EPA/600/3-88/008
February 1988
ROCKY MOUNTAIN ACID DEPOSITION MODEL ASSESSMENT
Evaluation of Mesoscale Acid Deposition Models for
Use in Complex Terrain
by
R. E. Morris
R. C. Kessler
S. G. Douglas
K. R. Styles
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 27711
ATMOSPHERIC SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 22771
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing!
REPORT NO.
EPA/600/3-88/008
3. RECIPIENT'S ACCESSION N(3 ...
fBSS 1G7481/AS
«. TITLE AND SUBTITLE
ROCKY MOUNTAIN ACID DEPOSITION MODEL ASSESSMENT:
Evaluation of Mesoscale Acid Deposition Models
for Use in Complex Terrain
5. REPORT DATE
February 1988
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
R. E. Morris, R. C. Kessler, S. G. Douglas,
and K. R. Styles
B. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Systems Applications, Inc.
101 Lucas Valley Road
San Rafael, CA 94903
10. PROGRAM ELEMENT NO.
N104/C/05/05-5145 (FY-831
11. CONTRACT/GRANT NO
68-02-4187
12. SPONSORING AGENCY NAME AND ADDRESS
Atmospheric Sciences Research Laboratory - RTP.NC
Office of Research and Development
U. S. Environmental Protection Agency
Research Triangle Park. NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/03
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This report includes an evaluation of candidate meteorological models and acid
deposition models.
The hybrid acid deposition/air quality modeling system for the Rocky Mountains
makes use of a mesoscale meteorological model, which includes a new diagnostic wind
model, as a driver for a Lagrangian puff model that treats transport, dispersion,
chemical transformation, and dry and wet deposition. Transport will be defined from
the diagnostic wind model based on the wind at the puff center. The treatment of
dispersion will be based on the parameterization in the PNL/MELSAR-POLUT, while re-
training the MESOPUFF-II dispersion algorithms as an option. Based on the evaluation
of the chemical mechanisms, the RIVAD.chemistry appears to be the most scientifically
sound, as well as consistent, with the Lagrangian puff model formulation. Dry depo-
sition will use the CCADM dry deposition module with some minor adjustments. Wet
deposition will be based on the scavenging coefficient approach as used in the ERT/
MESOPUFF-II.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
COSATi Field/Croup
18. DISTRIBUTION STATEMENT
19. SECURITY CLASS (Tins Keporij
IINr.l ASSTFTFPI
21. NO. OF PAGES
24i
RELEASE TO PUBLIC
20. SECURITY CLASS fTins page)
22. PRICE
EPA F«fi» 2220-1 (R.». 4-77) *HKVIOU» EDITION it OBSOLETE
1
-------
N O
I C
THIS DOCUMENT HAS BEEN REPRODUCED FROM THE
BEST COPY FURNISHED US BY THE SPONSORING
AGENCY. ALTHOUGH IT IS RECOGNIZED THAT C E K
TAIN PORTIONS ARE ILLEGIBLE, IT IS BEING RI
LEASED IN THE INTEREST OF MAKING AVAILABLE
AS MUCH INFORMATION AS POSSIBLE
-------
NOTICE
The Information 1n this document has been funded by the United States
Environmental Protection Agency under Contract No. 68-02-4187 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
The hybrid add deposition/air quality modeling system for the Rocky Moun-
tains makes use of a mesoscale meteorological model, which Includes a new
diagnostic wind model, as a driver for a Lagranglan puff model that treats
transport, dispersion, chemical transformation, and dry and wet deposi-
tion. Transport will be defined from the diagnostic wind model based on
the wind at the puff center. The treatment of dispersion will be based on
the parameterization 1n the PNL/MELSAR-POLUT, while retaining the MESO-
PUFF-II dispersion algorithms as an option. Based on the evaluation of
the chemical mechanisms, the RIVAD chemistry appears to be the most scien-
tifically sound as well as consistent with the Lagranglan puff model
formulation. Dry deposition will use the CCADM dry deposition module with
some minor adjustments. Wet deposition will be based on the scavenging
coefficient approach, as used in the ERT/MESOPUFF-II.
This modeling approach was guided by the comments of members of the
Western Add Deposition Task Force (WADTF) 1n response to a questionnaire
mailed 1n August 1986 and a meeting 1n May 1987 in Denver. The modeling
approach recommended by members of the WADTF was use of a Lagrangian acid
deposition model with a complex-terrain wind model to calculate long-term
source-specific deposition of nitrogen and sulfur. This modeling approach
must be cost effective, simple enough for use by the regulatory agencies,
and slmiliar to the existing regulatory models used for impact assess-
ment. If possible, 1t was desirable that the model have the ability to
calculate PSD Increment consumption of SO-, and TSP sources. We feel that
the hybrid modeling system described in this report meets these require-
ments 1n the most technically rigorous manner possible, subject to the
cost and complexity constraints. The modeling approach is not as compre-
hensive as the Eulerian model development effort (RADM) currently being
carried out by the National Center for Atmospheric Research and State
University of New York at Albany. However, this approach is more
technically rigorous than those currently used by regulatory agencies, and
will generate more defensible estimates of incremental impacts of acid
deposition and concentrations 1n regions of complex terrain 1n the Rocky
Mountains.
In a previous report we reviewed existing meterological and add deposi-
tion models, and reported on the selection and preliminary evaluation of
iii
-------
four candidate mesoscale meterologlcal models (CIT/WINDMOD, LANL/ATMOS1,
PNL/MELSAR-MET. AND SAI/CTWM) and four candidate acid deposition models
(ERT/MESOPUFF-II. PNL/MELSAR-POLUT, SAI/CCADM, and SAI/RIVAD).* This
report 1s a continuation of that report and Includes the following topics:
(1) a more detailed evaluation of the candidate meterologlcal models
over terrain within the Rocky Mountains;
(2) the design of a new diagnostic wind (DWM) model that uses com-
ponents of the candidate meteorological models;
(3) an evaluation of the new DWM using the same criteria used to
evaluate the candidate mesoscale meteorological models, then
comparing Its predictions with observations from the Rocky Moun-
tains, and then evaluating the DWM for two geographic
settings: a complex terrain/coastal environment and within a
large valley;
(4) a detailed evaluation of the candidate add deposition models;
and
(5) the design of a new add deposition/air quality based on compo-
nents 1n the candidate acid deposition models.
* R. E. Morris and R. C. Kessler, "Rocky Mountain Acid Deposition Model
AssessmentReview of Existing Mesoscale Models for use 1n Complex
Terrain" (Morris and Kessler, 1987).
iv
-------
CONTENTS
Abstract 111
Acknowledgments x11
1 INTRODUCTION 1
1.1 Background 1
1.2 Purpose of This Report 2
1.3 Overview of the New Hybrid Acid Deposition/
A1r Quality Modeling System for the Rocky Mountains 3
1.4 Report Organization 3
2 EVALUATION OF THE CANDIDATE METEOROLOGICAL MODELS 5
2.1 Evaluation with an Idealized Terrain Obstacle 5
2.1.1 CIT Wind Model 5
2.1.2 MELSAR-MET 5
2.1.3 ATMOS1 7
2.1.4 Complex-Terrain Wind Model 7
2.1.5 Conclusions 7
2.2 Evaluation with Terrain from the Rocky Mountains 8
2.2.1 CIT Wind Model 8
2.2.2 MELSAR-MET 8
2.2.3 ATMOS1 15
2.2.4 Remarks 15
3 EVALUATION OF THE CANDIDATE ACID DEPOSITION MODELS 21
3.1 Transport 21
3.2 Dispersion 22
3.2.1 Description of the Dispersion Algorithms 27
3.2.2 Evaluation of the Dispersion Algorithms 29
3.3 Chemical Transformation 40
3.3.1 Review of the Chemistry of Acid Deposition 49
3.3.2 Review of the Chemical Mechanisms 1n the
Candidate Models 51
3.3.3 Evaluation of the Chemical Mechanisms 53
3.3.4 Remarks 63
-------
3.4 Dry Deposition 69
3.4.1 MESOPUFF-II and CCADM Parameter!rations 70
3.4.2 Comparison of MESOPUFF-II and CCADM Performance 77
3.5 Wet Deposition 86
3.5.1 Review of the Wet Deposition Algorithms 1n
the Candidate Models 86
3.5.2 Evaluation of the Wet Deposition Algorithms 88
3.5.3 Remarks 92
4 DESIGN OF THE METEOROLOGICAL MODEL 93
4.1 The Diagnostic Wind Model 94
4.1.1 Design Overview 94
4.1.2 Model Formulation 94
4.2 Evaluation of the Diagnostic Wind Model 102
4.2.1 Flow over Idealized Terrain 102
4.2.2 Flow over Rocky Mountain Terrain 103
4.2.3 Evaluation of the new OWM Using Observations
from the Rocky Mountains 116
4.2.4 Evaluation of the DWM 1n a Complex Terrain/
Coastal Environment and Within a Large Valley 139
4.3 Specification of Other Meteorological Variables 144
4.3.1 Mixing Heights 148
4.3.2 Stability Classification 148
4.3.3 Friction Velocity 149
4.3.4 Convectlve Velocity 151
4.3.5 Monln-Obukhov Length 151
4.3.6 Temperature 151
4.3.7 Pressure 153
4.3.8 Relative Humidity 154
4.3.9 Precipitation Rate 154
5 DESIGN OF THE ACID DEPOSITION MODEL FOR THE ROCKY MOUNTAIN REGION 159
5.1 Transport 159
5.2 Dispersion 160
5.3 Chemical Transformation 160
5.4 Dry Deposition 161
5.5 Wet Deposition 161
5.6 Summary 161
6 SUMMARY AND RECOMMENDATIONS 163
References 165
Appendix: Dry Deposition Velocities Predicted by the
MESOPUFF-II and CCADM A-l
VI
-------
FIGURES
Number Page
2-1 Application scenario II mesoscale region containing the
Clear Creek shale oil plant and two PSD class I areas-
Flat Tops and Maroon-Bells Snowmass Wilderness 6
2-2 CIT model-generated winds over Rocky Mountain domain at
50 m above ground
2-3 MELSAR model-generated winds over Rocky Mountain domain at
50 m above ground 12
2-4 ATMOS1 model-generated winds over Rocky Mountain domain at
50 m above ground 16
3-1 Comparison of trajectories starting at 1600 at plume
heights of 10 m, 300 m, and 1000 m 23
3-2 Comparison of trajectories starting at 2200 at plume
heights of 10 m, 300 m, and 1000 m 24
3-3 Comparison of trajectories starting at 0400 at plume
heights of 10 m, 300 m, and 1000 m 25
3-4 Comparison of trajectories starting at 1000 at plume
heights of 10 m, 300 m, and 1000 m 26
3-5 Comparison of horizontal plume dispersion rates for
stability class A 30
3-6 Comparison of horizontal plume dispersion rates for
stability class B 31
3-7 Comparison of horizontal plume dispersion rates for
stability class C 32
3-8 Comparison of horizontal plume dispersion rates for
stability class D 34
VII
-------
Number Page
3-9 Sensitivity of the MELSAR MacCready horizontal
dispersion rate to terrain roughness 35
3-10 Sensitivity of the MELSAR MacCready horizontal
dispersion rate to height above terrain 36
3-11 Comparison of horizontal plume dispersion rates
for stability class E 38
3-12 Comparison of horizontal plume dispersion rates
for stability class F 39
3-13 Comparison of vertical plume dispersion rates
for stability class A 41
3-14 Comparison of vertical plume dispersion rates
for stability class B 42
3-15 Comparison of vertical plume dispersion rates
for stability class C 43
3-16 Comparison of vertical plume dispersion rates
for stability class D 44
3-17 Sensitivity of the MELSAR MacCready vertical dispersion
rate to terrain roughness 45
3-18 Sensitivity of the MELSAR MacCready vertical dispersion
rate to height above terrain 46
3-19 Comparison of vertical plume dispersion rates for
stability class E 47
3-20 Comparison of vertical plume dispersion rates for
stability class F 48
3-21 Sensitivity of the MESOPUFF-II and RIVAD chemical
mechanisms to solar intensity 55
3-22 Sensitivity of the daytime MESOPUFF-II and RIVAD
chemical mechanisms to temperature 56
3-23 Sensitivity of the nighttime MESOPUFF-II and RIVAD
chemical mechanisms to temperature 57
v i i i
-------
Number Page
3-24 Sensitivity of the daytime MESOPUFF-II and RIVAO
chemical mechanisms to relative humidity 58
3-25 Sensitivity of the nighttime MESOPUFF-II and RIVAD
chemical mechanisms to relative humidity 59
3-26 Sensitivity of the daytime MESOPUFF-II and RIVAD
chemical mechanisms to ozone concentration 61
3-27 Sensitivity of the nighttime MESOPUFF-II and RIVAD
chemical mechanisms to ozone concentration.... 62
3-28 Sensitivity of the daytime MESOPUFF-II and RIVAD
chemical mechanisms to NOX concentration 64
3-29 Sensitivity of the nighttime MESOPUFF-II and RIVAO
chemical mechanisms to NOX concentration 65
3-30 Sensitivity of the daytime MESOPUFF-II and RIVAD
chemical mechanisms to S02 concentration 66
3-31 Sensitivity of the nighttime MESOPUFF-II and RIVAD
chemical mechanisms to S02 concentration.... 67
3-32 Comparison of MESOPUFF-II and CCADM predicted S02
dry deposition velocities for three land use classes 79
3-33 Comparison of MESOPUFF-II and CCADM predicted sulfate
dry deposition velocities for three land use classes 81
3-34 Comparison of MESOPUFF-II and CCADM predicted NOX
dry deposition velocities for three land use classes 82
3-35 Comparison of MESOPUFF-II and CCADM predicted nitric acid
dry deposition velocities for three land use classes 84
3-36 Comparison of MESOPUFF-II and CCADM predicted nitrates
dry deposition velocities for three land use classes 85
3-37 Sensitivity of the MESOPUFF-II and RIVAD wet scavenging rates
to precipitation rates for (a) S02 and (b) sulfate 90
3-38 Sensitivity of the MESOPUFF-II and RIVAD wet scavenging rates
to precipitation rates for (a) NOX and (b) nitric acid 91
-------
Number Page
4-1 Winds generated by the Diagnostic Wind Model for
simulation Al 104
4-2 Winds generated by the Diagnostic Wind Model for
simulation A2 107
4-3 Winds generated by the Diagnostic Wind Model for
simulation A3 110
4-4 Winds generated by the Diagnostic Wind Model for
simulation Bl 113
4-5 Winds generated by the Diagnostic Wind Model for
simulation B2 117
4-6 Winds generated by the Diagnostic Wind Model for
simulation 83 120
4-7 DWM-generated wind fields at 0500 on 18 September 1984 124
4-8 DWM-generated wind fields at 1400 on 18 September 1984 130
4-9 Scatterplot and statistics of predicted versus observed
wind speeds at the three supplemental soundings 137
4-10 Histograms of deviations of predicted wind direction
from observations at the three supplemental soundings 138
4-11 Locations of the SCCAB and Central Valley modeling regions 140
4-12 DWM generated and observed surface wind fields for the
SCCAB region at 0400 PDT on 23 September 1987 142
4-13 DWM-generated and observed surface wind fields for the
SCCAB region at 1200 PDT on 23 September 1987 143
4-14 Depiction of wind circulation air floww and boundary
heights in the California Central Valley generated by
the two-dimensional primitive equation 145
4-15 DWM-generated surface and upper-layer wind fields for
the California Central Valley at 0400 on 7 August 1984 146
4-16 DWM-generated surface and upper-layer wind fields for
the California Central Valley at 1200 on 7 August 1984 147
-------
TABLES
Number Page
3-1 Suranertlme S02 canopy resistances used 1n the MESOPUFF-II
as a function of land use type and stability class 74
3-2 S02 canopy resistance used in the CCADM 75
3-3 Canopy resistances used 1n the CCADM assumed for dry-
deposited gases relative to S02 surface resistance 76
4-1 Slope and Intercept of temperature lapse rate correction
by Julian Day 152
XI
-------
ACKNOWLEDGEMENTS
The evaluation and design of a new add deposition model for the Rocky
Mountains 1s the result of a team effort that Involved personnel from the
U.S. EPA and other federal and state agencies. Members of the Western
Add Deposition Task Force were an Integral part of this effort. We would
like to thank 1n particular Mr. Larry Svoboda of the U.S. EPA Region VIII
and Mr. Al Rlbleu of the Bureau of Land Management for their contribu-
tions. Finally, we would like to acknowledge Mr. Alan Huber, the EPA pro-
ject officer, whose guidance helped focus the goals of this project.
xn
-------
1 INTRODUCTION
1.1 BACKGROUND
Add 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 1s 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 in
southwestern Wyoming. Several planned energy-related projects, Including
natural gas sweetening plants and coal-fired power plants, may consider-
ably 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 Air 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 Air 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. Air-quality-related concerns range from near-source plume
blight to regional-scale acid deposition problems. By law, the Federal
Land Manager of Class I areas has a responsibility to protect air-quality-
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 air-quality-related values will occur. EPA
Region VIII contains some 40 Class I areas in the West, including two
Indian reservations. Similar designation is being considered for several
of the remaining 26 Indian reservations in the region. State and federal
agencies, Industries, and environmental groups in the West need accurate
data concerning western source-receptor relationships.
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 source emis-
sions and receptor Impacts 1n the West. To address add deposition
problems 1n the East, the EPA has launched a major effort to develop a
state-of-the-art regional acid deposition modelRADM (NCAR, 1985).
According to the current plan, this model will undergo an Intensive model
evaluation during the period 1988-1989. Realistically, evaluation,
adaptation, and application of this sophisticated model to the West will
probably not occur until 1990 or beyond. Until that time, a practical
modeling tool with which the federal and state agencies can assess air
quality Impacts in the West is needed.
A1r quality modeling in this region 1s especially difficult because of the
complex air flow patterns over the Rocky Mountains and the difficulty of
predicting acid 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 model
evaluation.
1.2 PURPOSE OF THIS REPORT
This report discusses the development and initial evaluation of a meso-
scale add deposition modeling system designed for the Rocky Mountain
region for the Rocky Mountain Acid Deposition Modeling Assessment Project
under the auspices of the U.S. EPA. The primary objective of the project
1s to assemble a mesoscale air quality model based primarily on models or
modules currently available for use by federal and state agencies in the
Rocky Mountain region. To develop criteria for model selection and
evaluation, the EPA formed an atmospheric processes subgroup of the
Western Atmospheric Deposition Task Force, referred to as WADTF/AP. This
group comprises representatives from the National Park Service, U.S.
Forest Service, EPA Region VIII, the National Oceanic and Atmospheric
Administration, and other federal, state, and private organizations. On
the basis of our review of the modeling needs identified by the WADTF/AP,
the specific requirements of the model for this project are as follows:
-------
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, trans-
formation, and deposition of pollutants over complex terrain in the
Rocky Mountain region using relatively sparse NWS upper-air sound-
Ings.
1.3 OVERVIEW OF THE NEW HYBRID ACID DEPOSITION/AIR QUALITY
MODELING SYSTEM FOR THE ROCKY MOUNTAINS
The mathematical modeling system for the Rocky Mountain region described
1n this report consists of several components or modules. These com-
ponents can be divided Into two main categories: those that describe
meteorological processes (a mesoscale meteorological model) and those that
describe pollutant dispersion, chemical transformation, and deposition (an
add deposition/air quality simulation model).
The components of the Rocky Mountain modeling system were taken from
existing mesoscale meteorological and add deposition models that were
selected previously (Morris and Kessler, 1987). The components of these
candidate models were evaluated to determine which are the most scientifi-
cally sound yet internally consistent within the overall framework of a
Rocky Mountain acid deposition modeling system. The most technically
rigorous yet consistent components of the candidate models were Integrated
together to form the new modeling system. In the development of this
modeling each of the components has been evaluated separately. When new
components were designed that deviate significantly from the candidate
models, such as the new diagnostic wind model (DWM), then a rigorous
evaluation of these new components 1s made.
1.4 REPORT ORGANIZATION
The candidate mesoscale meteorological and acid deposition models are
evaluated 1n Sections 2 and 3, respectively. Section 4 describes the
-------
design and Implementation of a mesoscale meteorological model for the
Rocky Mountain region. The meteorological model contains a new diagnostic
wind model (DWM), which was subjected to a rigorous performance evaluation
using four different complex terrain regions; its predictions are compared
with observations from the Rocky Mountains. Section 5 describes the
design of the new Lagrangian acid deposition/air quality model. Finally,
Section 6 summarizes the work to date on the development of the modeling
system.
-------
2 EVALUATION OF THE CANDIDATE
METEOROLOGICAL MODELS
Four diagnostic wind models were considered for use 1n the Rocky Mountain
modeling systemthe CIT wind model, PNL's MELSAR-MET, SAI's Complex-
Terrain Wind Model, and LANL's ATMOS1. In this section we expand on the
preliminary evaluation of the four candidate models presented by Morris
and Kessler 1n their review of the Rocky Mountain modeling system
(1987). In that report the candidate models are compared and their per-
formance 1s evaluated 1n application to an Idealized terrain. Here we
briefly summarize the results of applications of the candidate models to
the terrain with 10 km resolution in the Rocky Mountain region depicted
1n Figure 2-1, and two new regions containing complex terrain.
2.1 EVALUATION WITH AN IDEALIZED TERRAIN OBSTACLE
As an Initial test of the candidate models, the models were exercised
using a three-dimensional bell-shaped mountain of a scale typically found
1n the Rocky Mountains using an Initial uniform flow field. The results
for each of the models can be summarized as follows.
2.1.1 CIT Wind Model
The CIT model can treat the kinematic effects of terrain on the airflow;
however, 1t lacks a provision for Froude number flow adjustment and thus
cannot simulate blocking effects 1f they are not defined by the Input wind
data. If Input data (wind observations) are plentiful and representative,
the flexibility of the CIT 1nterpolat1ve scheme 1s desirable; however,
when Input data are sparse, the model cannot simulate blocking and deflec-
tion.
2.1.2 MELSAR-MET
The MELSAR model 1s designed to simulate the blocking and deflection of
air flows typically found 1n the Rocky Mountain region under weak synoptic
conditions. However, due to the model's unique interpolation scheme used
-------
700
UTM Easting Zone 12
750 B00
650
550
700
750 800
UTM Eestine Zone 12
850
901
30B
FIGURE 2-1. 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).
-------
to define the grldded wind fields, spurious results are produced near the
boundaries of the modeling domain. Since the MELSAR assumes Its Initial
grldded wind field 1s mass consistent without additional constrained
adjustments, 1t 1s the most Inexpensive of the candidate models. If the
details of the vertical velocity field are unimportant, MELSAR may be suf-
ficient to represent blocking and deflection in the horizontal wind field.
2.1.3 ATMOS1
The ATMOS1 model lacks a Froude number adjustment term to treat blocking
and deflection but can provide a gross simulation of blocking that 1s
defined through a region-wide stability dependent parameter aj2 as user
Input. The ATMOS1 does adjust the wind fields to produce reasonable
vertical velocities.
2.1.4 Complex-Terrain Wind Model
The CTWM alone of the candidate models is designed to generate wind fields
using only a domain-mean wind input. It 1s also the only model that
attempts to simulate thermally generated upslope and downslope flows in
addition to deflection and blocking effects. However, the CTWM 1s also
the only candidate model formulated in Cartesian coordinates. The use of
a Cartesian coordinate system for simulating airflows 1n complex terrain
1s undesirable for the following reasons:
Airflows tend to follow the terrain.
The lower boundary condition is difficult to parameterize in
Cartesian coordinates.
Increased vertical resolution near the surface 1s needed to resolve
complex terrain airflows. In Cartesian coordinates this results 1n a
prohibitive number of vertical layers.
Also the ability of the CTWM to utilize more than one wind observation
within the model domain is unclear.
2.1.5 Conclusions
The comparative simulations of the mesoscale meteorological models using a
hypothetical terrain obstacle cannot by themselves serve as a basis for
recommending one model over another. Each of the models contain some
desirable attributes that would be warranted in a meteorological model for
the Rocky Mountain region. Although the CTWM contains several unique fea-
tures, notably the lack of a requirement of extensive input data and the
-------
treatment of upslope/dowr.slope winds, the formulation of the model 1n a
Cartesian coordinates 1s a serious drawback.
2.2 EVALUATION WITH TERRAIN FROM THE ROCKY MOUNTAINS
In order to evaluate model performance over typical Rocky Mountain ter-
rain, the CIT wind model, MELSAR-MET, and ATMOS1 were exercised over the
topographic domain depicted 1n Figure 2-1, which corresponds to the first
proposed application scenario discussed 1n the review report (Morris and
Kessler, 1987). The grid spacing used 1n these simulations was 10 km,
which resulted 1n a 25 x 25 array of grid cells for this region. The CTWM
was not Included 1n this series of experiments because of the problems
with the coordinate transformation demonstrated in the application to an
Idealized terrain obstacle.
In this series of experiments an initially uniform flow of 2 m/s from the
southwest (225°) was specified. Winds were generated at heights of 50,
200, 500, 1000, and 2000 m above ground on a 26 x 26 horizontal grid with
grid spacing of 10 km.
2.2.1 CIT Wind Model
As 1n the set of experiments using Idealized terrain, the CIT divergence
reduction procedure was exercised until maximum three-dimensional diver-
gence was reduced to 10~° s~*. Figure 2-2 depicts CIT model wind fields
at 50, 200, and 500 m above the ground for the Rocky Mountain domain. The
Initial wind field 1s minimally perturbed by the terrain. Note that the
characteristic terrain slopes 1n the Rocky Mountain domain are substanti-
ally smaller than those of the Idealized bell-shaped mountain; thus the
perturbations 1n this experiment should have smaller magnitude than those
1n the previous experiment.
2.2.2 MELSAR-MET
As 1n the Idealized terrain experiment, the atmosphere is assumed to be
uniformly isothermal. As recommended by Allwine and Whiteman (1985) the
spacing of the "Froude grid" is 50 km.
Figure 2-3 depicts the MELSAR wind fields at 50, 200, and 500 m above
ground. The directional variability exhibited by these fields has a hori-
zontal scale considerably larger than the characteristic terrain scales;
the Individual resolved terrain features do not seem to deflect the air-
flow. It 1s probable that the standard deviation of terrain height within
the Froude grid cells provides relatively low estimates of "obstacle
height" 1n this case. Also, several of the major terrain features are
aligned along the assumed domain-mean wind direction, minimizing deflec-
tion.
8
-------
NORTH
650
700
750
800
850
4500
4450
- 4400
4350
430C
700
750 800
SOUTH
850
4300
WNDMOD WIND VECTORS AT LEVEL
I""!""!""!
0 5 10 15
WIND SPEED (M/S)
- 1
FIGURE 2-2a. CIT mode]-generated winds over Rocky Mountain domain at
50 m above ground. Scaling of plotted winds is given at lower left.
Topography is contoured in meters. Horizontal grid spacing is 10 km.
-------
NORTH
650
700
750
BOO
850
430C
700
750
800
850
SOUTH
4500
4450
4400
4350
4300
WNDMOD WIND VECTORS AT LEVEL
in 11 ii ii ii ii ii I
0 5 10 15
WIND SPEED (M/S)
- 2
FIGURE 2-2b. At 200 m.
10
-------
NORTH
650
700
750
800
B50
4500-
4450 '-
440
4350-
430C
' ' / '
''," V, *,',""
4500
4450
4400
4350
700
750 800
SOUTH
850
4300
WNDMOD WIND VECTORS AT LEVEL
i n n ii ii iiii ii i
0 5 10 15
WIND SPEED (M/S)
FIGURE 2-2c. At 500 m.
11
-------
650
700
NORTH
750 800
850
450C2-
4500
445C t
4450
440
4350
430,
UJ
- 4400
4350
750 800
SOUTH
850
4300
MELSAR WIND VECTORS AT LEVEL - 1
I""!""!""!
0 5 10 15
WIND SPEED (M/S)
FIGURE 2-3a. MELSAR model-generated winds over Rocky Mountain
domain at 50 m above ground. Scaling of plotted winds is given
at lower left. Topography is contoured in meters. Horizontal
grid spacing is 10 km.
12
-------
NORTH
650
700
750
800
850
4450
500
- 4400
4350
430C
700
750 800
SOUTH
850
4300
MELSAR WIND VECTORS AT LEVEL
05 10 15
WIND SPEED (M/S)
FIGURE 2-3b. At 200 m.
13
-------
NORTH
650
700
750
800
850
Cfc 4450
i i i | i i /1 i
f 'P\/ S S S J /*>
f Vx^V / **-£ k S .
- 4500
« 4400
4350
700
750 800
SOUTH
850
4300
MELSAR WIND VECTORS AT LEVEL
I""!""!""!
0 5 10 15
WIND SPEED (M/S)
- 3
FIGURE 2-3c. At 500 m.
14
-------
Exaggerated flow appears along the northern and eastern boundaries of the
domain. This behavior may indicate a problem with the the MELSAR poly-
nomial interpolation scheme.
2.2.3 ATMOS1
ATMOS1 was exercised with the parameter a12 set equal to 0.02, and the
model run was halted after 20 iterations of the solution procedure.
Figure 2-4 depicts the ATMOS1 wind fields at 50, 200, and 500 m above
ground. Unlike the corresponding MELSAR-MET wind field, the ATMOS1 wind
field at 50 m displays a great deal of variability at horizontal scales
equal to or smaller than the characteristic terrain scales. Note that the
mass-consistent adjustment in ATMOS1 does not contain a smoother; thus
major differences between winds at adjacent grid points are preserved. In
general, maximum ATMOS1 wind speeds occur above the tops of major terrain
obstacles, consistent with potential flow theory.
2.2.4 Remarks
The comparative simulations discussed in this section Indicate that no one
of the candidate models is significantly better than another. A compre-
hensive model evaluation would involve tests of the ability of the models
to simulate actual observations 1n complex terrain. Another approach to
model evaluation is comparison of model results with analytic theory, as
discussed by Pielke (1984). As noted previously, Ross and Smith (1986)
demonstrate that ATMOS1 can reproduce analytic solutions for unstratified
potential flow over idealized obstacles. However, based on the analyses
of mountain-generated airflows by Smith (1979) and others, it 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 non-
hydrostatic model to reproduce analytic solutions for each of these types
of mountain waves. Additionally, Clark and Gall (1982) have utilized a
nonhydrostatic 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 is
capable of simulating either type of mountain wave disturbance unless the
disturbance is fully accounted for by input wind data. There may be cir-
cumstances under which mountain waves play a significant role in horizon-
tal and vertical transport of pollutants; primitive-equation numerical
simulations would be necessary to delineate this role.
15
-------
650
700
NORTH
750 800
850
4500-
4450-
440
435C
430;
f /" /* i' »i x1 /*' /' /' /i /' f /' 7'tu /'
/ «* A
ft* Sit t ///xi» x)*
«*
ft* Sit t ///xi»
'/ '^ ' ' ' ^ x -
' 'j » * + r r * r r ? ta '
/ ^
f / °9t »' » » x * f f f r\r
r 4500
4450
,- 4400
4350
700
750 800
SOUTH
850
4300
ATMOS1 WIND VECTORS AT LEVEL
i ii 1111111111111
0 5 10 15
WIND SPEED (M/S)
- 5
FIGURE 2-4a. ATMOS1 model-generated winds over Rocky Mountain
domain at 50 m above ground. Scaling of plotted winds is given
at lower left. Topography is contoured in meters. Horizontal
grid spacing is 10 km.
16
-------
650
700
NORTH
750 800
850
450C -
4450 -
LJ
440
4350=^
430£
r 4500
4450
r 4400
4350
700
750 800
SOUTH
850
4300
ATMOS1 WIND VECTORS AT LEVEL
0 5 10 15
WIND SPEED (M/S)
FIGURE 2-4b. At 200 m.
17
-------
NORTH
650
700
750
800
850
4500 -
4450-
440
43 5C ^
430C
? 4500
4450
700
750 800
SOUTH
850
-i 4400
4350
4300
ATMOS1 WIND VECTORS AT LEVEL
I""!""!""!
05 10 15
WIND SPEED (M/S)
FIGURE 2-4c. At 500 m.
18
-------
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 a-^- Ross and Smith (1986)
propose a scheme for calculation of a space-variable aj2 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 parameterize block-
Ing effects, the treatment produces somewhat questionable results.
The CPU time on a Prime 750 minicomputer required by each model for the
Idealized bell-shaped mountain simulations 1s as follows:
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
to be mass-consistent without additional adjustments. If the details of
the model's vertical velocity field are unimportant, MELSAR may be
sufficient to represent blocking and deflection of the horizontal wind
components by terrain, although the MELSAR obstacle height computation may
be open to question.
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 a12)t may be a better choice. The CIT
wind model 1s less desirable when Input data are sparse because 1t lacks
the ability to simulate blocking and deflection; however, if 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 that explicitly attempts to treat thermally generated up-
slope and downslope flows. The CTWM requires specification of several
arbitrary coefficients with little guidance. Accurate specification of
the coefficients would probably require a specific "tuning" of the
19
-------
coefficients for the Rocky Mountain region, based on available observed
wind data. Also, the CTWM does not have the ability to utilize randomly
spaced observations.
The CTWM 1s formulated in Cartesian vertical coordinates. We believe that
terrain-following vertical coordinates are strongly desirable for a wind
model 1n complex terrain. The transformation of the CTWM from Cartesian
to terrain-following coordinates is ambiguous when the slope flow treat-
ment is Included. Thus, as currently formulated in a Cartesian coordinate
system, the CTWM would not be an appropriate wind model to simulate air
flows over the complex terrain of the Rocky Mountains.
20
-------
EVALUATION OF THE CANDIDATE ACID DEPOSITION MODELS
Four add depos1t1on/a1r quality models were chosen for possible incorpor-
ation Into an acid deposition modeling system for application to the Rocky
Mountain region. These four models are the SAI/CCADM, a Lagrangian box
model; the ERT/MESOPUFF-II and PNL/MELSAR-POLUT. both Lagrangian puff
models; and the SAI/RIVAD, a Lagrangian plume segment model. These four
models contain different modeling approaches and parameterizatlons of the
processes that lead to add deposition and pollutant transport in complex
terrain. These four models were not chosen with the idea that any one of
the models would serve as the final acid deposition model, but that each
of the models contains modules and parameterizations that can be incor-
porated Into the final add deposition model.
The review of the existing models (Morris and Kessler, 1987) presented a
preliminary evaluation of the candidate acid deposition models' treatment
of the processes of transport, dispersion, chemical transformation, and
dry and wet deposition. Here we present a more detailed evaluation of the
candidate acid models' treatment of these processes. Based on this
evaluation, the most appropriate modules were chosen for Incorporation
Into the new acid deposition model for the Rocky Mountains, which is
described 1n Section 4.
3.1 TRANSPORT
The transport winds for the acid deposition model will be defined from the
multilayer terrain-following wind fields generated by the new diagnostic
wind model described 1n Section 4. Three of the candidate add deposition
modelS--MELSAR-POLUT, MESOPUFF-II. and RIVADdefine the plume trajectory
by using the wind at the plume centerline for advection. The fourth
candidate model, the CCADM, requires user input of the Lagrangian box
trajectory. As noted by Morris and Kessler (1987), use of the plume
centerline wind vector to advect the entire puff, whose vertical extent
may be over 1000 m, may not simulate the correct transport of the plume
mass, especially under conditions of decoupled flow regimes as occurs in
complex terrain. In this section we briefly examine the sensitivity of
air parcel trajectories 1n complex terrain to the height of the air parcel
above the ground.
21
-------
Figures 3-1 through 3-4 depict air parcel trajectories at plume heights of
10, 300 and 1,000 m for release times at 0400, 1000, 1600, and 2200. Wind
fields were generated by the diagnostic wind model for the September 17-
18, 1984 simulations in the Rocky Mountains. These simulations used sup-
plemental data collected for the ASCOT Brush Creek drainage flow experi-
ments (see Section 4.1.2.3). The stagnant flow conditions present during
these simulations should produce the maximum differences between trajec-
tories at different heights above ground since the complex-terrain wind
fields are not driven by synoptic forcing.
As seen 1n Figures 3-1 to 3-4, the surface trajectory (10 m) deviates
greatly from the two elevated (300 and 1,000 m) trajectories. In fact, as
can be seen by the symbols on the trajectories spaced six hours apart, the
maximum wind speeds occur 1n the surface trajectories when the air parcel
starts at the top of the 2400 m ridge during the two six-hour nighttime
trajectory segments (2200-0400 and 0400-1000). Thus 1t appears that
drainage winds dominate the surface trajectory paths on this day. This is
confirmed by the upper-air trajectories, which tend to be disorganized due
to the stagnant conditions.
This preliminary trajectory analysis illustrates the differences in trans-
port of air parcels at different heights in complex terrain:
The difference 1n transport characteristics between surface and ele-
vated releases confirms the need for multilevel wind fields in com-
plex terrain. The use of a surface wind speed with the power law
relationship with height cannot accurately characterize transport in
complex terrain.
When an emission release becomes well mixed, the advection of the air
parcel near the surface should Ideally be handled differently than
parcels aloft. Currently there are no Lagrangian models that treat
the vertical splitting of puffs. The acid deposition model for the
Rocky Mountains described 1n Section 4 has been formulated so that
this vertical splitting can be easily incorporated at a future time.
3.2 DISPERSION
The plume segment model, RIVAD, and the two Gaussian puff models, MESO-
PUFF-II and POLUT, all represent dispersion by expanding the plume seg-
ments or puffs in terms of the puff dispersion parameters ay and az. In
the CCADM there are two options for simulating diffusion. Either the
horizontal and vertical dlffuslvities are specified at edges of the
Lagrangian box, or the user specifies the size of the box as it moves
downwind. Either method requires that the user specify the dispersion
22
-------
HGRTH
650
->r.r\
750
eoo
45GC
445C
440!
4350
430;
700
750
800
4500
^350
350
SOUTH
FIGURE 3-1. Comparison of trajectories starting at 1600 at plume
heights of 10 m (D), 300 m (O), and 1000 m (A). Symbols on
trajectories are spaced at six-hour intervals.
23
-------
700
NORTH
750 SOO
450;
GO
UJ
44GC
435C
430
fc50
^450
700
750 £00
SOU'ri
£50
FIGURE 3-2. Comparison of trajectories starting at 2200 at plume
heights of 10 m (n), 300 m (O), and 1000 m (A). Symbols on
trajectories are spaced at six-hour intervals.
24
-------
650
700
800
S50
445C
oo
LJ
4^00
4350
430C*
^450
£400
^550
'650
700
750 300
SOUTH,
850
FIGURE 3-3. Comparison of trajectories starting at 0400 at plume
heights of 10 m (D), 300 m (o), and 1000 m (A). Symbols on
trajectories are spaced at six-hour intervals.
25
-------
NORTH
650
700
750
SCO
450C
445C
435C
430,
£50
700
750 BOO
SOUTH
4500
4450
4400
4350
850
4300
FIGURE 3-4. Comparison of trajectories starting at 1000 at plume
heights of 10 m (D), 300 m (O), and 1000 m (A). Symbols on
trajectories are spaced at six-hour intervals.
26
-------
rate; thus the CCADM methods for dispersion Involves excessive user Inter-
action. In the following sections we briefly review how the horizontal
and vertical dispersion parameters are calculated 1n the MESOPUFF-II,
RIVAD. and POLUT models and then evaluate these dispersion algorithms by
1ntercompar1ng the puff dispersion parameters predicted by the three
algorithms and comparing these predictions with the dispersion curves
estimates made by Pasquill, Gifford, and Turner (Turner, 1970).
3.2.1 Description of the Dispersion Algorithms
3.2.1.1 MESOPUFF-II
The MESOPUFF-II calculates oy and oz for distances out to 100 km using
formulas fitted to 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 dif-
ferential form:
oy(s + SA) = oy(s) +
do.
(3-D
S
AS
+ r
so that the puffs always grow with time. The Integral formulas for oy and
oz for travel distances less than 100 km are as follows:
oy(s) = a s°-9
oz(s) = c sd (3-2)
where a, c, and d are stability-dependent constants (Benkley and Bass,
1979b). For distances greater than 100 km, dispersion 1s based on time,
t (seconds), Instead of downwind distance, using the following formulas
(Heffter, 1965):
o (t + At) = o (t) + 0.5 At
J" J"
oz(t + At) = oz(t) + d-At//t (3-3)
where d 1s a stability-dependent parameter. The vertical extent of the
plume defined by oz is limited to the mixing depth.
3.2.1.2 RIVAD
Horizontal dispersion 1n the RIVAD accounts for the effect of vertical
wind shear using an approach suggested by Randerson (1972). On the basis
27
-------
of field measurements, Randerson found that diffus1v1t1es Increase rapidly
during a transition phase, which typically lasts about 10 hours. During
this phase
1.2
where oyiQ\ 1s oy at t = tQ, and t 1s the transport time.
This wind-shear-induced dispersion 1s several orders of magnitude greater
than eddy diffuslvHy. Beyond about 10 hours, dlffusivlty becomes con-
stant and we have the following limit on horizontal dispersion:
oy < (2 KHo> t)1/2 , (3-5)
where KH(D = 7 x 108 cm2/s.
Vertical dispersion 1n the RIVAD 1s handled 1n a somewhat similar way in
that dispersion 1s keyed to transport time:
where k 1s determined from Pasquill-Glfford 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).
3.2.1.3 MELSAR-POLUT
The horizontal dispersion 1n MELSAR-POLUT assumes that the square of the
total horizontal diffusion, a , is the sum of the squares of three com-
ponents: an Initial buoyancy-induced dispersion (Ay), diffusion resulting
from atmospheric turbulence (BJ, and diffusion resulting from horizontal
wind shear (Cy):
/2 2 2\1/2
oy - A2 + B2 + C* . (3-7)
28
-------
Similarly, the vertical diffusion coefficient is
(
2 A1/2
°z - K + B2 (3-8)
where A2 is the initial buoyancy-induced dispersion and B2 is the vertical
diffusion due to atmospheric turbulence. The formulas used in calculating
the diffusion coefficient in MELSAR-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-Obukhov
length, and friction velocity. These formulas are presented in Appendix A
of the report by Morris and Kessler (1987).
The user has two options for the calculation of horizontal and vertical-
dispersion due to atmospheric turbulence under neutral and stable condi-
tions in the POLUT model. The first option is a scheme proposed by Irwin
(1979); the second option uses an empirical relationship developed by Mac-
Cready, Baboul, and Lissman (1974), which accounts for effects of terrain
roughness on atmospheric turbulence. These parameterizations are
described in detail by Morris and Kessler (1987) and by Allwine and White-
man (1985).
3.2.2 Evaluation of the Dispersion Algorithms
The dispersion algorithms of the four candidate models are evaluated below
by comparing the calculated horizontal and vertical dispersion values with
each other and with the estimates of Pasquill, Gifford and Turner (PGT) at
different downwind distances and stabilities.
3.2.2.1 Horizontal Dispersion (oy)
Unstable Conditions
The growth of oy as a function of downwind distance as calculated by the
four models and the Pasquill, Gifford, and Turner (PGT) estimates for the
A, B, and C stability classes (unstable) are given in Figures 3-5, 3-6,
and 3-7. Note that the two methods in the MELSAR-POLUT model, those of
Irwin and MacCready, produce identical results for unstable conditions.
The MELSAR-POLUT and MESOPUFF-II horizontal dispersion algorithm produce
results very similiar to the PGT dispersion estimates for downwind dis-
tance under 10 km. The MELSAR-POLUT algorithm for horizontal dispersion
29
-------
Stability Class A
o
6
o
o
1-H
If.
i
,
^
j
^
^
i
i
,
i
^
t
i
/
/
x^l
/
J
/
^
^
^
/
f
/*
>
i
^
X^
'^
1
1
;
i
i
,
t
i
;
i
i
i
i
i
1
1
i
100.
1000. 10000.
Downwind Distance (meters)
FIGURE 3-5. Comparison of horizontal
plume dispersion rates for stability
class A.
100000.
PGT
MELSAR using IRWIN scheme
MELSAR using MacCREADY scheme
RIVAD
30
MESOPUFF
-------
o
o
o
o
o
o
o
!2
X
I
-------
Stability Class C
o
o
o
o
o
In
t-
O o
~v °
f~ ""*
£
1
X
/
!
i
i
/ "X
/ /
/ X
r
.
100.
1000. 10000.
Downwind Distance (meters)
FIGURE 3-7. Comparison of horizontal
plume dispersion rates for stability
class C.
32
100000.
PGT
MELSAR using IRWIN scheme
MELSAR using MacCREADY scheme
RIVAD
ME50PUFF
-------
due to atmospheric turbulence treats the near-field versus far-field dis-
persion effects by taking the maximum calculated dispersion rate of dif-
ferent formulas proposed by Draxler and Gifford. This produces a kink 1n
the oy that appears at approximately 10 km downwind for the unstable
cases. The MESOPUFF-II OY values match the PGT estimates better at
greater downwind distances until the MESOPUFF-II switches to the far-field
dispersion algorithms at 100 km, at which time the MESOPUFF-II oy curves
deviate from the PGT estimates. Since the near-field MESOPUFF-II disper-
sion algorithms are designed to match the PGT curves, 1t is not suprising
that they do. However, as noted by Gifford (1982), if typical short-range
diffusion coefficients are extrapolated to large downwind distances, the
results will fall short of both observed and theoretical values by amounts
ranging up to nearly an order of magnitude.
The RIVAD model calculates the largest horizontal dispersion parameters
for unstable conditions. At a distance of 10 km downwind the RIVAD hori-
zontal plume extent 1s approximately four times that of the other models
and the PGT estimate. This increased diffusion in the RIVAD is most
probably due to its parameterization of diffusion at the regional-scale,
which accounts for the effects of wind shear, while the MESOPUFF-II and
MELSAR-POLUT contain separate near-field and far-field algorithms.
Neutral Conditions
The horizontal dispersion parameters at different downwind distances under
neutral stability for the different methods are given in Figure 3-8. The
PGT, MESOPUFF-II, and MELSAR-POLUT, all of which use Irwin's scheme, pro-
duce similiar dispersion curves for neutral conditions. The MELSAR-POLUT
method, using MacCready's scheme, produces slightly higher horizontal dif-
fusion, while the RIVAD model produces the highest horizontal dispersion
parameters for neutral stability.
The MacCready algorithm in the MELSAR-POLUT model is the only algorithm 1n
which variations in complex terrain are used. As such, 1t is sensitive to
the terrain roughness and the height of the plume above ground. For the
dispersion curves shown in Figure 3-8, a terrain roughness value of 300 m
was specified, and the height above ground was 10 m. The terrain rough-
ness value of 300 m was the average terrain roughness for the mesoscale
region in the Rocky Mountain region depicted in Figure 2-1. The terrain
roughness for that region varied from 40 to 1000 m.
The sensitivity of the MacCready scheme to the prescription of terrain
roughness and height above ground for neutral conditions is shown in
Figures 3-9 and 3-10. As can be seen 1n Figure 3-9, the MacCready scheme
is very sensitive to the terrain roughness, where an increase by a factor
33
-------
Stability Class D
O ""
6
<^
o
6
o
«' i
o
1C
'/fi
~f
/
0.
u
x
/
/
,''
/
/
*
.
s
{.
1
'
/
10
'
'/
,r /'
/'
/
00.
/
/
/
'
/v
S
' /'
,.
f* /
i
i
/*
^
1
/
/
V
f
i
-------
Stability Class D
o
o
o
o
>,
I
w
FIGURE 3-9. Sensitivity of the MELSAR
MacCready horizontal dispersion rate
to terrain roughness (R).
1C
'/ .
/,
/ /
/ /.'
/
i
i
/
' /
1
1
t
\ i
t ;
' . "
.r i
' j .
\
R"
/
,
' ffi'
f
*
'
!
i
\
1
1
I
1
i !
j
i
;>
r
/
/
j
i
1 ,'' ,''
y ^*
' / s
/ ,'
' /
f\'
t
'
y
''
s
f
'
/
f*
/
,
,
1 1
.<' .'
\,-'
^
.f
/
:\ /"
/ s
*
f i
i i
i
1
1
|
i
1
f
1
/
/ y^
/ /
/
// i
/ / 1
/ / /
j/
1 ..'
'/"
b
. ' / ^
.' / \ ^^
' / s i
/ S '
S ' '
/
' : '
\ '
' i
i
f
-
\
'
'
X
1
//
/ /
y /
7
.>
/
i
\ \
\
i
i
\
i
i
' . ,
!
1
i
1
1 1 If
1
)0. 1000. 10000.
Downwind Distance (meters)
i
i
;
i
i
i
i
i
!
\
1 ;
100000.
PC-T
MacCREADY with R = 10 m
MacCREADY with R = ICO m
MacCREADY with R = SCO m
35
- MacCREADY with R = 1C DO
m
-------
Stability Class D
0
o
o
6
o
o
^H
w
(-
/
/"
/
s
ft
'//
4
V
/
/
^/J
^
/
/
^
/
V
^
/
^f
r*.i'
X
1 1
/
/
/
1
100.
1000. 10000.
Downwind Distance (meters)
FIGURE 3-10. Sensitivity of the MELSAR
MacCready horizontal dispersion rate
to height above terrain (H).
100000.
PC-T
MacCREADY with H = 13 m
MacCREADY with H = 5C m
MacCREADY with H = ICO m
MacCREADY with K = 530 m
36
-------
of 10 1n the terrain roughness results in an increase by a factor of 2 in
the horizontal plume extent. A terrain roughness value of between 10 and
100 m would give the best match with the PGT dispersion estimates; such
values are consistent with the values used in tthe experiments that led to
the development of the PGT curves. The MacCready scheme 1s less sensitive
to the height above ground, as shown in Figure 3-10; a terrain roughness
value of 300 m was used in the sensitivity tests. The behavior of the
MacCready algorithm to variations in terrain roughness and plume height is
as expected; the more complex terrain results in enchanced dispersion, and
the influence of the terrain on the dispersion is less as the plume height
above the terrain increases.
Stable Conditions
The horizontal dispersion parameters for stable conditions (classes E and
F) are given in Figures 3-11 and 3-12. These figures are similiar to
those produced for neutral conditions except that the MacCready scheme in
MELSAR-POLUT produces the largest horizontal dispersion parameters. This
1s not suprlsing since complex terrain will enhance dispersion and the
MacCready scheme 1s the only algorithm that takes into account this
enhancement. The effect 1s Increased in these figures because the terrain
roughness chosen for these experiments, 300 m, was from a region of very
complex terrain in the Rocky Mountains. Use of a lower value will produce
dispersion results closer to those produced by the other algorithms (see
Figure 3-9).
3.2.3.2 Vertical Dispersion (oy)
All of the models assume that the vertical expansion of the plume segments
or puffs is limited to the mixing height when the plume height lies below
the mixing depth. Since over long distances the plume will eventually
become uniformly mixed within the mixed layer, the characterization of
vertical dispersion 1s not as Important as that of horizontal disper-
sion. Rather, the correct calculation of the mixing depth and plume
height 1s required. When the plume centerline is above the mixing height,
the upward expansion of the plumes is at a rate for stable conditions
regardless of the stability within the boundary layer.
Unstable Conditions
The growth of the vertical dispersion parameter as a function of downwind
distance for A, B, and C stabilities is shown in Figures 3-13, 3-14, and
37
-------
o
o
o
o
o
o
o
X
I
CO
100.
Stability Class E
syu
1000. 10000.
Downwind Distance (meters)
FIGURE 3-11. Comparison of horizontal
plume dispersion rates for stability
class E.
'/
I
>7L\
/
100000.
PGT
MELSAR using IRWIN scheme
MELSAR using MacCREADY scheme
P.IVAD
ME50PUFF
38
-------
o
o
o
o
o
o
o
O
'
i
«
E
tjn
i/3
100.
Stability Class F
^%
IT
I
X
/
/
y
^
11
1000. 10000.
Downwind Distance (meters)
111 11
i
100000.
FIGURE 3-12. Comparison of horizontal
plume dispersion rates for stability
class F.
PGT
MELSAR using IRWIN scheme
MELSAR using MacCREADY scheme
RIVAD
- MESOPLTF
39
-------
3-15. As for the horizontal dispersion parameters under unstable condi-
tions, the RIVAD algorithms calculate the fastest vertical expansion, and
the MESOPUFF-II matches the PGT dispersion estimates well. On the other
hand, for class A stability, the MELSAR-POLUT algorithm (I.e., the Irwin
method, since the MacCready method is used only for neutral and unstable
conditions) produces the lowest vertical expansion rates. As the atmo-
sphere becomes less unstable, the ay values produced by the MELSAR-POLUT
algorithm tend toward those produced by the other parameterlzatlons. As
seen 1n Figure 3-15 for C stability, all of the algorithms show good
agreement with each other and with the PGT estimates. As noted above,
under unstable conditions the plume will eventually become well mixed in
the mixed layer.
Neutral Conditions
Under neutral conditions the MacCready scheme (1n MELSAR-POLUT) produces
the largest vertical dispersion rates, followed by the Irwin scheme (also
1n MELSAR-POLUT), the RIVAD, and the MESOPUFF-II. The MESOPUFF-II verti-
cal dispersion curves again match the PGT dispersion estimates (Figure 3-
16). As for the horizontal dispersion parameters calculated by the Mac-
Cready scheme, the vertical dispersion rates are very sensitive to the
specification of the terrain roughness (Figure 3-17) and a little sensi-
tive to the plume height above ground (Figure 3-18). Figure 3-17 shows
that the large vertical diffusion rates produced by the MacCready scheme
under neutral conditions (shown 1n Figure 3-16) are due to the terrain
roughness value of 300 m used in these experiments.
Stable Conditions
Due to the terrain roughness value used, the MacCready scheme in MELSAR-
POLUT produces the largest vertical diffusion rate for stable conditions
(Figure 3-19 and 3-20). The other methods all produce very low rates,
with oz 1n the range of 100 to 200 m under F stability at a downwind dis-
tance of 100 km. Even under E stability, at a downwind distance of 100
km, the oz produced by all of the algorithms except the MacCready scheme
oz are 1n the range of 200 to 400 m.
3.3 CHEMICAL TRANSFORMATION
Each of the four candidate acid deposition models contains different
methods for treating the chemical transformation of S02 to sulfates and
NOX to nitrates and nitric acid. The MELSAR-POLUT model does not treat
chemical transformation; the MESOPUFF-II uses an empirical fit to chemical
40
-------
Stability Class A
o
o
*
6
<->
o -
w
"o ° -
E*"1 _
i
a
E_
00
o
~"
/
{
t
1 J
]''/:
^
/
/
j
1
y
^
/
. /
'/'
/
r /
} r
)
)
1
.'/
/
/
/
/
/
;
/'
/ /
./
//
if
.-/
}'
V
,.''
1
^
/
,XX
,
.'
_,'
.'"
,X
r^
1
'.
I
1
j
I
^
^X'
^X"
X
X
i
^
i
i
i
1
i
,^
I
100.
1000. 10000.
Downwind Distance (meters)
100000.
FIGURE 3-13. Comparison of vertical
plume dispersion rates for stability
class A.
PGT
~ MELSAR using IRWIN scheme
MELSAR using MacCREADV scheme
- RIVAD
ME50FUFF
41
-------
Stability Class B
o
o
o
~*
o
<->
o
7)
C O
"S °
£""*
1
<8
o
1C
/ S
y
!
/
/
/
/ ^/
,-y
i
>0.
/
>
f
/
/
S
/
/
/
r j
1
1
/
/
V ,
*\
i
000.
/
/
/
/
X
/
/
/
jf
x
/
1
/
^
X
/r
^ !
^
X
i
i
'
L/
i
/
i
X
;
'
t
I
, f
IOC
*
/
\/
/
/
/
/
t
/
f
/
-------
o
o
Ol O
N
I
(0
E
t*
55
100.
Stability Class C
M
E3
~ZL
1000. 10000.
Downwind Distance (meters)
FIGURE 3-15. Comparison of vertical
plume dispersion rates for stability
class C.
PGT
11
i
100000.
MELSAR using IRWIN scheme
MELSAR using MacCREADY scheme
RPv'AD
ME50PUFF
43
-------
Stability Class D
n
fc.
o
«
o
~*
6
i->
o
6
o
u *
<->
1C
f'
sy /
y X"
//
//
X'x"
,x
/
x
1
,'
^
>
f
]
1
1
I
1,
,-f
'j"
v
\
0. 1
i
i
s
, '
s
s
^x
,/ x^
X'
1
X
-^
x
x^
,-
X
X-
/
^x
x
x
i
.-I1'
,-'f
X
^
.'
^
/
X^X"
X^ 1
\
\
1
.^
^
>
x
1
x
^x
sr
x
x
>
_jx
"X
S"
X
X
x
/
x
x
X
xc^
X
X
^
x
x
i
x
r"^
|
1
1
1
i
I
1
III 1
000. 10000. 10
x
*
_,
^S
^<
Jx' ^^J
^S^
^
0000
Downwind Distance (meters)
FIGURE 3-16. Comparison of vertical
plume dispersion rates for stability
class D.
PGT
MELSAR using IRWIN scheme
MELSAR using MacCREADY scheme
RIVAD
ME50PUFF
44
-------
o
o
o
o
o
o
o
CJ O
« £
I
£
K
100.
Stability Class D
1
X
X
-V-
X
X
X
x
X
x
X
X'
I T II I I I I
1000. 10000.
Downwind Distance (meters)
iIIITT
i
100000.
FIGURE 3-17. Sensitivity of the
MELSAR MacCready vertical dispersion
rate to terrain roughness (R).
PGT
MacCREADY with R
MacCREADY with R
MacCREADY with R
MacCREADY with R
1C m
ICO m
SCO m
1COO m
45
-------
Stability Class D
100.
! j
1000. 10000.
Downwind Distance (meters;
100000.
FIGURE 3-18. Sensitivity of the
MELSAR MacCready vertical dispersion
rate to height above terrain (H).
PC-T
MacCREADY with H = 10 m
MacCREADY with H = 5C m
MacCREADY with H = 130 m
MacCREADY with H = 533 m
46
-------
t-
0) O
N
I
o
M
53
LZ_
100.
Stability Class E
-Of
I
1000. 10000.
Downwind Distance (meters)
100000
FIGURE 3-19. Comparison of vertical
plume dispersion rates for stability
class E.
PGT
MELSAR using IRWIN scheme
MELSAR using MacCREADY scheme
RIVAD
- MESOPUFF
47
-------
o
o
o
o
o
in
u
0 o
N
I
w
100.
Stability Class F
XI
1000. 10000.
Downwind Distance (meters)
i r
100000
FIGURE 3-20. Comparison of vertical
plume dispersion rates for stability
class F.
PGT
MELSAR using 1P.WIN scheme
MELSAR using MacCREADY scheme
RIVAD
ME30PUFF
48
-------
box model simulations; the RIVAD uses a highly condensed chemical
mechanism that 1s an extension of the mechanism used 1n the PLUVUE-II
model; and the CCADM uses comprehensive gas-phase and aqueous phase non-
linear chemical kinetic mechanisms. The choice of one or more of these
chemical modules for use in a Lagrangian model for the Rocky Mountain
region 1s based on the following criteria:
At a minimum the mechanism must treat the oxidation of both S02 and
NOX.
The chemical mechanism must be consistent with the formulation of a
Lagrangian model. Any nonl1nearit1es within the chemical mechansim
must be based on conditions within the puff so that the puff super-
position principle will not be violated.
The chemical mechanism must be appropriate for the Rocky Mountain
region. A mechanism that 1s tuned for an urban atmosphere would
greatly exaggerate the oxidation rates in the Rocky Mountain region.
As with other components of the modeling system, the chemical
mechanism must be computationally efficient so that long-term
averages can be readily calculated.
The chemical reactions that lead to the formation of sulfates, nitrates,
and nitric add are briefly discussed below along with the chemical
mechanisms used by the candidate models.
3.3.1 Review of the Chemistry of Add Deposition
3.3.1.1 Sulfate Chemistry
The oxidation of S02 to sulfate 1n the atmosphere involves both gas- and
liquid-phase reactions (NRC, 1983). The most Important gas-phase
(homogeneous) reaction for the formation of sulfates 1s the oxidation of
S02 by the hydroxyl radical (OH-), which 1s formed mainly through ozone
photolysis:
S02 + OH- -* HSO-3 ... -* H2S04
There are several pathways for the formation of sulfates in the liquid
phase, Including the direct oxidation of S02 by ozone and metal-catalyzed
oxidation. The most Important pathway for the oxidation of S02 1n the
aqueous phase 1s the reaction with hydrogen peroxide:
49
-------
S02 + H202 --
S0 + 0 -
Fe, Mn
S02 + 02 ------ * H2S04
Although aqueous-phase oxidation of S02 can be very rapid, near the source
1t 1s generally limited to the amount of H202 available (oxldant limited).
Further downwind the oxidation of S02 by H202 may be limited by the amount
of S02 available (S02 limited).
3.3.1.2 Nitrate Chemistry
The formation of nitrate aeorosol and nitric add vapor from N02 occurs
mainly 1n the gas phase. During the day the reaction of NO? with the
hydroxyl radical forms nitric add at a rate almost seven times faster
than the reaction that forms sulfates:
N02 + OH+ -* HN03
Thus during the day the oxidation of S02 and N02 compete with each other
for the available hydroxyl radical. At night, nitrates and nitric add
are formed with a direct reaction with ozone:
N02 + 03 -* N03 + 02
N03 + N02 -* N205
N205 + H20 -* 2HN03
The relative concentrations of nitrates and nitric acid 1s dependent on
the amount of ammonia present. For the typical nitrate and ammonia con-
centrations found 1n the Rocky Mountains, most of the nitrates are con-
verted to nitric acid. Since both nitrates and nitric acid are scavenged
efficiently by precipitation, the distinction between these species can be
Ignored for purposes of modeling nitrogen deposition. When calculating
the pH of deposition or visibility Impairment, however, it is important to
distinguish between them. Information concerning the total nitrate and
ammonia concentration 1s required in order to split the nitrate species
between nitrates and nitric acid. Since the Lagrangian model being
developed here only has Information concerning concentrations from the
source 1n question, 1t cannot distinguish between nitrates and nitric
add.
50
-------
3.3.2 Review of the Chemical Mechanisms 1n the Candidate Models
3.3.2.1 MESOPUFF-II
The MESOPUFF-II contains five methods for treating the oxidation of S02 to
sulfates and two methods for treating the oxidation of NOX to nitrates and
nitric add. These methods are: user-specified rate constants, the ERT
theoretical method, and three methods for treating S02 oxidation based on
an analysis of air quality data by GUlanl (1981, St. Louis plume data),
Henry and Hidy (1982, St. Louis urban data), and Henry and Hidy (1981, Los
Angles urban data). Clearly those S02 oxidation methods based solely on
the analysis of urban aerometrlc data would not be appropriate for the
Rocky Mountains. Thus we restrict our discussion to the ERT method.
The ERT chemical transformation method produces rate constants for the
following reactions:
S02 -* S04
NOX -* HN03
NOX -* N03
NH3
HN03 «--* N03
The transformation rates for these reactions were developed by statisti-
cally analyzing hourly transformation rates produced by a box model using
the Atkinson and Lloyd chemical kinetic mechanism (Atkinson, Lloyd, and
Hinges 1982). These transformation rates were obtained by simulating the
dispersal of plume SOX/NOX Into background air containing ozone and reac-
tive hydrocarbons (RHC) over a wide range of environmental conditions
representing different solar radiation Intensities, temperatures, disper-
sion conditions, background ozone and RHC, and time of day. Stepwise
linear regression on the logarithms of the resultant concentrations was
performed to find the controlling variables. Linear regression techniques
were then performed on these variables to determine the transformation
rates for the above equations. Since the Atkinson and Lloyd mechanism
treats only homogeneous oxidation of S02, an empirically determined
heterogeneous S02 conversion term based on relative humidity (3 x 10"8
RHC) is added on to the homogeneous term with an imposed minimum value of
0.2 X/h.
The controlling variables for the homogeneous reactions for S02 were solar
radiation, atmospheric stability, and background ozone. For the oxidation
of NOX the controlling variables were atmospheric stability, background
ozone, and plume NOX concentrations. Although it is well known that
51
-------
photochemical activity (and hence SOo and NOX oxidation rates) increases
with Increasing temperature, the MESOPUFF-II chemistry module does not
account for the effects of temperature on oxidation rates. In addition,
the oxidation rates produced by the ERT mechanism will be relevant for the
background RHC levels used in the box model simulations. The dependence
of the NOX oxidation rate on the plume NOX concentrations presents a dif-
ferent problem. As stated by the model developers, when puffs overlap it
would be Incorrect to calculate a NOX oxidation rate for a puff based only
on the puff's own NOX concentration. Thus the MESOPUFF-II sums the NOX
concentrations from all overlapping puffs to obtain a single oxidation
rate for the puff 1n question. However, what 1s not stated by the model
developers 1s that a single puff from a NOX source will have the NOX con-
centrations 1n a Gaussian distribution around the plume centerllne; thus
the use of a single NOX oxidation rate for the entire puff may neverthe-
less be Incorrect.
3.3.2.2 RIVAD
The RIVAD model uses a highly condensed, simplified chemical mechanism to
calculate the chemical transformation rates for the formation of sulfates,
nitrates, and nitric add. The homogeneous oxidation of S02 and NOX comes
from the reaction of S02 with the hydroxyl radical (OH-)- The RIVAD model
estimates the concentration of the OH radical based on solar radiation
Intensity, ozone concentration, temperature, relative humidity, and N02
and SOo concentrations. A maximum possible calculated OH- concentration
1s defined based on numerous smog chamber simulations that used a complete
photochemical kinetic chemical mechanism. A constant S02 oxidation rate
of 0.2 fc/h 1s added on to the homogeneous rate estimated from the OH- con-
centration to take Into account any heterogenous reactions.
At night, a reduction 1n the hydroxyl radical reduces the rate of S02
oxidation 1n the RIVAD down to the heterogenous rate of 0.2 %h.
The RIVAD model uses the photo-steady relationship between NO, N02, and 03
1n order to determine the steady-state N02 and 03 concentrations. The
oxidation of N02 to nitric acid 1n the RIVAD depends on the estimated
hydroxyl radical concentration, as discussed above. At night, however,
nitric acid 1s formed through a direct reaction with the N02 and ozone
concentrations. For a more complete explanation of the RIVAD nitrate
chemistry mechanism, see Latimer, Gery, and Hogo (1986).
3.3.2.3 CCADM
The CCADM contains complete gas-phase and aqueous-phase chemical kinetic
mechamisms with associated mass transfer algorithms between phases (Gery
52
-------
et al., 1987; Morris and Kessler, 1987). This mechanism contains up-to-
date gas- and aqueous-phase reactions based on the literature as of Novem-
ber 1986. The mechanism 1s highly nonlinear and the reaction rates also
depend on the background concentrations. As such, this mechanism would
not be appropriate for use with a Lagrangian puff model treating a single
source without complete Information concerning the background concentra-
tions. Since these background concentrations within the Rocky Mountain
region are not available at this time, either through measurement programs
or modeling of regional photochemistry or add deposition, the explicit
CCADM mechanism cannot be used in a Lagrangian puff model for this region.
However, the CCADM can be used to generate a table of oxidation rates, as
has been done for the MESOPUFF-II and RTM-IINL models (Morris and Kessler,
1987). These oxidation rates would be obtained by repeated simulations of
the CCADM using different ambient and chemical conditions in the Rocky
Mountains. Although time constraints preclude development of this chemi-
cal mechanism for the Initial version of the Rocky Mountain model, its
development 1s currently underway and will be incorporated 1n later
versions of the model.
3.3.3 Evaluation of the Chemical Mechanisms
The chemical mechanisms used in the MESOPUFF-II and RIVAD models were
evaluated by calculating S02 and N02 oxidation rates for a variety of
ambient and plume conditions. The evaluation procedure consisted of
determining whether the mechanisms calculate reasonable oxidation rates
and respond to changes 1n environmental conditions 1n a fashion expected
by our knowledge of the chemistry of acid deposition. We started with the
following baseline conditions:
Ozone concentration = 40 ppb
NOX concentration = 1 ppb
S02 Concentration = 1 ppb
Relative humidity = 50 %
Temperature = 298 K
Solar declination angle = 25° daytime, 90° nighttime
Each of the Important environmental and plume parameters were then varied
across a range of values to determine the responses of the two chemical
mechanisms.
3.3.3.1 Solar Radiation
Increases in solar radiation cause Increases 1n ozone photolysis and hence
Increases 1n the hydroxyl radical and the oxidation of S02 and N02.
53
-------
Figure 3-21 shows the oxidation rates produced by the MESOPUFF-II and
RIVAD as a function of the solar zenith angle (90° minus solar zenith
angle) for the baseline environmental conditions. Both chemical
mechanisms behave slmiliarly 1n response to changes 1n solar Intensity.
The MESOPUFF-II oxidation rates Illustrate a steplike response to changes
1n solar Intensity that is a result of using the step function stability
1n Its parameterization of oxidation rates rather a continuous function,
such as the N02 photolysis rate or solar elevation. The most Important
difference between the two mechanisms is that at maximum solar intensity
the NOX oxidation rate calculated by MESOPUFF-II peaks at 3.7 %/h whereas
the rate calculated by RIVAD peaks at approximately 8.3 %/h. Since NOX
oxidation should be approximately seven times that of S02 (which is
approximately 1.6 to 2.0 %/h in Figure 3-2), we would expect NOX oxidation
rates from the models on the order of 10 %/h, which 1s about what the
RIVAD calculates.
3.3.3.2 Temperature
Figures 3-22 and 3-23 show the sensitivity of the chemical mechanisms to
temperature variations for daytime and nighttime conditions respec-
tively. The RIVAD chemical mechanism 1s highly sensitive to variations 1n
temperature, while the MESOPUFF-II algorithm does not respond at all to
temperature variations. At low temperatures the OH concentrations drop
partly because several species, such as PAN, become strong sinks for OH
under low temperatures. At night, under the ambient conditions shown 1n
Figure 3-23, the S02 oxidation rate for both the MESOPUFF-II and RIVAD
mechanisms 1s at the minimum rate defined by the 0.2 %/h heterogeneous
minimum value. The RIVAD NOX oxidation rate at night 1s also sensitive to
the ambient temperature; however, this is driven by the temperature-sensi-
tive N02, N205, N03, HN03 equilibrium rather than the hydroxyl radical,
which 1s zero at night. The temperature dependence of these oxidation
rates could be an Important advantage for an acid deposition model for the
Rocky Mountains because the high terrain produces low temperatures even 1n
the presence of sunlight.
3.3.3.3 Relative Humidity
The effect of relative humidity on the two chemical mechanisms for daytime
and nighttime conditions are shown in Figures 3-24 and 3-25 respec-
tively. It is Interesting to note that although the MESOPUFF-II and RIVAD
predict similar daytime S02 oxidation rates for the range of relative
humidity between 25 and 50 percent, they deviate from each other on the
two extreme ranges of relative humidity. Under the daytime environmental
conditions used 1n these tests, the RIVAD estimation of the OH+ concentra-
tion reaches the maximum allowable value at a relative humidity of about
54
-------
tn
2.0
10 20 30 40 50 60 70 80 90 10
1.5
ID
1.0
c
o
03
TD
X
CD
-------
en
cr>
>.u
50
260
270
280
290 300
o
x
o
0.5
o
GO
260 270 280 290
Temperature
300
310
2.0
1.5
1.0
0.5
310
0.0
FIGURE 3-22. Sensitivity of the daytime MESOPUFF-II and RIVAD
chemical mechanisms to temperature.
260
270
280
290
300
310
260 270 280 290
Temperature
1 [
300 310
KM VAD
07one -
SUZ
K'elotive Humidity
Solar Zenith Angle
Ntl,r I'hnt o lysis
39 ipphl
1 Ippbl
1 Ippb)
SO 17.1
= 0. 50S
-------
01
2.?'
2 1.5
.C
«J
^ 1.0
c
o
"D
X
°0.5
(NJ
O
C/l
n n
50 260 270 280 290 300 310
i 1 i 1 i 1 i 1 i 1 i 1
-
-
_
-
-
__
-
. 1 . 1 . l i 1 i l i I
2.0
260 270 280 290
300 310
.5
1.0
0.5
260 270 280 290
Temperature
300
310
0.0
c
o
03
o
X
O
X
CD
FIGURE 3-23. Sensitivity of the nighttime MESOPUFF-II and RIVAD
chemical mechanisms to temperature.
260
270 280
Temperature
290
300
MFSorurr-i
K ! VAD
I (ERT Methodl
Ozone
NUx
Rc I otivc Hum t dit y
Solar Zenith Angle
- 39
[pphl
I (pphl
I (ppbl
00 (7.1
90 Ideq.
O.O'/H
310
-------
1 0C
t . u
3.5
~ 3.0
JZ
~ 2.5
o
<->
^ -> n
A. . U
C
o
X
0 1.0
r\i
O
C/)
0.5
n n
) 10 20 30 40 50 60 70 80 90 1C
_ i i i I i I i 1 i 1 i 1 i 1 i 1 i 1 i _
/ ~
: / ~-
\ y \
~ l 1 l 1 1 1 1 1 1 1 l 1 1 1 l 1 l 1 l ~
)0 n
4 . U
3.5
3.0
2.5
2.0
1.5
1.0
0.5
n n
Re 1 ative Humidity
FIGURE 3-24. Sensitivity of the daytime MESOPUFF-II and
RIVAD chemical mechanisms to relative humidity.
10C
1 U
9
8
^ 7
4, 6
-------
UD
4.0(
3.5
~ 3.0
"2.5
Cl
*->
^2.0
c
o
£ i . 5
a
0 1.0
CM
O
0.5
n n
) 10 20 30 40 50 60 70 80 90 100 n
_ i 1 i I i I i I i | i 1 i 1 i 1 i 1 i _
-
-
-
-
-
- i 1 i 1 i 1 i 1 i 1 i 1 i 1 i 1 i 1 i '
3.5
3.0
2.5
2.0
1.5
1.0
0.5
n n
10 20 30 40 50 60 70 80 90
0 10
Re 1 ative Humidity
FIGURE 3-25. Sensitivity of the nighttime MESOPUFF-II and
RIVAD chemical mechanisms to relative humidity.
0 10
20 30 40 50 60
Relative Hum i d i ty
70 80 90 1
R I V A D
0 7 o n e
NOx
Solar Zenith Angle
NIK' I' h O t n I y s I S
T o m p o r o t u r c
- I I (TRT Method!
3Q ippb)
I (pph)
I i pph I
'io
0.000
? '.> M I
-------
25 percent, which causes both the S02 and NOX oxidation rates to flatten
out for relative humidities greater than 25 percent. On the other hand,
the MESOPUFF-II S02 oxidation rate starts Increasing for relative humidi-
ties greater than about 50 percent. This 1s due to the parameterization
of the heterogeneous oxidation of S02 oxidation based on relative
humidity, RH (3 x 10~8 RH), which is a surrogate for the aqueous oxidation
of S02. The reasoning behind this heterogeneous parameterization 1s
unclear and no explanation is offered by the model developers in the docu-
mentation of the model (Scire et al., 1983). The aqueous-phase oxidation
of S02 should depend mainly on the liquid water content, hydrogen peroxide
concentration, and solar Intensity, not the water vapor concentration
(relative humidity).
At night, for the environmental conditions used in these tests, both the
MESOPUFF-II and RIVAD produce their minimal allowable S02 oxidation rates
of 0.2 %/h. For NOX oxidation at night the MESOPUFF-II calculates its
minimal allowable value of 2 %/h, while the RIVAD model shows some sensi-
tivity to relative humidity because of the influence of water vapor 1n the
N02, N03, N205 equilibrium calculation.
3.3.3.4 Ozone Concentrations
Both the MESOPUFF-II and RIVAD require estimates of the background ozone
concentrations in order to estimate their oxidation rates. Figures 3-26
and 3-27 Illustrate the sensitivity of the two mechanisms to ozone concen-
trations for day and night conditions. During the day the MESOPUFF-II S02
and NOX oxidation rates are very sensitive to the specification of the
background ozone concentrations. For the environmental conditions used in
these tests, the RIVAD calculated the maximum value for OH- at a low ozone
concentration; hence its S02 oxidation rate is not very sensitive to back-
ground ozone. The RIVAD daytime N02 oxidation rate shows a curiously weak
dependency on the ozone concentration that cannot come from the reaction
with OH-, which has been set to the maximum allowable value for ozone
greater than about 15 ppb. As 1t turns out, this Increase 1n NOX oxida-
tion rate with Increasing ozone concentration 1s a result of the photosta-
tlonary state relationship of NO, N02, and 03 used in the RIVAD. Since it
1s N02 and not NO that converts to nitrates and nitric acid, the RIVAD
model apportions the NOX concentration to NO and N02, using the N02
photolysis rate value and the background ozone concentration, and then
calculates the amount of nitrate and nitric acid formed based on the N02
concentration. Thus this weak dependency of the daytime NOX oxidation
rate on ozone in the RIVAD mechanism is actually the result of more of the
NOX being apportioned into N02 as the ozone increases.
At night the MESOPUFF-II predicts Us nighttime minimum S02 and NOX oxida-
tion rates of 0.2 and 2.0 %/h. In the RIVAD the S02 oxidation rate at
60
-------
4.0<
3.5
£3.0
c.
x.
~ 2.5
o
*->
^2.0
c
0
*> 1 c:
03 l °
X
0 1.0
CNJ
CD
0.5
) 15 30 45 60 75 90 105 120 135 150
_ i i I i i I i i I i i | i i | i i 1 i i 1 i i 1 i i 1 i i _
'- ,/' ~-_
.' _
./
1 /' ~
- / / -
- / ~~-
~ / ;
~ 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 ~
15 30 45 60 75 90 105 120 135 IE
3.5
3.0
2.5
2.0
1.5
1.0
0.5
8-°
Ozone Concentration (ppb)
FIGURE 3-26. Sensitivity of the daytime MESOPUFF-II and
RIVAD chemical mechanisms to ozone concentration.
30
25
15 30 45 60 75 90 105 120 135 15
C-
.c
.-^ 20
4J
-------
,0 15 30 45 60 75 90 105 120 135
15 30 45 60 75 90 105 120 135 15
*» . u
3.5
~ 3.0
c.
"2.5
t>
^2.0
c
o
-------
night 1s also 0.2 %/h regardless of the ozone concentration, while the NOX
oxidation rate Increases with background ozone concentration because of
the effect of ozone on the N02, NOj, and N20g equllbrium calculation.
3.3.3.5 Nitrogen Oxide Concentration
The sensitivity of the oxidation rates to NOX concentrations are shown 1n
Figures 3-28 and 3-29 for daytime and nighttime conditions respectively.
Daytime NOX oxidation rates calculated by the MESOPUFF-II and RIVAD models
exhibit slmiHar responses to Increases 1n NOX concentration. However,
the MESOPUFF-II daytime S02 oxidation rate appears totally Insensitive to
changes 1n NOX concentrations. Since the oxidation of S02 and NOX during
the day revolves around competition for the hydroxyl radical, this lack of
sensitivity 1s somewhat disturbing. Thus in a plume containing both SOX
and NOX, as produced by a shale oil plant, the MESOPUFF-II will greatly
overpredlct the oxidation of S02 because 1t does not account for the com-
petition for the hydroxyl radical from the N02 reaction.
In both MESOPUFF-II and RIVAD the S02 oxidation rate at night 1s Insensi-
tive to changes 1n NOX concentrations and is at Us minimum value of 0.2
*/h. S1m1l1arly, the MESOPUFF-II produces Its minimum 2 X/h NOX oxidation
rate during the night regardless of NOX concentration. The RIVAD model,
however, produces a peak nighttime NOX oxidation rate at a NOX concentra-
tion of 20 ppb, with the rate reducing to zero for a NOX concentration of
40 ppb. This 1s because, for high NOX concentrations, the RIVAD assumes
all the NOX 1s NO. Thus at night, when the NOX concentration exceeds the
ozone concentration, all of the ozone is titrated out, resulting in no
more ozone to oxidize the N02.
3.3.3.6 Sulfur Dioxide Concentration
The effects of changes 1n S02 concentrations on the S02 and NOX oxidation
rates calculated by MESOPUFF-II and RIVAD are given in Figures 3-30 and
3-31. The MESOPUFF-II chemistry is totally insensitive to changes 1n S02
concentrations both during the day and night. The RIVAD mechanism 1s
Insenitive to changes in S02 concentration at night; however, during the
day both the S02 and NOX oxidation rates decrease as S02 concentrations
Increase. This 1s because of the limited availability of the hydroxyl
radical, which dominates the daytime oxidation of both S02 and NOX.
3.3.4 Remarks
3.3.4.1 Daytime Chemistry
The MESOPUFF-II daytime oxidation rates appear to be most sensitive to
changes 1n solar Intensity and background ozone concentrations, while the
63
-------
10 20 30 40 50 60 70 80 90 1QQ
10 20 30 40 50 60 70 80 90 10
-0.5
"0 10 20 30 40 50 60 70 80 90 100*
NOx Concentration Ippb)
0
FIGURE 3-28. Sensitivity of the daytime MESOPUFF-II and
RIVAD chemical mechanisms to NOX concentration.
10 20 30 40 50 60 70
NOx Concentration (ppb)
MESOPurr-ii
KI VAD
(CRT Method)
0 7 o n ti
St).?
K c I ,11 i v n Humidity
Solar Zenith Angle
NIK' I'hot.o I y s I ;-.
T omperaturo
40 (pphl
I (pphl
'.() (XI
i"j I ctc-g . I
O.SOS
."Hi IK)
Po
80 90 10
-------
2.0(
~ 1.5
£1
4J
03
^ 1.0
c
o
TD
X
°0.5
i i< *
SLI
olar Zenith Angle
NO.-.' I'ho t o 1 y s i
f ('mppr.it u r
40 50 60 70 80 90 100
1 i I i 1 i I i I i 1 i
-
-
-
V] i i i i i i i i i i i
40 50 60 70 80 90 1(
tratton (ppb)
PUFI- II If . RT Method)
0
o = 40 fpph)
/ - 1 ipphl
y - SO (X 1
1 = MO l di.'<9 . 1
s - 0.000
e - ;"'' H I k' 1
i
9
9
7
6
5
4
3
2
1
-------
en
2.0
~ 1.5
jd
1.0
c
o
*->
-------
2 0C
<_ VJ
~ 1.5
.c
0
.*->
5 o ioc
i. U 1 U
9
8
1.5 Z
^ 7
t> 6
"lo
1.0 ^5
c
o
->-> 4
1 3
0.5 °
X 9
CD c
Z.
1
1
n n n
) 10 20 30 40 50 60 70 80 90 1(
i I i I i I i I i I i I i I i I i I i
-
_
-
-
-
-
- -
_ _
-
-
i 1 i 1 i I i I i I i I i 1 i I i 1 i
)0
1
9
8
7
6
5
3
i
l
n
10 20 30 40 50 60 70 80 90 1
S02 Concentration (ppb)
0 10 20 30 40 50 60 70 80 90 101
S02 Concentration fppb)
MrSOPUhF- I I (ERT Method)
R [ V A f)
FIGURE 3-31. Sensitivity of the nighttime MESOPUFF-II and
RIVAD chemical mechanisms to S02 concentration.
Ozone - 3'J Inphl
NHx = 1 (pphl
Rolntivr Hi.imirJity - SO (7.1
Solar Zenith Angle ~ 1(o ideg. i
Nil/ I'lmtol y^-, i r. - 0.0/H
Temperdture = ?9B IK I
-------
RIVAD mechanism 1s less sensitive to changes in ozone concentrations dur-
ing the day, at least for the enviromental conditions used 1n these
tests. However, the RIVAD chemistry 1s more sensitive to changes 1n tem-
perature and NOX and S02 concentrations, and 1s also very sensitive to
solar Intensity. Of particular note 1s the ability of the RIVAD to cor-
rectly simulate the competition between S02 and NOX for the hydroxyl radi-
cal, which drives the daytime gas-phase oxidation of these species. Since
the MESOPUFF-II shows no sensitivity to changes 1n S02 concentrations, and
no sensitivity 1n its S02 oxidation rate to changes in NOX concentrations,
1t will overpredict the oxidation rates of these species for plumes near
the source.
3.3.4.2 Nighttime Chemistry
The MESOPUFF-II produces constant S02 and NOX oxidation rates of 0.2 X/h
and 2.0 X/h respectively. The RIVAD model also produces a constant 0.2
X/h S02 oxidation rate at night regardless of the enviromental condi-
tions. However, the oxidation rate of NOX 1s sensitive to changes in all
environmental conditions examined except changes in S02 concentrations.
This 1s due to the influences of ozone, temperature, and water vapor on
the N02, N205, N03, and HN03 equilibrium.
3.3.4.3 Aqueous Chemistry
Neither the MESOPUFF-II or the RIVAD chemical mechanisms contain explicit
treatment of aqueous chemistry. Both contain surrogate heterogeneous S02
oxidation rates with a minimum value of 0.2 X/h. The MESOPUFF-II ties
this heterogenous oxidation rate to relative humidity during daytime.
However, as noted above, aqueous-phase chemistry should be tied to liquid
water content, not water vapor concentrations; thus this parameterization
of surrogate aqueous-phase chemistry appears to be unjustified by current
knowledge.
3.3.4.4 Conclusions
For the initial version of the Rocky Mountain acid deposition model
(described 1n Section 5) both the MESOPUFF-II and RIVAD chemical
mechanisms will be included as optional treatments of chemical transforma-
tion. Based on the evaluation of the two chemical mechanisms, the RIVAD
parameterization 1s preferred over the MESOPUFF-II for the following
reasons.
68
-------
The RIVAD mechanslsm treats the competition between the S02 and NOX
species for the hydroxyl radical during the day.
The oxidation rates produced by the RIVAD show sensitivity to changes
in temperature, which can be important in the high terrain of the
Rocky Mountain region.
The RIVAD model treats the sensitivity of NOX oxidation to changes in
conditions at night, whereas the MESOPUFF-II uses constant values
regardless of the conditions.
The individual species of NO and N02 are treated separately by the
RIVAD model, while the MESOPUFF-II lumps NO and N02 together as
NOX. Since N02 is a criteria pollutant with an annual NAAQS, and
several western states and counties have their own standards--e.g.,
New Mexico has a 24-hour N02 standard, California has an hourly N02
standard, and Santa Barbara County has an hourly incremental N02
standard)the distinction between NO and N02 is important from a
regulatory perspective.
3.4 DRY DEPOSITION
Two of the candidate models, the MESOPUFF-II and the CCADM, use the more
technically rigorous resistance approach for the parameterization of dry
deposition. The RIVAD uses the dry deposition velocity concept, while the
POLUT does not consider pollutant loss from dry deposition.
The flux of pollutants to the ground due to dry deposition can be expres-
sed as:
Fd - V, c (3-9)
where Vj is the deposition velocity, and c 1s concentration at some refer-
ence height. In the RIVAD the deposition velocity is a function of land-
use type and is set to zero at night to account for the shielding effect
of the stable nocturnal boundary layer. During the day. however, the
deposition velocity is applied to the mixed-layer concentration, effec-
tively enhancing the rate of vertical diffusion of pollutants because mass
removed at the surface is immediately replaced with material from above.
In the resistance approach to dry deposition, the deposition velocity is
expressed as the inverse sum of the atmosphere, surface, and canopy resis-
tances:
Vd = <3
69
-------
In the MESOPUFF-II an option exists to treat vertically well-mixed puffs
with a three-layer model. This parameterization essentially removes the
enhanced rate of vertical diffusion by considering the loss of pollutants
only out of the surface layer.
The CCADM uses the dry deposition algorithm in the NCAR/RADM for gaseous
species (Walcek et al., 1986) and the algorithm in the ERT/ADOM for par-
ticulate species (Pleim, Venkatram, and Yamertino, 1984). It also uses a
surface layer for calculating atmospheric resistance to minimize the exag-
geration of depletion by instantaneous vertical mixing. The parameteriza-
tlons of dry deposition used by the MESOPUFF-II and the CCADM are compared
below.
3.4.1 MESOPUFF-II and CCADM Parameterizations
In the MESOPUFF-II the atmospheric (aerodynamic) resistance, r&, is given
by the following formula proposed by Wesley and Hicks (1977):
oj*)'1 [In(zr/z0) - <
The stability-dependent function 4>H is given by:
ra =
(3-11)
"H
-5 zr/L.
= exp{0.509 H- 0.39 -
ln(-zs/L) -0.090[ln(-zr/L)n
= 0
0 < zr/L < 1 (stable)
-1 < zr/L < 0 (unstable)
zr/L = 0 (neutral)
(3-12)
where:
zr -
2Q-
u*
K .
L >
the reference height (10 m)
the surface roughness (m)
the friction velocity (m/s)
the von Karman constant
the Monin-Obukhov length (m)
In the CCADM the surface resistance is given by the following equations
(Businger, 1973):
70
-------
- in
for c < °(unstable)
0 (stable)
~ In(zh/z0) for c - 0 (neutral)
(3-13)
*h(c) - 0.74 (1 - 9c)-1/2 for c < 0 (unstable)
0.74 + 4.7c for c = 0 (neutral)
= 0.74 for ; > 0 (stable) (3-14)
where c = z/L. These two representations of the atmospheric resistance
are similar; both are proportional to the Inverse friction velocity. Thus
as the wind speed Increases, the atmospheric resistance decreases.
For gaseous species the surface resistance (also known as the quasi laminar
sublayer resistance), rs, can be expressed as follows (Wesley and Hicks,
1977):
rs = (k uj kiT1 (3-15)
where B~* 1s the surface transfer coefficient. As suggested by Wesley and
Hicks (1977). a value of 2.6 is used for kB'1 for S02 and the other gases
(NOX and HN03) 1n the MESOPUFF-II. For the aerosol species 1n MESOPUFF-II
(sulfate and nitrate) the surface resistance is assumed to be 10 s/cm.
71
-------
In the CCADM the surface resistance 1s obtained using a species-dependent
formula taken from the ADOM/TADAP model:
rs . (3-16,
where Sc 1s the Schmidt number, defined as the ratio of the molecular dif-
fuslvlty of air (0.149) to the. molecular dlffusivity of the gaseous
species 1n question. As currently Implemented 1n the CCADM, all gaseous
species are assumed to have the same molecular dlffusivity as S02
(0.126). The value of a 1n Equation 3-16 1s set to 5 as recommended by
Hicks (1983).
For aerosol species 1n the CCADM the gravitational settling resistance
1/Vg acts 1n parallel to the other resistances:
v< = V ^T^vA (
The gravltlonal settling velocity 1n the CCADM is given by Stokes law:
V_ = P ? 3 C , (3-18)
where 9 18n
p = particle density = 1.0 g m"3
d = particle diameter = 10~6 m
g = gravitational acceleration = 9.8 m/s
n = dynamic viscosity coefficient for air = 1.83 10~4
and the value C 1s a correction factor for small particles, given by
C - 1 + * 1.257 + 0.4 exp[-0.55 d/x] (3-19)
where x is the mean free path of air molecules. Although the mean free
path 1s known to be dependent upon pressure, both the ADOM model (Pleim,
Venkatram, and Yamartlno, 1984) and the CCADM model (Gery et al., 1987)
assume a constant value of 6.53 x 10"6 cm for x.
72
-------
The quasi laminar sublayer resistance (surface resistance) for particles is
obtained from the friction velocity and collection efficiency as follows
(Pleim, Venkatram, and Yamartino, 1984):
rs - ^ (3-20)
where E is the collection efficiency given by
E = Sc"2/3 + 10'3/St (3-21)
2
and St is the Stokes number defined as St = T u*/n, where T is the
stopping time specified as 1.31 x 10" 5 m for a particle with a radius of
1 um and n is the dynamic viscosity of air.
In Equation 3-20 the value of a is 1.7, as recommended by Holler and Schu-
mann (1970), because it gives the best fit to measured deposition
velocities for particles.
The values for canopy resistance, rc, to S02 used in the MESOPUFF-II are
from Sheih and co-workers (1979), who estimate summertime canopy resis-
tance for S02 as a function of land use and stability class for summertime
conditions (Table 3-1). The canopy resistance to HN03 and the aerosol
species (S0~ and NOI) are assumed to be 0. For NOX the canopy resistances
(in s/cm) are defined as follows:
rc(NOx) = 1.3, unstable
= 5, neutral
= 15, stable
In the CCADM the canopy resistance to S02 varys diurnally and seasonally
and also varys if the surface is wet, as shown in Table 3-2 (Walcek et
al., 1986). The canopy resistances to other gaseous species are related
to the canopy resistance to S02 according to the multiplicative factors
given in Table 3-3.
For aerosol species in the CCADM the canopy resistance is 0. However, in
the equation for particulate deposition velocity there is a third resis-
tance, rarbVg, referred to as a virtual resistance in view of the fact
73
-------
TABLE 3-1. Summertime 502 can°Py resistances used in the
Mesopuff-II as a function of land use type and stability
class. (From: Sheih, Wesely, and Hicks, 1979).
Category Land Uae Type
1 cropland and pasture
2 cropland, voodland and grazing
land
3 irrigated crops
4 grated forest and woodland
5 ungrared forest and woodland
6 aubhumid grassland and semiarid
grazing land
7 open woodland grazed
8 desert shrubland
9 swamp
10 marshland
11 Metropolitan city
12 lake or ocean
Stability Class
z« A.B.C D E
0.20 100. 300. 1000.
0.10
0.20
0.30
0.20
0.50
1.0
100. 300.
100. 300.
200. 500.
50. 75.
75. 300.
1000. 1000.
0. 0.
0.
0.30
0.05
0.90
1.00
100.
100.
100.
100.
300.
300.
300.
300.
1000.
1000.
1000.
1000.
0
0
0
0
1000. 0.
1000. 0.
1000. 1000.
100. 0.
1000. 0.
1000. 0.
0. 0.
74
-------
TABLE 3-2. S02 canopw resistance, RSQ (s m" ), used in the
CCADM. (Source: Walcek et al., 1985.) 2
LAND USE
Urban
Agriculture
Range
Deciduous
forest
Coniferous
forest
Forested
swamp
Water
Swamp
Agriculture-
range mixture
SEASON
spring
summer
eartyfall
late fall
winter
spring
summer
early fall
late fall
winter
spring
summer
eartyfall
late fall
winter
spring
summer
early fan
late fall
winter
spring
summer
eartyfall
late fall
winter
spring
summer
eartyfall
toe tall
winter
spring
summer
eartyfall
fate (all
winter
spring
summer
eartyfall
late fall
winter
spring
summer
eartyfall
late fall
winter
INSOLATION (Watts m-2)
>400 200-400 0-200
1000
1000
1000
1000
200
50
70
500
50
100
100
100
500
500
100
100
60
1000
1000
1000
150
150
800
800
500
100
70
800
BOO
800
0
0
0
0
0
50
50
100
100
100
75
100
500
200
100
1000
1000
1000
1000
200
60
120
500
50
100
140
140
500
500
100
200
130
1000
1000
1000
240
240
800
800
500
200
140
800
800
800
0
0
0
0
0
60
60
100
100
100
100
140
500
200
100
1000
1000
1000
1000
200
75
200
500
50
100
200
200
500
500
100
400
300
1000
1000
1000
400
400
800
1000
500
400
300
800
1000
800
0
0
0
0
0
75
75
100
100
100
150
200
500
200
100
NIGHT
1000
1000
1000
1000
200
100
500
500
50
100
400
500
500
500
100
1000
1000
1000
1000
1000
1000
1000
800
1000
500
1000
1000
800
1000
800
0
0
0
0
0
100
100
100
100
100
250
500
500
200
100
WETTED
1000
0
1000
1000
200
0
0
100
50
100
0
0
100
100
100
0
0
500
500
1000
0
0
100
100
500
0
0
300
300
800
0
0
0
0
0
0
0
75
75
100
0
0
100
100
100
75
-------
TABLE 3-3. Canopy resistances used in the CCADM assumed for
dry-deposited gases relative to SO? surface resistance. (Source:
Chang et al., 1986).
Surface Resistance (snv1)*
Pollutant Over Land Surfaces Over wetted surfaces
NO
N02
OB
HNO3
HA
Aldehyde
HCHO
Methyl-hydrogen
peroxide
Peroxyacetic acid
HCOOH
NH3
RS02
RSO2
0.6RS02
0.0
0.1RS02
2.0RS02
0.5RS02
0.3RS02
0.3RS02
RSO2
0.2RS02
500
500
2000
0.0
0-1RS02
2.0RS02
0.5RS02
0.3RS02
0.3RS02
RSO2
0.2RS02
* Values for Rcn given in Table 3-2.
76
-------
that 1t 1s a mathematical artifact of th» equation manipulation rather
than a physical resistance (see Pleim, Venkatram, and Yamertlno, 1984; and
Gery et al., 1987).
The canopy resistance for the MESOPUFF-II 1s chosen from Table 3-1 based
on one of the 12 land-use classifications specified 1n the grid cell con-
taining the puff centrold. Clearly, when the puff 1s large and covers an
area that Includes several different land uses, this simplification may
Introduce some errors.
In the CCADM the fraction of coverage of each of the nine land-use clas-
sifications (1n Table 3-2) 1s specified for each grid cell. The CCADM
then calculates the fraction coverage across the base of the Lagrangian
box through weighted averaging of the grid cells covered by the box. Then
an average deposition velocity for the area covered 1s calculated using
the method proposed by Walcek and others (1986).
3.4.2 Comparison of MESOPUFF-II and CCADM Performance
The deposition velocities produced by the dry deposition algorithms in
MESOPUFF-II and CCADM are compared here using a variety of environmental
conditions and several land use classifications. Since the two algorithms
do not use the same land use classification scheme, the land use cate-
gories for the CCADM (Table 3-2) are adjusted to match the MESOPUFF-II
land use classes as closely as possible. The envlromental conditions that
vary are the surface wind speed and the exposure class, which 1s a measure
of Insolation as follows:
Ce =3, strong
= 2, moderate Daytime Insolation
= 1, slight
= 0, heavy overcast Day or night
4
= -1, >~s cloud cover
Nighttime cloudiness
= -2, < cloud cover
The stability class can be estimated from the exposure class and wind
speed using the method of turner (1970). Although the CCADM predicts
deposition velocities for many species, we compare only the deposition
77
-------
velocities for the five species 1n the MESOPUFF-II: S02, sulfate, NOX,
nitrate, and nitric add. Since there are very few measurements of dry
deposition, we cannot directly evaluate the two dry deposition algor-
ithms. Instead, the two methods will be compared against each other and
against the ranges of measured deposition velocities reported in the
literature.
3.4.2.1 Sulfur Dioxide Dry Deposition
The S02 deposition velocities predicted by the MESOPUFF-II and the CCADM
for three different land use classes are given in Figure 3-32. (A com-
plete set of predicted S02 deposition velocities for all land use classi-
fications is given in the Appendix.) The results for MESOPUFF-II and
CCADM are s1mH1ar for all three of the land use classes depleted 1n
Figure 3-32. For cropland and pasture, the MESOPUFF-II predicts deposi-
tion velocities that range from 0.1 to 1.0 cm/s, while the CCADM values
range from 0.1 to 0.7 cm/s. For the forest land use class, both models
predict slightly higher deposition velocities, ranging from 0.1 to 1.0
cm/s. The deposition velocities for these two land use classes produced
by the two models also have similiar characteristics as a function of
exposure class and surface wind speed.
The differences between the MESOPUFF-II and the CCADM for the positive
exposure class (daytime) can be attributed to differences 1n the methods
of representing stability 1n the two algorithms. The CCADM uses the expo-
sure class directly, whereas the MESOPUFF-II uses the Pasquill-Gifford
stability classification scheme 1n which stability (A-F) 1s a function of
exposure class and wind speed. At night the MESOPUFF-II appears to pro-
duce an anomalous S02 dry deposition velocity peak for clear skies and
wind speed around 2.5 m/s. These environmental conditions result 1n a
stability F classification, by which the MESOPUFF-II will assume a zero
canopy resistance to S02 (see Table 3-1). In general, under night condi-
tions the atmospheric resistance should be the dominant resistance, thus
the sensitivity to the canopy resistance under these conditions is ques-
tionable. Both the MESOPUFF-II and CCADM predict S02 dry deposition
velocities that are well within the range 0.04 to 2.8 cm/s for several
surface types cited in the literature (McMahon and Dennison, 1979).
For dry deposition over water, the MESOPUFF-II and the CCADM predict
remarkably similar patterns of SO? deposition velocities. The CCADM pre-
dicts values ranging from 0.5 to 3.0 cm/s, while the MESOPUFF-II values
from 0.1 to 2.0 cm/s. The reported measured values for S02 dry deposition
over water are 0.2 and 1.4 cm/s (Spedding, 1969), 0.9 and 0.5 cm/s (Owers
and Powell 1974), 2.2 cm/s (Whelpdale and Shaw, 1974), 2 cm/s (Prahm,
Tarp, and Stern, 1976), 0.41 cm/s (Garland, 1977), 0.5 cm/s (Smith and
78
-------
ucsopurr-ii
CCADU
1 23456789 10
.012345678 9 1C
I L
1 2 3 4 567 89 id" ~T> 1 2 3 4 5 6 7 8 9 10
Surfoct Wind Speed (m/«) Surfoce Wind Spe
-------
Hunt, 1978), and 0.4 to 4.0 cm/s (Sehemel, 1980). Sheih, Wesley, and
Hicks (1979) estimated S02 deposition velocities over the Atlantic Ocean
that ranged from 0.1 to 0.7 cm/s.
3.4.2.2 Sulfate Deposition
Predicted sulfate deposition velocities for three land use types-cropland/
pasture, forest/woodland, and waterare shown 1n Figure 3-33 (sulfate
deposition velocities for all land use classes are displayed in the appen-
dix). Although the MESOPUFF-II and the CCADM predict slmiHar patterns
for sulfate deposition, the MESOPUFF-II predicts much smaller values.
This result 1s directly related to the assumption in the MESOPUFF-II of a
constant surface (quasilamlnar sublayer) resistance of 10 s/cm for aero-
sols. Thus the maximum possible sulfate deposition velocity in the MESO-
PUFF-II 1s 0.1 cm/s. Measured values of sulfate deposition velocities
range from 0.03 to 1.0 cm/s (McMahon and Dennison, 1979); thus the upper
limit of 0.1 cm/s imposed by the MESOPUFF-II appears a little low. This
can be easily rectified by changing the assumed constant surface resis-
tance in the MESOPUFF-II. The CCADM predicts sulfate deposition veloci-
ties from 0.05 to 0.8 cm/s for the three land use classes in Figure 3-32;
the highest values occur for the forest land use class. For all land use
classes, the MESOPUFF-II predicts sulfate dry deposition values with
little variation at wind speeds above 1 m/s (0.070 to 0.1 cm/s).
3.4.2.3 Nitrogen Oxide Deposition
The CCADM predicts deposition velocities for NO and N02 separately, where-
as the MESOPUFF-II gives values for NOX. However, the NO and N02 deposi-
tion velocities predicted by the CCADM are Identical to each other since
the same canopy resistances are used for these two species (see Table 3-
3). Figure 3-34 compares the NOX deposition velocities predicted by the
MESOPUFF-II and CCADM for three land use classes. The CCADM predicts the
same dry deposition velocities for NOX as it does for S02. The MESOPUFF-
II deposition velocities for NOX resemble those calculated by MESOPUFF-II
for S02 without the anomalous peak for F stability. This is because the
MESOPUFF-II uses constant NOX resistances for the stable case.
For the cropland/pasture and forest/woodland land use classes the range of
NOX dry deposition velocities for the MESOPUFF-II is 0.1 to 0.7 cm/s and
for the CCADM 0.1 to 1.0 cm/s. The two models predict different deposi-
tion behavior over water; the CCADM-predicted dry deposition velocity for
NOX 1s as high as 3.0 cm/s. Since NO and N02 are not as soluble as S02,
the use of the same canopy resistance for these species over water seems
questionable. It should be noted that over wet surfaces, regardless of
80
-------
yCSOPurr-n
CCADU
017 5 416769 1Q .0123
3
t
«
S\
S
e t>
LJ
-
J
0
' o
g
. O
-
«
i
|-
i r 1 1 1 i i ;
e
§
3
r.
o
o
3
I
-
*
~
v
^?\\ \?
' \p\\ %
" \ n 1 °° \
: \1V\\ '
2 2
n
1 0 1
L>
I
0
0 a 0
LJ
-1 -'
_ 1 "
1
t
5
0
- 8
6
"
4 i 6 7 6 9 1
'
R
c
8
O
O
I
>
D
o 1
; U :
»
i
«
g
<
3
: A\\
3
5
s -
o -
R
O
O
y
f
>
3
?
~
s-
o -
O
s ;
O
£
1
i
-
Surtoce Wind Speed (m/») Surfoce Wind Spied (m/j)
SO4 Deposition Velocities (cm/s) (or CROPLAND AND PASTURE Lond Use Type
MESOPUT-ii CCADu
,0 1 234j»6769iq ,0 1 2345 6_ 78 9 1C
J
2
i i
O
9
o
£ 0
_ 1
*
c
c
o
- s
^J
; g
6
ills
i
0
o
s
0 | 0
u
-1 -'
1
f
c
o
1
1
3
c
c
0
*
O
(,
p
\ \
0 \
*
)
*
!
f*.
0
o
c
r\
- \ \ x« \
.\.\ \\
1
3
3
I
~
_
'f
B
o -
O
p
o
\ I
,
_
J
2
i
0
-1
-2
1 2345676
Surfoce Wind Speed (m/i)
1
Surface Wind Speed (m/§)
SO4 Deposition Velocities (cm/s) (or UNGRAZED FOREST/WOODLAND Land Use Type
MESOPuFT-n
1 ? 3 4 5 6 7 fl 9 1
CCAOu
3 « 9 6 7 B 9 1Q
- 2
2
S 1
o
£
i o
LJ
-1
-
s
°
o
6
>j
' 8
' O
-
-
8
6
P I H
fi I
:M } \
- § \ 8
il'l \ =
\\l\.\. V :
j j
2 2
.
i 5 i
u
|
0 | 0
u
-1 -1
-7 -:
o
o
^ 8
O
-
R
C
P
g
o
L»
o
w
' ° 1
: \
3
1 t
ri
*n _
0
O
is
R
o
S
6
- \ I 5
\ \ °
,1 \\ .
.
-
J
P
U
u>
p
15
-
:
I 2 3 4 5 6 7 6
Surface Wind Speed (m/i)
i 2 3 4 s i T i 5 fo'
Surface Wind Speed (m/«)
SO4 Deposition Velocities (cm/s) (or LAKE OR OCEAN Land Use Type
FIGURE 3-33. Comparison of MESOPUFF-II and CCADM predicted
sulfate dry deposition velocities for three land use classes
81
-------
ycsopurr-ii
CCADM
012 3 4 5 6 7 B 9 10. 01234 5 6 7 B 9 IQ
V ^ ^07^
t)1234567B910 t>1234S67B«10
N02 Deposition Velocities (cm/s) for CROPLAND AND PASTURC Lond Use Type
uESOPurr-ii
CCADu
0123456789 1Q .01234567891
t>1234S678»IO'' TJ12J45678SIO
Surface Wind Speed (m/>)
Surfoct Wind Speed (m/i)
N02 Deposition Velocities (cm/s) for UNGRA2ED TOREST/WOODLAND Lond Use Type
CCAOM
23456789
Surfoct Wind Speed (m/»)
10
1 23456789 10
Surfoct Wind Speed (m/s)
r-2
NO2 Deposition Velocities (cm/s) for LAKE OR OCEAN Lond Use Type
FIGURE 3-34. Comparison of MCSOPUFF-I! and CCADM predicted
NOX dry deposition velocities for three land use classes.
82
-------
land use class, the CCADM predicts NOX dry deposition velocities that
range from 0.1 to 0.2 cm/s, I.e., over a factor of 10 lower than for
water. Measurement of NOX deposition velocities 1s extremely difficult
because of the fast N0-03 reaction and because the release of nitrogen
compounds from the soil may result 1n a net negative deposition rate.
H111 and Chamberlain (1976) reported an N02 dry deposition velocity of 1.9
cm/s and a deposition velocity for NO of 0.1 cm/s. In a review of dry
deposition velocities from three studies, Sehtnel (1980) reported a range
of NOX deposition velocities from negative values to 0.5 cm/s. Studying
dry deposition 1n the Netherlands, van Aalst and co-workers (1983) mea-
sured values of the dry deposition velocities of NOX that ranged from -2.6
to 1.5 cm/s. Except for the NOX dry deposition velocity over water calcu-
lated by CCADM, both models predict numbers that are within the range of
the measurements. Since there are no reported measurements of deposition
of NOX over water at this time, one cannot discount the CCADM-predlcted
NOX deposition velocity; however, since NO and N02 are not as soluble as
S02, it 1s expected that over water the deposition velocity for NO and N02
would be lower than for S02. Thus the CCADM-predlcted deposition
velocities for NO and N02 over water are questionable.
3.4.2.4 Nitric Acid Deposition
NHric acid has a very high deposition rate compared with the other gases
studied because of its high reactivity. Both the MESOPUFF-II and the
CCADM assume a zero canopy resistance to nitric acid for all land use
categories. As shown in Figure 3-35 and the appendix, the two models also
predict remarkably slmiliar dry deposition velocities of nitric acid for
different enviromental conditions and land use characteristics. The ran-
ges of nitric add dry deposition velocities for the two models are
approximately 0.5 to 4.0 cm/s for cropland, 1 to 11 cm/s for forest, and
0.1 to 2.5 cm/s for water. There are very few measurements of the dry
deposition velocity of nitric acid; van Aalst (1983) reports a value of
0.6 cm/s. However, the fact that the two models agree on the dry deposi-
tion velocities for nitric acid gives us some confidence that the predic-
tions are reasonable.
3.4.2.5 Nitrate Deposition
Both the MESOPUFF-II and the CCADM predict similar dry deposition veloci-
ties for nitrate (Figure 3-36) as for sulfate (Figure 3-33). Thus the
discussion of sulfate dry deposition applies to nitrate. In particular,
the very low nitrate deposition velocities predicted by the MESOPUFF-II
(less than 0.1 cm/s) may be questionable.
83
-------
uCSOPurr-n CCAOM
y
2
0 1
O
i
1 o
*^
.- '
> 1 234567891
c
-
J
O
3
»
I
C
t»
D
\
\ \ \
1
r
j
>
n
*
t
t
o
rj
(N
0 \ \ » I
0- I 0 \
\ I \ \
\ \ ' \ \
^ '
f
«
>
J
r
i
^
o
c*
I
\ \ " \ 0 I
: , \\VlA\i
.
_.
t
-
-
"
-
J,-
P
^ 3<
2 2
i i i
0
i
o
0 £ o
u
-1 -1
-? -:
)1 234167891
. '
1
_ (
-
k
1
1
O
n
D
t
o
1
t
*
/ /
1 1 '
u
o
3
i,
e
*>
1
" ° 1 1 * 1
r \ 1 /
h \ \ f
1 \ *
: , 1, \, \,
,_
1
«
>
i
-
'
-
.
-
-
*^
/ ^
,/ J,
-
.
'
123456788 10'
Surtoce Wind Speed (m/i)
23456789
Surtoce Wind Speed (m/»)
HN03 Deposition Velocities (cm/s) for CROPLAND AND PASTURE Lond Use Type
MKOPUFT-II CCACU
.01 23456789 10. .0123456789 1C
J
2
i i
0
S
1 o
"
-1
~1
-
I
1
1
" \\\
1
1 \
-. \ \ V \
f
',
1
*
\\l
.
1
)
1
i
i
:
.
i
_
i
.
_
.
31 234567891
2 2
1 i i
0
I
0 | 0
UJ
-1 -1
o-2 ~\
-
_
:
1
|
)
\(( J
(
«
S
1
)
I \\ !/
i Me
p
>
1
1
w
)
'
>
i
i
_
-
_
.
i
51234567891
j
2
1
0
-1
o-2
Surtoc* Wind Speed (m/s) Surtoce Wind Spe«d (m/s)
HN03 Deposition Velocities (cm/s) for UNGRA2ED FOREST/WOODLAND Lond Use Type
MCSOrut t -n
' ' O 1 2 3 456 7 8 9 1Q
~T> I 2 3 4 5 6 7 8 9 10
Surfoce wind Sp««d (m/j)
I234567B910'
Surtoce Wind Speed (m/s)
HN03 Deposition Velocities (cm/s) for LAKE OR OCEAN Lond Use Type
FIGURE 3-35. Comparison of MESOPUFF-II and CCADrt predicted
nitric acid dry deposition velocities for three land classes
84
-------
ytsoPurr-n
CCADM
01 23456789 10. .01234567891
1)12345678610 T> 1 23 4 5 6 78 9 10
Surface Wind Speed (m/»)
Surfoc* Wind Speed (m/t)
N03 Deposition Velocities (cm/s) (or CROPLAND AND PASTURE Land Use Type
MCSOPurr-n CCADM
3,
2
*
0 1
u
I
o
n °
LJ
31234567691
!
c
c
o
- g
n
' 8
o
(
- i\
5 1
> 0
o
o
to
<£>
D
0
3 C
c
1 I 1 1 1
-
3
§'
'
D
: q
\\V\\ w
-
-
-
I
O
o.
" ff^L*\"*1 \ \ *SP
:ISvAV\ , \ , , , i :
<3 3,
2 2
>
i S i
o
0 K °
-1 -1
D 1 234567891
i
T
f
!
>*
O
o
-
1
1
1
0
i.
0
O
^
«o
'' v\
- \ \
\ ;
O
»
I
° 1
C
^
0
»
o
b
r
»
I
\ \ '" \
L I \ \\
e
5
0
1 I
-
-
e
o -
»
S
2
1
0
-1
t> 1 2345676910* t) 1 23456769 id"*
Surfoct Wind Speed (m/s) Surfoce Wind Speed (m/i)
N03 Deposition Velocities (cm/s) (or UNCRA2ED TOREST/WOODLAND Land Use Type
HESOPUFF-n CODu
0 1 2 3456789 iq .01 23456789 10
2
^
S \
0
i
i o
-i
-
1
1 1
5
0
o
s
o
_
"
,
1
'
-
8
O
'
1
' \\» \ \ J
:\VA :
-8\ i
AY \ :
j j
2 2
w
i S i
D
S
0 £ o
-i -i
.
>
3
o
- 8
O
_
c
' '
C
b
O
O
0
u>
3
b
; o i
; \ \
!
c
p
b
T
'\\\,
' '
0
o
is
i
i
,
.
~
0
i
P
yr
i
i
^
>
-
-
12345678910 t 1 2 345676810'
Surface Wind Speed (m/t) Surfoce Wind Speed (m/s)
NO3 Deposition Velocities (cm/s) for LAKE OR OCEAN Lond Use Type.
FIGURE 3-36. Comparison of MESOPUFF-II and CCADM predicted
nitrates dry deposition velocities for three land use classes
85
-------
3.5 WET DEPOSITION
The wet removal of pollutants consists of both in-cloud scavenging (rain-
out) and belovf-cloud scavenging (washout). Many factors contribute to the
scavenging rate of pollutants, Including pollutant type, cloud type and
history, and the precipitation rate. It 1s generally believed that the
scavenging of partlculates, such as sulfate and nitrate aerosols, is irre-
versible and the scavenging of gaseous species 1s reversible (NRC,
1983). Reversible scavenging refers to the release of contaminants from a
rain droplet back Into the atmosphere before the rain drop Impacts the
ground. We now briefly discuss the wet deposition parameterizations with-
in the three candidate models that treat wet scavenging. For a more com-
plete description of these algorithms and a review of the processes that
lead to wet deposition see Morris and Kessler (1987).
3.5.1 Review of the Wet Deposition Algorithms
1n the Candidate Models
3.5.1.1 CCADM
Of the candidate models, the CCADM 1s the only model with a wet deposition
algorithm to treat both reversible and Irreversible scavenging. The cal-
culation of ralnout for the gaseous species 1n the CCADM relies on the
gaseous-liquid equilibrium component of the aqueous-phase chemistry module
(Gery et al., 1987). Partlculate species are assumed to be totally in the
liquid state (I.e. complete nucleatlon for aerosols) within the cloud.
Washout of partlculates 1s parameterized using the algorithms of Scott
(1978). Gaseous species are washed out assuming that the species concen-
trations within the raindrop are in gaseous-liquid equilibrium with the
ambient air as the raindrop falls. Thus 1t 1s possible for some of the
gaseous species Inside the raindrop to be released back into the atmo-
sphere. The calculation of the gaseous-liquid equilbrium is dependent on
the species concentrations, the temperature, and the pH of the cloud
water. Since this algorithm requires knowledge of the total concentration
of all species that contribute to the cloud pH, and all species for which
scavenging rates are being calculated, use of this algorithm in a Lagran-
glan puff model 1s not appropriate.
3.5.1.2 MESOPUFF-II
The MESOPUFF-II and the RIVAD both contain simplified wet deposition
algorithms that are consistent with a Lagrangian puff model. The MESO-
PUFF-II uses the scavenging coefficient approach to calculate the wet
86
-------
deposition of r>02, S04, NOX, HN03 and N03. This approach assumes that the
loss of pollutant mass over one time step, At (s), due to a precipitation
rate, R (mm/h), 1s expressed as follows:
Q(t + At) = Q(t) exp(- A At)
where Q(t) and Q(t + At) represent the mass of the pollutant at the begin-
ning and end of the time step, and A (s~*) 1s the scavenging ratio expres-
sed as: A =
Here Rj 1s a reference rainfall rate (1 mm/h) and x 1s the scavenging
coefficient (s ) whose value depends on the species and whether the pre-
cipitation 1s liquid or frozen. The MESOPUFF-II uses the rainfall rate
and type from the nearest observation site.
3.5.1.3 RIVAD
The parameterization of wet deposition of gases in the RIVAD follows the
method suggested by Hales and Sutter (1973). The wet deposition rate con-
stant, W (1/h), for a gaseous species 1s defined as follows:
w _ 24.026 R
h(0.00667 R/V + SOL)
where R = precipitation rate (m/h)
h = plume parcel depth (m)
V = raindrop velocity (m/s)
SOL = species-dependent solubility parameter based on Henry's law
constant and the cloud pH (assumed to be 4.5 in the RIVAD)
The parameterization of wet deposition of particulates and sulfate aerosol
uses a method proposed by Scott (1978). The irreversible scavenging
algorithm is based on the assumptions that sulfate 1s scavenged within
clouds primarily by cloud droplet nucleation and beneath clouds the
largest aerosols are washed out by impaction. Scott's algorithm, which is
used 1n RIVAD, can be expressed as
14 Mc 0.75 Sn (1 - 4.41 x 10~2 R'0'88)
5 \J «* dk . /
x = 7O8 * (1.56 + 0.44 in R) + °'3 cl ma &tb (~
87
-------
where
x = ratio of sulfate mass to precipitation mass (gram sulfate/
gram
R = rainfall rate (mm/h),
mft = Concentration of subcloud sulfate aerosols greater than 1 vm
1n diameter, assumed equal to 0.1 SQ,
SQ = Subcloud sulfate hydrometeor concentration,
tjj = time required for hydrometeor to fall from cloud base to
ground,
GI = 5.2 x 10~3 for liquid hydrometeor
= 3.7 x 10~3 for frozen hydrometeor
M n 7* <: ovn (4.41 x IP"2 -R°-88)(435 R'0'71 + 1200)
Ms - 0.75 S0 exp (8450 - 2383 in R)
For convective clouds or clouds whose tops
are warmer than 0° C
= 0.1 SQ For layer clouds not dependent on Bergeron
process for rain initiation
= 0 For layer clouds dependent on Bergeron
process for rain initiation.
The RIVAD is currently configured for layer clouds dependent on Bergeron
process for rain initiation.
3.5.2 Evaluation of the Wet Deposition Algorithms
Of the three candidate models that treat wet deposition, both the MESO-
PUFF-II and the RIVAD wet deposition formulation are consistent with the
Lagrangian puff model framework. Although the two methods usedscaveng
1ng coefficient and solubility approachare basically different, they
have some similarities. First, sulfate and nitrate aeorosols are both
scavenged at the same rate by the MESOPUFF-II and the RIVAD. Second,
neither model treats gaseous scavenging as irreversible. Third, both
88
-------
parameterizations combine 1n-cloud scavenging (rainout) and below-cloud
scavenging (washout) into one scavenging rate. Finally, these parameteri-
zations are linear and the scavenging rate depends only on the precipita-
tion rate and species type and for MESOPUFF-II, whether the precipitation
1s liquid or frozen, not on the species concentrations.
3.5.2.1 Sulfur Dioxide
The S02 wet scavenging rates produced by the MESOPUFF-II and RIVAD models
as a function of precipitation rate are shown in Figure 3-37a. The MESO-
PUFF-II assumes that S02 1s not scavenged by a frozen hydrometer. Despite
the differences 1n their formulations, the shapes of the curves for the
two models are very similar. However, the MESOPUFF-II produced wet
scavenging rates that are approximately twice those of the RIVAD model.
Due to the lack of quantitative measurements, 1t cannot be determined
whether one algorithm 1s predicting a more accurate scavenging rate than
the other.
3.5.2.2 Sulfate
In both the MESOPUFF-II and the RIVAD the scavenging of particulates is
calculated based on the scavenging rate calculated for sulfate. Thus the
evaluation of the scavenging rates for sulfates also applies to nitrates
and particulate matter species. The wet sulfate scavenging rates for the
two models as a function of precipitation rate are given 1n Figure 3-
37b. Given the differences in their formulations, the similiarlty of the
sulfate scavenging rates produced by the two models for liquid precipita-
tion 1s quite encouraging. For precipitation rates below 0.1 1n/h, the
RIVAD produces a rate that is about 10 %/h higher than the rate given by
the MESOPUFF-II. The MESOPUFF-II wet scavenging rate for the frozen case
1s much lower than that for the liquid case, reflecting the fact that it
1s difficult for the particles to become embedded in ice crystals except
through the process of rimming.
3.5.2.3 Nitrogen Oxides
The MESOPUFF-II assumes that NOX is not scavenged through the process of
precipitation. This is verified by the RIVAD model, which produces very
low NOX scavenging rates (Figure 3-38a). This is because N02 and, even
more so, NO are both not very soluble and have a very low Henry's Law con-
stant.
89
-------
(a)
(b)
1 KJU
90
00
^ 70
\'t
rr. G0
I "
c 50
5 40
10
* 30
20
10
0.C
' 502
-
1' " 1 i i TIM
01 0.
Tree i p i
/'
/ /_
//:
/ /
/ / :
/ /
/ /
^^^ '
1 1 1 1 1 1 Ml 1 1 1 1 1 1 1 II
01 0.1 1 .
Lai i on Ralg ( i n/h )
1 V1V1 IWW
90 90
00 00
70 ^ 70
ID
C0 CT G0
50 ^ 50
40 ^ 40
30 £ 30
20 20
10 10
017
la
0.C
- 504
-
-
-
-
-
-
^:.^''- '
1 1 1 1 Mill
501 0.
P rec i p i
P
i\
M [
/ /
/ '
/ 1
/ ''
/
1 1 1 Mill
01 0.
Lalion Rale
i u n r\
1 V H I1
~ o n m i r r i i i
:
/ -
/ :
-
-
-
-
-
i i i M i M|
i i.
m/h)
. .-.,.. -i \
I 13 V}
90
00
70
G0
50
40
30
20
10
11 i r
FIGURE 3-37. Sensitivity of the MESOPUFF-II and RIVAD wet
scavenging rates to precipitation rates for (a) S02 and (b) sulfate.
-------
(a)
(b)
0.0008
0>
-l->
c
O>
0.0004
0.0002
0. 0000
NOx
0. 00
i I i i i i 11 l I I i i i i 11
0.01 0.1 1
Precipitation Rate (in/h)
-------
3.5.2.4 NHrlc Add
Nitric add is very soluble. The solubility parameter used for nitric
add 1n the RIVAD 1s approximately 14 magnitudes greater than for NOX and
8 magnitudes greater than for S02. Thus the RIVAD model calculates a wet
scavenging rate for nitric acid of 100 %/h for precipitation rates as low
as 0.0001 1n/h. The MESOPUFF-II produces scavenging rates that are com-
parable to those produced for sulfate (Figure 3-38b). The calculation of
a wet scavenging rate of 100 %/h regardless of the precipitation rate is a
little suspicious; however, because of the high solublity and reactivity
of nitric add 1t cannot be discounted.
3.5.3 Remarks
This discussion on wet deposition has deliberately been restricted to the
calculation of scavenging rates for given precipitation rates. The pro-
cess of wet deposition of pollutants Involves the complex interaction of
cloud physics, Including entrainment, venting, vertical tranport within
the cloud, and cloud mlcrophysics (phase changes of H20 between gas, 1ce,
cloud water, and rain water), aqueous- and gas-phase chemistry, as well as
wet scavenging. In addition, there are several other issues relating to
the modeling of wet deposition, Including the representation of the
patchlness of clouds and cloud ensembles. Current research, by groups
such as the Regional Add Deposition Model (RADM) team, 1s underway to
develop modeling techniques for dealing with these Issues. However, for
the purposes of developing an acid deposition model capable of estimating
annual add deposition Increments from specified sources 1n a cost-effec-
tive manner, rigorous treatment of all these processes is Impossible.
92
-------
DESIGN OF THE METEOROLOGICAL MODEL
The heart of the new mesoscale meteorological model 1s the wind field
generation algorithm. Of the diagnostic wind models reviewed, no one
model appears to be clearly superior to the other models. If there were 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-MET wind model 1s attractive because
of Us ability to represent blocking and deflection 1n a cost-effective
manner. However, the MELSAR, ATMOS1, and CIT wind models all require
meteorological measurements to infer any dynamic properties 1n the wind
field.
In addition to wind fields, an add deposition model requires other
meteorological Inputs, Including boundary layer heights, temperatures,
temperature lapse rates, relative humidities, stability, and such micro-
meteorological variables as friction velocity and Monin-Obukhov length.
The only candidate model that also generates fields of such meteorological
variables 1s the MELSAR-MET. The MELSAR-MET 1s coded in a highly modular
fashion, which allows for ease of addition, replacement, or modification
of any existing module. Thus the mesoscale meteorological model for the
Rocky Mountains will make use of the MELSAR-MET code as a basis for
generating wind fields and other meteorological variables needed for acid
deposition modeling 1n complex terrain.
Rather than adopt an existing wind model, we have elected to design a new
model that combines features from several existing diagnostic wind
models. This wind model would utilize all existing wind observations,
while simulating the effects of complex terrain in data-sparse sub-
regions. The design and formulation of the wind model 1s discussed 1n
detail 1n Section 4.1. A preliminary evaluation of the model is reported
1n Section 4.2. The model is first evaluated using the same tests as for
the candidate models; then the model predictions are compared against wind
observations from the Rocky Mountains; finally, the model 1s applied to
entirely different terrain settingsa large valley and a complex
terrain/coastal environment.
93
-------
4.1 THE DIAGNOSTIC WIND MODEL
4.1.1 Design Overview
The diagnostic model 1s used to generate grldded fields of the horizontal
wind components, u and v, at several user-specified vertical levels and at
a specified time. This model will use local surface and upper-air wind
observations, where available, while providing some Information on ter-
rain-generated air flows in regions where local observations are insuffi-
cient to account for terrain effects.
The diagnostic model requires grldded terrain heights, a mean wind value
for the modeling region, and region-average stability Information (dT/dz,
or Pasquill stability class). The model will also accept surface and
upper-air wind observations.
The generation of the wind field Involves two major steps. Step 1 is
based on the approach taken in the SAI Complex Terrain Wind Model, as
described by Liu and Yocke (1980). A mean wind value for the modeling
region is adjusted for the kinematic effects of terrain, thermodynamically
generated slope flows, and blocking effects based on a set of gross
parameterizations of these effects. Step 1 produces a spatially varying
gridded field of u and v at each vertical level.
Step 2 Involves the addition of observational information to the step 1
(u,v) field. An objective analysis scheme 1s used to produce a new
gridded (u,v) field. The scheme 1s designed so that the observations are
weighted relatively heavily 1n subregions where they are deemed represen-
tative of the mesoscale air flow, whereas 1n subregions where observations
are deemed unrepresentative the (u.v)-values from step 1 are weighted
heavily. Once the new gridded (u,v) field 1s generated, it can be
adjusted to mass consistency by the divergence-reduction procedure
described by Goodin and co-workers (1980).
4.1.2 Model Formulation
4.1.2.1 Vertical Coordinates
The diagnostic model is formulated in terrain-parallel vertical coordi-
nates. This allows computation of winds at constant heights above ground,
as well as variable vertical resolution. The horizontal position vari-
ables (x,y) and velocity variables (u,v) are invariant upon transformation
from Cartesian to terrain-parallel coordinates. If h denotes terrain
height, z denotes the Cartesian vertical position variable, and Z denotes
the terrain-parallel position variable, then
94
-------
Z = z - h(x,y) . (4-1)
If w denotes Cartesian vertical velocity, and W denotes terrain-parallel
vertical velocity, then
W = w - u dh/dx - v dh/dy . (4-2)
In terrain-parallel coordinates the incompressible conservation-of-mass
equation becomes
du/dx + dv/dy + dW/dZ = 0 (4-3)
4.1.2.2 Divergence Minimization Procedure
The divergence minimization procedure is utilized in both steps 1 and 2
and thus is described here. This procedure is nearly identical to the
procedure described by Goodin and co-workers (1980). The inputs to the
procedure are a "first-guess" three-dimensional (u,v) field and a three-
dimensional W field defined at points horizontally and vertically stag-
gered with the (u,v) levels. Assuming the W field is invariant, the pro-
cedure performs an iterative adjustment of the (u,v) field until the cen-
tered-difference approximation of the inequality,
du/dx + dv/dy + dW/dZ < c (4-4)
is satisfied at all grid points. E is the maximum allowable three-dimen-
sional divergence specified by the user.
The iterative adjustment is carried out as follows. At each grid point
(1,J,k) the three-dimensional divergence D(i,j,k) is computed,
n _ WiJ.k+l/2 " Wi.J.k-l/2 ui+l.j,k " Vl.j.k
AZ 2Ax
(4-5)
2Ay
where AX and Ay are the horizontal grid spacing in the x and y direction
(assumed constant) and AZ is the vertical layer thickness between k - 1/2
and k + 1/2. Note that W is defined at vertically staggered grid levels.
95
-------
Velocity components at the surrounding grid points are adjusted so that
D(1,jtk) is zero. The adjustment at a given grid point adds divergence at
surrounding grid points; thus the entire grid must be scanned iteratively
until the divergence minimization criterion is met at all points. The
adjustments take the form
j,k) + UT (4-6)
= u(i-lj.k) - U
+ VT
v(i,j-l,k) = v(ij-l.k) - VT
In making this adjustment, it is arbitrarily assumed that Uj = Vj; given
constant horizontal grid spacing, one can then show from Equation 4-5 that
UT = - D Ax/2 . (4-7)
4.1.2.3 Step 1 of Wind Field Generation
Kinematic Effects of Terrain. The treatment of the kinematic effects of
terrain follows the procedure described by Liu and Yocke (1980). Given a
mean wind, V, for the modeling region, and terrain height, h(x,y), a
terrain-forced Cartesian vertical velocity of the following form is
assumed:
w = (V - grad h) exp"kz (4-8)
where k 1s a coefficient of exponential decay that Increases with
atmospheric stability. Liu and Yocke assume that
k = N/|V| (4-9)
where N 1s the Brunt-Vaisala frequency, (g/e) (de/dz); Q is the potential
temperature; and |V| is the magnitude of the mean wind.
In the current model the Cartesian w of Equation 4-8 is transformed to a
terrain-parallel W, as 1n Equation 4-2, using the mean wind for the
modeling region. Thus dW/dZ = dw/dz. Assuming the mean wind as a first-
guess gridded (u,v) field, the divergence minimization scheme is exercised
to produce a gridded wind field, (u,v)k, adjusted for the kinematic
effects of terrain.
96
-------
Slope Flows. At each grid point 1n regions of complex terrain, the diag-
nostic model computes a slope flow vector (u,v) . This vector is added tc
the grldded wind field (u,v)k to obtain a new field (u,v)ks.
Let hx and hy denote 3h/dx and ah/dy, respectively. We define the slope
angle, a,
2 + h..2]** . (4-10)
X
The drainage direction, e^, is computed as shown by Allwine and Whiteman
(1985). An angle, B', 1s defined as
8' = tan-1(hy/hx) . (4-11)
A second angle, B", is defined from the following table:
Condition hx = 0 hx < 0 hx > 0
hu
y
hy
nv
y
= 0
< 0
> 0
*
270
90
B1 -
6' -
B1 -
H 180
t- 180
H 180
B1 +360
B + 360
B1
* Terrain is flat, no drainage direction.
The final definition of B^ (in degrees) 1s
8d = 90 - 6", 0 < B" < 90
(4-12)
Bd = 450 - B", 90 < B" < 360
The slope flow vector 1s oriented 1n the drainage direction. The speed of
the slope flow component is determined by the details of the parameteriza-
tion; in our discussion a positive speed denotes upslope flow.
Analytic solutions for downslope flows under highly idealized conditions
have been obtained by Prandtl (1942) and, more recently, by Mahrt (1982)
and Mtzjarrald (1984). However, analysis of upslope flow has received
much less attention, perhaps because the presence of turbulent mixing over
97
-------
a heated slope complicates the analysis. Although analytic solutions pro-
vide useful Insight into the physics of slope flow, their direct applica-
tion to complex-terrain situations 1s doubtful. Given the crudity of
other parameterlzatlons in the diagnostic model, and given that local air
flow 1s frequently Influenced by terrain features of horizontal scales
significantly smaller than model grid scales, we feel justified in formu-
lating a relatively arbitrary parameterization of slope flow effects.
The speed, S, of the parameterized slope flow 1s defined as follows:
S = S0 x f^t) x f2(Z,t) x f3(a) . (4-13)
S0 1s an arbitrarily specified slope flow "amplitude"; this 1s a region-
average parameter that 1s an estimate of the maximum speed of the slope
flow. The function f^ is a specified function of time of day that, in
general, will be assigned a value of -1 for fully developed downslope flow
and +1 for fully developed upslope flow. It may be allowed to vary during
periods of transition. The function f2 1s a vertical profile function.
Loosely guided by the Prandtl analytic solution for slope flow (as
presented by Rao and Snodgrass, 1981), the following expression is
proposed for f2:
f2(Z/ls) = A sin(Z/ls) exp(-Z/ls) (4-14)
Here ls 1s a vertical scale length for the slope flow, and
A « 0.707 exp(-it/4) (4-15)
normalizes f2 so that Its maximum value 1s 1. Note that we have substi-
tuted the terrain-parallel vertical coordinate Z for Prandtl's slope-nor-
mal coordinate n. Note also that the expression for f2 allows for an
overlying layer of reverse flow considerably weaker than the ground-based
slope flow. The depth of the ground-based slope flow layer is * x ls; the
maximum slope flow speed occurs at Z/ls = it/4.
Although an expression for ls is derived as part of Prandtl's solution,
for the purposes of this model ls will be specified arbitrarily as a func-
tion of time. For daytime upslope flow, ls should probably be set equal
to the estimated mixing height; for nighttime downslope flow, 1$ should
probably be set at 50-100 m based on the analyses and numerical experi-
ments of Rao and Snodgrass (1981), Arritt and Pielke (1986), and others.
If the vertical resolution of the model is coarse compared to the estimate
of 1$, f2 can be arbitrarily specified, independent of Equation 4-14, for
each model level as a function of time.
98
-------
The function fg describes the variability of the slope flow speed with
slope angle. The numerical simulations of Arritt and Pielke (1986) indi-
cate that the slope flow intensity is relatively insensitive to slope
angle when the angle is between 5 and 20 degrees, and that slope flow is
virtually absent for slope angles of 1 degree. On the basis of these
results, we propose for f3 the form
f? = a/an» a < an
3 . ° ° (4-16)
f3 = 1, a > a0
where aQ is an arbitrarily specified angle somewhere between 1 and 5
degrees. This slope flow parameterization does not at this time account
for nonlinear Interaction of slope flow with ambient flow. Again we
justify this omission on the basis of the one-dimensional downslope flow
simulations reported by Arritt and Pielke (1986); the range of results
obtained by Arritt for a variety of ambient flows seems well within the
uncertainty of this crude parameterization technique in complex terrain.
Terrain Blocking Effects. The treatment of the blocking effects of ter-
rain 1n the diagnostic model follows the procedure described by Allwine
and Whiteman (1985). From the gridded wind field, (u,v), the available
atmospheric stability information, and the gridded terrain heights, a
local Froude number,
Fr = S / N Ah (4-17)
1s computed at each grid point. Here S 1s the grid-point wind speed, N is
the Brunt-Vaisala frequency as defined in Equation 4-9, and Ah is the
"effective obstacle height" at the given grid point. If Fr is less than a
critical Froude number, Frc (usually equal to 1), and (u,v)ks at the given
grid point has an uphill component, (u,v)ks 1s adjusted so that the flow
1s 1n a terrain-tangent direction, with no change in speed. If Fr > Frc,
the flow is not adjusted. Thus a new gridded wind field (u,v)j is
obtained that reflects both terrain kinematic effects and thermodynamic
blocking effects.
We assume that
Ah(x.y.Z) = hmax(x,y) - z(x,y,Z) (4-18)
where hmax is the elevation (MSL or above some reference height) of the
"obstacle top" and z is the elevation of the grid point. The assignment
of a value to hmax in regions of complex terrain can be somewhat arbi-
trary. One option is to assume that hmax(x,y) is the largest value of the
99
-------
terrain height, h, within a specified radius of the given grid point; this
radius should be determined by the dominant horizontal scale of the ter-
rain. A second option is simply to subjectively assign a value of hmax to
each grid point. Both options will be available in this model.
4.1.2.4 Step 2 of Wind Field Generation
Step 2 of the diagnostic model combines the gridded wind field (u,v)i
generated in step 1 with available observed data to produce a "final"
gridded wind field (u.v^. This Involves four substeps: (1) objective
analysis; (2) smoothing of the analyzed field; (3) computation of a verti-
cal velocity field; and (4) minimization of three-dimensional divergence.
Objective Analysis. The objective analysis procedure is a modified
inverse-distance weighting scheme based on procedures utilized by Goodin
and others (1980), Godden and Lurmann (1983), and Ross and Smith (1986)
It is carried out separately for each model level. It is assumed that all
surface wind observations will be Incorporated at the lowest model
level. A preprocessor program will Interpolate upper-air observations
vertically and temporally so that "soundings" of u and v are defined at
all model levels at a given horizontal location.
For the purpose of discussion, (UQ.VQ)^ denotes an observed wind at sta-
tion k, and r^ denotes the horizontal distance from station k to a given
grid point. At each grid point, the wind vector is thus updated:
(u.v)' = Llr" (u.v)J + RTn (u.vjJ/ISr-" + Rn (4-19)
0.0 .
k J I k
This procedure weights the step 1 wind field, (u,v)j, heavily in regions
far removed from observations; the degree of influence exerted by (u,v)j
1s Inversely related to the value of the parameter R^. The exponent, n,
controls the relative influence of observations distant from a given grid
point. Goodin and co-workers suggest that n should be 2 for a relatively
dense set of surface observations, and 1 for a relatively sparse set of
upper-air observations.
Several constraints can be placed on the evaluation of Equation 4-19 at
the option of the user. A maximum radius of influence, Rmaxt may be
specified; if rk > R^x, the observation at station k is excluded from
Equation 4-19. If observations are densely spaced and representative of
the spatial variability of the air flow, Rmax should be relatively small;
otherwise, evaluation of Equation 5-18 may result in unwanted smoothing
effects.
100
-------
Alternatively, a parameter K,,,^ may be specified that allows inclusion in
Equation 4-19 of only the K^ closest stations. With this parameter the
effective maximum radius of influence can increase or decrease depending
on local observation density.
Finally, the user may construct barriers by specifying end points of line
segments in (x,y) space; if a specified barrier lies between a station and
a given grid point, that station is not considered when evaluating Equa-
tion 4-19. This technique can be used to reduce or eliminate deleterious
effects on the analysis of stations heavily influenced by local terrain
features (e.g., a canyon).
The parameters n, Rj, R^*, and 1^^ as just defined are specified separ-
ately for surface and upper-air observations. Each barrier specification
will include a specification of the maximum model vertical grid level at
which the barrier will be applied.
Smoothing of the Analyzed Wind Field. Goodin and co-workers (1980) indi-
cate the desirability of smoothing the gridded wind field resulting from
Equation 4-19. Thus, a simple five-point smoothing of the form
l.j) + A(i - l.j)
+ A(i,j - 1) + A(i,j + 1)] (4-20)
may be applied to (u,v)'. Although Goodin and co-workers indicate that
the amount of smoothing should be an increasing function of atmospheric
stability, we prefer to simply specify the number of smoothing passes
(usually no more than three) at each model vertical level. Smoothing of
the gridded wind field speeds up the divergence minimization procedure.
However, it should also be noted that overuse of such smoothing can obli-
terate important air flow features (e.g., a well-defined sea-breeze con-
vergence zone).
Computation of Vertical Velocity. An initial field of vertical velocities
in terrain-parallel coordinates, W, is computed from (u,v)' by integra-
ting the equation for incompressible conservation-of-mass (Equation 4-
3). The resulting three-dimensional velocity field is thus mass-consis-
tent. However, Godden and Lurmann (1983) note that vertical velocities
obtained from objectively analyzed (u,v) fields may be unrealistically
large near the top of the model domain. Godden and Lurmann utilize a pro-
cedure suggested by O'Brien (1970) to modify W:
W2(Z) = W'(Z) - (Z/Ztop) W^op (4-21)
Note that W2 is zero at the model top and is not mass-consistent with
(u,v)'. We believe that there may be situations in which utilization of
101
-------
the O'Brien procedure may not be desirable; for example, the model top may
pass through a well-resolved sea-breeze convergence zone within which a
large W value is legitimate (not an issue 1n the Rocky Mountains). Thus,
1n this model, imposition of Equation 4-21 is an option. If Equation 4-21
1s not Invoked, the final product of the model, (u,v)p, is equal to
(u.v)'.
Minimization of Three-Dimensional Divergence. If Equation 4-21 is
invoked, 1t 1s necessary to adjust the objective analysis product, (u,v)',
so that 1t 1s mass consistent with W2. The divergence minimization pro-
cedure described at the beginning of Section 4.1.2.2 is exercised with
(u,v)' as a first-guess horizontal wind field and with W2 held constant.
The adjusted horizontal wind field, (u,v)2, is the final product of the
diagnostic model.
4.2 EVALUATION OF THE DIAGNOSTIC WIND MODEL
We carried out a preliminary evaluation of the new diagnostic wind model
(DWM) 1n the same manner as our evaluation of the candidate wind models
(see Section 2). The DWM was exercised on both the hypothetical bell-
shaped mountain described in the review report (Section 5.1.2; Morris and
Kessler, 1987) and the Rocky Mountain terrain (see Section 2 here). For
these tests the model was run without Input of mesoscale observational
data, with an initially uniform flow.
The new DWM 1s also evaluated using observations from the Rocky Mountain
region. These simulations illustrate the ability of the new DWM to
assimilate observational data into its mesoscale wind field. Simulations
are carried out with all wind observations, and then without some of the
observations so that the model predictions can be compared with observa-
tions not used 1n Its wind field generation procedure.
As an Illustration of the applicability of the new DWM, the DWM was exer-
cised for two completely new and different terrain configurations. The
first 1s a coastal/complex terrain environment with a dense network of
wind observations; the second is within a large valley where the wind pat-
tern 1s dominated by complicated slope flows. In the latter simulation
the flexibility and adaptability of the new DWM is further illustrated by
the use of output from two-dimensional simulations of a primitive equation
mesoscale meteorological model used as input to the DWM.
4.2.1 Flow Over Idealized Terrain
Three simulations of flow over a bell-shaped mountain were carried out.
In simulation Al an Isothermal atmosphere is assumed and slope flow
102
-------
effects are excluded. Simulation A2 is Identical to simulation Al except
that slope flow effects are added, with the slope flow parameterization
tuned to produce maximum downs lope flow of approximately 2 m s~* at the 50
m level only. Simulation A3 is similar to simulation A2 except that the
atmosphere is assumed to be neutrally stratified and the slope flow para-
meterization is tuned to produce maximum upslope flow of approximately 2 m
s"1 at the 50 m level.
As in the previous tests, an initially uniform flow of 2 m s'1 from the
West (270°) was specified. The model grid specifications are identical to
those reported in the review report (Morris and Kessler, 1987).
Figure 4-1 shows the wind fields predicted by DWM under simulation Al at
50, 200, and 500 m above ground. The parameterization of blocking effects
produces marked deflection of flow upwind of the mountain top at each of
these levels, while the mountain top acceleration due to terrain kinematic
effects 1s most apparent at the 50 m level. Although the blocking
parameterization in DWM 1s essentially that of MELSAR, the results in
Figure 4-1 differ substantially from those of the MELSAR simulation
(depicted in Figure 5-2 of the review report) for two reasons: (1) the
methods of computing obstacle heights 1n the two models are different; and
(2) the MELSAR polynomial representation of the wind field tends to act as
a smoother.
Figure 4-2 shows the results of simulation A2. At the 50 m level results
are qualitatively similar to those obtained with the CTWM for downslope
flow (Figure 5-6 of the review report) although the slope flow magnitude
1s weaker. However, at the 200 m and 500 m levels the results of simula-
tion A2 are nearly Identical to those of simulation Al; the spurious
"return" flow produced by the transformed CTWM does not appear 1n the cor-
responding DWM simulation.
Model results for simulation A3 at the 50 m level reflect the combination
of the kinematic effects of terrain and the imposed upslope flow (Figure
4-3). The levels above are minimally effected; blocking effects are
absent given the assumed neutral stratification.
4.2.2 Flow Over Rocky Mountain Terrain
We carried out three DWM simulations of flow over the Rocky Mountain
domain depicted in Figure 2-1. Simulations Bl, B2, and B3 are identical
respectively to simulations Al, A2, and A3 in idealized terrain except
that the Rocky Mountain terrain depicted in Figure 2-1 is substituted for
the bell-shaped mountain. Grid specifications and initial conditions are
as described in Section 2.2.
Results of simulation Bl are depicted in Figure 4-4. Simulation Bl is
most directly comparable to the MELSAR and ATMOS1 simulations depicted in
103
-------
0 5 10 15
NORTH
20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0 3 VD 15T20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
SOUTH
95
DWM WIND VECTORS AT LEVEL -
I"11!""!""!
0 5 10 15
WIND SPEED (M/S)
FIGURE 4-la. Winds generated by the Diagnostic Wind Model for
simulation Al at 50 m above ground. Scaling of plotted winds
is given at lower left. Topography is contoured in meters.
Horizontal grid spacing is 5 km.
104
-------
NORTH
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
3 ID 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
DWM WIND VECTORS AT LEVEL - 2
0 5 10 15
WIND SPEED (M/S)
FIGURE 4-lb. At 200 m.
105
-------
NORTH
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 BO
90 95
9 10 15 20 25> 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
DWM WIND VECTORS AT LEVEL
I""!""!""!
0 5 10 15
WIND SPEED (M/S)
FIGURE 4-lc. At 500 m.
106
-------
0 5 10 15 20 25 30
NORTH
35 40 45 50 55 60 65 70 75 BO S5 90 95
0 3 1t> 15 26 25 30
35 40 45 50 55 60 65 70 75 60 85 90 95
SOUTH
DWM WIND VECTORS AT LEVEL - 1
0 5 10 15
WIND SPEED (M/S)
FIGURE 4-2a. Winds generated by the Diagnostic Wind Model for
simulation A2 at 50 m above ground. Scaling of plotted winds
is given at lower left. Topography is contoured in meters.
Horizontal grid spacing is 5 km.
107
-------
NORTH
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
9 1t) 15 20 25. 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
DWM WIND VECTORS AT LEVEL
0 5 10 15
WIND SPEED (M/S)
FIGURE 4-2b. At 200 m.
108
-------
NORTH
0 5 10 15 20 25 30 35 4Q 45 50 55 60 65 70 75 80 85 90 95
25 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
DWM WIND VECTORS AT LEVEL
I Illl Mil II I I III
05 10 15
WIND SPEED (M/S)
- 3
FIGURE 4-2c. At 500 m.
109
-------
NORTH
0 5 10 15 20 25 30 35 40 45 50 55 60
65 70 75 BO 85 90 95
"0 5 10 15 20 25 30 35 40 45 50 55 60
SOUTH
65 70 75 80 85 90 95
DWM WIND VECTORS AT LEVEL - 1
I11"!"11!""!
0 5 10 15
WIND SPEED (M/S)
FIGURE 4-3a. Winds generated by the Diagnostic Wind Model for
simulation A3 at 50 m above ground. Scaling of plotted winds
1s given at lower left. Topography is contoured in meters. i
Horizontal grid spacing is 5 km.
110
-------
NORTH
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
i i i i i i i i i i i i
::v:::\
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
SOUTH
DWM WIND VECTORS AT LEVEL - 2
0 5 10 15
WIND SPEED (M/S)
FIGURE 4-3b. At 200 m.
Ill
-------
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
DWM WIND VECTORS AT LEVEL
I""!""!""!
05 10 15
WIND SPEED (M/S)
- 3
FIGURE 4-3c. At 500 m.
112
-------
650
NORTH
700
750
BOO
850
450C
4450
en
440i
435
430C
t ~f\' ' ~ ' r ' <£. r r f f
- »( / \ r *-t /.//////
4500
4450
- 4400
4350
700
750 800
SOUTH
850
4300
DWM WIND VECTORS AT LEVEL - 1
0 5 10 15
WIND SPEED (M/S)
FIGURE 4-4a. Winds generated by the Diagnostic Wind Model for
simulation Bl at 50 m above ground. Scaling of plotted winds
is given at lower left. Topography is contoured in meters.
Horizontal grid spacing is 10 km.
113
-------
650
700
NORTH
750 800
850
| ' ' /
t - />\ f * / / r r *£
/ 'i /
» X <" 9f X -»->« V^ff
4500
-^ 4450
- 4400
4350
430,
750 800
SOUTH
850
4300
DWM WIND VECTORS AT LEVEL - 2
I""!""!""!
05 10 15
WIND SPEED (M/S)
FIGURE 4-4b. At 200 m.
114
-------
650
700
NORTH
750 BOO
850
4500
-^ 4450
- 4400
4350
T50
700
750 800
SOUTH
850
4300
DWM WIND VECTORS AT LEVEL
I II II [I II Mil III
05 10 15
WIND SPEED (M/S)
- 3
FIGURE 4-4c. At 500 m.
115
-------
Figures 2-3 and 2-4, respectively. Like ATMOS1, and unlike MELSAR, the
DWM appears to respond on the characteristic horizontal terrain scales.
The differences 1n the ATMOS1 and DWM solutions are most apparent to the
lee of terrain obstacles, for two reasons: (1) the DWM includes a direct
parameterization of blocking that operates upstream but not downstream of
an obstacle; and (2) the DWM includes a smoothing operation, while ATMOS1
does not.
Results of simulations B2 and B3 are depicted 1n Figures 4-5 and 4-6
respectively. In simulation B2 downslope flow vectors are added at the 50
m level to the corresponding field obtained in simulation Bl; at upper
levels the wind fields from simulations Bl and B2 are nearly identical.
In simulation B3 the lowest level reflects the addition of upslope flow
vectors. Upper levels are essentially undisturbed; the parameterlzations
of terrain kinematic effects and blocking effects are essentially inopera-
tive in the assumed neutral atmosphere.
4.2.3 Evaluation of the New DWM Using Observations
from the Rocky Mountains
Using the same mesoscale domain 1n the Rocky Mountains that was used in
the previous tests, we exercised the new DWM with surface and upper-air
measurements. The DWM was exercised from 1600 on 17 September 1984 to
1500 on 18 September 1984 to produce hourly grldded wind fields in six
vertical layers. This period was selected because of the availabllty of
supplementary radiosonde observations at three sites within the mesoscale
modeling domain. These three supplemental observations (Meeker, Rangely,
and Rifle, CO) were collected as part of the Atmospheric Studies in Com-
plex Terrain (ASCOT) Brush Creek experiments. Although many more meteoro-
logical measurements were available within Brush Creek Canyon Itself, this
canyon 1s very narrow and measurements made within it are only applicable
to the canyon.
The DWM was exercised twice for each hour of the 24-hour period, once
using the routine NWS data only and once with the supplemental data. In
this manner the DWM can be evaluated by qualitatively comparing the wind
fields generated with and without the supplemental data and performing a
quantitative performance evaluation of the DWM by comparing the predicted
wind speeds and wind direction from the simulation without the supple-
mental data with the supplemental data.
The ASCOT Brush Creek experiments were designed to study drainage winds
in the Brush Creek canyon. The formation of drainage winds generally
requires clear stagnant nights. If there is significant synoptic flow
1t will over power the drainage winds.
116
-------
650
700
NORTH
750 800
B50
4500 ^
445C
440
4350
430C
- 4500
4450
- 4400
700
800
850
SOUTH
4350
4300
DWM WIND VECTORS AT LEVEL - 1
I""!""!""!
0 5 10 15
WIND SPEED (M/S)
FIGURE 4-5a. Winds generated by the Diagnostic Wind Model
for simulation B2 at 50 m above ground. Scaling of plotted
winds is given at lower left. Topography is contoured in
meters. Horizontal grid spacing is 10 km.
117
-------
NORTH
650
700
750
800
850
4500
4450 «-
en
440
4350
430C
4500
^ 4450
- 4400
4350
700
750 800
SOUTH
850
4300
DWM WIND VECTORS AT LEVEL - 2
0 5 10 15
WIND SPEED (M/S)
FIGURE 4-5b. At 200 m.
118
-------
650
NORTH
700
750
800
850
4500 -
4450 -
440
4350
430£
4500
4450
- 4400
4350
700
750
600
850
4300
SOUTH
DWM WIND VECTORS AT LEVEL
i ii 1111111111111
0 5 10 15
WIND SPEED (M/S)
- 3
FIGURE 4-5c. At 500 m.
119
-------
NORTH
650
700
750
800
850
4500
4450L
435C *
430C
4500
4450
440C s^-
- 4400
4350
700
750 BOO
SOUTH
850
4300
DWM WIND VECTORS AT LEVEL - 1
05 10 15
WIND SPEED (M/S)
FIGURE 4-6a. Winds generated by the Diagnostic Wind Model for
simulation B3 at 50 m above ground. Scaling of ploted winds
is given at lower left. Topography is contoured in meters.
Horizontal grid spacing is 10 km.
120
-------
650
700
NORTH
750 BOO
850
4500
4450 -
440
4350 t
430C
4500
4450
LJ
- 4400
4350
700
750
800
850
4300
SOUTH
DWM WIND VECTORS AT LEVEL - 2
0 5 10 15
WIND SPEED (M/S)
FIGURE 4-6b. At 200 m.
121
-------
650
700
NORTH
750 BOO
850
4500 '-
4450 £
440
435C fr
430£
f /JS *****!**&**** s\ * t\
*(* * -*~J ₯* * fi * t * f }> f f
- * V«. * J ****** *^<
4500
4450
- 4400
4350
700
750 800
SOUTH
850
4300
DWM WIND VECTORS AT LEVEL
I""!1"1!"11!
0 5 10 15
WIND SPEED (M/S)
- 3
FIGURE 4-6c. At 500 m.
122
-------
4.2.3.1 Qualitative Evaluation
Figure 4-7 Illustrates the DWM-generated wind fields at the six vertical
levels at 0500 on 18 September 1984. Figure 4-7 shows the wind fields
generated by the DWM 1n the simulations that used all meteorological
observations and the simulations that used just the routine NWS meteoro-
logical data. The routine NWS data within the mesoscale modeling domain
consisted of an upper-air sounding at Grand Junction (GNDJ) and a surface
site at Eagle (EAGL). The three supplemental radiosonde observation sites
were located at Meeker (MKR), Rifle (RFL) and Rangely (RNG), Colorado.
Additional meteorological observations outside of the mesoscale domain
were used as Input: Lander, Wyoming (to the north) and Denver, Colarado
(to the east).
Surface wind fields generated by the DWM with and without the supplemental
radiosonde observations are Identical. This is because the supplemental
observations did not Include any observations near the surface. The sur-
face wind fields show significant downslope flow from all of the major
terrain features.
For the higher levels the effects of the supplemental data on the DWM wind
fields can be seen. Of particular note 1s that the Rangely sounding
appears to be calm from 100 to 1,500 m. The Meeker observation shows
winds coming from the southeast at approximately 5 m/s at 100-300 m and
then becoming calm until 1500 m. All the upper-air soundings Indicate low
winds from the southwest at 1500 m. Clearly the power law relationship
used to extrapolate surface wind speeds to wind speeds aloft 1n EPA-
approved models 1s not valid for this time period and location.
Examples of DWM-generated wind fields with and without the supplemental
data at 1400 on 18 September are shown in Figure 4-8. Again the surface
wind fields with and without the supplemental data are identical, only
this time there 1s upslope flow around the terrain obstacles. For the
upper levels the DWM wind fields without the supplemental data are domina-
ted by the Grand Junction sounding, which is recording calm winds away
from the surface. These calm winds are not reflected in the Rangely and
Meeker supplemental observations; thus the wind fields above the surface
are very different 1n the simulations with and without the supplemental
data.
123
-------
With Supplemental Observations
Without Supplemental Observations
INJ
-C.
650
700
NORTH
750 500
850
1500
4450
'3.
700
750 300
SOUTH
SbO
4400
4350
4300
650
700
NO^TH
75,0 300
350
b50
4300
SOOTH
OWM wiNO VECTORS AT LEVEL - 1 (10M)
05 10 15
WINO SPEED (M/3)
Reproduced from
bosl available copy.
FIGURE 4-7a. DWM-generated wind fields at 0500 on 18 September 1984: 10 m.
-------
With Supplemental Observations
Without Supplemental Observations
NORTH
550
4500 -
700
750
300
ro
tn
750 800
SOUTH
- 4500
4450
4400
4350
4300
550
700
POO
B50
r/1
uJ
440^
V.^c'/v^ \V
\ S^\ * ". \ \ 3t.
. *- V Y \ X!, W V '
. x x~^A X -CVx v xx xvx < v. A «> v. K N v
-M:vxxx<^S:->»ooSr>^Sr3~^NX-^vX/< «s ^3)\
.vv(^<^-^r -
15CO
4450
4400
4350
DWM W'MD VECTORS AT LEX/EL - 2 (lOOVt)
0 5 10 '5
WIND iPEE'D (V>I/S)
FIGURE 4-7b. DWM-generated wind fields at 0500 on 18 September 1984: 160 m.
-------
With Supplemental Observations
Without Supplemental Observations
ro
01
650
700
NO3TH
750 200
850
650
700
800
550
450
>^.» ?« L^ v v v -^c-'Ac^ v v v Y* i *
« */,
«. m. V V V f. V *. V **^£- *- * V. "/ f
- -'- - -»«.«. v v vXv^v^-Os^*.|N^vlvJJv r
^^^ , , , V V VV VV VV 4,^ V ^J\
V VK>, *^,'^-
4500
4450
< UJ
LJ i V
4400
4350
4JOO
\
435C
'i\'' \ ' ^' ' ' ».' ^ ^\ ^-*S* * '/A
-^nj:::::^
V ;. V V "s *- f- \ Vc!kV V Y\ \\\ .
4500
4450
4400
4350
700
SOUTH
/"50 3OO
SOU'w
DWM WIND VECTORS AT LEVEL - 3 (500V)
0 5 10 15
WIND SPEED (M/3)
FIGURE 4-7c. DWM-generated wind fields at 0500 on 18 September 1984: 300 m.
-------
With Supplemental Observations
Without Supplemental Observations
550
700
NORTH
750 SOO
NORTH
850
\ «3 ' *x>*
y >_>.2i'w^x ,j - \\~~>^~7r^^ /* )\ '^_^
/ XLv-rr-v ''r^.v «JVr-<-zft>jf 'CO1 * ' '
* { * 1 -\o
-^ x-v ~^-
\ »
^
x V \ ><^\\ /'>^W
O/C^)-
r :>- ^f^r^-^i^^ \l
^VAC1^^' r ^~
v^torr^ \^-,.
:' >, ^
4450
4400
4J50
'650
IV,
,'bO
ajCi
DWV WIND VECTORS Af LEX/EL - 4 (600M)
I""!"11!""!
0 5 10 15
WIIIO SPEED C
FIGURE 4-7d. DWM-generated wind fields at 0500 on 18 September 1984: 600
m.
-------
With Supplemental Observations
Without Supplemental Observations
ro
oo
<>50
700
NORTH
750 800
350
650
700
730
200
350
- 4500
_*>>/ *'i »"»\ » » fc » * » /» ~~** * *
, i , «",» ..!?*< ^ « ( » « \« -
/ x^y A v \ /
&&<*>
/ '-N:x
V^x* --V>X.V
/bO 800
SOUTH
SOU"-1
OV/W WIND VECTOPS AT LE^L - 5 (tOOOM)
I11"!""!""!
0 5 10 15
WIIIO SPEED (W./S)
FIGURE 4-7e. DWM-generated wind fields at 0500 on 18 September 1984: 1000 m.
-------
With Supplemental Observations
Without Supplemental Observations
f\)
IO
650
700
NOSTH
730 300
850
450C
445C
440C
435C
' \ i r '
,-,»., .* i*,'
x' x"v' x' ,' X"' ' }4 .j U ,' .
/^X^OrVviK ^-
x* /" S [S / / ^y'\J' Y .' .
S SSS/S/SS// x^°XX X M /» A,/<
x- ^J V '
x* X* x* x* x* x* x» x^x* x* x' X X X 3" x» p*^f
?,X./'j>^ X x- x» xTxX~X<^tx^/' x* SS^~~S S JS\ / V /-N.
I "V ' ^--^"^jJ ^ A ^ -"* ^* ^ -^--^ ''&fr~>*~'{ S (s S \ ' .
&ye&/fe3»>
650
70C
NOP'H
750 300
4500 450.'
4450 445;
0) '/I
< LJ
4400 440:
4350 435C
650
'50
800
850
4300
'' /'// LSS ? f f* V f
'V--' /;/ >'\U<' ' ' ' ;',,/;../ *<$> ' > > tV. ' '
L-/ /V ////// /,'/ / / /v .' / / / /^/^ ir - .
« t
' f f f /'X/ / \
SCO
650
4450
4400
4350
4.300
3OU"H
DWM WIND veCr
I"1 !'"'|"''|
0 5 10 15
WIMO SPEED 'M/3)
S AT LEVEL - 6 (I500M)
FIGURE 4-7f. DWM-generated wind fields at 0500 on 18 September 1984: 1500 m.
-------
to
O
650
With Supplemental Observations
700 750 800 850
Without Supplemental Observations
450C
445C
01
UJ
$
j,..^-^* ,^^Y#V' '/T
. ^ sr»J. ^* s f*J ^ \.J/s\t* A / / /!>.
"~* -^-^ ~ '
650
700
NORIX
750 500
440C
43 5C
430
D'.vvt WIND VECTORS AT LEVEL - ) (10M)
OS 10 15
WIND SPEED (M/3)
- 45CO 450r
- 4450
4400
4350
4300
4500
- 4450
c 4400
^350
43CO
SOUTH
FIGURE 4-8a. DWM-generated wind fields at 1400 on 18 September 1984: 10 m.
-------
With Supplemental Observations
Without Supplemental Observations
650
700
NORTH
750 300
350
450'.
- « ^f* »*««»«*b»
650
7CO
'50
850
- 4500 450C
4450 445C
^ 4400 44-OC
4350 435C
430C
550
;oo
/50 800
SOUTH
3bO
4300 430P
v. ^'
S */ 3-Q * Is
*s~ * \" ^v?A ** ' '!'('
45CO
4450
4400
4350
ffbO
300
DWM WIND VECTORS Ar LEVEL - 2 (I COM)
05 10 15
WIMD
FIGURE 4-8b. DWM-generated wind fields at 1400 on 18 September 1984: 100 m.
-------
With Supplemental Observations
Without Supplemental Observations
OJ
ro
650
700
450C
4450
440C
43 5C
430;
N03TH
750 SOO
850
650
700
MOPTH
750 300
850
» *JJW * ?-» '» ) ** )'* "* -^ -* -* '
4500 450C
4450 445C r
10
<
uj
4400 44QC- - -v/^
4350 4350 f-
- 4500
700
/50 SCO
SOUTH
350
4300 4300
'^ \) « ^L *^) )» * * * * */ X*^^* " * ^ *\ ^ *
530
700
SOUTH
DV/W V/INO VECTORS AT LEVEL - 3 (500M)
0 <> 1015
WIMD SPEED (M/3)
FIGURE 4-8c. DWM-generated wind fields at 1400 on 18 September 1984: 300 m.
-------
With Supplemental Observations
Without Supplemental Observations
CO
OJ
650
700
NORTH
750 BOO
350
450C
44 5C
440C
435C
430C
f?'? - --C . - -xt If-
650
700
750
300
850
4500 45GC
4450 4450
4j50 435C
700
800
35':
4JOO
,
>^7^^>i u:
I v O l
' % s ^A / x-
»5» *f*» '«* ** »\ (^* * fr ( -
o \ 'V "V i / [
-^/. . \/,i A. . ,-i.i.
4500
4450
4400
4350
50
700
SOUTH
,"50 300
SOUTH
3 DO
4300
DWM WITiD VECTORS AT LEVEL - 4 (600M)
) 5 10 15
WHID SPEED (M/S)
FIGURE 4-8d. DWM-generated wind fields at 1400 on 18 September 1984: 600 m.
-------
With Supplemental Observations
Without Supplemental Observations
650
700
NORTH
750 800
950
450'.
. . . . ./ >-^»* " * »
650
700
MOPTH
750 300
350
.1 - - - 4500 450C
-4450 4450
7) '/i
< uj
LJ J
- 4400 44.0C
4350 435C
4300 4300
V>> J ;/ U,- '
w^ 1 >"^1- o- - - ^-
4500
4450
4400
4350
700
0U "H
WIND VECTOP3 AT LEVEL - 5 (1000M)
0 5 '0 15
WIND SPEE3 (M/
FIGURE 4-8e. DWM-generated wind fields at 1400 on 18 September 1984: 1000 m.
-------
With Supplemental Observations
Without Supplemental Observations
CO
tn
550
700
NORTH
750 300
350
650
700
NORTH
750 300
850
450C
445;
y>
LJ
440C
4350
430
^V /,i£~>*-^~x / A -' ^ > - --r'^ -r:
! / *,xr^' >.' ( >J^1 Q T ' /v) I t_I - - ' 'tl
4500 450C
4450 445C
LJ i
4400 440C
4350 435C
/oo
/SO ''jO'.
SOU TH
330
1300 430't
l^) rs- '>.:::
-
-------
4.2.3.2 Quantitative Evaluation
A preliminary quantitative evaluation of the DWM was made by comparing the
wind vectors predicted 1n the simulation without the supplemental data and
the winds observed 1n the supplemental soundings. A scatter plot of the
predicted versus observed wind speeds 1s shown 1n Figure 4-9. The DWM
underpredlcts the observed wind speed by 0.6 m/s out of an average
observed wind speed of 2.1 m/s. This 1s because the Grand Junction sound-
Ing dominates the upper-level flows predicted by the DWM. Grand Junction
1s located 1n a valley and thus records lower wind speeds. As seen 1n the
scatterplot, the predicted and observed wind speeds do not correlate well
(correlation coefficient 0.037). The stagnant nature of the period simu-
lated 1s shown by the large number (50 percent) of calm winds (< 1 m/s) 1n
the predictions and observations.
The predicted and observed wind directions are compared 1n Figure 4-10.
Figure 4-10a shows the deviation of the predictions from the observations;
Figure 4-10b 1s similar but the calm wind data points have been removed.
As seen 1n Figure 4-10a, for all data, the positive and negative devia-
tions from the observations exactly cancel each other out, resulting in a
zero bias. When the calm wind conditions are removed (Figure 4-10b) there
1s a higher percentage of deviations near zero although there 1s also a
net negative bias of approximately -10 degrees.
The percentage of predicted wind directions within 30 degrees of the
observations 1s 28 percent for all data and 43 percent when the calm winds
are removed. The number of predictions within 60 degrees of the observa-
tions 1s 51 percent for all data and 72 percent without the calm winds.
4.2.3.3 Remarks
This preliminary evaluation of the DWM using observations from the Rocky
Mountains 1s probably the most stringent test that can be designed for the
model. The stagnant conditions that existed during these tests results 1n
slope flows dominating the surface winds while local wind eddies and
fluctuations dominate the observed winds aloft. The lack of good agree-
ment between the predicted and observed wind speeds 1s somewhat dis-
appointing, but the model generally did replicate the stagnant conditions,
with both observed and predicted wind speeds under 4 m/s.
More encouraging was the model's ability to predict wind directions. The
centering of the predicted-observed wind direction residuals on the zero
line with a Gaussian-!1ke profile indicates that deviations of the predic-
tions from the observations are not systematic.
136
-------
4.00 -
Ci
UJ
cc
I.00 -
1.00 2.00 3.00
CESERVED (tr./s)
. 00
CF THE PROBABILITY DENSI'Y FLNCTICN
AVERAGE
DEVIATICN
SKEUNESS
rUR'CSIS
C'hER MEASURES
UFFER CCAR'ILE
LUER CUAR'ILE
VALUE
VALUE
OBSERVED
2.14554
i 1.23589
C. 42619
-C. 73276
1.E97CO
3.C61CO
1. J4CCO
C. 141CC
4.974CG
PREDICTED
1.56241
C.9-68C
1.25574
1.22461
1.3C4CO
2.CCCCC
0.6940C
0.42 4 DC
4.6d9CO
SriUU OF PREDICTION PAPAME'ERS
CCPREUATICN COEFFICIENT OF FREDIC'EC
VERSUS CESERVED C.C32
T|-E BCUfiDS OF "hF CCRREL'.'ICf, AT ThE
CONFIDENCE LEVEL OF C.C5C ARE
LC^ ECLND -0.162 HIGH ECUND 0.225
R.A'IC CF OVER TC I NC-ER FREDIC'ICf.S C . 4 5 I
PEFCEK" OF OVER FREDICTIOKS
GRE*"ER THAN 2CO FERCEN7 OF 'HE
OESER/ED 16.5C5
FERCEK" CF i;.:ER FREDIC'IONS
UESS 'HAN 5C FEROEN' CF ThE
CESERvED 25.922
FIGURE 4-9. Scatterplot and statistics of predicted versus observed
wind speeds at the three supplemental soundings (N=103).
137
-------
(a)
0.20 -
I i I I I
ThE
-.21 ECLAi: 29.CCC
0.20 -
£ O.OS -
(b)
-130.00 -76.CC -2e.CC
RESiClAL (CES-FFEL,
TI-E E:KS;ZE EC^ALS 26.ceo
25.03 73.00 I3C.CO
FIGURE 4-10. Histograms of deviations of predicted
wind direction from observations at the three
supplemental soundings: (a) all data (N = 101),
(b) with calms removed (N = 5).
13C
-------
This evaluation further Illustrates the Importance of having a dense
array of meteorological observations 1n order to reproduce observed wind
fields. Even though the new DWM was designed to run with a sparse set of
observations through the parameterizing of the physical processes that
drive air flows in complex terrain. At any given time these parameteriza-
tlons need to be tied to observations to replicate the observed condi-
tions.
4.2.4 Evaluation of the DWM 1n a Complex Terrain/
Coastal Environment and Within a Large Valley
The new DWM was evaluated using two different modeling regions. The first
1s an area along the California coast that Includes complex terrain. The
second region consists of part of the California Central Valley. The
locations of these two regions are shown 1n Figure 4-11.
4.2.4.1 Complex Terrain and Coastal Environment
Increased activities 1n oil exploration and drilling off of the coast of
California near Santa Barbara has raised questions concerning the effects
of these activities on air quality 1n the South Central Coast Air Basin
(SCCAB) of California. This concern has resulted 1n several federal,
state, and county agencies joining together to sponser a massive meteoro-
logical and air quality measurement collection program known as the South
Central Coast Cooperative Aerometric Monitoring Program (SCCCAMP). This
comprehensive measurement program collected several types of meteorologi-
cal and air quality data during periods of the summers 1n 1984 and 1985.
One of the purposes of this program was to characterize air flow patterns
1n the region to aid 1n the analysis of impacts on air quality in the
SCCAB from future emission sources. The characterization of these air
flow patterns is made particularly difficult because of the combined
effects of the complex terrain of the Santa Ynez, San Rafael, and Santa
Monica mountains and land-sea breezes generated by the Pacific Ocean.
This difficulty 1s further compounded by the fact that most of the wind
measurements are along the coast, thus there are regions with a dense
array of measurements (the coastline) and regions with sparse data (inland
and out to sea). Thus the DWM developed under the auspices of the Rocky
Mountain Acid Deposition Model Project was identified as the most
appropriate diagnostic wind model for predicting wind flow patterns in the
region because of its ability to predict wind flows in areas with and
without measurements in a cost effective manner.
139
-------
550 600 650 700 750 BOO 850 900 950
4300
4250 "-
K/-r OCO
4250
,IV -r 4200
4150
- rLr;r~j;.'.', '' ~~S
. r.T:r'."j:r'~ ' r .
Central Valley
Modeling Region
SCCAB Modeling Region
3800
-- 5650
3?00
550 600 650 [700 750 BOO 650 900 I 950
UTM Easting (Zone 10)
FIGURE 4-11. Locations of the SCCAB and Central Valley modeling
r.egions.
140
-------
The Minerals Management Services, the federal agency responsible for
managing the oil deposits, has funded a study that uses the new DWM to
characterize wind flow pattterns in the region. In addition, other groups
who have interests in the area, such as the Western Oil and Gas Associa-
tion, have also funded efforts to use the new DWM with the SCCCAMP data to
predict gridded wind fields in the SCCAB. At this time, the new DWM has
been used to generate hourly wind fields in the region for 11 case days
from the 1985 SCCCAMP study, and four case days from the 1984 SCCCAMP
study. For the 1985 simulations approximately 80 surface and 20 upper-air
wind observation sites were used as input into the DWM. The 1984 SCCCAMP
data base has fewer observation sites; approximately 20 surface and five
upper-air sites are available.
An example of two hours of the surface wind fields produced by the DWM and
the observations used for the SCCAB region are depicted in Figure 4-12 .and
4-13. Figure 4-12a shows the wind field for 0400 PDT on 23 September
1985. The simulation shows significant downslope flow, which is also
evident in the observations. Also evident 1s the sea breeze coming from
the northwest, which is deflected further south by the downslope winds
coming off of the terrain features inland.
The wind fields for the SCCAB region for 1200 PDT on 23 September are
shown 1n Figure 4-13. During the afternoon both the DWM and the observa-
tions reflect upslope winds 1n the complex terrain region. The sea breeze
circulation around Gavlota Pass (middle left of figure) has formed the so-
called Gaviota eddy. The DWM wind fields match the observations quite
well, which 1s not surprising since they are used as Input Into the model.
These simulations Illustrate the ability of the DWM to make use of many
observations in Its generation of wind fields yet still produce major flow
features (e.g., slope flows and terrain deflection) away from the observa-
tions. This ongoing work effort will be reported on in early 1988.
4.2.4.2 Simulations 1n a Large Valley
In 1986 the U.S. National Park Service sponsored a scoping study to deter-
mine whether ozone concentrations produced by urban areas and oil produc-
tion 1n the California Central Valley could be transported to national
parks 1n the Sierra Nevada mountains (Yosemite, Sequoia, and Kings Canyon
national parks). The regional oxidant model, the RTM-III (Liu, Morris,
and Killus 1984), was deemed the most appropriate tool for this task.
One of the most Important inputs for any regional or mesoscale air quality
model 1s the wind fields. In the California Central Valley elevated ozone
concentrations are usually associated with stagnant conditions in which
141
-------
(a) DWM Wind Fields
NORTH
MO*
J609
7*- 'VWJ **"/, t rz-«,»'»'»'' f s"vvv"^*'V>-*;
: ,:j"-A-^^s.'i'-'-<'>--*^i-'''>>^£.*»">?-?
,--* .' ' ' 'VV^4-;-^---v -'-;-. . .-^^f^f^r^-'J'r,' '&+ S
' - '^' *x~i_;>,..: C\>!\, x^...-.'-;»- - ^--^^cx * f fi^J/ ,-^v
^i^7:C-"t±>'^i-'.'' -7' 1 vv^ii1^" cwr*^.', 'K-' ''^''
x U \ i ' - - - xW^^i^X'k* >/-^' ''"
\\\\\\ \« '* * \ v ;«'''-''» ./VV^^^^-^x^ > / > i--»-»~Vr f^!^i> - ' '-'--'
^ V V V>\S>v W^XA^ \* '*''*'''* ''f' ' ''*'** I I f/X N'\" *4>^^"' >b8
SOUTH
(b) Observed Winds
NORTH
?rM ?NR -VIM .fH J4H
2?B 24B
SOUTH
J:B us
0 '0 ?0 JO 4O
(KM)
0 S 10 15
WIND SPCCD
FIGURE 4-12. DWM generated and observed surface wind fields (10 m) for the
SCCAB Region at 0400 PDT on 23 September 1987.
142
-------
(a) DWM Wind Fields
NORTH
's ^*^ "*~^~ I* ~ * * ^* * * _ - _^^fc-*-y.-^J^*^-*^^ . ..
vv'«.~\^^^-^-^..^;^.-ii=:»=i^=«^rf^»--^.«-» ' ^f^j."^, "^^~»~~»~-«"~""-»'~«~>.v»v»v»>»^»'^.s*'» -
!fs ^ ^xSis ^^ ^'"rff
JOS
JJB
SOUTH
(b) Observed Winds
NORTH
248
768
JOB
32B
J-iB
FIGURE 4-13. DWM-generated and observed surface wind fields (10 m) for the
SCCAB Region at 1200 POT on 23 September 1987.
143
-------
slope flows dominant the flow regime. Thus 1t was initially determined
that 1n order to simulate these complicated slope flows a three-dimen-
sional primitive equation model was required.
A three-dimensional primitive equation mesoscale meteorological model
(Pielke, 1974) was exercised for several hours of the selected oxldant
episode. Although the model produced the slope flows, cost considerations
precluded Its use for generating three-dimensional wind fields for this
study. Thus the primitive equation model was exercised in Its two-
dimensional form to reproduce vertical cross sections of upslope and down-
slope wind fields and boundary layer heights across the valley as repre-
sented 1n Figure 4-14.
Because of the flexibility of the formulation of the initial version of
the DWM, the results of the two-dimensional primitive equation model could
be Input Into the DWM as psuedo-observational soundings. Examples of the
layer 1 and 3 wind fields generated for the California Central Valley at
0100 and 1200 are shown Figures 4-15 and 4-16. As for the Rocky Mountain
and SCCAB regions, the model produced both nighttime downslope and daytime
upslope winds. In addition, due to the psuedo-soundings from the primi-
tive equation model, the model was also able to produce the return flows
In the third layer (see Figures 4-15 and 4-16). Details of the applica-
tion of this Initial version of the DWM to the California Central Valley
have been reported by Moore, Morris, and Daly (1987). The finalized
version of the DWM and the RTM-III are presently being applied to an
expanded region containing the California Central Valley and San Francisco
Bay Area under the sponsorship of the NPS; results are to be reported in
early 1988.
4.3 SPECIFICATION OF OTHER METEOROLOGICAL VARIABLES
The add deposition modeling system will require grid-point estimates of a
number of meteorological variables, based on rather sparse surface and
upper-air observational data. These variables include
mixing height
Pasqu1ll-G1fford stability class
friction velocity
convectlve velocity
Monin-Obukhov length
surface temperature
144
-------
(a) Daytime Upslope Flows
(b) Nighttime Downslope Flows
m
a.
t-
L.
V
it n »
Number of grid cell across valley (easting)
Number of grid cell across valley (easting)
FIGURE 4-14. Depiction of wind circulation air flows and boundary heights in the
California Central Valley generated by the two-dimensional primitive equation.
-------
CTi
» X VrVcM »»»>« «J«. .
»»»»*.»% »<«Y"*--"^'.
V' '"' '^
»\\\>«»>» ^v»sx»\SWJ;>-
i * -^^>»>» -x \ \ \ \ \ v » I I / ^^ £
[* "» ^.A.>x>. XN>>> \>:-« / *"*- * 9
;V-v
20-
19
'.-4^. - - ' «,< »..»»>
^
^ « « *; V H V
*'r ' * ^ ''^.N ^
* f > f'/ / t f » » i » j
..'.» * //./>» f ».«»'<
».«'«
'
10
13
I')'I'I'M< flr»l-L«r»r Wind Vector fl«ld« (m/»)
0 2 4 8 10 for 100 on 8 / 7 / B4
flHD SPtfD (It/ft
m O ( ~ ' .
J J f > X X X X X XXX'AX'^^X'.X.XX'
/" * " i " ? <*' '
.*% . ."..- \ ?
4. «.. «'-.«-. « * *W ««««S«««'T'S
.,,,..., r ^. ./«»>
' I
13
10
Thlrd-Ur«r Wind V«c«or Dvldi (m/»)
for 100 on a / 7 / 84
zo
FIGURE 4-15. DWM-generated surface and upper-layer wind fields for the
California Central Valley at 0400 on 7 August 1984.
-------
30 :
29-
19
10-
*» *. ^ »
. V X «
. «» V »
''.**
, i^t > /W
% 1 1 t f- t :S *+
t t t f /: /. X S-
tf
*-*- ,
hr^
*-V>y
's/ t
' //
' V /
' /N/ ;;j
N' / M j
.v ,./,;./ 11
\ \Vv
\\
' M > \ \
J \ \\\\\ .
\ \\\\ \\-VS
\ \ \ \\\>^
V\ \VN^^.-
\'\ \ \ \^ ^ r»
( V X V VA ^ r'" '
1111 Vi \i Vi v \i S *^
^ -. ^ s s
-, - s J
10
13
20
I'l'l'l'l'l
0 Z 4 6 B 10
HtfD SPCKD (H/S)
ririt-l«y«r Wind Vector Fleldt (m/t)
for 1200 on B / 7 / 84
, ^ > rv^T
'N--VrV|
1
29
20-
19-
10-
F t 'i <'» » «.; *"'»'.:»
I i . IJerctd / .
ll <»«»»>»««».* »VYt>!4iT.h<}M
f f f «. » . . . . . . . ^. % .A »* YYJ
/ » f ....,,....,, t £t
.t I » '
K-T
r0)111/ . . . « . > <, r < >«,,-
...... to«di«r,
* * * Jl ' '
* .X «T *%/':
-,-,» *.»« < '.
, , -> / ^ '.....«»
,,,',,,. . . . Y"»»1 < i
-//- &"
* + , , » » «
»>.,».»»« t »..+
' *i<.» t * Vs
v&o' ' '
fit ft « »..>' »' »:A A VVV<^' "*
' > >..' » * » .«' \pli \f\ V V V v v.
%'
'/-/» >
T» t I -f"r *V
t / r'-a »
> t t t «
v \ r i V«
\ \ < \ *
> » « ^ \?\ \ > N v v. ».
% . « \ \ \ V \ V \ N ^/x"
«. « i\ \ \ \ \ \~VA ^,'»
Thlrd-Larer Wind Vector Fl*ld« (m/t)
TOT 1200 on 8 / 7 / 84
FIGURE 4-16. DWM-generated surface and upper-layer wind fields for the
California Central Valley at 1200 on 7 August 1984.
-------
surface pressure
relative humidity
precipitation rate
The MELSAR-MET model served as the basis for the development of the new
meteorological driver for the Rocky Mountain acid deposition model. Much
of the following discussion on the prescription of meteorological inputs
1s abstracted from the technical description of MELSAR-MET (Allwine and
Whlteman, 1985).
4.3.1 Mixing Heights
The grldded mixing heights are computed for each hour using surface
weather observations and upper-air observations. The hourly mixing height
at a grid point is the maximum of a convectlve mixing height or a mechani-
cal mixing height. The convectlve mixing height 1s set equal to zero dur-
ing the night. The mechanical mixing height 1s computed as 53 x 10~4 Ug,
where Uq 1s the free stream wind speed (m/s). This formulation for the
mechanical mixing height 1s given by Benkley and Bass (1979).
The convectlve mixing height 1s computed using a technique described by
Benkley and Schulman (1979). The hourly mixing height at a weather sta-
tion 1s estimated by determining the height of the Intersection of the
surface potential temperature and the morning potential temperature sound-
Ing. The technique accounts for warm or cold air advection into the
region by adjusting the hourly surface potential temperature values
according to an advection rate. The advection rate 1s determined from the
difference 1n potential temperature between the afternoon and morning
sounding at a height above the convectlve mixing height. The technique
also makes adjustments for differences between the temperature at the sur-
face station and the surface temperature at the radiosonde station, or
makes adjustments 1f the minimum surface temperature occurs before the
morning sounding. This is accomplished by adjusting the morning sounding
to fit the minimum surface temperature observation.
Once the mixing height is computed at each weather station, the mixing
height at a grid point is determined by an inverse-distance-square weight-
Ing of the station mixing heights to the grid points.
4.3.2 Stability Classification
The Pasqu1ll-Gifford-Turner (PGT) stability classes are determined for
each grid cell for each hour using the approach given by Turner (1970).
148
-------
Given the wind speed at the surface, the solar elevation angle, and the
fractional cloud cover, the PGT stability class can be determined from the
following table (from Turner, 1970):
Day Night
Wind Speed
at 10 m
<2 m/s
2-3
3-5
5-6
>6
Incoming Solar Radiation
Strong Moderate SI
A A-B
A-B B
B B-C
C C-D
C D
ight
B
C
C
D
D
Some
Clouds
E
E
D
D
D
Few
Clouds
F
F
E
D
D
4.3.3 Friction Velocity
The surface friction velocity, u* (m/s), 1s computed for each grid cell
for each hour using surface weather observations. The approach used is a
modification of the approach given by Scire and co-workers (1984). The
surface friction velocity for unstable conditions can be estimated by the
method described by Wang and Chen (1980):
{1 + a In [1 + b Q /QJ} (4-22)
00
0* in /^^ \ <4-23)
- zms
Q0 = H/(p cp) (4-25)
H = A0R + HQ (4-26)
HQ = 2.4CQ - 25.5 (4-27)
149
-------
-3
e u^
Q = (4-28)
0 kgz
m
a = 0.128 + 0.005 In (z0/zm), zQ/zm « 0.01
(4-29)
= 0.107, z0/zm > 0.01
b = 1.95 + 32.6 (z0/zm)0*45 (4'30)
where
k = the von Karman constant (-0.4)
Cp = the specific heat of air at constant pressure (996 m2/s2 deg)
um = the wind speed (m/s) measured at height zms (m)
ZQ = the surface roughness (ml
p = the density of air (kg/m3)
g = acceleration due to gravity (9.81 m/s2)
e = surface potential temperature (K)
R = Incoming solar radiation (W/nr)
AQ = fraction of R converted to sensible heat flux
CQ = opaque cloud cover (tenths).
During stable conditions, u* 1s determined by the following method (Venka-
tram, 1980a):
= 5^1 [i + c°'5] (4-31)
'ON - In (zm/z0)
(4-32)
C = 1 V- c - ° (4'33)
CDNUm
o Y Z_
u2 (4-34)
U0 " k A { '
where y and A are constants with values of 4.7 and 1100, respectively, and
CQN is the neutral drag coefficient.
150
-------
4.3.4 Convectlve Velocity
The convectlve velocity scale, w* (m/s), 1s computed for each grid cell
for each hour using surface weather observations. The approach used is
that given by Scire and co-workers (1984). During convectlve conditions,
w* is calculated from Its definition:
(4-35)
where TQ is the surface air temperature (K), Q0 is from Equation 4-25, and
2^ is the mixing height from Section 4.3.1. For QQ less than zero, w* is
equal to zero.
4.3.5 Monin-Obukhov Length
The Monin-Obukhv length, L (meters), 1s computed for each grid cell for
each hour using surface weather observations. For unstable conditions it
1s computed from Its definition:
u T
whose terms have been defined earlier. During stable conditions, L is
given by Venkatram (1980b) as
L = 1100 u* (4-37)
4.3.6 Temperature
The surface temperature, TQ (kelvins> is computed for each grid cell for
each hour using surface observations and a seasonal empirical relationship
between surface temperature and elevation (surface temperature lapse rate)
derived from analysis of climatological data in western Colorado (PEDCO,
1981). PEDCO analyzed up to 40 years of surface temperature observations
for nine stations in western Colorado. They determined the average tem-
perature change with elevation for each month of the year. These monthly
surface temperature lapse rates were plotted versus Julian day and the
points connected with straight lines. The slopes and intercepts of these
lines are given 1n Table 4-1. The hourly temperature observation at each
surface station 1s Interpolated to each grid point using an Inverse-
distance-squared weighting factor. The interpolated temperature from each
151
-------
TABLE 4-1. Slope and intercept of temper-
ature lapse rate correction by Julian Day.
Julian Day (d) Slope (m) Intercept ~(b)
1 < d < 16 -0.0152 -1.6950
16 < d < 75 -0.0607 -0.9592
75 < d < 105 0.0093 -6.2100
105 < d < 136 -0.0552 0.5619
136 < d < 166 0.0180 -9.3880
166 < d < 197 -0.0135 -4.1510
197 < d < 228 0.0116 -9.1077
228 < d < 258 0.0037 -7.2960
258 < d < 289 0.0832 -27.8223
289 < d < 319 0.0500 -18.2200
319 < d < 350 0.0261 -10.6052
350 < d < 366 -0.0152 3.8600
152
-------
weather station 1s corrected for altitude differences between the weather
station and the grid point. The corrected temperature 1s
TG = TS +
-------
o = P(O) l -
ps = R~T~ (4'41)
then solving the following equation for e using Newton's method:
(4-42)
\ "a/
4.3.8 Relative Humidity
The Lagranglan acid deposition model requires relative humidity in order
to estimate water vapor concentrations used 1n the calculation of chemical
transformation rates. The Interpolation of relative humidity is an uncer-
tain process because of Its dependence on temperature. Thus the mesoscale
meteorological model interpolates dew point in space and time, from which
a three-dimensional distribution of relative humidity can be obtained.
The procedures used to Interpolate dew point are very similar to those
used for temperature. First a surface dew point field is obtained by
Interpolating the measurements from the surface sites using an elevation
adjustment derived from an analysis of upper-air soundings from the Rocky
Mountains. At each upper-air station, dew point lapse rates 1n the free
atmosphere are calculated above and below the mixing height. These dew
point lapse rates are then Interpolated onto the grid using an Inverse
distance squared weighting procedure. The Lagranglan acid deposition
model then calculates the relative humidity at any three-dimensional point
1n the modeling domain by first calculating the temperature, T, and dew
point, Tp, at the point using the surface values and lapse rates, and then
calculating the relative humidity using the following equation:
7.5 T 7'5 TD
RH - 100 10 237'3 + T - (4-43)
4.3.9 Precipitation Rate
Modeling of wet deposition requires estimates of precipitation at each
grid cell of the modeling region. For the Rocky Mountain model it is
expected that one-hour or 3-hour precipitation averages will be required
depending on the user-defined update Interval. The spatial extrapolation
of precipitation at such short time scales represents a special challenge
154
-------
over complex terrain such as the Rocky Mountains. Precipitation often
occurs as the result of terrain-forced lifting of air masses above the
point where condensation takes place. After air masses pass over such
terrain obstacles, there 1s no moisture for precipitation on the lee side
of the mountain. As a result, there 1s often augmented precipitation on
the windward side of ridges, and rain shadows (minimums) 1n the lee of the
ridges. Terrain height and slope are obvious factors 1n determining how
much precipitation will fall.
Unfortunately, simple rules for spatially Interpolating precipitation
using terrain Information do not always hold. For example, there may be
substantial channeling of the paths that moist air masses may take. This
channeling may be other than west-east along which most synoptic-scale
storms travel over the western Rockies. Therefore such factors as north-
south canyons and unresolved terrain features may act to produce precipi-
tation variations that do not obey simple relations for precipitation
estimates, such as functions of elevation and east-west terrain slope.
The goal of the present analysis was to obtain a year of short-term pre-
cipitation data at a grid resolution of approximately 5-10 km. Observed
data was used as a starting point 1n the Interpolation process. Contribu-
tions of precipitation due to terrain effects are added to precipitation
estimates obtained by spatial Interpolation of observed data. By making
extensive use of the observations we guided the spatial Interpolation pro-
cess so that 1t did not produce spurious precipitation amounts.
Two sources of precipitation data are availabledaily precipitation
totals and hourly precipitation rates. In the Interpolation methodology
described 1n the following paragraphs we attempt to use all of the data to
provide realistic fields of precipitation, thereby avoiding some global
assumptions that leave the Interpolated precipitation fields with serious
departures from reality, such as the phenomena of "popping" rainfields
where rainfall maxlmums appear simultaneously over the modeling region.
4.3.9.1 Interpolation of Dally Precipitation Data
The first step in the Interpolation methodology is to estimate the dis-
tance of the (1,j) grid cell from the kth observation site. The following
Inverse distance weight is estimated for each grid cell and observation
site:
w1(j(k) = 1.0/D3 = 1.0/[(X1§j - Xk)2 + (Y1fj - Y/l1'5
155
-------
The largest two weights for each (1,j) grid cell are retained. A
coefficient C1 1s estimated that will make the w1 k(k) over all 24-hour
precipitation observation sites sum to one. Th1s'coeff1c1ent 1s then used
to estimate new weights that sum to one, I.e.,
Only the highest and second largest weights are used to estimate the
Interpolated value at the (1,j) grid cell. The remaining portion of the
concerning the underlying terrain.
The weight used to factor 1n the regressed value of precipitation 1s
determined from
B., j(k) = 1>0 ' twi j(1ar9est) + wi j(2nd largest))
The geographically Induced portion of the precipitation, X_e-, 1s computed
within the sphere of Influence of each site. The sphere of Influence for
the kth observation site 1s defined as all grid cells that Identified the
kth site as having the largest weight. The Interpolated precipitation
estimate, X^ 4, 1s computed from the following equation:
where k denotes the site with the largest weight; 1 denotes the site with
the second largest weight; «k 1 = 1 1f Xk j< 0, 6k 1 = 0 if Xk = 0.
The Xqeo depends on whether the east-west slope S1 j is either positive or
negative. The slope 1s computed using the 10 km average terrain, H, 1n
the following manner:
SU = H1+U " H1.J *
If the slope at the (1,j) cell 1s positive and the slope at the kth obser-
vation site 1s also positive, the following equation is used:
Xgeo=0-01 (H1J-Hk>+Xk
If the slope 1s negative then a different set of regression equations
applies. If the slope Sk is greater than zero then the following relation
1s used:
156
-------
If the slope Sk 1s less than zero then the following relation 1s used:
If the slope Sk 1s negative, but the slope at S1 4 1s positive, then we
estimate the precipitation from ground level as follows:
V°-01HU
These formulas can occasionally produce rainfall fields that 1n places are
relatively discontinuous and may be somewhat spurious. To deal with these
cases we first smooth the precipitation field with N passes of a simple
five-point filter, I.e.,
^ = °'5 X1,J + °'125 (X1+U + X1-l.j + X1,J+1 + X1,J-l)
The number of passes 1s generally less than 10 to avoid excessively smooth
precipitation fields.
The Interpolated rainfall using this procedure was well behaved enough so
that no smoothing was required (N = 0). In order to avoid the Intrusion
of precipitation Into areas without rain, we take the following precau-
tions. First, all dally precipitations of less than 0.01 inch are set
equal to zero. Secondly, where the kth observation site does not show any
significant precipitation, interpolated precipitation values less than
0.05 Inches are set equal to zero within the "dominant" sphere of influ-
ence of that station.
4.3.9.2 Distribution of Rainfall Within the Day
At each of the hourly observation sites the total precipitation within
each day was weighted so that the sum over all hours of the day equals
1.0. Weighted precipitation within a day was created even for days when
no precipitation occurred at the site. This was done to provide weights
when rain might have occurred at adjacent sites. These weights were
created by linearly Interpolating between days when precipitation did
occur. The linear Interpolation was done in such a way that the sums of
these weights over the day still sum to 1.0.
The Inverse distance weights for each (i,j) grid cell and kth observation
site were computed 1n the same fashion as was done for the daily data.
These weights were then multiplied by the weights for calculating the dis-
tribution of the dally rainfall. The resulting product of weights had to
157
-------
be such that the distributed daily total precipitation must sum to the
dally observed or Interpolated precipitation, I.e.,
X
1 ,00 = C, ^k) w. M PI .(k)
1 j i , j ' , j ' » j
where P^j(k) 1s the weight for the distribution of precipitation at the
kth s1te'w1th1n a single day. The C^ j(k) 1s the constant required for
recovery of the observed or Interpolated daily precipitation at the (i,j)
grid cell. When the distributed precipitation 1s less than one, the pre-
cipitation 1s assumed to be zero.
158
-------
DESIGN OF THE ACID DEPOSITION MODEL
FOR THE ROCKY MOUNTAIN REGION
Our evaluation of the final four candidate acid deposition models indi-
cates that no one of these models is the best choice for calculating
source-specific acid deposition impacts in the Rocky Mountain region
(Morris and Kessler, 1987). Our evaluation also indicates that the most
flexible modeling approach would be the Gaussian puff model formulation.
However, neither of the candidate Gaussian puff models (the MESOPUFF-II or
MELSAR-POLUT) appears to be superior in all processes that lead to acid
deposition. The MELSAR-POLUT model appears to describe transport and dis-
persion 1n complex terrain better than the MESOPUFF-II; however, the MEL-
SAR-POLUT does not treat chemical transformation or scavenging. In this
section we describe a new acid deposition model that uses the most scien-
tifically sound components of the candidate models.
5.1 TRANSPORT
All of the candidate acid deposition models, except the CCADM, use the
wind at the plume centerline to advect the puff or plume. The CCADM
relies on user Input for its trajectory definition. The analysis of
transport as a function of height (Section 3.1) indicates that the
resultant trajectory 1n complex terrain is very dependent on the height of
the air parcel above the ground. The differences in trajectories at dif-
ferent heights were magnified by the stagnant conditions that existed in
the evaluation tests; however, those tests did verify that defining tra-
jectories in complex terrain is very uncertain.
Use of the wind at the plume centerline is consistent with an actual simu-
lated air parcel trajectory at some height and with the formulation of the
new Rocky Mountain Lagrangian acid deposition model. The use of a verti-
cally vector averaged wind for advecting a puff through complex terrain
may result in an impossible trajectory if sufficient wind shear exists.
In the formulation of the Rocky Mountain acid deposition model considera-
tions are given for the future implementation of allowing vertical shear-
ing of the Lagrangian puffs when decoupled flow conditions exist. How-
ever, for this initial version of the Rocky Mountain model, the Lagrangian
puff is advected as a cohesive unit.
159
-------
5.2 DISPERSION
Of the four candidate models, the MELSAR-POLUT parameterization of disper-
sion Includes the most complete description of diffusion over complex ter-
rain (see Appendix A in Morris and Kessler, 1987; or Allwine and Whiteman,
1985). Thus the MELSAR-POLUT dispersion algorithm has been implemented
1n the new Rocky Mountain model. The performance evaluation of this
algorithm showed that 1t reacts as expected to changes 1n terrain rough-
ness (see Section 3.2).
Since the Rocky Mountain model 1s intended to be used for the calculation
of Impacts as part of the PSD permitting process, the user should have the
option of calculating dispersion 1n a manner similiar to EPA-approved
models. Thus, the MESOPUFF-II dispersion algorithms have also been
Implemented in the Rocky Mountain model. The MESOPUFF-II parameterization
of the Oy and oz curves attempts to duplicate the values suggested by Pas-
quill, Gifford and Turner (Turner 1970), which are used in most EPA-
approved models.
5.3 CHEMICAL TRANSFORMATION
Of the candidate models, the CCADM contains the most comprehensive chemis-
try module. However, the computational requirements of CCADM and the
model's need for an ambient field of background concentrations preclude
Its use in the Rocky Mountain acid deposition model. In the evaluations
of the parameterized pseudo-first-order chemistry mechanisms used in the
MESOPUFF-II and RIVAD (see Section 3.3), the RIVAD chemistry responded as
expected to changes in environmental and concentration conditions. The
MESOPUFF-II oxidation rates were totally insensitive to changes in tem-
peratures, which may be Important in the higher elevations of the Rocky
Mountains. In addition, the MESOPUFF-II chemical mechanisms appear to be
designed for the urban or polluted atmosphere of the East Coast. The
RIVAD model has been applied to the western states, including the Rocky
Mountains, and evaluation of the model's performance shows quite good
agreement between the predicted and observed ambient concentrations.
Thus, the RIVAD chemical mechanism has been Implemented in the Rocky Moun-
tain model. In order to give the user other options for chemical trans-
formation, the MESOPUFF-II theoretical chemical mechanism has also been
Implemented in the Rocky Mountain model as an option.
The modular design of the Rocky Mountain model will easily allow the
Insertion of new chemical mechanisms as they become available. Future
mechanisms could be developed for the Rocky Mountain region by exercising
160
-------
a model with a sophisticated chemical kinetic mechnism, such as the RADM
or CCADM, and parameterizing the chemical reaction rates in terms of a
look-up table, as is done in the RTM-IINL, or through regression equa-
tions. In addition, when the engineering version of the RADM becomes
available, the chemical mechanism therein may be appropriate for the Rocky
Mountain model.
5.4 DRY DEPOSITION
The preferred approach for the modeling of dry deposition involves the
resistance concept. The deposition velocities produced by the two candi-
date models that use the resistance approach, the MESOPUFF-II and the
CCADM, were compared and evaluated (see Section 3.4). The deposition
velocities calculated by these models were generally consistent with
available measurements; notable exceptions were the deposition velocities
for the aerosol species (sulfates and nitrates) calculated by the
MESOPUFF-II, and the deposition velocity for N02 over water produced by
the CCADM. The CCADM calculates an areal average deposition velocity over
the region occupied by the plume, while the MESOPUFF-II bases its deposi-
tion velocity on to a single land use category located at the puff's
center. Thus 1t seems more appropriate to use the CCADM dry deposition
module within the Rocky Mountain acid deposition model. The CCADM dry
deposition algorithm is also the one most like the parameterization in
RADM. The CCADM algorithm was extended to include dry deposition of
pollutants over snow. In addition, the surface resistance of NOX over
water has been increased. This modified version of the CCADM has been
implemented in the Rocky Mountain model.
5.5 WET DEPOSITION
Evaluation of the wet deposition algorithms (see Section 3.5) shows that
the MESOPUFF-II scavenging coefficient approach is the most flexible and
consistent with the Lagrangian puff formulation of the Rocky Mountain
model. The ability to easily Incorporate the different scavenging
characteristics of liquid versus frozen precipitation is especially
Important in the high elevation regions of the Rocky Mountains. Thus wet
scavenging within the Rocky Mountain model is parameterized in terms of
the scavenging coefficient approach. As new scavenging ratios are
reported 1n the literature, they can be easily incorporated in the model.
5.6 SUMMARY
The acid deposition/air quality module of the Rocky Mountain model incor-
porates the most appropriate and scientifically sound components of the
161
-------
candidate acid deposition models. The model uses the Lagrangian puff
model formulation and parameterizes the major processes as follows:
Transportuses the wind vector from the new diagnostic wind model at
the plume centerline height above ground.
Dispersionuses the MELSAR-POLUT complex terrain dispersion formulas
with the MESOPUFF-II parameterization of the PGT disperiosn curves
also Implemented as an option.
Chemical Transformationuses the RIVAD peusdo first-order chemical
reaction rate mechanism; the MESOPUFF-II theoretical rate expressions
are implemented in the model as an optional mechanism.
Dry Depositionuses the CCADM resistence approach with the cell-
averaging procedure currently implemented in the RADM.
Wet Depositionuses the scavenging coefficient approach as used in
the MESOPUFF-II model.
162
-------
6 SUMMARY AND RECOMMENDATIONS
A new hybrid mesoscale add deposition/air quality model has been designed
specifically for calculating Incremental Impacts of deposition of nitrogen
and sulfur species and concentrations of PSD pollutants 1n the complex
terrain region of the Rocky Mountains. The modeling system contains two
principal components: a mesoscale meteorological model and an acid depo-
sition/air quality model. The formulation of the new model combines the
most technically advanced components of existing candidate mesoscale
meteorological and add deposition models that are consistent with the
overall design of the modeling system. The selection of the candidate
models and a preliminary evaluation was presented 1n a previous report
(Morris and Kessler, 1987). The overall design of the modeling system was
guided by the recommendations of the potential users represented by the
members of the Western Add Deposition Task Force.
A preliminary evaluation of the modeling system was conducted. The new
Diagnostic Wind Model (DWM) was evaluated using a hypothetical terrain
obstacle, terrain from the Rocky Mountains, a complex terrain/coastal
environment with a dense observational network, and within a large valley;
finally, the DWM predictions were compared with observations from the
Rocky Mountains. These evaluations of the new DWM Illustrated the flexi-
bility of the DWM 1n simulating air flows over a variety of complex-
terrain configurations 1n areas with and without observations.
The acid deposition/air quality component of the hybrid modeling system
was evaluated by evaluating the individual modules and components that
comprise the model.
Although the hybrid modeling system was constructed by using state-of-the-
art components from existing models, the modeling system was designed to
be flexible and easily expanded. Instead of selecting a single component
for Insertion Into the modeling system, the model was configured with
several options to treat major processes, such as transport, dispersion,
and chemical transformation. In addition, as our understanding of these
processes increases, the Insertion of new modules i-nto the modeling system
can be easily accomplished.
163
-------
The delivery of the Initial version of the model 1s scheduled for the end
of 1987. Along with the model code, a user's guide and a document
describing the technical formulation of the modeling system 1s also
scheduled for delivery. Other documents under development as part of the
Rocky Mountain Model Assessment Project are a protocol for evaluating the
performance of the new modeling system and a report describing the evalua-
tion.
The performance evaluation of the new model 1s required 1n order to have
confidence 1n the model predictions and Identify any areas of the modeling
system that need Improvement. In particular, 1t must be demonstrated that
the model adequately predicts source-receptor relationships of add depo-
sition 1n complex terrain. Unfortunately, there are currently no data
bases for the evaluation of add deposition source-receptor relation-
ships. The best data bases available for evaluation source-receptor rela-
tionships consist of several tracer experiments. The model evaluation
protocol will contain a review of all pertinent tracer data bases avail-
able for evaluating the new Rocky Mountain model and will select a few for
the evaluation. In addition, the protocol will also recommmend ways in
which the new model's ability to predict add deposition Impacts can be
evaluated.
164
-------
REFERENCES
Allwlne, K. J.. and C. D. Whiteman. 1985. "Green River A1r Quality Model
Development: MELSARA Mesoscale Air Quality Model for Complex Ter-
rain. Volume 1Overview, 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).
Arrltt, R. W., and R. A. Pielke. 1986. Interactions of nocturnal slope
flows with ambient winds. Boundary-Layer Meteorol.. 37:183-195.
Atkinson, R., A. C. Lloyd, and L. Wlnges. 1982. An updated chemical
mechanism for hydrocarbon/NOx/SOx photooxldatlons suitable for Inclu-
sion in atmospheric simulation models. Atmos. Environ., 16:1341.
Benkley, C. W., and A. Bass. 1979a. Development of Mesoscale Air 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.
Benkley, C. W., and L. L. Schulman. 1979. Estimating hourly mixing
depths from historical meteorological data. J. Applied Meteor.,
18:772-780.
Businger, J. A. 1973. "Turbulent Transfer 1n the Atmospheric Surface
Layer." In Workshop on Micrometeorology. American Meteorological
Society, Boston, Massachusetts.
Chang, J. S., R. A. Brost, L.S.A. Isaksen, P. Middleton, W. R. Stockwell,
and C. J. Walcek. 1986. RADM, a three-dimensional Eulerian acid
deposition model. Part I: Physical concepts and model formula-
tion. Submitted to J. Geo. Res.
165
-------
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.
Drake, R. L., C. H. Huang, and W. E. Davis. 1981. "Green River Ambient
Model Assessment Program - Progress Report for the Regional and Meso-
scale Flow Modeling Components." Pacific Northwest Laboratory, Rich-
land, Washington (PNL-3988).
Durran, D. R., and J. B. Klemp. 1982. The effects of moisture on trapped
mountain lee waves. J. Atmos. Sci.. 39:2490-2506.
Durran, D. R., and J. B. Klemp. 1983. A compressible model for the simu-
lation of moist mountain waves. Mon. Weather Rev.. 11:2341-2361.
EPA. 1985. Interim Procedures for Evaluating Air Quality Models:
Experience with Implementation. U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina (EPA-450/4-85-006).
EPA. 1986. Guideline On Air Quality Models (Revised). U.S. Environ-
mental Protection Agency, Research Triangle Park, North Carolina
(EPA-450/2-78-027R).
FUzjarrald, D. R. 1984. Katabatic wind in opposing flow. J. Atmos.
Sc1., 41:1143-1158.
Fox, D. G. 1981. Judging air quality model performance: A summary
of the AMS Workshop on dispersion model performance. Bull. Am.
Meteorol. Soc.. 62:599-609.
Garland, J. A. 1977. The dry deposition of sulfur dioxide to land and
water surfaces. Proc. R. Soc. Lond. A, 354:245-268.
Gery, M. W., R. E. Morris, S. M. Greenfield, M. K. Liu, D. A. Stewart,
G. Z. Whitten, and J. L. Fieber. 1987. "Development of a Comprehen-
sive Chemistry Acid Deposition Model (CCADM)." Systems Applications,
Inc., San Rafael, California (SYSAPP-87/061).
Glfford, F. A. 1982. Horizontal diffusion in the atmosphere: A
Lagranglan-dynamical Theory. Atmos. Environ., 16(3):505-512.
G1llan1, N. V., S. Kohll, and W. E. Wilson. 1981. Gas-to-particle 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.
166
-------
Godden, D., and F. Lurmann. 1983. "Development of the PLMSTAR Model and
Its Application to Ozone Episode Conditions 1n the South Coast Air
Basin." Environmental Research & Technology, Inc., Westlake Village,
California (ERT P-A702-200).
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.
Hales, J. M., and S. L. Sutter. 1973. Solubility of sulfur dioxide in
water at low concentrations. Atmos. Environ.. 7:997-1001.
Hayes, S. R. 1979. Performance Measures and Standards for Air Quality
Simulation Models. U.S. Environmental Protection Agency, Research
Triangle Park. North Carolina (EPA-450/4-79-032).
Heffter, J. L. 1965. Variations of horizontal diffusion parameters with
time for travel periods of one hour or longer. J. Appl. Meteorol.,
4:153-156.
Henry, R. C., and G. M. Hidy. 1981. Author's reply to discussion of
"Multlvariate analysis of particulate sulfate and other air quality
variables. Part I. Annual data from Los Angeles and New York."
Atmos. Environ.. 15:425-426.
Henry, R. C., and G. M. Hidy. 1982. Multlvariate analysis of particulate
sulfate and other air quality variables by principal components.
Part II. Salt Lake City, Utah and St. Louis, Missouri. Atmos.
Environ.. 16:929-943.
Hicks, B. B. 1983. In Critical Assessment Document on Acid Deposition
(Chapter VII). Atmospheric Turbulence and Diffusion Laboratory,
NOAA, Oak Ridge, Tennessee (File No. 81/24).
H111, A. C., and E. M. Chamberlain. 1976. "Removal of Water Soluble
Gases from the Atmosphere by Vegetation." Proc. Symp. on Atmosphere-
Surface Exchange of Particles and Gases. ERDA Symposium Series NTIS.
HUlyer, M. J., S. D. Reynolds, and P. M. Roth. 1979. Procedures for
Evaluating the Performance of Air Quality Simulation Models. U.S.
Environmental Protection Agency. Research Triangle Park, North Caro-
lina (EPA-450/4-79-033).
167
-------
Irwln, J. S. 1979. "Estimating Plume DispersionA Recommended Generali-
zed Scheme." Proceedings of Fourth Symposium on Turbulence. Diffu-
sion, and Air Pollution. American Meteorological Society.
Latimer, D. A., M. W. Gery, and H. Hogo. 1986. "A Theoretical Evaluation
of the Role of Nighttime Nitrate Formation 1n the Formation of
Layered Haze." Systems Applications, Inc., San Rafael, California
(SYSAPP-86/012).
L1u, M. K., R. E. Morris, and J. P- Killus. 1984. Development of a
regional oxidant model and application to the northeastern United
States. Atmos. Environ.. 18:1145-1161.
L1u, M. K., and M. A. Yocke. 1980. Siting of wind turbine generators 1n
complex terrain. J. Energy. 4(1):10-16.
MacCready, P. B., Jr., L. B. Baboolal, and P.B.S. Ussaman. 1974. "Dif-
fusion and Turbulence Aloft Over Complex Terrain." AMS Symposium on
Atmospheric Diffusion and Air Pollution, Santa Barbara, California
Mahrt, L. 1982. Momentum balance of gravity flows. J. Atmos. Sci.,
39:2701-2711.
McMahon, T. A., and P. J. Denlson. 1979. Empirical atmospheric deposi-
tion parametersa survey. Atmos. Environ., 13:571-585.
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-87/155).
Morris, R. E., and R. C. Kessler. 1987. "Rocky Mountain Add Deposition
Model Assessment." Systems Applications, Inc., San Rafael, California
(SYSAPP-87/201, EPA-600/3-87-013; NTIS PB87-180584-AS)
Morris, R. E.t M. K. L1u, and S. D. Reynolds. 1987. "The Systems Appli-
cations, Inc. Regional Transport Model: Current Status and Future
Needs." A1r Pollution Control Association International Specialty
Conference on the Scientfic and Technical Issues Facing Post-1987
Ozone Control Strategies. Hartford, Connecticut.
NCAR. 1985. "NCAR Eulerian Regional Acid Deposition Model." National
Center for Atmospheric Research, Boulder, Colorado (Technical Note
NCAR/TN-256+STR).
168
-------
NRC. 1983. Add Deposition; Atmospheric Processes in Eastern North
America. National Research Council. National Academy Press, Wash-
ington, DC.
Owers, M. J., and A. W. Powell. 1974. Deposition velocity of sulphur
dioxide on land and water surfaces using a 35S tracer method. Atmos.
Environ.. 8:63-67.
PEDCO. 1981. "Colorado's Climate, Meteorology and Air Quality." PEDCO
Environmental, Inc., Cincinnati, Ohio.
Plelke, R. A. 1974. A three-dimensional numerical model of the sea
breezes over southern Florida. Hon. Weather Rev.. 102:115-139.
Plelke, R. A. 1984. Mesoscale Meteorological Modeling. Academic Press.
Pleim, J., A. Venkatram, and R. Yamartino. 1984. "ADOM/TADAP Model
Development Program, Volume 4: The Dry Deposition Module." Environ-
mental Research and Technology, Inc., Concord, Massachusetts (P-B980-
520).
Prahm, L. P., U. Torp, and R. M. Stern. 1976. Deposition and transforma-
tion rates of sulphur oxides during atmospheric transport over the
Atlantic. Tellus 28:355-372.
Prandtl, L. 1942. Fiihrer durch die Stromungstehre. Verlag Vieweh und
Sohn, Braunschweig, Germany.
Randerson, D. 1972. Temporal changes in horizontal diffusion parameters
of a single nuclear debris cloud. J. Appl. Meteorol., 11:670-673.
Rao, K. S., and H. F. Snodgrass. 1981. A nonstationary nocturnal
drainage flow model. Bound. Layer Meteorol.. 20:309-320.
Ross, D. G., and I. Smith. 1986. "Diagnostic Wind Field Modelling for
Complex TerrainTesting and Evaluation." Centre for Applied Mathe-
matical Modelling, Chisholm 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 Acid Rain Test. World Resources Institute
(Research Report II).
Scire, J. S., and F. W. Lurmann. 1983. "Development of the MESOPUFF II
Dispersion Model." Sixth Symposium on Turbulence and Diffusion,
American Meteorological Society (paper PI.11).
169
-------
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.
Sdre, J. S., F. W. Lurmann, A. Bass, and S. R. Hanna. 1984. "User's
Guide to the MESOPUFF II Model and Related Processor Programs."
Environmental Research & Technology, Inc., Concord, Massachusetts.
Scott, B. C. 1978. Parameterization of sulfate removal by precipita-
tion. J. Appl. Meteorol.. 17:1375.
Sehmel, G. A. 1980. Particle and gas dry deposition: a review. Atmos.
Environ., 14:983-1011.
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.
Smith, R. B. 1979. The influences of mountains on the atmosphere. Adv.
Geophys.. 21:87-230.
Smith, F. B., and R. D. Hunt. 1978. Meteorological aspects of the trans-
port of pollution over long distances. Atmos. Environ., 12:461-477.
Spedding, D. J. 1969. Uptake of sulphur dioxide by barley leaves at low
sulphur dioxide concentrations. Nature. 224:1229-1231.
Turner, D. B. 1970. Workbook of Atmospheric Dispersion Estimates. U.S.
Environmental Protection Agency, Research Triangle Park, North Caro-
lina (AP-26).
Van Aalst, R. M.t H.S.M.A. Diederen, J. H. Duyzer, and G. M. Meyer.
1983. "Dry Deposition of Acid Precursors in the Netherlands." TNO,
Delft, The Netherlands.
Venkatram, A. 1980a. "Estimation of Turbulence Velocity Scales in the
Stable and the Unstable Boundary Layer for Dispersion Applica-
tions." Eleventh NATO-CCSM International Technical Meeting on Air
Pollution Modeling and Its Application, pp. 169-179.
Venkatram, A. 1980b. Estimating the Monin-Obukhov length in the stable
boundary layer for dispersion calculations. Boundary-Layer
Meteorology. 19:481-485.
170
-------
Walcek. C. J., R. A. Brost, J. S. Chang, and M. L. Wesley. 1986. SO?,
sulfate, and HN03 deposition velocities computed using regional land-
use and meteorological data. Atmos. Environ.. 20(5):949-964.
Walcek, C. 0., and W. R. Stockwell. 1986. The contribution of aqueous-
phase radical chemistry to sulfur and ^2 chemistry 1n cloudy
environments. Submitted for publication.
Wang, I. T., and P. C. Chen. 1980. Estimations of heat and momentum
fluxes near the ground. Proc. 2nd Joint Conference on Applications
of Air Pollution Meteorology.
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. Air Pollut. Control Assoc.. 27:1110-1116.
Whelpdale, D. M., and R. W. Shaw. 1974. Sulphur dioxide removal by tur-
bulent transfer over grass, snow and water surfaces. Tellus, 26:195-
204.
Wlllmott, C. J. 1982. Some comments on the evaluation of model per-
formance. Bull. Am. Meteorol Soc.. 82:1309.
171
-------
APPENDIX
Dry deposition velocities (cm/.s) predicted by the MESOPUFF-II and the
CCADM for
Sulfur dioxide (S02)
Sulfate (S04)
NOX (N02)
NHrlc Add (HN03)
Nitrate (N03)
-------
MESOPUFF-I
CCADM
,0123
ro
I I I I I I
2345678
Surface Wind Speed (m/s)
9 10
T) 1 2 5 4 5 6 7 8 9 10~
Surface Wind Speed (m/s)
-2
SO2 Deposition Velocities (cm/s) for CROPLAND AND PASTURE Land Use Type.
-------
MESOPUFF-I
CCADM
,0 1 2
U>
01 234567891
1 2 3 4 6 7 8 9 o T)1 2 3 4 5 6 7 8 9 10
Surface Wind Speed (m/s) Surface Wind Speed (m/s)
-2
SO2 Deposition Velocities (cm/s) for CROPLAND/WOODLAND/GRAZING Land Use Type.
-------
MESOPUFF-I
CCADM
,0 1 2
12345678911
2345678
Surface Wind Speed (m/s)
9 10
"5 1 2 3 4 5 6 7 8 9 10
Surface Wind Speed (m/s)
-2
SO2 Deposition Velocities (cm/s) for IRRIGATED CROPS Land Use Type.
-------
MESOPUFF-II
CCADM
,0123456
01
I I I I
2345678
Surface Wind Speed (m/s)
9 10
0
-1
0 1 23 4 5 6 7 8 9 10
Surface Wind Speed (m/s)
-2
S02 Deposition Velocities (cm/s) for GRAZED FOREST/WOODLAND Land Use Type.
-------
MESOPUFF-II
CCADM
.012345678911
"15 1 2 4 5 6 7 8
Surface Wind Speed (m/s)
0.1
0.2
0.1
I I
\ I
I I I
0
-1
0 1 2 3 4 5 6 7 8 9 10
Surface Wind Speed (m/s)
-2
SO2 Deposition Velocities (cm/s) for UNGRAZED FOREST/WOODLAND Land Use Type.
-------
MESOPUFF-I
CCADM
,01234567891
1 23456789 11
'-0 1
23456789
Surface Wind Speed (m/s)
io
2345678
Surface Wind Speed (m/s)
9 1i
SO2 Deposition Velocities (cm/s) for SEMIARID GRAZING Land Use Type.
-------
MESOPUFF-
oo
,0
T> 1 23456789 10
2345678
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
SO2 Deposition Velocities (cm/s) for OPEN WOODLAND GRAZED Land Use Type.
-------
MESOPUFF-I
CCADM
,01 234567891
0 1
i
UD
(ft
I/)
o
o
in
O
a.
x
LJ
'-0 1
2345678
Surface Wind Speed (m/s)
2345678
Surface Wind Speed (m/s)
SO2 Deposition Velocities (cm/s) for SWAMP Land Use Type.
-------
MESOPUFF-II
CCADM
2345678
Surface Wind Speed (m/s)
10~2 ~'
1 23456789 10
Surface Wind Speed (m/s)
-2
S02 Deposition Velocities (cm/s) for MARSHLAND Land Use Type.
-------
MESOPUFF-II
CCADM
,0 1
o 1
o
0)
o
Q.
x
LJ
-1
8 9 li
1 1 1 I 1 I 1 I
2345678
Surface Wind Speed (m/s)
,0123
tn
1 Z 1
O
0)
0
to
0
g-
LJ
0
1
9 11
8 9 1
1 1 1 1 1
1 1 1 1
2345678
Surface Wind Speed (m/s)
0
-1
9 10
SO2 Deposition Velocities (cm/s) for METROPOLITAN CITY Land Use Type.
-------
MESOPUFF-I
CCADM
,01234567891
in
O
O
0)
LJ
0
- o
-1
'15 1 2 3 4 5 6 7 8
Surface Wind Speed (m/s)
01 234567891
to
O
ZJ
U)
o
0 X °
LJ
o
2
1
0
-1
I
-
u
c
-
-
-
-
^
c
, r ,
o
T
D
C
I
^
en
D
~ /
)
en
~ \
-
I
in
: \ I
- , , \ ,
N
C
C
1 1
^
e
j
3
u
r
)
-------
MESOPUFF-I
CCADM
CO
3(
2
w
o 1
o
o
b
o
0
_ 1 00
_ i_ V
Vi\
\
\
o
- \ \jj \ \ 'O
\ l^M \ *C
, lot \ \
..
1
b
\~A
To
1\ "\
\ \, \ ,
4567891
I I I 1 I I
-
-
-
_
__
_
-
V~
-
.llllll
% 3
2 2
en
1 0 1
O
0)
L
CO
o
0 g- 0
LJ
-1 -1
31 234567891
1
-
I
1
o
o
tn
- o
b
-
_
_
-
_
1
1 II
in
IN
b
o
IN
b
o
.
U1
0
o
o
o
CP
1
1
o -
b -
in
b
p
OJ
o
in
(N
b
o
CN
b
\ °
\ _.
\ V
, \\,
1
-
-
_
(~} ^
Tf
b -
p
OJ
01
o
OJ
c
1
;>
1
1
_
-
_
-
t> 1 2345678910" t) 1 234567891
%
2
1
0
-1
o-2
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
SO4 Deposition Velocities (cm/s) for CROPLAND AND PASTURE Land Use Type.
-------
MESOPUFF-II
CCADM
,012345678910, ,01234567891
o
2
v>
(/)
0 1
O
0)
:5
O
Q. 0
x w
LJ
1
-
-
-
1 1 1 1 1 1 1 1
-
o
b
to
ui
0
O)
o
o
_
"
"\
\
_
~
V
\\\ ° 1
- \\\ 0
\\\ &
- \\\ &
\\\° \
~\\\'° \
\\\1 234567891
j j
2 2
M
1 0 1
O
0)
13
O
00-0
LJ
-1 -1
o-2 ~{
I
-
I I
I
TN
O
- O
^
_
~
1 1
1 1
1
in
d>
Tf
O
.
O
(M
C5
k
\
O
_
\
\ \
-
_
in
c
o
Jv
-\
\ \
: \ \ I
- , A , \ \ ,
1 1
D
_
_
_
1
31 234567891
- 2
- 1
- 0
1
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
S04 Deposition Velocities (cm/s) for CROPLAND/WOODLAND/GRAZING Land Use Type.
-------
MESOPUFF-II
CCADM
(ft
(/)
O
O
0)
v_
u
CO
O
Q.
X
LJ
01234567891
^
2
1
0
-1
-
in
03
o
b
o
_ b
CD
o
:
-
-
c
1
c
1 1 1 1 1 1 i
-
3
b -
co
en
o
01
0
b
3
b
I 00
- \p\ V
\ vM
\ 1
1*1
\\~-M
; Wfl
~
~
-
-
O
\ 0
\ ^
\ 1 234567891
- 2
- 1
- 0
o
o
cu
LJ
- -1
3
2
1
0
-1
31 234567891
-
"
1
1
o
b
p
~ b
en
_
-
_
-
""
c
1
C
r
1
1
in
CN
b
D
SI
o
O
'
en
3
0
L .
" ° \
b \
" 1 \
*-
U
0
1
1 1 1
O
"^
(
in
to
b
O
b
->
si
b
«*
o
CsJ
i
\ °
\
\ ^ \
\\
C
s
c
p
OJ
c
o
c
\ , \, \,\\
3
1
n
1
D
1
o
tn _
b
in
b
D
1-
D
_
-
_
-
~
o
en -
c
in
c
1
3
1
3
~
-
<3
2
1
0
-1
t> 12345678910^
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
S04 Deposition Velocities (cm/s) for IRRIGATED CROPS Land Use Type.
-------
MESOPUFF-II CCADM
,012345678910, ,012345678910,
j
2
CO
CO
o 1
O
9)
3
CO
o
Q. n
x w
LJ
-1
1
i i i i i i i i
p
- b -
10
-
*~
~
_
_
A
n
-
-
_
_
a
'U%
\ o \
~ \ \ S \
V \ *° \
; \ \ °, \ ;
\\i\ \ i \ i i i i i i i
_> j
2 2
W
V)
1 0 1
o
0)
^
CO
o
0 g- 0
LJ
-1 -1
1
-
1
1 1
CM
(D
- p
-
-
-
_
_
_
_i
1
1 f 1
o
c
0
4
O
LH
\N3
\
- \
1 \
^ \ \
^
ji
u
1
-
-
D
o
in
c
O
_N.
+
\
.*-*
\ \ ^
\ \ ^
: . \ ,\.\
\,
6
u_
_
_
_
-
T
c
1
D
1
J
2
1
0
1
i
t> 1 2345678910^ t) 1 2345678910^
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
SO4 Deposition Velocities (cm/s) for GRAZED FOREST/WOODLAND Land Use Type.
-------
MESOPUFF-II
CCADM
,0
7
8
i
t«
^>j
JD
O
3
1
O
c
V
D
C
c
o
ID
C
\
D
1
D
-
-
~
:
0
D -
-
-
-
J
2
1
0
-1
n
t> 123456789 10 ^
Surface Wind Speed (m/s)
SO4 Deposition Velocities (cm/s) for UNGRAZED FOREST/WOODLAND Land Use Type.
-------
MESOPUFF-II
CCADM
oo
,01 2345678910, ,01 234567891
3
2
co
10
0 1
O
0)
3
CO
0
o. n
X
LJ
-1
1
CO
o
~
-
_
~
_
J
1 1 1 1 1 1 1 1
-
_
_
o
b
IP
o
o
CO
_
-
\ ~
- x \ \ H
0 \ \
_ \ \
o \ \
- m \
\ \ »
S 1 1 °
" °\ 1 1 \
\ \l \ \ 1 \ 1 1 1 1 1 1
j j
2 2
in
tn
1 0 1
O
V
u
to
o
0 g- 0
Ld
-1 -1
1
-
i
1
o
»
o
in
- o
C)
_
-
_
-
-
_
1
1 1 1
in
CN
o
o
CM
c
p
1
cn
o
O
o
'o
D
1
-
-
_
p
OJ
o
CN
d
o
CN
c
in
\ ^.
\ T
\
\ ,
D
1
_
-
_
_
_
p -
OJ
c
1 1
3
1
t> 1 2345678910" t) 1 234567891
- 2
- 1
- 0
1
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
SO4 Deposition Velocities (cm/s) for SEMIARID GRAZING Land Use Type.
-------
MESOPUFF-II
CCADM
,012345678910, ,01234567891
o
2
in
0 1
6
0)
^
o
o. n
x u
LJ
-1
1 1
m
00
o
c
-
_
6
r
i i i i i i i
~
o
b
to
en
0
O)
o
b
0
b
i 0°
en
i
"~
~
_
i
u U\ \ \ J
^A\ °
:m % :
-\|\
\ ^^ \
- i\\\
\\ \
'o \
\ \
-ll,\ A 1
*j j
2 2
w
1 0 1
o
9)
=J
O
0 X °
Ld
-1 -1
I
-
1 1
i i
Q
io
- o
~
~
"
_
I I
I
in
c
*
b
fT)
C
o
K i
r>
\
u O
\~
iJ
y *
\ '
1 1 \ 1
D
:>
-
~
~
_
in
b -
0
*.
T
D
I I
I
~
-
;
-
t» 1 2345678910^" t> 1 234567891
- 2
- 1
0
1
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
SO4 Deposition Velocities (cm/s) for OPEN WOODLAND GRAZED Land Use Type.
-------
I
ro
o
MESOPUFF-II CCADM
012345678910, ,01234567891
j
2
in
0 1
O
o>
0
x 0
LJ
-1
'I
in
CO
0
b
~
_
1
1 1 1 1 1 1 1
"
o
b
to
cn
0
CD
O
b
o
b
_ 00
1
"
_
-
1
- V\\ v -\
r\\ *?>
\ \c? I \ *§»
\ol o V
- \o\ b \
I 1 1 1 ^>v \
- \M
- \\\
w \
°\ \
-tU A 1
^ w>
2 2
0)
(/)
1 0 1
O
OJ
ZJ
o
0 g- 0
LJ
-1 -1
o -:
-
"
\\ \
o
<
b
O
b
cn
~
"
_
-
-
c
I
1 1 1
p
N)
cn
0
M
Q
o
cn
3
o
* I
° \
o V
T \
\
1
1 1 1
O
^
O
in
n
b
o
OJ
0
in
b
O
CN
b
1
" \ ° I
\ ^
1 \ \ I
\\,
~
"
_
o
b
in
.
o
0
c
1
3
1
1
_
~
_
1
t) 1 2345678910*" t) 1 234567891
- 2
- 1
- 0
1
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
S04 Deposition Velocities (cm/s) for SWAMP Land Use Type.
-------
MESOPUFF-I
CCADM
,012345678910, ,01234567891
2
(/)
(/)
0 1
o
0)
D
CO
o
Q. n
X
bJ
-1
I
_
1 1 1 1 1 1 1 1
~
_
o
- b -
co
L
-
~
_
_
A
n
-
_
_
o
- \ \s
- \ o \
\ CD \
- \Y\ \ -
\ \ l\ 1 I \ I 1 1 1 1 1 1
j *j
2 2
(/)
CO
1 0 1
o
0>
^J
CO
0
0^0
LJ
-1 -1
I
"
_
-
1
f
1
SJ
C)
- 0
-
-
_
_
-
±
1
1
1
p
c
O
j
O
GJ
o
\fO
\
9 \
^
Jl
u
1
1
O
pn
u
3
CD
in
c
. Th
- 1 \ \ °
\ \
(
\ \ ^ \
\ \ X \
\\ \\
3
U
-/
-
-
-
-
_
_
CO
C
0
"X
D
C
\
D
J
1
3
-
-
-
"
-
"0 1 2345678910" t> 1 234567891
- 2
- 1
0
1
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
S04 Deposition Velocities (cm/s) for MARSHLAND Land Use Type.
-------
MESOPUFF-I
CCADM
,0
8
o
O
0
-1
o
CO
CD
"'O
I
2345678
Surface Wind Speed (m/s)
1 S 1
o
o
3
D
3
1
O
c
^
6
0
c
0
(D
C
\
D
xj
I
D
-
-
-
0
6 -
-
-
_
~
-
t> 1 234567891
- 2
- 1
0
1
Surface Wind Speed (m/s)
S04 Deposition Velocities (cm/s) for METROPOLITAN CITY Land Use Type.
-------
MESOPUFF-II
CCADM
ro
CO
V)
o
O
0)
L_
D
O
LJ
,012345678910, ,01234567891
J
2
1
0
-1
_
-
i
1 1
00
o
b
o
b
*N
- ID
O
" b
-
_
-
- \
j
c
c
1 1 I 1 1 I
_
r>
D
d
o
b
on
_
_
X ~~
\ \
\\o \ \
- \v° \ \
- \V\\ \
\ \ \ *
- \9\ \ o.
I 0 \ 0
:ni\\ \ '-
ul\,\, V -
*j o
2 2
V)
1 0 1
O
0)
i_
to
O
0 g- 0
LJ
-1 -1
I
I
I
o
d
m
- 0
d
-
~
-
_
_
-
_
I I
I I
in
CN
d
p
io
c
o
Ul
o
o
o
b
~ \
- \
If
3
1 1
in
r> _
c
O
Ol
O
in
CN
d
o
CN
c
-)
: \ t
- , I \
\ ,
5
I
D
-
~
_
_
_
p
CjJ
c
o
OJ
c
I
>
1
71
_
_
-
t) 1 2345678910^ t> 1 234567891
- 2
- 1
- 0
1
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
SO4 Deposition Velocities (cm/s) for LAKE OR OCEAN Land Use Type.
-------
MESOPUFF-I
CCADM
,0123456
4567891!
J»
r\>
2345678
Surface Wind Speed (m/s)
2345678
Surface Wind Speed (m/s)
9 10
NO2 Deposition Velocities (cm/s) for CROPLAND AND PASTURE Land Use Type.
-------
MESOPUFF-I
,0 1 2
I
ro
on
,0123
CCADM
456
7891!
J I
"'O
0
-1
1 23456789 10
Surface Wind Speed (m/s)
0 1 2 3 4 5 6 7 8 9 10
Surface Wind Speed (m/s)
-2
N02 Deposition Velocities (cm/s) for CROPLAND/WOODLAND/GRAZING Land Use Type.
-------
MESOPUFF-II
CCADM
1 23456789 1i
I
ro
O
O
0)
i_
3
10
O
Q.
x
LJ
2345678
Surface Wind Speed (m/s)
2345678
Surface Wind Speed (m/s)
9 10
-2
N02 Deposition Velocities (cm/s) for IRRIGATED CROPS Land Use Type.
-------
MESOPUFF-II
CCADM
,0123
1 234567891!
3>
I
ro
^j
23456789
Surface Wind Speed (m/s)
10
2345678
Surface Wind Speed (m/s)
9 10
-2
NO2 Deposition Velocities (cm/s) for GRAZED FOREST/WOODLAND Land Use Type.
-------
MESOPUFF-II
CCADM
,012345
4567891!
ro
oo
1 23456789 10
Surface Wind Speed (m/s)
1 23456789 10
Surface Wind Speed (m/s)
-2
NO2 Deposition Velocities (cm/s) for UNGRAZED FOREST/WOODLAND Land Use Type.
-------
MESOPUFF-I
CCADM
,0
r\j
VXD
2345678
Surface Wind Speed (m/s)
9 10
2345678
Surface Wind Speed (m/s)
9 10
-2
N02 Deposition Velocities (cm/s) for SEMIARID GRAZING Land Use Type.
-------
MESOPUFF-I
CCADM
CO
CD
67891
,01234567891
1 23456789 10
Surface Wind Speed (m/s)
T) 1 23456789
Surface Wind Speed (m/s)
NO2 Deposition Velocities (cm/s) for OPEN WOODLAND GRAZED Land Use Type.
-------
MESOPUFF-II
CCADM
CO
<=y
2345678
Surface Wind Speed (m/s)
2345678
Surface Wind Speed (m/s)
NO2 Deposition Velocities (cm/s) for SWAMP Land Use Type.
-------
MESOPUFF-I
CCADM
,0
CO
ro
O
0)
V)
O
CL
x
LJ
2345678
Surface Wind Speed (m/s)
9 10
2345678
Surface Wind Speed (m/s)
9 10
-2
NO2 Deposition Velocities (cm/s) for MARSHLAND Land Use Type.
-------
MESOPUFF-II
CCADM
,0
CO
oo
t/>
(/)
o
O
0)
v_
D
10
O
Q.
X
Ld
'o
2345678
Surface Wind Speed (m/s)
7
I
I
I
I
8
I
2345678
Surface Wind Speed (m/s)
9 1
0
-1
9 10
-2
NO2 Deposition Velocities (cm/s) for METROPOLITAN CITY Land Use Type.
-------
MESOPUFF-II
CCADM
,0 1
CO
1 234567891
I I I
in
o
q
/
M
b
2345678
Surface Wind Speed (m/s)
9 10
'O1
in
I] L
JO
UI
in
CN
I / I I
0
-1
2345678
Surface Wind Speed (m/s)
NO2 Deposition Velocities (cm/s) for LAKE OR OCEAN Land Use Type.
-------
MESOPUFF-II
CCADM
CO
in
3<
2
w
0 1
O
0)
D
(/)
O
LJ
-1
D 1 2 3 4
- c
q
->
en
-
_
_
;
1
1
to
b
.*
01
*>
'0
\
\
1 \ \
0 \ \ ^
'CP 0
- \ U \
\ \ "1
\ \ T
: \ \\,
t> 1 2 3 4
567891
11
'
in
rj
c
o
^4-
OJ
c
3
n
m
r>j
\
\
n
c
^
OJ
cn
\ °*
\°\
, \\
_
_
-
in
"*
D
i
_
_
-
-
rt
U i
^ -
-
% 3(
2 2
to
to
1 D 1
O
0)
^
to
o
0 Q. 0
Ld
-1 -1
5678910^ \
1 23456789 11
O
2
1
0
-1
1
-
i
1
b
- m
0
-
-
-
-
-
c
t-
0
In
\
3
\ \ -
1
.»
cn
/
: \T
- , 1 , V
\,
i
h
t
to
b
u
r
o
-------
MESOPUFF-II
CCADM
CO
(
2
10 -i
0 1
o
Q)
t
u
0
o. n
X
Ld
-1
31 234567891
1
-
-
-
_
-
I
c
_*
\
I I
M
r
o
: \ \
\ \
I
j
T
4
i
\ \ -
: , ,\A\
I
L
k
L
^
,
1
I
1
I
1
1
-
)
_
-
-
3
1
0 (
2 2
1 0 1
O
0)
:5
o
0^0
LJ
-1 -1
T "
31 234567891
1
-
-
,
-
I
r
_t
\ '
I 1
M
f
g
: \ \
\
\ \
1
^
T
J
^»
: , 1 .\,\.
1
L
*
L
!k
1
1
T
<1
T
(.
1
1
_
-
_
-
-
1
*
2
1
0
-1
T
t> 1 2345678910^ t) 1 2345678910^
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
HN03 Deposition Velocities (cm/s) for CROPLAND/WOODLAND/GRAZING Land Use Type.
-------
MESOPUFF-II
CCADM
,012345678910, ,01234567891
2
to
V)
5 1
0
u
to
0
^- 0
X
LJ
-1
- c
L
-
-
-
-
_
O
r
j
n
c
c
cn
b
\ \
o \
\
1 n
h
1
K)
a
D
vj
i
1
u
r<
o
n
M
cn
j
\\
\ \ . \
\l
\\\
c
1
1
^.
C
)
)
3
1
_
-
in
^ -
O
^
^
c
D
ro
i
\
71
1
t
u
<
1
-
-
_
-
-
T ~"
i
1
J J
2 2
V)
1 o 1
O
l_
u
V}
o
o g- o
LJ
-1 -1
I
_
-
-
-
-
-
_
-
-
i i
c
A
N
1
1 '\
1 1
M
1
^
f)
^
J
\
-f
- \ \ T
i i \ i \ iii
1 1
L
J-
1
-
-
-
-
-
_
in -
^
1
-
-
-
~
I
t) 1 2345678910" t) 1 234567891
- 2
- 1
- 0
1
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
HNO3 Deposition Velocities (cm/s) for IRRIGATED CROPS Land Use Type.
-------
MESOPUFF-I
CCADM
,012345678911
,012345678911
U)
oo
o 1
o
o>
D
o
Q. f)
X
UJ
-1
OB
cn
CD
T>1 2 34 5 6 7 8
Surface Wind Speed (m/s)
J
2
(/>
0 1
O
0)
U
o
8- o
Ld
-1
1
I
-
-
-
-
-
_
-
I
K
^
\ F
i
3
f
/
: \ \
\
: \
I
T
"->
^
I
u
t
i
t-
/
\(
I
>
i
-)
(.
, I A \\,/ ,
I
fs
1
3
"S
3
1
1
c
c
j
1
3
-
-
-
-
-
_
D -
-
-
"~
O
2
1
0
-1
->
312345678910^
Surface Wind Speed (m/s)
HNO3 Deposition Velocities (cm/s) for GRAZED FOREST/WOODLAND Land Use Type.
-------
MESOPUFF-II
CCADM
,012345678911
,01234567891!
CO
o 1
o
0)
O
9- 0
-1
ii
CD
(N
I I
U)
%
O
0)
0^0
LJ
I I I I I I
0
-1
1) 1 2 3 4 56 7 8 9 10 ^ "5 1 2 3 4 5 6 7 8 9 10
Surface Wind Speed (rn/s) Surface Wind Speed (m/s)
-2
HNO3 Deposition Velocities (cm/s) for UNGRAZED FOREST/WOODLAND Land Use Type.
-------
MESOPUFF-I
CCADM
,01 234567891
i
o
o 1
O
0
$
LJ
I I
o
p
In
I I
n
in
I I \ I \ I I I
in
ri
Ol
,01234567891
1 o 1
o
0
0
9-0
yv
Ld
""01 23456789 10
Surface Wind Speed (m/s)
in
I I i
in
m
M
b
1 1
en
q
r>
0
-1
1 23456789 10
Surface Wind Speed (m/s)
-2
HN03 Deposition Velocities (cm/s) for SEMIARID GRAZING Land Use Type.
-------
MESOPUFF-II
CCADM
*>
i
(
2
in
0 1
0
0)
D
O
LJ
-1
D 1 2 3
I
-
_
-
! 1
r
.»
\ - '
\i
0
\
,
t) 1 2 3
4567891
1
c
i
I
j
j
-i
\ \
| \ 1 1 1
1 1
L
i
I
i
I ,
1
>
-
_
T
_
-
-
1
, ,
2 2
1 S 1
o
0)
r>
0
0 g. 0
LJ
-1 -1
D 1 234567891
1
-
_
-
"
1 1
r
.*
\ '
V
1
M
P
M
\ V
1 1
*i
T
J
\ \ v
1 1 \ 1 \ \ 1
1
U
h
t
w
1
1 1
>
i
T
I
1
-
J -
_
-
_
-
-> ~
-
45678910^ tl 1234567891
- 2
- 1
- 0
1
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
HN03 Deposition Velocities (cm/s) for OPEN WOODLAND GRAZED Land Use Type.
-------
MESOPUFF-I
CCADM
3
2
10
0 1
O
0)
D
0
Q. O
X w
LJ
-1
j 1 234567891
T
-
-
-
-
i i
r
_»
\ '
1 1
si
p
3
\
\, \
: \ I
1 1
-1
)
-1
\
"\
i i \ i \ i n i
1 1
t
^
t
V
1 1
T
-
-
-
__
T
-
-
a (
2 2
w
1 0 1
O
0)
D
O
0 g- 0
LJ
-1 -1
_-? _:
01 234567891
""0 1 23456789 10
Surface Wind Speed (m/s)
o
2
1
0
-1
-
_ c
e
-
-
^
'
-
I
q
3
r
c
I
c
r
en
3
-
I 1
i- p I
^ \
\
I
in
3
si
h
c
1
t
r
D
rri
J
en
o
c\
__, I
K I \ U
\, \, \,\,
c
1
I
O
n
n
c
T
in
r*
3
?
, ,
t
1
-
-
-
-
D
-
_
-
I
1
0
-1
T> 1 23456789 10
Surface Wind Speed (m/s)
c-2
HNO3 Deposition Velocities (cm/s) for SWAMP Land Use Type.
-------
MESOPUFF-II
CCADM
,012345678911
,012345678911
I
CO
o 1
O
QJ
O
Q.
X
Ul
-1
00
1 f
O>
00
o
cn
O
0)
i_
ZJ
O
0 g- 0
LJ
i i r
03
cn
03
01
0
-1
~T) 1 2 3 4 5 6 7 8 9 10 'O 1
Surface Wind Speed (m/s)
23456789 10
Surface Wind Speed (m/s)
HNO3 Deposition Velocities (cm/s) for MARSHLAND Land Use Type.
-------
MESOPUFF-II
CCADM
,012345678911
,01234567891*
f 5 1
* O
O
g-
UJ
-1
I I
en
r\i
'15 1 2 3 4 5 6 7 8
Surface Wind Speed (m/s)
1 S 1
O
O
0^0
LJ
en
CT)
9 1~0 ^ "5 1
0
-1
2 3 4 5 6 7 8 9 10
Surface Wind Speed (m/s)
-2
HNO3 Deposition Velocities (cm/s) for METROPOLITAN CITY Land Use Type.
-------
I
tn
MESOPUFF-II CCADM
C
2
V)
0 1
O
i_
U)
O
o. n
X w
LJ
) 1 2
(N
C
-
-
_
_
_
D
r
c
in
0
3
>
c>
o
" ^\
- 9 \
o
'y
i n
- 1 V\
\ °\
- \
- \
t> 1 2
34567891
(L
I
3
6
\
OD
C
0
^
D
^
T
-
b
O)
O
O
v\
p
\o
A'01
\\\\
\
\
I
_.
j
n
*~
_.
io
_»
\f
\- \
M
O)
c
\\\
s>
u
_
V
-.
cn
->
T-'
_>
J
-
r
i.
j
M
C
O)
r
03
"~
r^.
r
-
c
"" \ \ -*
N
\v\
n
\\\
3 -
_
-
-
~
-
_
_
O
f
.-"
u
OD
r
\
D
vi
-
-
_
-
C
- in
o
-
"
-
_
_
-
D
1
_,
U1
O
T
o
1 "\
I
:
"' /
/
I
in
i T
1 , 1, \ ,1 ,
345678910^ t) 12345
67891
I I
1
KJ
t
M
b
ji
1
"
-
o
ri
in
r
o
r
si
_
_
-
-
-
/ °-
I
/i i
1
67891
*7
- 1
- o
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
HNO3 Deposition Velocities (cm/s) for LAKE OR OCEAN Land Use Type.
-------
MESOPUFF-I
CCADM
3
2
i/?
0 1
O
0)
rj
o
8- o
Ul
31 23456789 10.
"
A
00
o
6
" p
- b
CD
. o
_
_
_
_
1
\
o
b
to
Ul
o
O)
o
b
0
b
_ 00
\ W
Y^ \
\
O
1 1 1 1 1 1
-
-
_
:m % 1
,\0\\ 0
\\°\\ %
: \\\\ \ °\
-1 \, u
,
\ :
\i i i i i i
*j
2
In
1 O
O
123456789 10 *
3
2
s
o
01
O
S- o
LU
-1
^ l*_V^T,^V//l_>^/I
1
-
~
1
1
p
o
in
- o
o
-
-
-
-
_
c
1
1 U
in
(N
c
o
CM
C
3
Ul
o
o
o
'o
~ »
- \
i C
D
U
3
1
1 i
o -
Tf
C
in
c
O
c
c
~>
CM
b
O
CN
b
3
1 \ 1-
- \\
r , \, \\,
1
f*
3
3
C
D -
I
_
-
-
-
D -
Tj-
b -
a
Ul
p
OJ
c
1
>
1
1
_
-
_
-
0
-1
Surface Wind Speed (m/s)
2345678
Surface Wind Speed (m/s)
10
-2
NO3 Deposition Velocities (cm/s) for CROPLAND AND PASTURE Land Use Type.
-------
MESOPUFF-II CCADM
3
2
i/>
v>
0 1
O
0>
3
(0
O
Q. n
X
LJ
-1
31 234567891
"
_
1 I 1 1 1 1 1 1
"
o
b
CO
o
O)
o
*
_
r\
V
~
_
_
^
\\\ ° i
- \\\ o
\\\ "ft
\ \ ' °^
V \ o \
-\\ o \
\ \ \ ID \
\\\f\ \ \
\\\\\ \i K i i i i i i
% 31
2 2
ID
(/)
1 0 1
O
0)
D
(/)
O
0^0
UJ
-1 -1
31234567891
I
~
~
I I
I
CM
6
0
~
-
_
I I
1 1
1
in
d
*
d
^
d
CN
d
\
o
\ \
\ \
_
~"
-
'
_
_
in
c
p
-\
\ \
" \ \ \
I I \ I l\ \ I
1 1
D -
-
_
I
^
2
1
0
-1
n
t) 123456789 10" t> 1 23456789 10 ^
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
NO3 Deposition Velocities (cm/s) for CROPLAND/WOODLAND/GRAZING Land Use Type.
-------
MESOPUFF-II
CCADM
,012345678910,
J
2
to
o 1
O
0>
1
o
°- 0
X
LJ
-1
~
in
CD
O
b
~ o
_ b
CD
o
_
-
1
c
I I 1 I 1 1 1
~
>
b -
ID
01
o
en
q
b
o
b
00
h V\\
- \ p> \
" \°\
- \o\
:\W
~
-
«
-
A.
o ~
o* -
o \
o \
ID \
a \
\ \
\\ill \ i \ i \i i i i i i
j
2
w
1 0
o
CM
b
o
CM
<
\ °
j
(
\ £ \
I \ 1 \
, , \
3
C
D
1
-
o
in _
c
in
b
3
^r
C
o
Ol
c
o
b
\
1
n
1
3
D
_
_
-
p
en -
c
in
b
1
1
>
-
-
J
2
1
0
-1
o
t> 123456789 10^
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
NO3 Deposition Velocities (cm/s) for IRRIGATED CROPS Land Use Type.
-------
MESOPUFF-I
CCADM
,01 23456789 1Q, ,01 23456789 1Q
2
w
GO
D 1
O
£
to
0
o. n
X W
LJ
-1
1
"
i i i i i i i i
*~
p
- b -
ID
C
-
_
_
_
A
n
-
-
_
_
i
o
- \ V ° H
\ \ «g,
\ 0 \
- 1 Vs \
- \\1 \ "
\\i\ 1 i \ i i i i i i i
o o
2 2
(O
(/I
1 D 1
O
0)
CO
0
0 X °
LJ
-1 -1
I
~
-
i
I l
CM
6
- o
-
-
-
_
_
:
^
1
1 |l 1
O
c
o
4
o
{j* ,
Q
\NJ
\
°- \
\ \
\
^
ji
i
1
-
D
o
in
c
0
K.
+
1
O
\ \ \
t , \ \ \
\,
i
_
_
-
-
~
to
c
1
D
1
O
2
1
0
-1
T
"0 1 2345678910^ t> 1 2345678910^
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
N03 Deposition Velocities (cm/s) for GRAZED FOREST/WOODLAND Land Use Type.
-------
MESOPUFF-I
CCADM
,0
8
91
en
o
o
o
0)
l_
3
tO
O
LJ
-1
p
b
CD
CD
t>
i
2345678
Surface Wind Speed (m/s)
,0
8
9 1i
2 2
to
1 o 1
O
0)
l_
=J
o
0 X °
LJ
91
2 '
0 Z
I
- c
-
-
~
:
-
-
I
r
c
3
.k
c
I
^4
D
c
I
C
j
3
?
\to
\
- ° \
I
1
o
c
D
i
I/
C
\ *
: i \ \
I \ ° \
\ \ °; \
.\\ \\
;i
o
1 1
r
c
[>
0
->
3
1
o
c
"V
D
0
c
p
CD
C
\
J
1
1
D
-
-
:
o
D -
-
-
O
2
1
0
1
^
312345678910^
Surface Wind Speed (m/s)
NO3 Deposition Velocities (cm/s) for UNGRAZED FOREST/WOODLAND Land Use Type.
-------
MESOPUFF-I
CCADM
3
2
w
0 1
o
1 2345678910^ t) 1 234567891
- 2
- 1
- 0
1
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
N03 Deposition Velocities (cm/s) for SEMIARID GRAZING Land Use Type.
-------
MESOPUFF-I
CCADM
,0 1
^
2
in
0 1
i O
in
ro o>
1^
V)
O
g- 0
LU
-1
1
- OQ
0
CD
-
~
"~
-
_
o
o>
o
Q
O
o
_ 00
- b\\
\
\o \
\cP \
\O I
- o\
- v°\
\^l '
: \\\v
-
2345678910, ,01234567891
1
1 1 1 1 1 1 1
_
-
o
b
CD
en
~
~~
\
O
k
i O ~
1 ^
o \
\
o \
iD \
O \
\ \
\ l\ 1 \ 1 1 1 1 1 1
j j
2 2
(0
1 0 1
O
0)
13
O
0 g- 0
X
LJ
-1 -1
I
_
-
i i
I 1
p
M
- o
-
"
~
-
_
.
1 1
1
_
L
(
*
0
fo
o
O
ro
i
o
\
\
n
3
-
-
~
-
_
in
o -
o
tf*l
\ \ :
: \ \
i i \ i i \ i
3
1 1
1
_
-
~
_
-
t> 1 2345678910" *01 234567891
- 2
- 1
0
1
0
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
NO3 Deposition Velocities (cm/s) for OPEN WOODLAND GRAZED Land Use Type.
-------
MESOPUFF-I
CCADM
CTI
CO
CO
CO
£>
O
0)
i_
CO
o
QL
x
LJ
(
2
1
o
-1
)1 234567891
M
in
CO
o
b
-
'
_
1
1 1 1 1 1 1 1
^
o
b ~
ID
en
O
O5
o
b
0
b
_ CD
V (J1 i
- ?o\\
_
i
\ra\\ °
- U\\ -co
- \ \ . \ \ (J1
\O\ 1 o \
- \°\
. \-J\
- \Wl
i . \
0 \
IP \
o \
-\\\\\\,\ A '
% 3<
2 2
CO
CO
1 0 1
O
0)
3
CO
0
0 £ 0
X
LJ
-1 -1
31 234567891
-
~
|| |
0
b
o
~ b
e
"
_
n
I
^
1 1 1
p
fo
e
-»
o
o
en
-*
r~^
\
o
b \
i \
\
u
n
1
1 1
o
^]
c
in
n
cr
o
CJ
c
-)
CN
b
o
CN
b
\ o
\ \ '* \
'-
\\\,
3
D
C
D
-
-
*
b
in
t^i
b
o
b
1
1
1
_
-
_
~
1
%
2
1
0
i
T
t) 1 23456789 10 " t) 1 2345678910^
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
N03 Deposition Velocities (cm/s) for SWAMP Land Use Type.
-------
MESOPUFF-II
CCADM
in
-b
co
co
co
O
LJ
3(
2
1
°
-1
31234567891
1
c
__ ^
u
c
-
_
_
:\
i i i i i i i i
3
3
D
n
-
_
_
\
o
- \ b
:\s\ :
\ (D \
: \ \ °. \ :
\ \ i\ 1 i \ i i i i i i i
S 3C
2 2
CO
1 S 1
O
9)
to
0
0 x °
UJ
-1 -1
t> 123456789 10 ^ t
1 23456789 11
1
"
1
r
c
- p
-
~
~~
'
-
c
1
M
D
C
1
1
1
O
t
O
3
OJ
->
\Vo
\
- O \
\
(1
C
^ *
1 \ \ °
\ \ \
\ \ ° V
\ \ w» \
: ,\ \ \\
Jl
(J
1
r
c
3
0
1
D
1
O
C
Xi
D
a
c
p
vj
(D
C
\
D
1
1
D
-
-
~
:
3
D
-
-
_
-
O
2
1
0
_ i
1
312345678910^
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
NO3 Deposition Velocities (cm/s) for MARSHLAND Land Use Type.
-------
MESOPUFF-II
CCADM
,0
8
in
cn
V)
l/> .
o 1
o
0)
o
Q.
X
LJ
-1
O
b
CD
CD
t)
l
2 3 ' 4 5 6 7 8
Surface Wind Speed (m/s)
,0
8
91
0
u
V)
O
a.
x
LJ
v>
1 % 1
O
CD
0
1 -1
2 '
10
p
01
in
o
CD
O
CD
b
p
-vl
1
2345678
Surface Wind Speed (m/s)
p
bo
-------
<
in
o>
MESOPUFF-II CCADM
,0 123456789 1Q ,01 23456789 1Q,
^
2
V)
V)
O
o>
OT
O
o. n
x w
LiJ
i
,_
,
-
- ti
I
1 1
(D
O
6
o
b
v
D
o
CD
_
_
-
k
- \
-1
r
1 1 1 1 1 1
_
^
\s*
o
o
o
b
00
_
_
\
i \ -1
\\0 \ \
\ \ . \ A fc
1 \ O V \
- \t \
- Ip \ i
\ 0 \ 0
:1H\ t :
-Ul\,\, V -
J J
2 2
(n
V)
1 0 1
0
0)
l_
V)
o
0 g- 0
UJ
-1 -1
I
_
1
1
o
6
in
- o
b
_
"
_
_
_
C
i r
1 1
in
(N
c*
p
M
c
3
!_,
m
o
fi
o
o
~ \
- \
u
D
_y
1 1
_
in
**? -
c
o
OJ
c
in
CN
CJ
0
CM
C
~>
\ \ *":
: \ v
F , 1 \\ ,
3
1
3
D
_
~
_
_
_
O
OJ
c
p
OJ
c
1
D
1
n
_
-
o
2
1
0
-1
i
t) 1 2345678910^ t> 1 2345678910^
Surface Wind Speed (m/s)
Surface Wind Speed (m/s)
NO3 Deposition Velocities (cm/s) for LAKE OR OCEAN Land Use Type.
------- |