 03 N02 . (R-3)
1s unaffected at night. Hence, NO and 03 cannot readily coexist 1n the
dark since the predicted concentration of at least one species rapidly
becomes zero, and later reactions of that species are then unimportant.
Finite-difference solutions become Inefficient when simulating a rapid
zero asymptote because the time step required to avoid negative concentra-
tions can become small. To eliminate this condition 1n CCADM, we perform
a computational tltratlon of NO and 03 prior to use of the nighttime
chemistry subroutine. As noted, this results In either a NO or 03 concen-
tration of zero. Thus, two different chemical regimes can be entered Into
the nighttime chemistry subroutine. The concentrations of these species
are then tested to determine which regime 1s Important, and the chemistry
1s solved with a code that 1s efficient for that regime.
For cases 1n which the NO concentration 1s high and the 03 concentration
approaches zero, only the reactions among NO, N02, HN02, and H20 (plus N02
formation from PAN and PNA decomposition) are Important. The chemistry of
this regime can therefore be simply represented by reactions 17, 18, 23,
25, and 50 In Table B-l. PAN and PNA decomposition 1s assumed to produce
only NO*. Since no steady-state approximations need be made and only a
few state species are Important, determination of the Crank-Nicholson
Jacoblan elements 1s a relatively simple process.
The alternative regime, 1n which the NO concentration approaches zero,
must have the ability to determine the nocturnal radical concentrations
produced from reactions with 03. Although the radical concentrations pro-
duced at night are minor compared to those produced during photolytlc
reactions, they cannot be Ignored since they are the only radical sources 1n
the mechanism. Therefore, even though the NO reactions can be Ignored 1n
the nighttime 03 regime, the consideration of radical chemistry requires a
more formidable reaction set than does the alternative nighttime regime.
In addition to the larger reaction set required 1n the nighttime ozone-
dominated regime, nitrogen mass balance must be carefully considered
because the rapid reaction between NO and N03 does not readily occur
(when [NO! « 0), resulting 1n the Initial formation of high N03 and NoOg
concentrations by the reaction of NO? and 03 (reaction 7). The sum of N03
and twice N205 can be of the same relative concentration as N02 under
B-37
-------
these conditions. Since steady-state determination of N03 and N20c con-
centrations does not account for the mass 1n those species, a nighttime
chemistry routine must decrement the Initial NOX concentration by the
amount of steady-state nitrogen produced (N03 and twice N205) to correctly
describe this portion of the chemistry. In a similar manner, the Initial
ozone must be decremented by the number of ozone molecules required to
produce the steady-state N03 and N20g.
The steady-state equations for (N03]_s and lN2Og]ss were presented
earlier. Since no nocturnal production of 0 Is predicted by the mechanism
and reaction 29 can be temporarily Ignored, these equations can be modi-
fied 1n the following manner:
[NOJl03]k7 + [N?OJ klfi
'"°3]7+ I.J ; C
[N205]ss » IN03]SS[N02] • RKX (B-48)
where
RN(J » [FORM]k38 + [ALD2]k46 + lOLE)k6() + iTOLjkgg + [PHENlk^
+ [XYL]k6g + [ISOP]k74 (B-49)
RKX - k/(k + k) . (B-50)
As explained previously, the Initial NOX of the 03-dom1nated regime 1s
equal to Initial N02 since NO and 03 are titrated prior to Input. The
concentrations of N02 and 03 after determination of steady-state concen
trations for N03 and N2C>5 can be represented by
[N02]Q - [N03]ss - 2
l°3Us
B-38
-------
where
[N02Jo and [03)0 » Initial N02 and 03 concentrations before the
nighttime chemistry calculations
[N021SS and [03]ss « steady-state N02 and 03 concentrations, which
equal the N02 and 63 concentrations decremented
by the amount of steady-state nitrogen 1n N03
and N20g.
Substitution of these equations Into the N205 and N03 steady-state
expressions yields
- 'N03'ss
(NO 1 .
3'ss (IN02|0 - IN03ISS - 2|H205ISS) (k13
Now substitution of [N205]ss Into the [N03]ss equation will yield an equa-
tion describing the steady-state N03 concentration 1n terms of known vari-
ables. Omitting the appropriate algebra, the cubic equation for (N03]ss
1s
[RKX[4 - RNQ • RKX - 2(k13 + kM - klfi - RKX) - k?]
• [4 • RNQ - RKX + (2[M02]0 RKX - I)(k13 + kw RKX) + k7(2[03l0 . RKX
• [RNQ + lN02]Q(k13 + kM - k16 RKX) * k14([N02l0 + [03l0
* [N02]Q RKX([N02]0 - 2(03l0))]
1 " ° (B'55)
B-39
-------
This equation can be solved for (N03LS through use of a Newton-Raphson
solution of the cubic equation since the mathematical constraints of the
system make It possible to Identify the correct root. Convergence 1s
usually achieved 1n two or three Iterations. After this calculation 1s
made, the estimate of IN03]SS can be substituted Into Equation B-54 to
calculate [^05)55. These values are used to determine the N02 and 03
concentrations for subsequent calculations within the current call to the
chemistry submodule by substitution Into Equations B-51 and B-52.
Concentrations for the remaining four steady-state species (OH, HOg, CoOj,
and X0£) are calculated using the Newton-Raphson method described earlier,
though the number of terms and the order of the steady-state expressions
1s lower because of the more limited nocturnal reaction set. Once cover-
gence 1s achieved for the steady-state concentrations, the rates of change
for state species and the JacobIan elements that.represent the changes 1n
one state species with respect to another can be determined. These are
used 1n a Crank-Nicholson (predictor-corrector) solution of the chemistry
of state variables 1n the current time step.
DEMONSTRATION OF CHEMISTRY SUBMOOULES AND MASS BALANCE
The chemistry submodules were designed and tested 1n a separate, stand-
alone box model. Although such a model cannot accurately simulate the
meteorology of the atmosphere, It provides the developer with a method for
Isolating and verifying the chemical calculations. A careful choice of
the Input parameters can allow the box model calculations to mimic real
conditions. Example box model simulation results are presented 1n Figures
B-2 and B-3 to demonstrate some of the chemical features of the chemistry
submodules. The Important consideration 1n examination of these results
1s whether or not, given conditions similar to those expected 1n the atmo-
sphere, the submodule predicts reasonable chemical changes and mass 1s
conserved In the formulation.
Two test scenarios were selected from the work of Walcek and Stockwell
(1986). The Initial conditions and fixed parameter values for these
scenarios, designated Ocean and Adirondack, are shown 1n Table B-7.
Results for cloudwater pH, formaldehyde, formic add, hydrogen peroxide
and S(VI) are shown 1n Figure B-2. Nitrogen and sulfur mass balance plots
are presented 1n Figure B-3. These plots are cumulative summations of
successive nitrogen and sulfur species; for Instance, the S(IV)
concentration versus time is the lowest line on each sulfur budget plot
and the S(IV) + S(VI) concentrations are summed under the second line,
with the S(VI) concentration represented as the difference between
B-40
-------
n. roa
9.0
8.0
4.0
3.0
8.0
1.0
0.0
ADIRONDACK
OCEANIC
J L
10
SO
inoi •». me
1.0
1.S
»•<»
0.5
0.0
ADIRONDACK
OCEAMC
'
-L 1 1 u
to
30 40
SIMULATION TOtt (mlnutM)
SO
60
FIGURE B-2a. Cloud water pH and Hgt^ concentration versus time for
Oceanic and Adirondack submodule simulations.
B-41
-------
HCOOH va. TDM
150.0
KCHOW.T1U1
360.0
aoo.0
250.0
K 200.0
'150.0
100.0
50.0
0.0
ADIRONDACK
OCEANIC
10
30 30 40
SOfUUTIOM TDM (mlnuUi)
80
FIGURE B-2b. Concentration versus time profiles for formic add and
formaldehyde 1n the Oceanic and Adirondack submodule simulations.
B-42
-------
s(vi)
120.0
00.0
•04)
0.0
ADIRONDACK
OCEANIC
80
30
40
SO
aOfUUTBN TIKE (mlnutra)
FIGURE,B-2c. S(VI) concentration versus time for Oceanic and Adirondack
submodule simulations (the S(IV) profiles can be obtained from the sulfur
budget versus time plots).
B-43
-------
MTBOCf N BUDGET v>. TIME (OoMole MM)
0.30
I
I
8
0.20
0.10
0.00
Other N-products-)
100
PAN
78
NO
SO
28
HMO,
i I I I
0 6 10 IB- 80
90 96 40 45 80 SB
SUITOR BUDGET m. HUE (Oounic MM)
0.»2
100
• 28
0.00
8 10 18 30 28. 30 36 40 4S 80 SB
•nnfunoN TDU
•tf
FIGURE B-3a. Nitrogen and sulfur budgets for Oceanic submodule
simulation.
B-44
-------
NRMGCM BUDOCT *•. HUE (Adirondack
1.28
1.00
1
0.79
0.50
0.00
100
MN
78
SO
88
NNOj
j i
0 81018
303640458056
•is
mnru* IUDOCT «•. not
1J8
1.00
S
0.78
0.60
0.88
0.00
100
$(»!)
79
90
aa
SUV)
0 8 10 18
88 30 38
SWUUTIOH m« (adautM)
48 80 88
FIGURE B-3b. Nitrogen and sulfur budgets for Adirondack submodule
simulation. B"45
-------
TABLE B-7. Chemical species and Initial conditions used 1n
simulations.
Organic Species (ppb)
ETH
OLE
PAR
TOL
XYL
FORM
ALD2
FACD
MHP
PAN
PAA
CH4
Inorganic Species (ppb)
NO
NCU
HN03
NH,
°3
SOo
CO
C02
H20
Other Species
H2S04 (ug/ra3)
Fe(+3) (M)
Mn(+2) (M)
Na(+) (M)
Meteorological Conditions
ki (min"1)
T (<) ,
Cloud Water (g/mj)
Cloud Transmisslvity (%)
Cloud Drop Radius (urn)
Initial
Adirondack
0.026
0.001
2.50
0.02
0.001
0.29
0.092
0.062
0.85
0.60
0.080
1400.
0.016
0.039
0.55
0.60
53.2
0.95
0.980
200.
340xl03
8500.
0.80 ,
4xio-;
2xio-;
IxlO'7
0.6
278.
0.4
50
10
Conditions
Ocean
0.001
0.001
0.33
0.001
0.001
0.24
0.018
0.001
0.30
0.09
0.007
1400.
0.028
0.069
0.09
0.15
47.7
1.20
0.065
100. ,
340xl03
8500.
0.15
9x10'°
4x10"°
2xlO'8
0.6
278.
0.4
50
10
B-46
-------
the two lines. Finally, Figure B-4 shows the rate of loss of S(IV) as a
function of oxidizing species and time (note the differences 1n ordlnate
scales).
The Adirondack scenario always has higher Initial conditions (except for
NO, N0£, and h^C^), often by an order of magnitude (Table B-7). The plots
In Figure 3-2 show the effects of this difference 1n lower cloudwater pH
(due mainly to larger Initial H2S04), more rapid decay of H202 and a
higher production rate of S(VI) (there 1s about 15 times more Initial S02
1n this case), and greater production of formaldehyde and formic add
(from much larger concentrations of gas-phase organlcs). The Oceanic case
1s organic- and S02- limited, producing far less S(VI) or oxidized organlcs
at a higher pH.
Figure B-2 shows that after the Initial depletion of oceanic S02 the rate
of S(IV) decay drops to near zero. More Important, because the pH 1s
maintained near 5, oxidation of S02 by other species than h^O? 1s
competitive. This 1s less the case 1n the Adirondack simulation where pH
Is less than 4 and the rates of all major oxidizing species other than
decrease.
Apparent sulfur mass loss 1s trivial throughout these simulations because
only one reaction produces a final product that 1s not followed by further
chemical reactions. This product 1s hydroxylamlne dlsulfonate (Oblath et
al., 1981) and Its formation occurs 1n reaction 233b of Table B-4. The
complete reaction 1s
2 HS03" + N02" ----- > OH" + HON(S03)22' . (R-233b)
Sulfur and nitrogen are both permanently lost from the respective totals
because the HON(S03)2 is not accounted for. In the Ocean and Adirondack
scenarios shown, 6 x 10"9 and 1.5 x 10"8 ppm were unaccounted for after
one hour (0.0061 and 0.0013 percent). The sulfur budget plots (Figure B-
4) show that variation from 100 percent Initial sulfur cannot be visually
discerned after one hour.
Apparent nitrogen losses are more significant 1n the current formulation
because additional unaccounted produce losses can occur 1n reactions 60
and 227 (Tables B-l and B-4):
PHO + N02 ----- > nUrophenol (R-60)
S(IV) + PAN ----- > S(VI) + N-product (R-227)
We have not Included these product species because they have no known,
significant reactions and therefore would only Increase the size of the
B-47
-------
OCEANIC CASE
r 1 1 r— —r- —r- —r- —r~ —i
00
0 10 IS 20 29 30 39 40 49 50 59 «0
ADIRONDACK CASE
50.0
1 1 '—\ r- —r- —r- —r
50.0
40.0
40.0
1
30.0
•X 20.01
10.0
0.0
III
30.0
20.0
10.0
III I
• I
•"0 9 10 15 20 29 JO 39 40 49 90 » 60
SMUATON TMC (minute.)
FIGURE B-4. Rate of S(IV) loss versus time divided Into specific
oxidizing species.
8-48
-------
state species solution matrix. It would also be misleading to Include the
nitrogen-containing products as 1f they were other species. Hence, some
additional apparent loss of nitrogen results. In the Ocean and Adirondack
scenarios, 3.9 x 10"6 and 2.8 x 10"6 ppm were unaccounted for after one
hour. This represented 1.41 and 0.23 percent of Initial nitrogen (0.28
and 1.21 ppb). Although this seems somewhat high, we note that the
Initial nitrogen was rather low (0.28 and 1.21 ppb), with PAN accounting
for 32 and 50 percent of that mass. Although these Initial conditions may
be unrealistic, they show the effect of unaccounted product mass,
especially that of reaction 227. Linear Integration of the three reaction
throughput rates for each minute over the one-hour simulation provides an
estimate of the unaccounted mass, which 1s equal to the differences noted
(within Integration error). From this we conclude that the mathematical
representation 1n the chemistry submodules 1s mass conservative (1f one
considers the three unaccounted products) to within the roundoff error
limits of the computer.
B-49
-------
REFERENCES
Allen, J. H.( J. D. Mallck, and S. M. Serebreny. 1984. "Modeling the
Vertical Distribution of Clouds." SRI International, Menlo Park,
California.
Anbar, M., and P. Neta. 1967. A compilation of specific blmolecular rate
constants for the reactions of hydrated electrons, hydrogen atoms and
hydroxyl radicals with Inorganic and organic compounds 1n aqueous
solution. Int. J. Appl. Radlat. Isot.. 18:493-523.
Atkinson, R. 1985. Kinetics and mechanisms of the gas-phase reactions of
the hydroxyl radical with organic compounds under atmospheric condi-
tions. Chem. Rev.. 85:69-201.
Atkinson, R., and A. C. Lloyd. 1984. Evaluation of kinetic and mechanis-
tic data for modeling of photochemical smog. J. Phys. Chem. Ref.
Data. 13:315-444.
Atkinson, R., and W.P.L. Carter. 1984. Kinetics and mechanisms of the
gas-phase reactions of ozone with organic compounds under atmospheric
conditions. Chem. Rev.. 84:437-470.
Bahnemann, D., and E. J. Hart. 1982. Rate constants of the reaction of
the hydrated electron and hydroxyl radical with ozone 1n aqueous
solution. J. Phys. Chem.. 86:252-255.
Bass, A. 1985. Personal communication with Arnold M. Bass, National
Bureau of Standards, Galthersburg, Maryland.
Baulch, D. L., R. A. Cox, P. J. Crutzen, R. F. Hampson, Jr., J. A. Kerr,
J. Troe, and R. T. Watson. 1982. Evaluated kinetic and photochemi-
cal data for atmospheric chemistry: Supplement I. J. Phys. Chem.
Ref. Data. ll(2):327-466.
Baulch, D. L., R. A. Cox, R. F. Hampson, Jr., J. A. Kerr, J. Troe, and
R. T. Watson. 1984. Evaluated kinetic and photochemical data for
atmospheric chemistry: Supplement II. CODATA Task Group on gas
phase chemical kinetics. J. Phys. Chem. Ref. Data. 13(4):1259-1380.
B-50
-------
Behar, D.t G. Czapskl, J. Rabanl, L. M. Dorfman, and H. A. Schwartz.
1970. The add dissociation constant and decay kinetics of the per-
hydroxyl radical. J. Phys. Chem.. 74:3209-3213.
Behar, D., G. Czapskl, and I. Duchovny. 1970. Carbonate radical 1n flash
photolysis and pulse radlolysls of aqueous carbonate solutions. J..
Phys. Chem.. 74:2206-2210.
B1e1sk1, B.H.J. 1978. Reevaluatlon of the spectral and kinetic proper-
ties of H02 and OjJ free radicals. Photoehem. Photoblol.. 28:645-649.
Botne, E., and D. Schulte-FrohUnde. 1980. Reaction of dlhydroxymethyl
radical with molecular oxygen 1n aqueous solution. Z. Naturforsch.
B. Anorg. Chem. Org. Chem.. 35:1035-1039.
Boyce, S. D., and M. R. Hoffmann. 1984. Kinetics and mechanisms of the
formation of hydroxymethanesulfonlc add at low pH. J. Phys. Chem..
88:4740-4746.
Bray, W. C. 1938. The Interaction of ozone and hydrogen peroxide 1n
aqueous solution. J. Am. Chem. Soc.. 60:82-87.
Buhler, R. E., J. Staehelln, and J. Holgne. 1984. Ozone decomposition 1n
water studied by pulse radlolysls, 1, H09/OZ and HCL/OI as Intermedi-
ates. J. Phys. Chem.. 88:2560-2564. c c 3 J
Byers, H. R. 1974. General Meteorology. McGraw H111, New York.
Calvert, J. G. 1980. "The Homogeneous Chemistry of Formaldehyde Genera-
tion and Destruction Within the Atmosphere." Proceedings of the NATO
Advanced Study Institute on Atmospheric Ozone. U.S. Department of
Transportation, F.A.A., Office of Environment and Energy, High Alti-
tude Program, Washington, D.C.
Cantrell, C. A., W. R. Stockwell, L. G. Anderson, K. L. Busarow,
D. Perner, A. Schmeltekopf, J. G. Calvert, and H. S. Johnston.
1985. Kinetic study of the NO^-CH20 reaction and Its possible role
1n nighttime tropospheHc chemistry. J. Phys. Chem.. 89:139-146.
Chameldes, W. L. 1984. The photochemistry of a remote marine stratiform
cloud. J. Geophys. Res.. 89:4739-4755.
Chameldes, W. L., and D. D. Davis. 1982. The free-radical chemistry of
cloud droplets and Its Impact upon the composition of rain. J. Geo-
phys. Res.. 87:4863-4877.
B-51
-------
Chen, S., V. W. Cope, and M. Z. Hoffman. 1973. Behavior of COZ radicals
generated 1n the flash photolysis of carbonatoamlnes complexes of
cobalt(III) 1n aqueous solution. J. Phys. Chem.. 77:1111-1116.
Chrlstensen, H., K. Sehested, and H. CorfHzen. 1982. Reactions of
hydroxyl radicals with hydrogen peroxide at ambient and elevated tem-
peratures. J. Phys. Chem.^ 86:1588-1590.
Oamschen, 0. E.t and L. R. Martin. 1983. Aqueous aerosol oxidation of
nitrous add by 02, 03, and H202. Atmos. Environ.. 17:2005-2011.
Dodge, M. C., and R. R. Arnts. 1979. A new mechanism for the reaction of
ozone with oleflns. Int. J. Chem. K1net.. 11:399-410.
Oog11ott1, L.t and E. Hayon. 1967. Flash photolysis of persulfate Ions
1n aqueous solutions. Study of the sulfate and ozonlde radical
Ions. J. Phys. Chem.. 71:2511-2516.
EPA. 1986. U.S. EPA Workshop to Develop a Simple Model for Add Deposi-
tion. Atmospheric Research Sciences Laboratory. Office of Research
and Development, U.S. Environmental Protection Agency, Research Tri-
angle Park, North Carolina (EPA-600/9-86/017).
Erlckson, R. E., L. M. Yates, R. L. Clark, and D. McEwan. 1977. The
reaction of sulfur dioxide with ozone 1n water and Its possible atmo-
spheric significant. Atroos. Environ.. 11:813-817.
Farhatazlz, and A* B. Ross. 1977. "Selected Specific Rates of Reactions
of Transients from Water 1n Aqueous Solution, III: Hydroxyl Radical
and Perhydroxyl Radical and Their Radical Ions." U.S. Department of
Commerce, Washington, DC (NSRDS-NBS 59).
Gear, C. W. 1971. Numerical Initial Value Problems 1n Ordinary Dif-
ferential Equations. Prentice-Hall, New Jersey.
Graedel, T. E., and C. J. Weschler. 1981. Chemistry within aqueous atmo-
spheric aerosols and raindrops. Rev. Geoohys. Space Phys.. 19:505-
539.
Graedel, T. E., and K. I. Goldberg. 1983. Kinetic studies of raindrop
chemistry, 1, Inorganic and organic processes. J. Geophys. Res..
88C:10,865-10,882.
Graham, R. A., and H. S. Johnston. 1978. The photochemistry of N03 and
the kinetics of the N205-03 system. J. Phvs. Chem.. 82(3):254-268.
B-52
-------
Gratzel, M.t A. Hengleln, and S. Tanlguchl. 1970. Pulsrad1olyt1sche Beo-
bachtungen iiber die Reduktlon des NO^-Ions und u'ber BUdung und Zer-
fall des persalpetrlgen Saure 1n wassrlger Losung. Ber. Bundsenqes.
Phys. Chem. 74:292-298.
Gunn, K.t and 6. D. Klnzer. 1949. J. Meteorol.. 6:243.
Hagesawa, K.t and P. Neta. 1978. Rate constants and mechanisms of reac-
tion for Clg radicals. J. Phys. Chem.. 82:854-857.
Hales, J. M., and D. R. Drewes. 1979. Solubility of ammonia In water at
low concentrations. Atmos. Environ.. 13:1133-1147.
Haruta, K.( and T. Takeyama. 1981. Kinetics of oxidation of aqueous bro-
mide 1on by ozone. J. Phys. Chem.. 85:2383-2388.
Hicks, B. B. 1983. In Critical Assessment Document on Add Deposition
(Chapter VII). ATDL contribution File No. 81/24. Atmospheric Turbu-
lence and Diffusion Laboratory, NOAA, Oak Ridge, Tennessee.
Hitchcock, D. R., L. L. Splller, and W. E. Wilson. 1980. SulfuHc add
aerosols and HC1 release 1n coastal atmospheres: evidence of rapid
formation of sulfurlc add partlculates. Atmos Environ.. 14:165-182.
Hoffmann, M. R., and J. G. Calvert. 1985. "Chemical Transformation
Modules for EuleHan Add Deposition Models. Vol. 2—The Aqueous-
Phase Chemistry." U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina (EPA/600/3-85/017).
Holgne, J., and H. Bader. 1983a. Rate constants of reactions of ozone
with organic and Inorganic compounds 1n water, 1, Non-dissociating
organic compounds. Water Res.. 17:173-183.
Holgne, J., and H. Bader. 1983b. Rate constants of reactions of ozone
with organic and Inorganic compounds 1n water, 2, Dissociating
organic compounds. Water Res.. 17:185-194.
Hu1e, R. E., and P. Neta. 1984. Chemical behavior of SO* and SOg
radicals 1n aqueous solutions. J. Phys. Chem., 88:5665-5669.
Hu1e, R. E. and P. Neta. 1986. Rate constants for some oxidations of
S(IV) by radicals 1n aqueous solutions. Submitted to Atmos. Environ.
B-53
-------
Jacob, D. J. 1986. Chemistry of OH 1n remote clouds and Us role 1n the
production of formic add and peroxymonosulfate. J. Geophys. Res.,
91(D9):9807-9826.
Jayson, G. G., B. J. Parsons, and A. J. Swallow. 1973. Some simple,
highly reactive, Inorganic chlorine derivatives 1n aqueous solu-
tion. Trans. Farad. Soc.. 69:1597-1607.
Jet Propulsion Laboratory. 1983. "Chemical Kinetics and Photochemical
Data for Use 1n Stratospheric Modeling." Evaluation Number 6. Pasa-
dena, California (83-62).
Jet Propulsion Laboratory. 1985. "Chemical Kinetics and Photochemical
Data for Use 1n Stratospheric Modeling." Evaluation Number 7. Pasa-
dena, California (85-37).
Kerr, J. A., and J. G. Calvert. 1985. "Chemical Transformation Modules
for Eulerlan Add Deposition Models. Volume I: The Gas-Phase
Chemistry." National Center for Atmospheric Research, Boulder,
Colorado (EPA-600/3-85-015).
Kessler, E. 1969. On the distribution and continuity of water substance
1n atmospheric circulations. Meteor. Monoqr.. No. 32, 84 pp.
Koenlg, L. R., and F. W. Murray. 1976. Ice-bearing cumulus cloud evolu-
tion, numerical simulation and general comparison against observa-
tions. J. Appl. Meteorol.. 15:747-766.
Kosak-Channlng, L. F., and G. R. Helz. 1983. Solubility of ozone 1n
aqueous solutions of 0-0.6 M 1on1c strength at 5-30°C. Environ. Sc1.
Techno!., 17(3):145-149.
Latlmer, W. M. 1952. The Oxidation States of the Elements and Their
Potentials 1n Aqueous Solutions. Prentice-Hall, New York (pp. 70-
89).
Le Henaff, P. 1968. Methodes d'etude et propHetes des hydrates, heml-
acetals et hemlthloacetals derives des aldehydes et des cetones.
Bull. Soc. Ch1m. France. 4687-4700.
Ledbury, W., and E. W. Blair. 1925. The partial formaldehyde vapour
pressure of aqueous solutions of formaldehyde. Part II. J. Chem.
Soc., 127:2832-2839.
B-54
-------
Lee, Y-N. 1984a. "Kinetics of Some Aqueous-Phase Reactions of Peroxy-
acetyl Nitrate." In 6as-L1qu1d Chemistry of Natural Waters, Vol. 1,
21/1-21/7, Brookhaven National Laboratory (BNL 51757).
Lee, Y-N. 1984b. "Atmospheric Aqueous-Phase Reactions of Nitrogen
Species." In Gas-L1qu1d Chemistry of Natural Waters. Brookhaven
National Laboratory.
Lee, Y.-N., and J. A. L1nd. 1986. Kinetics of aqueous-phase oxidation of
nltrogen(III) by hydrogen peroxide. J. Geophys. Res.. 91(02):2793-
2800.
Lee, Y-N., and S. E. Schwartz. 1981. Reaction kinetics of nitrogen dio-
xide with liquid water at low partial pressure. J. Phys. Chem..
85:840-848.
Lee, Y-N., and S. E. Schwartz. 1983. "Kinetics of Oxidation of Aqueous
Sulfur(IV) by Nitrogen Dioxide." In Precipitation Scavenging. Dry
Deposition, and Resuspenslon. Volume 1, H. R. Pruppacher, R. G.
Semonln, and W.G.N. SUnn, eds., Elsevler, New York.
Le Henaff, P. 1968. Methodes d1 etude et propHetes des hydrates, heml-
acetals et hemlthloacetals derives des aldehydes et des cetones.
Bull. Soc. CMm. France. 4687-4700. .
Lewellen, W. S., and Y. P. Sneng. 1980. "Modeling and Dry Deposition of
SO? and Sulfate Aerosols." Electric Power Research Institute, Palo
Alto, California (Report EA-1452).
Lille, J., A. Hengleln, and R. J. Hanrahan. 1978. "Reactions of the Car-
bonate Radical Anlon with Organic and Inorganic Solutes 1n Aqueous
Solution." Presented at the 176th meeting of the American Chemical
Society, Miami Beach, Florida.
L1nd, J. A., and G. L. Kok. 1986. Henry's law determinations for aqueous
solutions of hydrogen peroxide, methylhydroperoxlde, and peroxyacetlc
add. J. Geophvs. Res.. 91(07):7889-7895.
Livingston, R., and H. Zeldes. 1966. Paramagnetic resonance study of
liquids during photolysis—hydrogen peroxide and alcohols. J. Chem.
Phys.. 44:1245-1259.
Markovlc, V., and K. Sehested. 1972. "Rad1olys1s of Aqueous Solutions
of Some Simple Compounds Containing Aldehyde Groups, I, Formalde-
hyde." "Proc. Tlhanv Symp. on Radlat. Chem.. 3d edition, J. Dobo and
P. Hadvlg, eds., Akadem1n1 K1ado, Budapest (pp. 1269-1280).
B-55
-------
Marsh, A.R.W., and W. J. McElroy. 1985. The dissociation constant and
Henry's law constant of HC1 1n aqueous solution. Atmos. Environ..
19:1075-1080.
Martell, A. E., and R. M. Smith. 1977. Critical Stability Constants.
Vol. 3; Other Organic Ugands. Plenum Press, New York, 495 pp.
Martin, L. R. 1984. "Kinetic Studies of Sulflte Oxidation 1n Aqueous
Solution." In S02. NO and N02 Oxidation Mechanisms: Atmospheric
Considerations. J. 6. Calvert, ed., Butterworth, Boston, Massachu-
setts (pp. 63-100).
Martin, L. R., and D. E. Damschen. 1981. Aqueous oxidation of sulfur
dioxide by hydrogen peroxide at low pH. Atmos. Environ.. 15:1615-
1621.
Martin, L. R., D. E. Damschen, and M. S. Jude1k1s. 1981. The reaction of
nitrogen oxide with SOp In aqueous aerosols. Atmos. Environ..
15:191-195.
Martin, L. R., D. E. Damschen, D. J. Goulston, and H. S. Jude1k1s.
1983. "Atmospheric Aerosol Chemistry of Sulfur and Nitrogen
Oxides." The Aerospace Corporation, El Segundo, California.
McArdle, J. V., and M. R. Hoffmann. 1983. Kinetics and mechanism of the
oxidation of aquated sulfur dioxide by hydrogen peroxide at low pH.
J. Phys. Chem.. 87:5425-5429.
McElroy, W. J. 1986. The aqueous oxidation of S02 by OH radicals.
Atmos. Environ.. 20:323-330.
Hunger, J. W., C. Tiller, and M. R. Hoffmann. 1986. Identification of
hydroxymethanesulfonate 1n fog water. Science. 231:247-249.
Moller, U., and G. Schumann. 1970. Mechanisms of transport from the
atmosphere to the earth's surface. J. Geophys. Res.. 75:3013-3019.
NBS. 1965. "Selected Values of Chemical Thermodynamlc Properties."
National Bureau of Standards. 1, NBS Technical Note 270-1, 124 pp.
Oblath, S. B., S. S. Markowltz, T. Novakov, and S. G. Chang. 1981.
Kinetics of the formation of hydroxylamlne dlsulfonate by reaction of
nitrite with sulfltes. J. Phys. Chem.. 85:1017-1021.
B-56
-------
Perner. D., Schmeltekopf, A., R. H. Wlnkler, H. S. Johnston, J. G. Cal-
vert, C.A. Cantrell, and W. R. Stockwell. 1985. A laboratory and
field study of the equilibrium N20g * NO- + N02. J. Geophys. Res..
90(D2):3807-3812.
PerHn, D. D. 1982. Ion1zat1on Constants of Inorganic Adds and Bases 1n
Aqueous Solution, 2d edition. Pergamon Press, New York.
Plum, C. N., E. Sanhueza, R. Atkinson, W.P.L. Carter, and J. N. Pitts,
Jr. 1983. OH radical rate constants and photolysis rates of a-
dlcarbonyls. Environ. Sc1. Techno!.. 17(8):479-484.
Pruppacher, H. R., and J. D. Klett. 1978. M1crophys1cs of Cloud and Pre-
cipitation. D. Reldel, Dordrecht, The Netherlands, 714 pp.
Rettlch, T. R. 1978. Some photochemical reactions of aqueous nitric
add. D1ss. Abstr. Int. B.. 38:5968.
Ross, A. B., and P. Neta. 1979. "Rate Constants for Reactions
of Inorganic Radicals 1n Aqueous Solution." U.S. Department of Com-
merce, Washington, DC (NSRDS-NBS 65).
Sander, S. P., and C. C. Klrcher. 1986. Temperature dependence of the
reaction NO + N03 * 2N02. Chem. Phvs. Letters. 126(2):149-152.
Saxena, P., C. Seigneur, and A. K. Pollack. 1986. A critical assessment
of existing data bases for add deposition modeling studies. J. A1r.
Pollut. Control Assoc.. 36, 1n press.
Schmidt, K. H. 1972. Electrical conductivity techniques for studying the
kinetics of radiation-Induced chemical reactions 1n aqueous solu-
tions. Int. J. Radlat. Phys. Chem.. 4:439-468.
Scholes, G., and R. L. WHlson. 1967. y-rad1olys1s of aqueous thymlne
solutions. Determination of relative reaction rates of OH radi-
cals. Trans. Faraday Soc.. 63:2982-2993.
Schwartz, S. E. 1983. "Mass-Transport Considerations Pertinent to Aque-
ous-Phase Reactions of Gases 1n Liquid-Water Clouds." Presented at
NATO Advanced Study Institute on Chemistry of Multiphase Atmospheric
Systems, 26 September-8 October 1983, Corfu, Greece (BNL-34174).
Schwartz, S. E. 1984a. Gas- and aqueous-phase chemistry of H0£ 1n liquid
water clouds. J. Geophys. Res.. 89:11,589-11,598.
B-57
-------
Schwartz, S. E. 1984b. "Gas-Aqueous Reactions of Sulfur and Nitrogen
Oxides In Liquid-Water Clouds." In S02, NO. and N02 Oxidation
Mechanisms: Atmospheric Considerations J. G. Calvert, ed., Butter-
worth, Boston, Massachusetts (pp. 173-208).
Schwartz, S. E., and J. E. Freiberg. 1981. Mass-transport limitation to
the rate of reaction of gases 1n liquid droplets: application to
oxidation of S02 1n aqueous solutions. Atmos. Environ.. 15(7):1129-
1144.
Schwartz, S. E., and W. H. White. 1981. Solubility equilibrium of the
nitrogen oxides and oxyadds 1n dilute aqueous solution. Adv.
Environ. Sc1. Enq.. 4:1-45.
Sehested, K., J. Holcman, E. Bjergbakke, and E. J. Hart. 1984. A pulse
radlolytlc study of the reaction OH + 03 1n aqueous medium. J. Phys.
Chem.. 88:4144-4147.
Sehested, K., J. Holeman, and E. J. Hart. 1983. Rate constants and pro-
ducts of the reactions of e ", 01, and H with ozone 1n aqueous solu-
tions. J. Phys. Chem.. 87:1951-1954.
Sehested, K., 0. L. Rasmussen, and H. Frlcke. 1968. Rate constants of OH
with HOo, 0" and H.ot from hydrogen peroxide formation in pulse-
Irradiated oxygenated water. J. Phys. Chem.. 72:626-631.
Seigneur, C., and H. M. Barnes. 1986. Technical considerations 1n
regional aerosol modeling. In A1r Pollution Modeling and Its Appli-
cations. 5, C. de Wlspelaere, ed., Plenum Press, New York, in press.
Seigneur, C., and P. Saxena. 1984. A study of atmospheric add formation
1n different environments. Atmos. Environ.. 18:2109-2124.
Shapllov, 0. D., and Y. L. Kostyukovsk11. 1974. Reaction kinetics of
hydrogen peroxide with formic add 1n aqueous solutions. K1net.
Katal.. 15:1065-1067.
Shelh, C. M., M. L. Wesely, and B. B. Hicks. 1979. Estimated dry depo-
sition velocities of sulfur over the eastern United States and sur-
rounding regions. Atmos. Environ.. 13:361-368
Shepherd, J. G. 1974. Measurements of the direct deposition of sulfur
dioxide onto grass and water by the profile method. Atmos. Environ..
8:69-74.
B-58
-------
Smith, R. M.t and A. E. Martell. 1976. Critical Stability Constants.
Vol. 4; Inorganic Complexes. Plenum Press, New York.
Sorensen, P. E., and V. S. Andersen. 1970. The formaldehyde-hydrogen
sulphite system 1n alkaline aqueous solution: kinetics, mechanism,
and equilibria. Acta Chem. Scand.. 24:1301-1306.
Staehelln, J., R. E. Buhler, and J. Holgne. 1984. Ozone decomposition 1n
water studied by pulse radlolysls, 2, OH, and HO^ as chain Intermedi-
ates. J. Phys. Chem.. 88:5999-6004.
Staehelln, J., and J. Holgne. 1982. Decomposition of ozone 1n water:
rate of Initiation by hydroxide Ions and hydrogen peroxide. Environ.
Sc1. Techno!.. 16:676-681.
Strehlow, H., and I. Wagner. 1982. Flash photolysis 1n aqueous nitrite
solutions. Z. Phys. Chem. Wiesbaden. 132:151-160.
Su, F., J. G. Calvert, and J. H. Shaw. 1979. Mechanism of the photooxl-
datlon of gaseous formaldehyde. J. Phys. Chem.. 83(25):3185-3191.
Taube, H., and W. C. Bray. 1940. Chain reactions 1n aqueous solutions
containing ozone, hydrogen peroxide, and add. J. Am. Chem. Soc..
62:3357-3373.
Trelnln, A., and E. Hayon. 1970. Absorption spectra and reaction
kinetics of NO?, N203, and NgO* 1n aqueous solutions. J. Am. Chem.
Soc.. 92:5821-5828.
Walcek, C. J., and G. R. Taylor. 1986. A theoretical method for compu-
ting vertical distributions of acidity and sulfate production within
cumulus clouds. 3. Atmos. Sc1.. 1n press.
Walcek, C. J., and W. R. Stockwell. 1986. The contribution of aqueous-
phase radical chemistry to sulfur and H202 chemistry 1n cloudy
environments. Submitted for publication.
Weeks, J. L., and J. Rabanl. 1966. The pulse radlolysls of deaerated
aqueous carbonate solutions, I, Transient optical spectrum and
mechanism; II, pK for OH radicals. J. Phys. Chem.. 70:2100-2106.
We1nste1n, J., and 6.H.J. B1elsk1. 1979. Kinetics of the Interaction of
H02 and 01 radicals with hydrogen peroxide; The Haber-Uelss reac-
tion. J. Am. Chem. Soc.. 101:58-62.
B-59
-------
Wesely, M. L., and B. B. Hicks. 1977. Some factors that affect the depo-
sition rates of sulfur dioxide and similar gases on vegetation. JL
A1r Pollut. Control Assoc.. 27:1110-1116.
Whitten, G. Z., J. P. Klllus, and R. G. Johnson. 1985a. "Modeling of
Auto Exhaust Smog Chamber Data for EKMA Development." Systems Appli-
cations, Inc., San Rafael, California (EPA-600-3/85-025).
Whltten, G. Z., J. P. Klllus, and R. G. Johnson. 1985b. "Development of
a Chemical Kinetic Mechanism for the U.S. EPA Regional Oxldant
Model." U.S. Environmental Protection Agency, Research Triangle
Park, North Carolina (EPA-600/3-85-026).
Whltten, G. Z., and M. W. Gery. 1985. "Development of CBM-X Mechanisms
for Urban and Regional AQSMs." U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina (EPA-600/3-86-012).
B-60
------- |