United Statea
Environmental
Protection
Agancy
Off tea of Air Quality
Planning and Standarda
Raaaareh Triangla Park, NC 27711
EPA-450/4-90-006B
Sl 1990
AIR
SEPA
URBAN AIRSHED MODEL
STUDY OF FIVE CITIES
Demonstration of Low-Cost Application of
the Model to the City of Atlanta and the
Dallas-Fort Worth Metroplex Region
-------
EPA-450/4-90-006B
URBAN AIRSHED MODEL
STUDY OF FIVE CITIES
Demonstration of Low-Cost Application of
the Model to the City of Atlanta and the
Dallas-Fort Worth Metroplex Region
By
Ralph E. Morris
Thomas C. Myers
Edward L. Can-
Marianne C. Causley
Systems Applications, Inc.
101 Lucas Valley Road
San Rafael, CA 94903
EPA Project Officers:
Richard D. Scheffe, Office of Air Quality Planning and Standards
John C. Chamberlin, Office of Policy Planning and Evaluation
OFFICE OF AIR QUALITY PLANNING AND STANDARDS
U. S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
APRIL 1990
-------
Disclaimer
This material has been funded wholly or in part by the United
States Environmental Protection Agency. It has been subject to
the agency's review, and it has been approved for publication as
an EPA document. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
-------
Contents
List of Figures iv
List of Fables vii
1 INTRODUCTION 1
Use of the Urban Airshed Model 1
The "Five Cities" UAM Study 2
2 PLANR PROCEDURES FOR PREPARING UAM INPUTS 4
Episode Selection 6
Data Assessment and Meteorological Characterization 7
Preparation of UAM Input Files 7
Diagnostic Simulations to Define the Base Case 21
3 APPLICATION OF THE UAM TO THE CITY OF ATLANTA 25
Episode Selection 25
Meteorological Characterization 27
Modeling Domain Definition 28
Routine Data Available for Atlanta 28
Preparation of UAM Input Files 32
Definition of Base Case 43
H APPLICATION OF THE UAM TO THE DALLAS-FORT WORTH
MEXTROPLEX AREA 56
Episode Selection 56
Meteorological Characterization 57
Modeling Domain Definition 68
Routine Data Available for Dallas-Fort Worth 68
Emissions Preparation 71
Preparation of UAM Input Files 74
Tracer Simulation 95
Evaluation Simulations 96
-------
5 SUMMARY AND CONCLUSIONS ,
Twice-Daily Upper-Air Soundings
Surface Wind Speeds ,
Air Quality Data ,
Emissions
Conclusions
References
Appendix A: Percent Contribution of Initial Concentrations,
Boundary Conditions (Four Lateral Faces Plus Top
Boundary) Anthropogenic Area Source Emissions,
Point Source Emissions, and Biogenic Emissions
to Hourly Tracer, NOX, and VOC Concentrations for
Atlanta on * June 198*
Appendix B:
Appendix C:
Appendix D:
Appendix E:
Appendix F:
Hourly Predicted Ozone Concentrations for Atlanta
Diagnostic Run 1
Hourly Predicted Ozone Concentrations for Atlanta
Diagnostic Run 2
Sensitivity Analysis of the Calculations of Ozone
Concentrations to Wind Speed Specification
Memorandum of 10 May 1989 on the Selection of a
UAM Modeling Episode for the Dallas-Fort Worth
Metroplex Area
Initial Condition Tracer Simulation for Dallas-
Fort Worth
Appendix G: Predicted Hourly Ozone Concentrations for Dallas-
Fort Worth Diagnostic Run 1
Appendix H: Predicted Hourly Ozone Concentrations for Dallas
Fort Worth Diagnostic Run 2
17*
176
177
177
179
180
iii
8910%rl
-------
Figures
2-1 PLANR procedures for applying UAM ............................. 5
2-2 Flow diagram for creating the UAM CHEMPARAM input file ......... 10
2-3 Flow diagram for creating the UAM TERRAIN input file ............. 12
2-4 Flow diagram for creating the UAM DIFFBREAK and REGIONTOP
input files ..................................... . ............... 13
2-5 Flow diagram for creating the UAM METSCALARS input files ........ 15
2-6 Flow diagram for creating the UAM TEMPERATUR input file ........ 18
2-7 Flow diagram for creating the UAM WIND input files ............... 19
2-8 Flow diagram for creating the UAM 1985 base case low-level
emissions (EMISSIONS) and elevated emissions (PTSOURCE)
files [[[ 22
3-1 UAM modeling domain for Atlanta ................................ 29
3-2 Emissions for the Atlanta 1984 base case .......................... 37
3-3 Observed and predicted ozone concentrations for the Atlanta
diagnostic run 1 ................................................ 45
3-4 Observed and predicted ozone concentrations for the Atlanta
evaluation run 2 ................................................ 48
3-5 Observed and nearest-neighbor predicted ozone concentrations for
the Atlanta diagnostic run 2 ..................................... 49
3-6 Observed and nearest-neighbor predicted ozone concentrations for
the Atlanta diagnostic run 2 ..................................... 50
3-7 Scatterplot residual analysis and model performance
statistics for predicted and observed hourly ozone
concentrations for the Atlanta diagnostic run 2 ..................... 51
-------
4-2 Constant pressure heights at 0600 on 31 August 1985 ................ 60
4-3 Constant pressure heights at 1800 on 31 August 1985 ................ 62
4-4 Observed daily maximum ozone concentrations in the Dallas-
Fort Worth area ...................................... , . . , ...... 64
4-5 Locations of upper-air meteorological observation sites and
UAM modeling domain for the Dallas-Fort Worth metroplex area ..... 69
4-6 Estimated diurnal emissions of monoterpenes and isoprene
in Tampa, Florida, for an average of summer days .................. 75
4-7 1985 emissions for Dallas-Fort Worth ............................. 76
4-8 Time series of predicted and observed hourly ozone
concentrations for the Dallas-Fort Worth diagnostic run 1 ........... 97
4-9 Scatterplot, residual analysis, and model performance
statistics for predicted and observed hourly ozone
concentrations on 30 August 1985 in Dallas-Fort Worth
for diagnostic run 1 ............................................. 103
4-10 Scatterplot, residual analysis, and model performance
statistics for predicted and observed hourly ozone
concentrations on 31 August 1985 in Dallas-Fort Worth
for diagnostic run 1 .... ......................................... 105
4-11 Time series of predicted and observed hourly NO concentrations
in Dallas-Fort Worth for diagnostic run 1 ..... . .................... 108
4-12 Time series of predicted and observed hourly N©2 concentrations
in Dallas-Fort Worth for diagnostic run 1 .......................... 113
4-13 Time series of predicted and observed hourly CO concentrations
in Dallas-Fort Worth for diagnostic run 1 .......................... 118
4-14 Isopleths of predicted daily maximum ozone concentrations
with superimposed observations on 30 August 1985 for the
Dallas-Fort Worth diagnostic run 2 ............................... 124
4-15 Time series of predicted and observed hourly ozone concentrations
for the Dallas-Fort Worth diagnostic run 2 ......................... 125
4-16 Scatterplot, residual analysis, and model performance
statistics for predicted and observed hourly ozone
concentrations on 30 August 1985 in Dallas-Fort Worth
for diagnostic run 2 ............................. . ............... 131
v
-------
4-17 Isopleths of predicted daily maximum ozone concentrations
with superimposed observations on 31 August 1985 for
diagnostic run 2 ................................................ 133
4-18 Scatterplot, residual analysis, and model performance
statistics for predicted and observed hourly ozone
concentrations on 31 August 1985 in Dallas-Fort Worth
for diagnostic run 2 .............................................
4-19 Time series of predicted and observed hourly NO concentrations
in Dallas-Fort Worth for diagnostic run 2 .......................... 137
4-20 Time series of predicted and observed hourly NO2 concentrations
in Dallas-Fort Worth for diagnostic run 2 .......................... 142
4-21 Time series of predicted and observed hourly CO concentrations
in Dallas-Fort Worth for diagnostic run 2 .......................... 147
4-22 Time series of predicted and observed hourly ozone concentrations
for the Dallas-Fort Worth diagnostic run 2 using a one-cell
search nearest neighbor analysis .................................. 153
4-23 Time series of predicted and observed hourly NO concentrations
(ppb) for the Dallas-Fort Worth diagnostic run 2 using a
one-cell search nearest neighbor analysis .......................... 157
4-24 Time series of predicted and observed hourly N©2 concentrations
(ppb) for the Dallas-Fort Worth diagnostic run 2 using a
one-cell search nearest neighbor analysis .......................... 160
4-25 Time series of predicted and observed hourly ozone concentrations
for the Dallas-Fort Worth diagnostic run 2 using a two-cell
search nearest neighbor analysis ..................... . ............ 163
4-26 Time series of predicted and observed hourly N©2 concentrations
(ppb) for the Dallas-Fort Worth diagnostic run 2 using a
two-cell search nearest neighbor analysis .......................... 167
4-27 Time series of predicted and observed hourly ozone concentrations
for the Dallas-Fort Worth diagnostic run 2 using a two-cell
search nearest neighbor analysis .................................. 170
vx
-------
Tables
2-1 Surface roughness and deposition factors based on studies by
Argonne National Laboratories 11
2-2 Exposure class (CE) classification based on cloud cover and
solar zenith angle 16
3-1 Highest ozone days in Atlanta, 1984-1987 26
3-2 Surface meteorological observation sites, locations, and variables
in the Atlanta region 30
3-3 Upper-air observation sites, locations, and distance from
Atlanta used in this study 31
3-4 Ozone monitor names and locations 31
4-1 Surface meteorological observation sites at the continuous
air monitoring stations 70
4-2 Upper-air meteorological observation sites in the vicinity of
Dallas-Fort Worth 72
4-3 Surface air quality monitors in the Dallas-Fort Worth area in
operation during 21-31 August 1988 73
4-4 Summary of emission totals for the Dallas-Fort Worth
modeling domain 90
Vll
a i na,* i
-------
1 INTRODUCTION
The job of reducing ozone concentrations to levels below the National Ambient Air
Quality Standard (NAAQS) has proven to be far more difficult than was thought when
the Clean Air Act was passed and amended. The level of ozone precursor emissions
remains too high; either emission reductions have been too small or have been
required of the wrong sources, or both.
A plethora of technical explanations has been offered for failure to attain the ozone
standard. These include perceived weaknesses in the attainment planning process
(Federal Registrar, Vol. 52, No. 226, November 24, 1987; OTA 1988a,b,c), incomplete
understanding or recognition of the anthropogenic and natural factors that cause ele-
vated tropospheric ozone levels (Science, 1988), the failure to consider the effects of
natural emissions (Chameides et al., 1988; Morris et ai., 1989), use of a simplistic
modeling approach (OTA, 1988a; Seinfeld, 1988a; Burton, 1988), and failure to reduce
the amount of emissions intended, either through overestimates of the effectiveness
of control technology or failure to account for certain categories of emission
sources. The EPA, after lengthy consideration, has proposed a comprehensive policy
that includes major changes in the planning process for reducing ozone concentra-
tions (Federal Registrar, Vol. 52, No. 226, November 24, 1987). These changes
include improvements in modeling practices and requirements for improving the data
to support improved modeling practices. The EPA is now evaluating public com-
ments on the proposed policy.
USE OF THE URBAN AIRSHED MODEL
The EPA recommends that states use the Urban Airshed Model (UAM) for the model-
ing of ozone and photochemical reactive pollutants in urban areas (EPA, 1986). An
alternative approach, the Empirical Kinectics Modeling Approach (EKMA), has been
accepted for demonstrating attainment of the ozone standard in most State Imple-
mentation Plans (Federal Registrar, Vol. 52, No. 226, 1987). The UAM and EKMA
are quite different types of models; the UAM is a three-dimensional grid model while
the EKMA is a trajectory box model.
A reluctance to use the UAM in the past is based on the perception that it requires
using data from costly intensive measurement studies and requires extensive compu-
tational resources. Most of the cost of applying the UAM is attributed to the prac-
tice of conducting an extensive evaluation of UAM performance, which usually
8910trl 2
-------
entails many diagnostic simulations. This evaluation enables us to understand why
the UAM performs as it does for a particular application and, if deemed necessary,
to take actions to improve model performance. Historically, it has been expected
that the UAM will calculate hourly ozone concentrations to within approximately 15
to 20 percent of the observed peak value (Seinfeld, 1988a; Burton, 1988). More
recent applications of the UAM to the Los Angeles basin have used routinely avail-
able meteorological data and predicted observed ozone levels with a high degree of
skill (Seinfeld, 1988a; Burton, 1988; Hogo, Mahoney, and Yocke 1988). A recent
application of the UAM to the New York metropolitan area used simple inputs, i.e.,
constant wind fields and mixing depths (Rao 1987).
This simplified use of the UAM, relying on routinely available data and reducing the
requirement for strict evaluations of model performance, offers air quality managers
a practical air quality assessment tool for identifying emission control strategies
that demonstrate attainment of the ozone NAAQS. This simplified approach is cal-
led Practice-for-Low-Cost-Airshed-Application-for-Nonattainment-Regions
(PLANR). The PLANR use of the UAM requires almost the same quantity and
quality of inputs as EKMA, and the overall application cost is substantially reduced.
The possible exception is the emissions inventory, which in PLANR applications
should contain the same spatially (horizontally and vertically) and temporally varying
emissions used in standard UAM applications (such detail is necessary to account for
the differing reactivities of VOC emissions). However, local agencies generally have
emissions inventories at hand; in addition, UAM input inventories can be readily esti-
mated from existing national emissions inventories (e.g., the National Acid Precipi-
tation Program 1980 and 1985 inventories). Knowledge of current emission rates is
needed to estimate the emission controls required to achieve attainment of the
ozone NAAQS.
The PLANR use of the UAM may not be appropriate for all nonattainment regions.
When attainment is expected to be imminent, improved methods for using EKMA
may be adequate. In other, more complex situations, such as the Los Angeles basin,
the Houston region, and the New York Metropolitan area among others, the com-
plexity of meteorological conditions and the emissions distribution and the severity
of the ozone attainment problem probably require a more detailed application of the
UAM. The application of UAM to these more complex situations, called Practice-of-
Airshed-Application-in-Complex-Regions (PACR), would involve more extensive
model performance requirements and hence more diagnostic simulations, and a
resultant increase in costs. However, even for a complex nonattainment region, the
PLANR approach would probably be more comprehensive and reliable than EKMA for
estimating the controls needed to achieve ozone attainment.
THE "FIVE CITIES" UAM STUDY
The EPA has funded a study of the PLANR approach in five urban areas in the U.S.
(New York, St. Louis, Atlanta, Philadelphia, and Dallas-Ft. Worth). The main objec-
tives of this "Five Cities" study are to:
8910>»rl 2
-------
(1) Demonstrate the usefulness of PLANR for air quality planning;
(2) Determine the effects of alternative fuels and alternative Reid vapor
pressure values for fuels on urban ozone concentrations;
(3) Demonstrate the use of PLANR to evaluate SIP control strategies; and
(4) Transfer the UAM model, modeling data bases, and applications tech-
nology to the states for use in future SIPs.
In addition, the study includes two city-specific analyses:
(1) For the St. Louis and Philadelphia areas, comparison of the PLANR use of
the UAM (i.e., using only routinely available data) with applications of
the UAM that use an extensive data base; and
(2) The effects of biogenic emissions on anthropogenic emission reductions in
the Atlanta area.
Previous reports on the "Five Cities" study have documented the PLANR use of UAM
and the evaluation of alternative fuel emission scenarios for the New York metro-
politan area and the city of St. Louis (Morris et al., 1989a); analyses of the effects of
biogenic emissions on the anthropogenic VOC emission reductions required to meet
attainment of the ozone NAAQS for the city of Atlanta (Morris et al., 1989c); and an
evaluation of the PLANR use of the UAM for the city of St. Louis (Morris, Myers,
and Carr, 1989).
This report describes the PLANR application of the UAM to the city of Atlanta and
the Dallas-Fort Worth metroplex area.
89104rl 2
-------
2 PLANR PROCEDURES FOR PREPARING UAM INPUTS
The Urban Airshed Model (UAM) is a three-dimensional grid model designed to calcu-
late the concentrations of both inert and chemically reactive pollutants by simula-
ting the physical and chemical processes in the atmosphere that affect pollution con-
centration. The basis for the UAM is the atmospheric diffusion or species continuity
equation. This equation represents a mass balance in which all of the relevant emis-
sions, transport, diffusion, chemical reactions, and removal processes are expressed
in mathematical terms. The model is usually applied to an 8- to 72-hour period dur-
ing which adverse meteorological conditions result in elevated pollutant concentra-
tions of the species of interest.
The basic UAM model formulation is discussed elsewhere (Ames et al., 1985a,b;
Reynolds, Tesche, and Reid 1979). The version of the UAM employed in this study
differs from the previous versions in that it includes the latest version of the Carbon
Bond mechanism, the CB-IV (Gery, Whitten, and Killus 1988), and the advection/dif-
fusion algorithm has been updated (Smolarkiewicz, 1983). The description of these
updates to the UAM are discussed elsewhere (Hogo, Mahoney, and Yocke, 1988;
Morris et al., 1989a,b).
Since the release of the first version of the UAM in 1979-1980, there have been vari-
ous improvements not only in the UAM itself but in the procedures used to develop
UAM inputs. The 1979-1980 UAM preprocessors rely on extensive measurement data
and then interpolate the measured data to provide the gridded fields of input data
required by the UAM. More recently, procedures have been developed for generating
fields of meteorological inputs for air quality models using limited data availability
(Morris et al., 1988, 1989a,b,c). These routines have been incorporated into the
PLANR procedures for preparing UAM inputs.
As shown in Figure 2-1, the PLANR procedures for applying the UAM involves the
following activities:
(1) Identification of the modeling episode.
(2) Assessment of available data (meteorological and air quality) for the
episode.
(3) Characterization of the meteorological conditions during the episode to
develope a conceptual picture of the conditions that led to the elevated
ozone concentrations.
8910«*rl 2 4
-------
Episode selection
Meteorological characterization
Domain definition
Emissions preparation
Data assessment and adjustment
to represent the conceptual
picture of the meteorological
conditions of the episode
UAM diagnostic simulation
UAM model
performance evaluation
Is
UAM
model performance
Project emissions to
future year
Evaluate alternative emission
control strategies
FIGURE 2-1. PLANR procedures for applying UAM.
EEE89104
-------
fa) Definition of the modeling domain, including spatial extent, horizontal
grid spacing, and number of vertical layers.
(5) Preparation of the TERRAIN UAM input file.
(6) Preparation of the basfe case emissions inventory (the EMISSIONS file
and input to the PTSOURC preprocessor of UAM)
(7) Adjustment of the UAM preprocessor meteorological and air quality
input data to represent the conceptual picture of the meteorological
conditions.
(8) Preparation of the UAM meteorological input files.
(9) Simulation of an inert tracer simulation to define starting time
(optional).
(10) Run the point source emissions preprocessor to prepare the PTSOURCE
file and exercise the UAM for a diagnostic simulation.
(11) Evaluate whether the diagnostic simulation is deemed adequate as a
base case. If model performance is considered inadequate, go back to
(7).
(12) Project emissions to future year(s) (optional)
(13) Analyze alternative emission control strategies.
EPISODE SELECTION
The first activity of any UAM modeling study is to select an episode of high pollution
levels. The model's ability to replicate the measured pollutant levels will then serve
as a test of its ability to simulate accurate pollutant levels under the same condi-
tions but with different emission parameters. The episode for an ozone analysis
should satisfy the following criteria:
High and widespread ozone concentrations
No atypical meteorological conditions
Completeness of routine data
Recent occurrence of one of the highest observed ozone concentrations
8910%rl 2
-------
The selection of a UAM modeling episode consists of the following activities:
Ranking of the days from several recent years based on the maximum ozone
concentration for each day with ozone concentrations greater than the ozone
NAAQS;
•
Definition of possible modeling episodes (one day or several days) from group-
ings of the highest observed ozone days;
Examination of the number of station-hours (i.e., total number of hours at any
station) within each possible modeling episode and number of ozone monitors
exceeding the ozone NAAQS for each episode. Elimination of days without
widespread elevated ozone conditions (i.e., those episodes with isolated
localized ozone peaks);
Examination of the meteorological conditions for each candidate episode, and
elimination of those episodes experiencing atypical meteorological conditions;
Assessment of data availability, and elimination of candidate episodes for
which sufficient routine data are not available;
Consideration of other factors, such as other motivations for the study;
Conferring with state, local, and federal agencies for the final selection.
DATA ASSESSMENT AND METEOROLOGICAL CHARACTERIZATION
After a modeling episode has been selected, the next activity is the assessment and
acquisition of all surface meteorological data in the vicinity of the region and upper-
air data within about 500 km of the area of interest. By analyzing this data and daily
weather maps, a conceptual picture of the meteorological conditions that led to the
elevated ozone concentrations is developed. This meteorological characterization is
necessary because this conceptual picture will ultimately decide how the routine
meteorological data will be interpreted and interpolated for input into the UAM pre-
processors.
PREPARATION OF UAM INPUT FILES
The following 13 input files are required for UAM modeling analyses.
DIFFBREAK contains the daytime mixing height or nighttime inversion height for
each column of cells at the beginning and end of each hour of the simulation.
8910>»rl 2
-------
REGIONTOP contains the height of each column of cells at the beginning and end of
each hour of the simulation. If this height is greater than the mixing height, the cell
or cells above the mixing height are assumed to be within an inversion.
WIND contains the x and y components of the wind velocity for every grid cell for
each hour of the simulation. The maximum wind speed for the entire grid and
average wind speeds at each boundary for each hour are also included in this file.
METSCALARS contains the hourly values of the meteorological parameters that do
not vary spatially. These scalars are the NC>2 photolysis rate constant, the concen-
tration of water vapor, the temperature gradient above and below the inversion base,
the atmospheric pressure, and the exposure class.
AIRQUALITY contains the initial concentrations of each species for each grid cell at
the start of the simulation.
BOUNDARY contains the location of the modeling region boundaries. This file also
contains the concentration of each species that is used as the boundary condition
along each boundary segment at each vertical level.
TOPCONC contains the concentration of each species for the area above the model-
ing region. These concentrations are the boundary conditions for vertical integra-
tion.
TEMPERATUR contains the hourly temperature for each surface layer grid cell.
EMISSIONS contains the ground-level emissions of NO, NO2» seven carbon bond cate-
gories, and CO for each grid square for each hour of the simulation.
PTSOURCE contains the point source information (stack height, temperature and
flow rate, the plume rise, the grid cell into which the emissions are emitted) and the
emissions rates for NO, NO2> seven carbon bond categories, and CO for each point
source for each hour.
TERRAIN contains the value of the surface roughness and deposition factor for each
grid square.
CHEMPARAM contains information about the chemical species to be simulated,
including reaction rate constants, upper and lower bounds, activation energy, and
reference temperature.
SIMCONTROL contains the simulation control information, such as the time of the
simulation, file option information, default information, and information on integra-
tion and chemistry time steps. .
Each of these files is explained in detail below.
8910trl 2
-------
CHEMPARAM
The CHEMPARAM file contains the information required by the numerical
algorithms used to solve the Carbon Bond IV chemical kinetics mechanism in the
UAM(CB-IV). The file only needs to be created once for all UAM simulations. It is
created by the chemistry preprocessor program (cprep.f) using the CHEMCB4.INP
file as input (see Figure 2-2).
TERRAIN
Information in the TERRAIN file is created by analyzing USGS topographical maps
and/or land-use maps if available. Each grid cell of the modeling domain is assigned
an integer between 1 and 13 that represents the most dominant land use category as
listed in Table 2-1. This two-dimensional array of integers is then processed by the
UAM terrain preprocessor (creter.f), which assigns to each grid cell a surface rough-
ness and vegetation factor based on one of the 13 land use categories and is format-
ted into the proper format for the UAM (see Figure 2-3).
DIFFBREAK and REGIONTOP
The preparation of the DIFFBREAK file is a three-step procedure that uses surface
meteorological data and a representative upper-air sounding and excercising three
preprocessor programs (see Figure 2-4): (1) the mixing height program (e.g.,
mixht.f), which calculates the maximum daily mixing height at the location of each
surface meteorological observation site; (2) the RAMMET program (rammet.f),
which calculates hourly mixing heights at each surface meteorology site based on the
maximum daily mixing height at the site, the height of the base of the nocturnal
inversion from the morning sounding, and the surface data at the site; and (3) the
UAM diffusion break preprocessor (dfnsbk.f), which spatially interpolates the mixing
heights from each surface meteorology site to form the gridded field required by the
UAM for each hour of the simulation. The REGIONTOP for the PLANR applications
performed to date is defined as 50 - 100 meters above the maximum mixing height.
Some care must be taken in chosing a representative upper-air sounding for the cal-
culation of mixing heights. The location of the sounding should not be separated
from the modeling region by a large terrain feature or by such meteorological condi-
tions as fronts. In addition, since ozone episodes usually occur during limited mixing
conditions, the calculated mixing heights should be compared with climatological
mixing heights for the region interest. If the calculated mixing heights greatly
exceed the average summer season maximum mixing height for the region, and a
UAM diagnostic simulation tends toward underprediction of the observed peak ozone,
8910Hrl 2
-------
CHEMCB.4INP
( cprep. f J
CHEMPARAM
FIGURE 2-2. Flow diagram for creating the UAM CHEMPARAM
input file.
EEE89104
ID
-------
TABLE 2-1. Surface roughness and deposition factors
based on studies by Argonne National Laboratories.
Land Use Surface Roughness Deposition
Category (meters) Factor
1
2
3
4
5
6
7
8
9
10
11
Urban
Agricultural
Range
Deciduous Forest
Coniferous Forest
including wetland
Mixed Forest
Water
Barren land
Nonforest Wetlands
Mixed Agricultural
and range
Rocky (low shrubs)
3.00
0.25
0.05
1.00
1.00
1.00
0.0001
0.002
0.15
0.10
0.10
0.2
0.5
0.4
0.4
0.3
0.3
0.03
0.2
0.3
0.5
0.3
8910U 3
11
-------
USGS topographic and/or
land use maps
Objective
assignment of each
UAM grid cell to one of
13 land use
categories listed in
Table 2-1
( crcter. f J
^
TERRAIN
FIGURE 2-3. Flow diagram for creating the UAM TERRAIN
input file.
EEE89104
12
-------
Surface
meteorological data
Upper-air
meteorological data
mixht. f
.
1
r
Maximum daily mixing
height at a representative
surface meteorological site
Objective
identification of
' nighttime DIFFBREAK ^
using top of nocturnal
inversion from
morning sounding
Minimum daily
mixing height
JLf.
I rammet. f J
Hourly mixing heights
at each surface site
( dfnsbk.f J
DIFFBREAK
REGIONTOP
FIGURE 2-4. Row diagram for creating the UAM DIFFBREAK
and REGIONTOP input files.
EEE89104
13
-------
then the mixing heights should be adjusted downward. In the PLANR applications
performed to date, the mixing height algorithm of Kelley (1981) (mixht.t) was used,
although other algorithms may be used at the user's discretion.
METSCALARS
Several hourly varying but spatially invariant meteorological parameters need to be
defined. The atmospheric pressure in atm (ATMOSPRESS) and water vapor concen-
trations (CONCWATER) are defined as the average value of the hourly observations
from all of the surface meteorological sites. Note that if atmospheric pressure is
not reported then a value of 1.0 is adequate.
The NO2 photolysis rate (RADFACTOR) is usually calculated based on the solar
zenith angle for the location (latitude/longitude) and time (hour and day) of interest
using the sunfunc.f program (see Figure 2-5). Adjustments for the attenuation of
solar radiation and consequently the NO2 photolysis rate due to clouds can also be
accounted for. The NO2 photolysis rate can also be directly estimated from mea-
sured insolation data, but this data is usually not routinely available. The exposure
class (EXPCLASS) is calculated from cloud cover observations and the solar zenith
angle, which is also given in the sunfunc.f program following the procedures given in
Table 2-2.
The temperature gradients below (TGRADBELOW) and above (TGRADABOVE) the
inversion (mixing height) are estimated from the representative upper-air soundings
and the observed surface temperatures. An interpretation of the upper-air sounding
and surface temperatures must be made to determine the temperature gradient
inputs. This interpretation is especially difficult when using only the twice-daily
routine upper-air data. The need to represent the typically complex vertical struc-
ture of the atmosphere with just two temperature gradients will demand some com-
promises in determining their values. First, the modeling episode is split into two
regimes: a daylight regime, during which DIFFBREAK represents the convective
mixed-layer, and a nightime regime, when the DIFFBREAK represents the top of a
surface-based inversion.
For the daylight regime, TGRADBELOW is based on the afternoon sounding of the
day to be modeled and the exposure class. For example, for the Atlanta episode the
temperature gradient from the surface to the mixing height ranged from -0.0096 to
-0.0107 (K/m). Thus for exposure class 2 and 3 (most unstable) a value of -0.0105
was used for TGRADBELOW. An exposure class of zero represents neutral condi-
tions, thus TGRADBELOW was set to -0.0098 during zero exposre class. Finally, dur-
ing periods of exposure class of 1 an intermediate value of -0.010 was used for
TGRADBELOW.
8910»*rl 2
14
-------
Surface
meteorological data
i
Day and
location
ATOMPRES CONCWATER Cloud cover
1
Ujjper-air
meteorological data
f sunfunc. f J
RADFACTOR
EXPCLASS
^ ^
METSCALERS
TGRADBELOW TGRADABOVE
1
FIGURE 2-5. How diagram for creating the UAM METSCALERS input files.
EEE89104
15
-------
TABLE 2-2. Exposure class (CE)
classification based on cloud
cover and solar zenith angle.
Solar Zenith
Angle
(degrees)
Cloud
Cover
(percent) CE
> 85
> 85
< 30
< 30
30 < 6 < 55
30 < 8 < 55
55 < 8 < 85
55 < 6 < 55
> 90
< 50
> 50
< 50
> 50
< 50
> 50
< 50
> 50
0
-2
-1
3
2
2
1
1
0
89101* 3
16
-------
To estimate TGRADABOVE during the daylight hours, the temperature at the top of
the region (Ttop) is obtained from the afternoon sounding and is assumed to be con-
stant during the daylight regime. Then, for each hour of the day the temperature at
the mixing height (Tm^x) is estimated from the surface temperature (Tsurf) and
TGRADBELOW as follows:
Tmix ~ Tsurf * DIFFBREAK * TGRADBELOW
The TGRADABOVE can be calculated:
TGRADABOVE = (Tt - Tmix) / (REGIONTOP - DIFFBREAK)
During the nighttime regime the DIFFBREAK is fairly constant, ranging from 100 to
500 meters. It is usually assumed that the temperature at the DIFFBREAK will
remain constant at night. Thus, using the temperature at the DIFFBREAK from the
morning sounding (Tm^x) and the surface temperature of the hour in question (Tsur^),
the temperature gradient below the inversion can be estimated as follows:
TGRADBELOW = (Tmix - Tsurf) / DIFFBREAK
As an additional constraint, TGRADBELOW is not allowed to exceed the temperature
gradient from the surface to the DIFFBREAK from the morning sounding. During
the nighttime regime the value for TGRADABOVE is assumed to be constant and is
obtained using the temperature gradient from the DIFFBREAK to the REGIONTOP
from the morning sounding (following the night in question).
TEMPERATUR
The hourly varying fields of surface temperatures needed by the UAM are obtained
by interpolating the hourly measured surface temperatures to the modeling grid using
the tmprtr.f UAM preprocessor (Figure 2-6). The tmprtr.f preprocessor is excercised
using the "stat interp" option (i.e., 1/r interpolation); a large radius of influence is
specified so that each grid cell is influenced by all of the surface measurements
weighted by the inverse distance from the station to the grid cell in question.
WIND
When only routinely available data are used in the UAM, the wind fields are one of
the most uncertain inputs. The wind fields are created by exercising the Diagnostic
Wind Model (DWM) (Douglas and Kessler 1988; Morris et al., 1988) to obtain 14 verti-
cal levels of winds at constant heights above ground. The winds are then vertically
averaged to the UAM vertical layer structure and the O'Brien adjustment scheme is
used to eliminate the vertical velocity through the top (REGIONTOP) of the model-
ing domain. This last operation is accomplished using the uamwnd.f program (see
Figure 2-7).
8910trl 2
17
-------
Surface
meteorological data
( tmprtr. f J
TEMPERATUR
FIGURE 2-6. Flow diagram for creating the UAM TEMPERATUR
input file.
EEE89104
IP
-------
Day and
location
Upper-air
meteorological data
Surface
meteorological data
Meteorological
characterization, development
of conceptual picture
of meteorological conditions
of the ozone episode
Interpolation of
meteorological data to
hourly inputs and
adjustments to represent
conceptual pictures
FIGURE 2-7. How diagram for creating the UAM WIND input files.
EEE89104
19
-------
An important component of the PLANR use of the UAM is how the observed wind
data is input into the DWM so that it represents the conceptual picture of the
meteorological conditions of the episode. The DWM creates an hourly wind field in
two steps. (1) A domain-mean wind representative of synoptic air flows of that hour
is modified by the model to account for the presence of the terrain to cpeate the
multilevel wind fields; (2) The observations are objectively combined with the wind
field created in step 1. Since routine data tend to be sparse, the domain-mean wind
will have a large influence on the resultant wind field and must be selected with
care.
Typically only the twice-daily upper-air soundings from one or two sites within
approximately 400 km of the modeling domain and a handful of surface meteorologi-
cal sites within and in the immediate vicinity of the modeling domain will be avail-
able for generating wind fields for the PLANR UAM. As a first guess, the hourly
varying domain-mean wind is usually defined by linearly interpolating the twice-daily
soundings from the nearest upper-air site. When a high-pressure system or front
passes through the modeling domain, the domain-mean wind must be chosen more
carefully. The timing of the passage of weather system can usually be obtained by
examining when the wind shifts in the hourly surface wind observations. Examples of
this more creative use of the routinely available upper-air data will be given in the
discussion of the PLANR application of the UAM for Atlanta (Section 3).
Ozone episodes for an urban area are typically characterized by hot stagnant
meteorological conditions with limited mixing. There are usually two types of sur-
face meteorological observations: hourly average values from a state or local
agency network, or instantaneous (one-minute averages) collected by observers at
Federal Aviation Administration (FAA) sites. Recent analysis of collocated hourly
average and instantaneous FAA wind observations in the South Central Coast Air
Basin of California have indicated that, under stagnant conditions, the hourly
average wind speeds are only 44 to 98 percent of the concurrent FAA values (depend-
ing on the site) (see Appendix D). Based on seven collocated FAA and hourly average
wind speed monitoring sites, the FAA wind speeds were on average a factor of 1.8
times the hourly average wind speeds. This result is not surprising for two reasons.
First, as noted by Killus and Gutfreund (1984), vector averaging over an hour to
obtain an hourly average wind observation inherently results in a lower wind speed
than is obtained from an instantaneous or one-minute average. This fact is
especially true during periods of light and variable winds, which typically occur
during ozone episodes. Second, FAA observations are mainly used to advise pilots of
any adverse wind conditions and wind directions during take-offs and landings. The
pilots need information concerning the possibility of cross runway winds, not
information concerning the hourly average wind. The UAM, on the other hand,
requires information concerning the vector average transport conditions of the
hour. Thus, if initial diagnostic UAM simulations that use FAA data without adjust-
ments indicate that the predicted ozone peak is too low and occurs too far down
8910"»rl 2
-------
wind, then elimination of the unrepresentative FAA wind observations should be
considered. Reduced FAA wind speeds might be used for model application in the
absence of more representative data.
AIRQUALITY, BOUNDARY, and TOPCONC
The initial concentrations (AIRQUALITY) and boundary conditions for the lateral
boundaries (BOUNDARY) and aloft (TOPCONC) are assumed to be "clean" unless
higher concentrations are known or expected. This can be determined through mea-
sured air quality data or from knowledge that the area is known to be affected by
transported pollutants. The effects of both anthropogenic and biogenic emissions
need to be reflected in the initial and boundary concentrations. The "clean" values
recommended for use in the PLANR UAM are as follows:
VOC = 25 ppbc (using EKMA default speciation)
ISOP = 0.0001 ppb
NOX = 1 ppb (3/4 NO2, 1/4 NO)
O3 = 40 ppb
CO = 200 ppb
The UAM is usually initiated well before the ozone episode of interest to minimize
the effects of initial concentrations. Thus "clean" values are usually sufficient for
initializing the model. The first diagnostic simulations for Atlanta and Dallas both
used these "clean" values for boundary conditions. However, because of the presence
of large amounts of biogenic emissions in the vicinity of Atlanta, the VOC lateral
boundary conditions were increased in subsequent diagnostic simulations for Atlanta.
EMISSIONS and PTSOURCE
The emission inventories for the PLANR application of the UAM to Atlanta and
Dallas-Fort Worth were developed from the 1985 NAPAP annual continental U.S.
emissions data base in several steps (Figure 2-8). Since the PLANR base cases for
Atlanta and Dallas both used 1985 emission inventories, the creation of the inven-
tories is more straight forward than if data for a different base year were needed.
Thus there was no need to make adjustments for growth, effects of Reid Vapor Pres-
sure, and any possible emission controls (e.g., Stage I and II). The reader is referred
to SAI (1989) for more details on the procedures for preparing emission files for use
in the UAM.
DIAGNOSTIC SIMULATIONS TO DEFINE THE BASE CASE
An important feature of the PLANR use of UAM is that the number of diagnostic
simulations for improving model performance to an acceptable level is limited. As
-------
1985 NAPAP Annual County Area Source Category
and Point Source Emissions
Mobile
sources
Evaporative/
exhaust separation
factors
Annual average
to episode-specific
VMT adjustment
Annual to
episodic temperature
and MOBILES to
MOBILE4 adjustment
including addition
of running losses
Other
gasoline
marketing
1985
uncontrolled
refueling
losses for
Aldehyde
adjustment
Mobile
exhaust
Mobile
evaporative
Assignment of
diurnal profiles
by source
category
Point
sources
Assignment of
diurnal profiles
using NAPAP
1
r
Point
sources
FIGURE 2-8. Row diagram for creating the UAM 1985 base case low-level emissions
(EMISSIONS) and elevated emissions (PTSOURC) files.
EEE89104
-------
Mobile exhaust,
evaporative,
and refueling
1
Other
area sources
1
Griddedby
' emission surrogates ^
(population, landuse,
etc.) and industrial
source category
Griddedby
NAPAPUTM
Hata
t
Mobile exhaust
evaporative,
and refueling
1
I
Other area
sources
1
Speciation
ofVOCandNOx
into CB-IV species
by source type
Allocation
to hours of
episode
DIFFBREAK
REGIONTOP
WIND
METSCALERS
HGURE2-8. Concluded.
EEE89104
23
-------
will be seen, the diagnostic simulations usually differ in the methods used in the
interpolation of the meteorological data and how it is input into the UAM preproces-
sors so that they represent the conceptual picture of the meteorological conditions
of the ozone episode. For the Atlanta application, one diagnostic simulation was
carried out with only biogenic emissions in order to determine the level of back-
ground VOC concentrations due to only biogenic emissions for use in the boundary
conditions.
Diagnostic simulations are an integral part of any UAM application, not only for
improving model performance but also for obtaining indications that the model is
getting the right answer for the right reasons. Adjustments to the UAM inputs were
based on interpretation of the observed data so that the UAM inputs better reflected
our conceptual picture of the meteorological conditions of the episode. Thus,
adjustments to the UAM inputs to improve model performance should be based on
reasonable physical interpretation of the observed data rather than random modifica-
tions of the data.
8910"»rl 2
24
-------
3 APPLICATION OF THE UAM TO THE CITY OF ATLANTA
There were two main objectives in the PLANR application of the UAM to the city of
Atlanta: (1) demonstration of the PLANR use of the UAM for Atlanta; and (2)
analysis of the effects of biogenic emissions on VOC emission control scenarios in
the Atlanta region. In this report we discuss the demonstration PLANR application
of UAM to Atlanta. The analysis of the effects of biogenic emissions on VOC emis-
sion control strategies in Atlanta is described in another report (Morris et al.,
1989b).
As discussed in Section 2, PLANR involves the following steps: episode selection,
meteorological characterization, domain definition, emissions preparation, UAM
model input development, and performance of diagnostic simulations to achieve a
statisfactory base case. The following sections discuss these activities.
EPISODE SELECTION
Table 3-1 lists all days from the years 1984 to 1987 in Georgia (either Atlanta or
Columbus) in which the maximum daily ozone concentration at any monitor in the
AIRS data base exceeds the National Ambient Air Quality Standard of 120 ppb. Of
the 62 days where ozone concentrations in Georgia are greater than or equal to 120
ppb, 61 occur in Atlanta and one (18 March 1984) in Columbus (note that on other
days Columbus may have an exceedence of the ozone standard but the ozone con-
centrations are less than recorded in Atlanta). The most striking feature in the
ozone record for Atlanta are the months of July and August, 1987, where in a 21-day
period (19 July to 7 August 1987) there were 15 exceedences of ozone NAAQS. This
period also included the very highest observed ozone concentration in Atlanta (201
ppb on 31 July 1987). On most days during this period, most ozone monitors in
Georgia, including Columbus, recorded high ozone concentrations. This seems to
indicate a region-wide buildup of ozone concentrations over an extended length of
time and area. Thus, the elevated ozone concentrations in Atlanta during July and
August, 1987 may be attributable to regional-scale as well as urban-scale ozone
formation.
The UAM analysis of Atlanta analyzed the effects of biogenic emissions on urban
ozone formation and on the effectiveness of anthropogenic VOC emissions strategies
designed to reduce ozone. A previous study on the effects of biogenic emissions for
Atlanta used the EKMA model to analyze ozone formation on 4 June 1984
8910trl H
25
-------
TABLE 3-1. Highest ozone days in Atlanta, 1984 - 1987.
Rank
I
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
Peak
03
201.
169.
166.
165.
164.
164.
160.
157.
155.
155.
155.
152.
151.
150.
150.
149.
147.
147.
146.
145.
145.
144.
142.
140.
140.
137.
137.
136.
136.
136.
135.
135.
133.
133.
133.
132.
132.
131.
131.
130.
130.
130.
130.
129.
129.
128.
127.
127.
126.
125.
125.
125.
125.
125.
123.
123.
122.
122.
122.
121.
120.
120.
Mon i tor
130890002
130890002
130970002
132470001
130890002
130890002
1305'J0002
132470001
132470001
130970002
130970002
132150008
130890002
132470001
130970002
131210053
132470001
132470001
130890002
132470001
132470001
132470001
130890002
130970002
132470001
130890002
132470001
130890002
130890002
130890002
130890002
130890002
130970002
130970002
132470001
132470001
132470001
132470001
130890002
132470001
130890002
132150008
132470001
132470001
130890002
132470001
130890002
132470001
130890002
132470001
132470001
132470001
132470001
130890002
130890002
130890002
132470001
132470001
130890002
130890002
130890002
130970002
Maximum Daily Ozone
Date 130890002 130970002 131210053 132150008 132470CJ1
Panthers. Sweetwat. HLK Columbus Conyer?
DKLB SWTR MLKM COLO CNYR
7/31/87
8/ 1/87
7/24/87
6/26/86
7/18/86
7/23/86
6/27/86
8/ 2/87
7/10/84
7/23/87
7/25/87
3/18/84
7/31/86
6/ 9/87
7/30/87
7/26/87
6/ 4/84
8/22/84
8/ 3/87
6/26/85
7/12/85
6/10/87
4/26/86
8/21/87
6/16/86
7/22/86
8/ 2/86
8/ 4/87
8/ 5/87
6/28/84
6/ 6/85
7/29/87
6/11/87
9/ 3/87
9/14/84
6/ 2/87
9/ 2/87
8/ 1/86
5/13/85
6/ 4/85
6/24/87
7/27/87
B/ 7/87
8/ 4/86
7/19/86
6/ 3/87
8/,18/84
6/ 5/85
7/19/87
8/23/87
6/ 1/84
6/ 3/84
7/11/85
7/ 8/86
8/16/86
7/21/86
7/20/85
8/ 1/85
7/26/86
6/22/86 .
7/22/87
8/20/87
201.
169.
140.
163.
164.
164.
160.
138.
115.
154.
113.
60.
151.
129.
112.
129.
130.
118.
146.
115.
118.
137.
142.
99.
118.
137.
112.
136.
136.
136.
135.
135.
103.
91.
105.
109.
108.
114.
131.
120.
130.
108.
129.
114.
129.
103.
127.
92.
126.
115.
97.
105.
118.
125.
123.
123.
115.
95.
122.
121.
120.
100.
98.
130.
168.
-99.
-99.
-99.
-99.
88.
-99.
155.
155.
-99.
-99.
73.
150.
100.
-99.
-99.
118.
-99.
-99.
95.
-99.
140.
-99.
-99.
-99.
100.
95.
-99.
-99.
128.
133.
133.
-99.
68.
88.
-99.
-99.
-99.
75.
108.
78.
-99.
-99.
65.
-99.
-99.
115.
78.
-99.
-99.
-99.
-99.
-99.
-99.
-99.
-99.
-99.
-99.
98.
120.
141.
138.
155.
-99.
-99.
-99.
-99.
128.
-99.
153.
118.
-99.
-99.
103.
113.
149.
-99.
-99.
128.
-99.
-99.
115.
-99.
104.
-99.
-99.
-99.
113.
103.
-99.
-99.
118.
103.
95.
-99.
86.
89.
-99.
-99.
-99.
104.
94.
107.
-99.
-99.
72.
-99.
-99.
109.
96.
-99.
-99.
-99.
-99.
-99.
-99.
-99.
-99.
-99.
-99.
116.
108.
118.
80.
113.
95.
90.
92.
67.
70.
53.
100.
90.
152.
108.
75.
93.
98.
83.
78.
80.
72.
65.
85.
100.
68.
108.
80.
103.
75.
100.
55.
67.
88.
120.
78.
103.
63.
90.
85.
83.
-99.
53.
130.
63.
103.
90.
75.
83.'
67.
73.
83.
75.
70.
63.
60.
78.
90.
83.
85.
78.
78.
68.
75.
162
158
90.
165.
142.
111.
131.
157.
155.
85.
94.
-99.
114.
150.
93.
88.
147.
147.
129.
145.
145.
144.
105.
98.
140.
84.
137.
116.
120.
122.
12?.
131'.
86.
80.
13?.
132
132.
131 .
72.
130.
118.
120.
130.
129
112.
128.
100.
127.
75.
125.
125.
125.
125.
123.
93.
110.
122.
122.
74.
85.
82.
96.
8910**
26
-------
axis of the high pressure system moved near the modeling region, a slight easterly
component to the wind field developed ( a result of the circulation about the high-
pressure system). This flow was first seen in the upper levels (> 600m) on the morn-
ing of the 4th, but by mid morning was seen in the surface layer. After the passage
of the high-pressure system, the winds became southwesterly. The weather condi-
tions during the modeling period were clear, hot, and humid. Maximum temperatures
were in the upper 80s and dew point temperatures in the low 60s.
MODELING DOMAIN DEFINITION
The modeling domain for the 3-4 June 1984 modeling episode should include a fairly
large fetch upwind of Atlanta to account for sufficient amounts of biogenic emis-
sions upwind of the urban core of Atlanta. The modeling domain should also include
enough area downwind of Atlanta to capture the maximum ozone concentrations that
will be formed. Grid spacing should be sufficiently small in order to resolve the
anthropogenic emission distribution in Atlanta. To determine the modeling domain,
synoptic weather maps along with upper-air data from Athens and Waycross,
Georgia; Nashville, Tennessee; and Centreville-Brent, Alabama, and surface data
from South Dekaib were analyzed to determine the mean flow conditions that existed
on 3-4 June 1984.
Figure 3-1 gives a region-wide perspective of the modeling domain for Atlanta. This
modeling domain consists of a 40 by 40 array of 4 km square grid cells. The domain
origin is located at UTM coordinates 660 km easting, 3665 km northing in zone 16
and extends 160 km in the east and north direction. As discussed in the modeling
protocol (SAI, 1988) five vertical layers were used in the UAM: two below the mix-
ing height and three above. The region top was based on the maximum mixing height
occurring during the modeling episode.
To accommodate the analysis of the effects of biogenic emissions on urban ozone
formation in Atlanta, there is a region 75 to 100 km wide upwind of Atlanta. This
allows for a 4 to 12 hour loading of biogenic emissions into the atmosphere upwind
from the outskirts of Atlanta.
ROUTINE DATA AVAILABLE FOR ATLANTA
On 3-4 June 1984 there were six surface meteorological observation sites operating
in and around the city of Atlanta (Table 3-2). There were no upper-air observation
sites located within the modeling domain. Thus, we made use of upper-air observa-
tions from five sites that surround the modeling domain for generating wind and
mixing height inputs (Table 3-3). The closest upper-air site to Atlanta is Athens,
Georgia, approximately 100 km to the east-north-east. Air quality data in and
around Atlanta during June 1984 consisted of three monitoring sites (Table 3-4).
8910i*rl <»
28
-------
(Chameides et al., 1988). It was highly desirable that this day also be chosen for the
present analysis as long as no other reasons, such as excessive wind shear or other
unusual meteorological phenomena, precluded its selection. On 4 June 1984 the
maximum daily ozone concentration was 147 ppb at Conyers, Georgia. For the two-
day period of 3-4 June 1984, winds are predominantly from the northwest, with some
southwesterlies late on 4 June. Maximum daily ozone concentrations on 4 June are
78 ppb at Dallas upwind of Atlanta, and 130 ppb at DeKalb and 147 ppb at Conyers,
at distances of 15 km (DKLB) and 40 km (CNYR) downwind of Atlanta (see Figure
3-1). Thus, it appears that there is not a large amount of transported ozone into the
Atlanta region on 4 June 1984 and the ozone exceedence is predominantly due to
emissions from the Atlanta area.
Since the 4 June 1984 ozone event is not overly influenced by transported ozone and,
as will be shown, does not contain any unusual meteorological conditions, it appears
to be an appropriate day for the analysis of biogenic emissions and a demonstration
of the PLANR use of the UAM. Since it is desirable to minimize the effects of
initial conditions in any UAM simulation, the UAM modeling episode was initialized
on 3 June 1984 and terminated the evening of 4 June 1984. Thus, UAM modeling
inputs were prepared for 3 and 4 June 1984 with the actual starting time on 3 June
determined from an inert tracer simulation. This type of inert tracer simulation
starts with an initial concentration of 100 and zero boundary conditions and emis-
sions. The inert model is exercised with several different starting times; the starting
time of the inert model simulation that eliminates the initial condition inert tracer
concentrations from the region of interest on the morning of 4 June is the starting
time chosen for the photochemical model simulation.
It should be noted that other days listed in Table 3-1 may be more appropriate for
demonstrating ozone attainment in a State Implementation Plan (SIP). Generally, it
would be desirable for a SIP to pick a day with a higher ozone concentration, a more
recent episode, a day in which the elevated ozone concentrations can be attributed
to Atlanta, and a day that represents typical meteorological conditions that produces
ozone exceedences in Atlanta. The very highest ozone concentration in Table 3-1
(201 ppb observed on 31 July 1987) appears to be an anomalous event since it is
almost 20 percent higher than any of the other high ozone events over the four year
period.
METEOROLOGICAL CHARACTERIZATION
On 3 and 4 June 1984 a high-pressure system passed over the Atlanta region. The
500 mb height contours at 0700 EST on 4 June 1984 shows the high-pressure ridge
approaching Atlanta. This high-pressure ridge passed through the region during the
evening of the 4th of June; the axis of the surface high-pressure system passed
through the modeling region during the afternoon of the 4th. On the 3rd of June,
daytime low-level winds (<600m) were from the west at 3-6 ms throughout the
modeling region, while winds aloft were in the same direction, but stronger. As the
89104rl if
27
-------
66.0 85.5 85.0 64.5 64.0 63.5 63.0 62.5
35.0 -
34.5 -
34.0 -
33.5 -
33.0 -
32.5 -
35 i
34.!
34.1
33.!
33.(
32.!
66.0 65.5 65.0 64.5 64.0 83.5 63.0 62.5
FIGURE 3-1. UAM modeling domain for Atlanta. Modeling domain consists
of 40 by 40 array of 4. km grid cells with an origin at UTM coordinates
660 km easting, 3665 km northing, zone 16.
8910'+
29
-------
TABLE 3-2. Surface meteorological observation sites, locations, and
variables in the Atlanta region.
Location UTM (Zone 16)
Site Name
Atlanta Hartsfield
International Airport
Dobbins Naval Air Station
Fulton County (Charlie
Brown) Airport
Dekalb Peachtree Airport
Conyers Monastery Monitor
South Dekalb Panthersville
Monitor
UTMX
739.580
729.590
729. 947
749.724
772.248
752.780
UTMY
3726.148
3755.498
3740.709
3752.306
3719.869
3730.991
Variables Measured
WS, WD, T, TD, P
WS, WD, T, TD, P
WS, WD, T, TD
WS, WD, T, TD
WS, WD
WS, WD, T, TD
8910U 3
30
-------
TABLE 3-3. Upper-air observation sites, locations, and distance from Atlanta
used in this study.
Location
Site Name
Athens, GA
Nashville, TN
Greensboro, NC
Waycross, GA
Centerville-Brent, AL
Lat
35°
36°
36°
31°
32°
57'
15'
5'
15'
54'
UTM Coordinates (Zone 16)
Long
83° 19'
86°
79°
82°
87°
34'
57'
24'
15'
UTMX
839.
538.
1134
937.
475.
644
182
.426
394
841
UTMY
3763
4012
4016
3467
3640
.429
.482
.946
.152
.638
Distance from
Atlanta (km)
109
'199
491
328
. 275
TABLE 3-4. Ozone monitor names and locations,
UTM Location (Zone 16)
Monitor JD
130890002
130970002
131210053
132150008
132470001
Monitor Name
DeKalb Jr. College (DKLB)
Sweetwater Creek State Park (SWTR)
MLK Marta Station (MLKM)
Columbus Airport (COLO)
Conyers Monastery (CNYR)
UTMX
752.78
719.28
743.10
693.15
772.25
UTMY
3730.99
3736.10
3737.15
3799.91
3719.87
Years Available
(1984 - 1987)
1984 - 1987
1987
1987
1984 - 1987
1984 - 1987
89101* 3
31
-------
There were no air quality data available near the boundaries of the modeling domain
for use in prescribing boundary conditions.
PREPARATION OF UAM INPUT FILES
The following paragraphs describe how the UAM input files were prepared for the
PLANR application of the UAM to Atlanta.
DIFFBREAK
This file contains the daytime mixing height or nighttime inversion height for each
column of cells at the beginning and end of each hour of the simulation. Hourly mix-
ing heights were estimated at several surface meteorological sites through use of the
hourly surface measurements of temperature and the twice-daily upper-air observa-
tions from a representative upper-air site (Athens, Georgia). For three of the sur-
face meteorological observation sites that recorded temperature (Fulton County
Airport, Dobbins Air Force Base, and South Dekalb) the diurnal variations in mixing
heights were calculated using the RAMMET meteorological processor. These hourly
values were then input into the standard UAM mixing height interpolation program,
DFSNBK (Ames et al., 1985b) using the 1/r interpolation option to produce the hourly
spatially varying fields of mixing heights. The maximum daily mixing height varied
from approximately 1200 to 1350 meters above ground across the region. Two of the
surface observation sites that recorded temperature were not used in the generation
of the DIFFBREAK file; one (Dekalb Peachtree Airport) had missing temperature
data for 4 June 1984 and data from the other site (Atlanta Hartsfield International
Airport) were not available in time to perform the mixing height analysis.
REGIONTOP
This file contains the height of each column of cells at the beginning and end of each
hour of the simulation. If this height is greater than the mixing height, the cell or
cells above the mixing height are assumed to be within an inversion. For the applica-
tion of the UAM(CB-IV) to Atlanta a constant 1,400 m AGL region top was used.
This value was picked because it is 50 m above the maximum mixing height; thus all
five vertical layers of the UAM are contained within the well mixed layer, offering
the maximum vertical resolution possible with the five-layer UAM configuration.
WIND
This file contains the x and y components of the wind velocity for every grid cell for
each hour of the simulation. There are two steps in creating the wind fields for
Atlanta: (1) application of the Diagnostic Wind Model (DWM) (Morris et al., 1988,
8910>»rl >» 32
-------
1989a; Douglas and Kessler, 1988) using 15 vertical levels and data from the six sur-
face and five upper-air meteorological observation sites; and (2) vertical interpola-
tion of the 15-layer hourly wind fields into the five-layer UAM configuration used for
the Atlanta application (see Morris et al., 1989a).
The DWM creates an hourly 15-layer wind field by first creating a wind field by
adjusting a domain-mean wind (based on upper-air observations) for terrain effects
(kinematic effects, thermodynamically generated slope flows, blocking, and deflec-
tion). Then the observational information is added to the wind field by weighting the
initial wind field heavily away from the observations and the observed winds heavily
in the vicinity of the observations.
The 15-layer wind fields created by the DWM are then vertically averaged into the
UAM five-layer configuration. Since the surface observations provide a more
detailed representation of the air flows in the vicinity of Atlanta and their presence
is only felt within the lowest layer of the DWM, the first UAM vertical layer was
assumed to have the same wind field as the first DWM layer. The DWM first layer is
also weighted 50 percent in the second UAM vertical layer. The wind fields for the
remainder of the UAM vertical layers were obtained from the DWM wind fields using
weighted averaging. The resultant five-layer wind field was then modified using the
procedure suggested by O'Brien (1970), which minimizes the vertical velocity out of
the top of the region. In this manner the boundary concentration assumed to exist
above the region top (TOPCONC) does not greatly influence concentrations within
the modeling domain.
For the application of the UAM to Atlanta, three separate wind fields were created
for use in diagnostic simulations 1 and 2 and a sensitivity test reported in Appendix
D. As will be discussed later, these wind fields were created using the procedures
described above but differed in definition of the domain-mean wind and the surface
observation sites used.
METSCALARS
This file contains the hourly values of the meteorological parameters that do not
vary spatially. The NO2 photolysis rates were calculated for the CB-IV mechanism
using procedures described by Schere and Demerjian (1977) and actinic flux data
collected by Bass and co-workers (1980). The concentration of water was based on
measurements of temperature and dewpoint at three of the surface meteorological
observation sites. These three values were averaged to obtain the hourly input for
the UAM. Water concentrations ranged from approximately 14,000 to 18,000 ppm.
An atmospheric pressure value of 1 atm was used.
Exposure class is a measure of near surface stability: +3 during high solar intensity
to -2 at night with no clouds. Exposure class was assigned based on the solar inten-
sity: a value of -2 at night and daytime values of from 0 (one hour day/night transi-
8910»»rl if
33
-------
tion period) to 3. The temperature gradients below (TGRADBELOW) and above
(TGRADABOVE) the inversion were based on the twice-daily upper-air soundings
taken at Athens, Georgia, and hourly surface temperatures observed at Charlie
Brown Airport. Values for TGRADBELOW from the surface to the mixing height
(approximately 1300 m AGL) from the Athens evening sounding (1900 LST) are
-0.0096, -0.0103, and -0.0107 K/m for June 2 through 4, respectively. Based on these
measurements, values for TGRADBELOW were -0.0105 K/m for exposure classes 2
and 3, -0.01 K/m for exposure class 1, and -0.098 for exposure class 0.
The Athens morning sounding (0700 LST) measured values for TGRADBELOW of
0.018 and 0.030 K/m on June 3 and 4 respectively, assuming a mechanical mixing
height of 250 m AGL. Observed temperatures at 250 m on 3 and 4 June were,
respectively, 295.6 and 297.8 K. Nighttime hourly values for TGRADBELOW were
then calculated for June 3 and 4 based on the surface temperature at the Charlie
Brown Airport and the measured value at 250 m from the 0700 LST sounding.
For predawn hours, the temperature gradient from 250 m to the region top (1,400 m)
from the 0700 sounding at Athens was used, resulting in values for TGRADABOVE of
-0.0025 and -0.0075 °K/m for June 3 and 4, respectively. For daytime hours a tem-
perature at the mixing height was calculated for each hour of the day using the sur-
face measurement at the Charlie Brown Airport and the value for TGRADBELOW.
Then the hourly temperature gradient between the mixing height and the region top
was estimated using the calculated value at the mixing height and the measured
temperature from the 0700 sounding at Athens at 1,400 m AGL (290.8 and 290.7 K on
June 3 and 4, respectively)
AIRQUALITY
This file contains the initial concentrations of each species for each grid cell at the
start of the simulation. Since the UAM was initiated at 12 noon on 3 June 1984,
most of the material from the initial conditions were advected out of the region by
the morning of 4 June; thus, initial concentrations do not influence ozone formation
on 4 June. Accordingly, initial concentrations were assigned "clean values" as fol-
lows:
VOC 25 ppbc (using EKMA default speciation)
ISOP 0.001 ppb
NOX 1 ppb (split 3/4 NO2, 1/4 NO)
O^ 40 ppb
CO 200 ppb
8910«trl it
34
-------
BOUNDARY
This file contains the location of the modeling region boundaries. It also contains the
concentration of each species that is used as the boundary condition along each
boundary segment at each vertical level. For the first application of the UAM to
Atlanta (diagnostic run 1), the minimum of a one-cell buffer of unsimulated cells was
used (i.e., the boundary conditions), resulting in a simulation region of 152 km by 152
km. After examining the predicted ozone concentrations in diagnostic run 1 and the
emissions distribution, we determined that the computational region for Atlanta
could be reduced by about one-third without affecting the results. Thus for diag-
nostic run 2 and the sensitivity test (Appendix D) an 8-, 4-, 1-, and 3-cell buffer was
used on the north, south, east, and west boundaries, respectively.
As will be discussed, two sets of boundary conditions were derived for evaluation
runs number 1 and 2. For evaluation run number 1, "clean" values (listed previously)
were specified for boundary conditions. However, these clean values may under-
estimate background concentrations because of the biogenic emissions which
increase background VOC and ISOP (isoprene) concentrations above the clean back-
ground level. Thus in order to determine a better estimate of background conditions,
a UAM simulation was carried out with no anthropogenic emissions (i.e., biogenic
emissions and clean boundary concentrations only). During the day, isoprene and
VOC concentrations ranged from approximately 0.5 to 2.0 ppb and 35 to 75 ppbC,
respectively. Thus for the second diagnostic simulation, a 1 ppb ISOP boundary
condition was prescribed and VOC boundary conditions were set at 40 ppbC; the
increase from the clean value of 25 ppb was mainly (92%) due to the lower reactive
PAR species and the remainder of the increase was due to increases in OLE.
Boundary conditions for ozone (40 ppb) and NOX (1 ppb) were the same as used in
diagnostic run 1.
TOPCONC
This file contains the concentration of each species for the area above the modeling
region. Since the wind fields are processed to eliminate the vertical velocity through
the region top, the model results are not sensitive to TOPCONC. Accordingly, the
clean concentration values listed previously were used.
TEMPERATUR
This file contains the hourly temperature for each surface layer grid cell. Hourly
spatial varying temperatures were obtained by using 1/r interpolation from the sur-
face meteorological observations.
8910trl f
35
-------
EMISSIONS
This file contains the ground-level emissions of NO, NO2> CB-IV VOC species cate-
gories, and CO for each grid square for each hour of the simulation. Anthropogenic
emission estimates were obtained from the 1985 NAPAP county-wide emissions
inventory through a multistep process that includes updating mobile source emissions
from MOBILE-3 to MOBILE-4 based emission factors, applying summer weekday and
episodic temperature adjustments, gridding of data using a known surrogate distri-
bution, and speciation of the VOC and NOX emissions into the CB-IV mechanism
species (see Figure 2-8).
The NAPAP motor vehicle emission categories were split into exhaust and evapora-
tive emissions using splitting factors from the MOBILE-3 emissions program and the
same average temperature as was used in generating the NAPAP inventory. The
MOBILE-4 emissions program was then exercised for the temperature conditions on 4
June 1984 to obtain day-specific evaporative and exhaust emissions. Estimates of
running loss emissions (which are not included in the NAPAP inventory) were then
obtained by applying the ratio of the running loss and exhaust emission factors to the
mobile exhaust VOC emission rate from the NAPAP inventory.
The county-wide area source emissions were then mapped to the gridded modeling
domain by assigning each area source Source Classification Code (SCO to a known
surrogate distribution (see SAI, 1989). Known surrogate distributions include:
agriculture, urban, rural, and water from the national Geographical Information Ser-
vices (CIS) land-use data base; population from the 1980 census; airports and limited
access roadways based on digitizing the locations of airports and freeways from
standard USGS maps; and spatial coverage based on the fractional coverage of each
grid cell in each county. The gridded area source emissions were then adjusted to
hourly emissions for a summer weekday and speciated into CB-IV species based on
their SCC codes.
Estimates of biogenic VOC emissions were obtained from EPA's Atmospheric
Research and Exposure Assessment Laboratory (EPA/AREAL) on a 1/4° longitude by
1/6° latitude grid for the entire Southeast. The EPA/AREAL had biogenic emission
estimates for the entire Southeast for a two-week period during 1980. These emis-
sion estimates consider the effects of episodic temperature and solar intensity on the
biogenic emission estimates. Based on analysis of surface temperatures, it was
determined that 26 August 1980 provided the best match with episodic temperature
and light intensity conditions of i 3une 1984. The biogenic emissions from the 1/4°
by 1/6° grid were gridded onto the 4 km square grid cells used in the UAM modeling
based on spatial covering of grid cells. The biogenic emissions from EPA/AREAL
were already speciated into CB-IV species.
The final EMISSIONS file was obtained by merging the gridded NAPAP area source
emissions with the low-level point sources and the biogenic emissions. The distribu-
tion of the resultant anthropogenic and biogenic emissions are given in Figure 3-2.
8910»»rl t
36
-------
1 i i i ,- i i i i i I i ( I I \J I I I I
10-
665
FIGURE 3-2a. Anthropogenic area source and low-level point source NOX
emissions (kg/day) for the Atlanta 1984 base case.
89104
37
-------
10-
FIGUPE 3-2b. Anthropogenic point source NOX emissions (kg/day) for the
Atlanta 1984 base case.
89104
38
-------
ATLANTA — 1985 base — anthropogenic only
RHC
(Kg/day)
FIGURE 3-2c. Anthropogenic area source and low-level point source VDC
emissions (kg/day) for the Atlanta 1984 base case.
89104
39
-------
10
745
3705
2&LANTA 30
ROG
o3665
FIGURE 3-2d. Biogenic VOC emissions (kg/day) for the Atlanta 1984 base case.
40
89101+
-------
10-
10
'<* \
FIGURE 3-2e. Anthropogenic point source VOC emissions, (kg/day) for the
Atlanta 1984 base case.
89104
41
-------
Of particular note is that biogenic VOC emissions account for approximately 60
percent of the VOC emissions within the modeling domain.
PTSOURCE
This file contains the point source information (stack height, temperature and flow
rate, plume rise), the grid cell into which the emissions are emitted, and the emis-
sions rates for NO, NO2> CB-IV VOC categories, and CO for each point source for
each hour. The 1985 NAPAP.point source emissions file was separated into low-level
and elevated points based on plume rise estimates. Those point sources whose plume
rise was less than 25 m for typical atmospheric conditions were considered low-level
sources and were merged with the EMISSIONS file after applying the summer week-
day and diurnal emission profiles supplied as part of the NAPAP point source
inventory.
Elevated point sources from the 1985 NAPAP point source file were adjusted to
summer episodic conditions with diurnal variation using the adjustments given in the
point source file. The emissions were then speciated into CB-IV species based on
their SCC and SIC codes. The resultant data was run through the UAM point source
preprocessor for input to the UAM. Of particular note is that point source NOX
(including electrical utilities) emissions account for almost 50 percent of the total
NOX emissions within the modeling domain.
TERRAIN
This file contains the value of the surface roughness and deposition factor for each
grid square. Each 4 km grid cell was assigned to a land-use category based on a
digitized standard USGS map of the modeling region. The land-use categories were
then converted to roughness lengths and vegetation factors according to data pub-
lished by Argonne National Laboratory (see Morris et al., 1989a).
CHEMPARAM
This file contains information regarding the chemical species to be simulated, includ-
ing reaction rate constants, upper and lower bounds, activation energy, and reference
temperature. Reaction rate constants correspond to those given in the report docu-
menting the CB-IV mechanism (Gery, Whitten, and Killus, 1988), except that reac-
tions for methanol (MEOH) and ethanol (ETOH) have also been added to the
mechanism (see Morris et al., 1989a).
-------
DEFINITION OF BASE CASE
One of the key components of the PLANR use of the UAM is the development of a
satisfactory base case simulation within a limited number of diagnostic simula-
tions. However, the UAM base case must show some skill in replicating the observa-
tions, otherwise one would not have confidence in the model's responses to alterna-
tive emission control strategies. A minimal performance goal in the past was to
have the predicted regional maximum ozone concentration be within 30 percent and
in the general location of the peak observed value. Model performance has been
considered good if the predicted peak ozone is within. 15 percent. Clearly, when
model inputs are based only on sparse routine data, rather than intensively measured
data as in the past, model performance cannot be expected to always be as good in
the past. However, there should be some minimal expectations of model perfor-
mance since incorrect characterization of base case ozone concentrations may lead
to incorrect calculations of ozone reductions due to alternative emission inputs.
The protocols for several recent UAM studies of the impacts of California offshore
drilling emissions (Haney et al., 1986; Yocke et al., 1985) defined a minimal model
performance standard as follows: (1) the UAM-predicted regional maximum ozone
concentration should be within 20 percent and in the general location of the observed
maximum, and (2) the UAM predicted maximum at the location of the observed
maximum should be within 30 percent of the observed value.
We adopted this model performance standard as a performance goal for the PLANR
application of the UAM for Atlanta and Dallas-Fort Worth. However, model evalua-
tion should always include consideration of whether the right answer is being
obtained for the right reason. Here we will also discuss approaches that may
improve UAM performance but which also may deviate slightly from the demonstra-
tion of the PLANR use of the UAM. The PLANR use of the UAM for Atlanta
involved three diagnostic simulations before a base case was obtained that satisfied
the minimal performance goal. A sensitivity test was also performed (Appendix D).
Diagnostic Run 1
In the first diagnostic simulation, the UAM was exercised with clean boundary condi-
tions (i.e., 40 ppb ozone, 25 ppbc VOC, 0.001 ppb ISOP, and 1 ppb NOX), anthropo-
genic and biogenic emissions, and a first estimate of the wind field. The wind field
was created using the DWM: the vertically varying domain-mean wind was defined as
the vector average from the five upper-air soundings surrounding the modeling
domain. The upper-air data obtained from the National Climatic Data Center
(NCDC) on the TD9743 format was missing data from the lowest 1,500 m AGL of the
0700 June 4 sounding at Athens, Georgia. In addition, surface wind data was used
from all of the surface sites, except Atlanta Hartsfield Airport, which was
inadvertently left out of the initial analysis.
8910Hrl i»
43
-------
Hourly average predicted ozone concentrations for diagnostic run 1 are contained in
Appendix B. A comparison of predicted versus observed hourly ozone concentrations
for the three monitors within the UAM modeling domain is shown in Figure 3-3. As
seen in Appendix B and Figure 3-3, the regional maximum predicted ozone concen-
tration is 13.54 pphm, within 8 percent of the observed maximum of 14.7 which was
recorded at the Conyers Monastery (CNYR) ozone monitor. However, the predicted
regional maximum is approximately 28 km (7 grid cells) to the north-north-west of
the observed value. The maximum predicted ozone at the CNYR site for diagnostic
run 1 is 8.79 pphm, an approximate 40 percent underprediction of the observed
maximum at that site. Model performance is better at the South Dekalb ozone
monitor, where the maximum daily observed value of 13.0 pphm is reproduced to
within 15 percent (11.1 pphm). However, concentrations are underpredicted at the
Dallas (DLLS) ozone monitor, which lies upwind of the city of Atlanta. The maxi-
mum daily observed ozone concentration at DLLS of 7.8 pphm is underpredicted by
about 60 percent (4.6 pphm).
Diagnostic Run 2
Improvements Over Diagnostic Run 1
An analysis of the observed ozone features for 4 June 1984 indicates two major
deviations of the predictions from the observations for diagnostic run 1: (1) the
gross underprediction of the upwind monitor indicates that the ozone and ozone pre-
cursor loadings upwind of Atlanta are too low; and (2) the predicted cloud of eleva-
ted ozone concentrations appears to be too far north during the afternoon of 4 June
1984.
The fact that diagnostic run 1 is underpredicting the ozone at the upwind (DLLS)
monitor is not surprising since clean background concentrations were used for the
upwind boundary conditions. In particular, given the presence of biogenic VOC emis-
sions in the area, the boundary VOC concentrations used in diagnostic run 1 (25 ppbc
VOC; 0.001 ppb ISOP) are probably too low. In order to determine the magnitudes of
VOC concentrations to use as boundary conditions, a UAM simulation without the
anthropogenic emissions (i.e., biogenic emissions and clean boundary conditions only)
was performed. This simulation of biogenic emissions only produced a maximum
buildup of ISOP and VOC concentrations that ranged across the modeling domain
from, respectively, 0.5 to 2.0 ppb and 35 to 75 ppbC. Thus we chose values a little
lower than the midpoint of the maximum buildup, 1 ppb for ISOP and 40 ppbC for
VOC, as boundary conditions in diagnostic run 2.
An examination of the observed ozone concentrations at the DLLS monitor on 4 June
1984 (see Figure 3-3) reveals that the observed maximum daily ozone concentration
of 7.8 pphm occurs fairly early in the day (12 noon). This suggests that the peak
observed ozone concentration at DLLS is mainly due to entrainment of ozone con-
centrations aloft as the mixing height rises rather than to local production of ozone
8910trl H
44
-------
160
i i i i i I i i i i i i > i i i i i i i i i i
- DKLB 03
24 0
160 160
6
12
18
24
T i i r i | i i i i i | i i i i i | i i i i r
h CNYR 03 q,BSERVCD n
CD PREDICTED -=- -
12
TIME (HOURS)
18
- 120 120
m
H 80 g: 80
-40 40
24
Q-
160
120
80
4O
12 18
TIME (HOURS)
24
160
12
18
120
80
8
1 1 1 1 1 1 1 1 1
- DLLS 03
-
CD
jcDijjcnciicncpc] i i
i i i i i i i i i i i i i i
OBSERVED CD
PREDICTED — -
m _
0 :
i i i i i i i i i CDCDCDCD
12 18
TIME (HOURS)
160
120
80
40
24
FIGURE 3-3. Observed and predicted ozone concentrations (ppb)
for the Atlanta diagnostic run 1.
STSTEMS APPLICATIONS. INC.
89104
-------
due to local sources. Thus the boundary concentrations for ozone, and possibly NOX,
could justifiably be increased. However, we elected to keep the fairly low values for
ozone and NOX boundary conditions (40 and 1 ppb, respectively) for diagnostic run 2.
An examination of the observed winds, the predicted wind fields in diagnostic run 1
(see Morris et al., 1989b), and patterns of predicted hourly ozone concentrations
(Appendix B) indicates that a southerly component in the wind field may occur too
early in the afternoon. The persistence of a more northerly component in the winds
in the late morning and early afternoon would result in the predicted cloud of ele-
vated ozone concentrations impacting the CNYR monitor, producing a higher peak at
that site. However, the wind measurements at the CNYR site observed southerly
winds from about noon on. To determine whether this southerly wind was due to
local effects or larger-scale flow features, the Georgia State meteorologist was con-
sulted. The CNYR wind observation site is located near a grove of trees and may not
be representative of mesoscale wind flows, and the Atlanta Hartsfield Airport wind
observation site (which was inadvertently left out of the original analysis) was better
sited (D. Kemmerick, personal communication, 1989). Thus, the CNYR wind mea-
surement was eliminated and the Atlanta Hartsfield Airport winds added for the next
diagnostic run.
A new definition of the domain-mean wind was also used for the diagnostic run 2.
Instead of a linear interpolation to the hour in question of a vector average of the
five twice-daily upper-air soundings that surround Atlanta, most of which are over
200 km away, the Athens Georgia upper-air sounding alone was used to define the
domain mean wind. Wind data for the lowest 1,500 m of the 4 June sounding at 1900
LST was supplied by the state of Georgia.
A further modification was made on the boundary definition in diagnostic run 2. An
examination of the spatial distribution of emissions (Figure 3-2) and the predicted
hourly ozone concentrations from diagnostic run 1 (Appendix B) indicated that
portions of the northern, southern, and western portions of the modeling domain
could be eliminated without affecting the results. The boundary was thus modified
and diagnostic run 1 was rerun to verify that the new boundary did not affect the
computations.
Results of the Simulation
Diagnostic run 2 differed from run 1 in that (Da higher VOC value was used
compared to 25 ppbC) and higher ISOP boundary conditions were used (1 compared to
0.001 ppb); (2) the domain-mean wind for the DWM came from the Athens upper-air
sounding rather than a vector average from the five upper-air soundings; (3) the
CNYR surface wind measurement was eliminated and the Atlanta Hartsfield wind
measurement added to the analysis; (4) a minor mistake in the emissions inventory
was corrected; and (5) the boundary was modified. The new UAM layer 1 wind fields
generated for diagnostic run 2 is discussed in Morris et al., (1989b). The resulting
46
-------
predicted hourly ozone concentrations for evaluation run 2 are given in Appendix C.
The predicted regional maximum ozone concentration is 13.2 pphm, within approxi-
mately 10 percent of the observed maximum. The predicted maximum ozone is
approximately 22 km to the north of the observed maximum ozone. However, as
shown in Figure 3-4, at the location of the peak observed ozone, the maximum obser-
vation is replicated to within less than 30 percent. In addition, the daily maximum
observed ozone concentrations at the DKLB and DLLS monitor are reproduced to
within, respectively, 18 and 36 percent.
Comparison of UAM grid cell-average concentrations with the point observations at
the site of the observed ozone is a particularly stringent test. A slight incorrect
characterization of the wind field, as is very likely when sparse data sets are used,
will result in the placement of the elevated ozone plume away from the ozone moni-
tor resulting in poor model performance statistics. However, this displacement of
the ozone plume may not affect the model's response to emission control require-
ments. Thus it has been suggested that concentrations in adjacent grid cells (i.e.,
nearest neighbor) also be compared with the observed value in order to determine
whether the ozone cloud is just slightly displaced (Seinfeld, 1988a; Burton, 1988).
Figures 3-5 and 3-6 compare the observed and predicted hourly ozone concentrations
resulting from a one-cell and two-cell search of the predicted concentrations closest
to the observed values. At the CNYR monitor the predicted maximum daily ozone
concentrations for the point, a one-cell search, and a two-cell search match the
observed value of 14.7 pphm to within, respectively, 29, 25, and 22 percent. How-
ever, at the other two monitors (DKLB and DLLS), the nearest-neighbor analysis also
produced improvements in model performance; the point, one-cell search, and two-
cell search match the peak observed observation to within 18, 8, and 5 percent for
the DKLB monitor, and 36, 32, and 23 prcent for the DLLS monitors. For the CNYR
monitor, it appears the elevated cloud of ozone concentrations is too far north. At
the DKLB monitor the predicted ozone concentrations in the general vicinity (i.e.,
within two grid cells) are in very good agreement with the observed maximum values.
Figure 3-7 shows a scatter plot and model performance statistics between predicted
and observed hourly ozone concentrations for diagnostic Run 2. Also contained in
Figure 3-7 are analyses of the residuals of the predicted and observed ozone con-
centrations. As indicated in Figure 3-7 and the time series plots in Figure 3-4,
diagnostic run 2 tends to overpredict the nighttime ozone concentrations and under-
predict the peak daytime values. As a result, the bias of predicted hourly ozone
concentrations is very low (less than 1 percent). However, the gross error (average
absolute error) is fairly high (43 percent). The predicted hourly ozone concentrations
for diagnostic run 2 follow the diurnal and spatial variations of the observations well,
as indicated by the high correlation coefficient of 0.9. Nevertheless, even when all
time and space constraints are removed, the simulated peak ozone value (13.2 pphm)
is 10 percent less than the observed peak (14.7 pphm).
8910«»rl •* 47
-------
160
12
18
I i i i i i i i i r [ T i i i i ( i i i i r
OBSERVED Q
PREDICTED -=-
120
24 0
160 160
12
18
24
I I \\ I | I I I I I | i I I I I | I I I 1 T
- CNYR OBSERVED m ,
D PREDICTED — ~
°a
160
120
80
40
12
TIME (HOURS)
18
12
TIME (HOURS)
18
160
6
12
18
I i i i I i i i i i I i I i I i j i I I i r
- DLLS
OBSERVED UJ
PREDICTED —
120 -
m
8
80 -
40 -
6 12 18
TIME (HOURS)
160
120
80
40
24
FIGURE 3-4. observed and predicted ozone concentrations (ppb) for the Atlanta
diagnostic run. 2.
48
SYSTEMS APPLICATIONS. INC.
8910U
-------
160
12
18
I I I I I I 1 I 1 I I I I I I I I I I I I I I
'" OBSERVED [JJ
PREDICTED
120
24 0
160 160
12
18
24
I I i I i | I I T
- CNYR
i i i i i i i r
13 PREDICTED
Qa
n
6 12 18
TIME (HOURS)
12
TIME (HOURS)
160
120
80
40
18
24
160
6
12
18
24
120
CD
80
4O
T II T ] I I I f I | I I ' I l| I T I
- DLLS
OBSERVED [
PREDICTED -
6 12 18
TIME (HOURS)
160
120
80
24
FIGURE 3-5. Observed and nearest-neighbor predicted (one-cell search) ozone concentrations
(ppb) for the Atlanta diagnostic nan 2.
SYSTEMS APPLICATIONS, INC.
-------
160
12
18
I I I i I I I i I I | I I I I I | I I I I r
DKLB
OBSERVED
PREDICTED
160
120
- 120
24 0
160 160
I I I I I I I I I I I I I I I I I I I I I I !
- CNYR
l-OBSERVED CD
PREDICTED —
12
TIME (HOURS)
18
12
TIME (HOURS)
18
24
160
12
18
160
- 120
i i i i i I i i i i i I i i i i i I i i i i i
' DLLS - m H
OBSERVED CO
6 12 18
TIME (HOURS)
24
FIGURE 3-6. Observed and nearest-neighbor predicted (two-cell search) ozone concentration
(ppb) for the Atlanta diagnostic nan 2.
50
SYSTEMS APPLICATIONS. INC.
89m
-------
120.00
90.00
a.
a
LJ 60.00
o
LLJ
CC
a.
30.00
I I I I | I I T
I I I I
•\l I I I I I I I I I I I I I I I I I I I I I I I
30.00 60.00 90.00
OBSERVED (ppb)
120.00
MOMENTS OF THE PROBABILITY DENSITY FUNCTION
OBSERVED PREDICTED
AVERAGE
STANDARD DEVIATION
SKEkNESS
KURTOSIS
OTHER MEASURES
MEDIAN
UPPER OUARTILE
LOKER OUARTILE
MINIMUM VALUE
MAXIMUM VALLE
50.39999
47.63355
0.51320
-1.22432
40.00000
88.00000
5.00000
5.00000
147.00000
50.88372
29.03820
0.36600
-0.88317
44.24000
72.92999
27.54000
0.53000
- 107.30000
SKILL OF PREDICTION PARAMETERS
CORRELATION COEFFICIENT OF PREDICTED
VERSUS OBSERVED 0.896
THE BOUNDS OF THE CORRELATION AT THE
CONFIDENCE LEVEL OF 0.050 ARE
LOW BOUND 0.831 HIGH BOUND 0.937
RATIO OF OVER TO UNDER PREDICTIONS 1.069
PERCENT OF OVER PREDICTIONS
GREATER THAN 200 PERCENT OF THE
OBSERVED 38.333
PERCENT OF UNDER PREDICTIONS
LESS THAN 50 PERCENT OF THE
OBSERVED 3.333
FIGUPE 3-7a. Scatterplot and model perfonrance statistics for predicted and observed
hourly ozone concentrations for the Atlanta diagnostic run 2 (N = 60).
8910H
51
-------
0.20 -
-40.00 -20.00 0.00
RESIDUAL (OBS-PRED)
THE B1NSIZE EQUALS 10.000
20.00
40.00
RESIDUAL ANALYSIS
AVERAGE -0.48384
STANDARD DEVIATION 25.16143
Sr.EWNESS 0.20968
ICURTOSIS -1.08769
OTHER MEASURES
MEDIAN -0.21000
UPPER QUART1LE 18.80000
LOWER QUART1LE -22.82000
MINIMUM VALUE -42.45000
MAXIMUM VALUE 57.07001
BIAS CONFIDENCE INTERVAL
AT THE 0.0500 LEVEL
LOWER BOUND -12.5186
UPPER BOUND 11.5509
STD RESIDUAL CONFIDENCE INTERVAL
AT THE 0.0500 LEVEL
LOWER BOUND 480.3294
UPPER BOUND 879.6230
THE MEASURES OF GROSS ERROR
THE ROOT MEAN SQUARE ERROR IS 24.96
THE AVERAGE ABSOLUTE ERROR IS 21.88
VARIOUS MEASURES OF RELATIVE VARIABILITY
OBSERVATION COEFFICIENT OF VARIATION
0.9451
RESIDUAL COEFFICIENT OF VARIATION
0.4992
RATIO OF RESIDUAL TO OBSERVED ST. DEV.
- 0.5282
FIGURE 3-71. Residual analysis of observed minus predicted hourly ozone
concentrations for evaluation run 2 (N = 60).
52
891DU
-------
77.3
58.0
40.9
31.8
20.7
32.7
-0.5
-13.7
-21.7
-32.7
-41.9
oo
w -59.0
cc
-78.2
cc
0_
m
o
I I I I
I I I
I I I
- Q
n
n
D
DO
Q
Q
Q
0
Q
D
D
1 1 1 1 1 1 1 1 1
1 1 1 1 1 I 1 1 1 1 1 1 I I
5.00
HOUR
10.00
15.00
20.00
THE LINEAR MODEL PARAMETERS
THE CORRELAflON IS 0.6804
THE LOWER BOUND IS 0.5155
THE UPPER BOUND IS 0.7967
AT THE 0.0500 PERCENT LEVEL
THE Y-X LINEAR MODEL INTERCEPT IS 10.576
THE Y-X LINEAR MODEL SLOPE IS 0.157
THE X-Y LINEAR MODEL INTERCEPT IS -31.398
THE X-Y LINEAR MODEL SLOPE IS 2.944
FIGURE 3-7 c. Plot of residuals versus tine of day for
evaluation run 2 (N = 60).
89104
53
-------
Possible Improvements to Diagnostic Run 2
Improvements in model performance from diagnostic run 2 could be obtained by rais-
ing the VOC, NOX, and ozone boundary conditions, which may be justified based on
the underprediction of the ozone at the upwind monitor (DLLS) and the low values
current being used. The peak observed ozone concentration at the upwind monitor
occurs fairly early in the day (12 noon), which indicates it is probably due to entrain-
ment of an elevated ozone concentration aloft rather than local chemistry.
Improvements can also be made in the wind field. The wind field still has a signifi-
cant southerly component in the late morning and afternoon, which appears to advect
the elevated ozone cloud away from the CNYR monitor. Some wind shear is also
present between UAM layers 1 and 2, which broadens the ozone peak from the urban
emissions, resulting in a lower predicted regional maximum ozone concentration.
Since in the afternoon UAM layers 1 and 2 are both contained within the well-mixed
layer, one would not expect there to be significant amounts of wind shear between
these layers. As noted earlier, during 4 June 1984 a high-pressure ridge passed
across Atlanta, which resulted in a turning of the winds from the NNE to the SSW.
This turning is present in the 0700 and 1900 LST upper-air soundings at Athens, but
the exact hours of the turning of the upper-level winds cannot be determined from
the hourly surface wind measurements, which are subjected to local influences. The
use of linear interpolation of the 0700 and 1900 LST soundings at Athens results in a
continuous turning of the wind throughout the day, when in actuality it probably
occurs within a few hours as the high-pressure ridge passes through. If the upper-
level winds persist with a more northerly component during the morning hours, as is
present in the 0700 Athens sounding, and then turns to the SSW in the afternoon, the
elevated ozone cloud could then possibly impact the CNYR monitor at the proper
time. In addition, use of identical wind fields in layers 1 and 2 during periods of
rapid vertical mixing to eliminate the wind shear below the mixing height may result
in a sharper predicted peak concentration; the Athens upper-air sounding does sup-
port the presence of some wind shear in the mixed-layer.
Another possible explanation for the underprediction of the observed ozone peak at
Conyers is that the 1985 NAPAP inventory may underestimate the amount of VOC
emissions. There are still many uninventoried sources whose individual emissions
may be small, but when combined together may produce a significant impact In
addition, there is a very large uncertainty in the mass amount of biogenic emissions,
the speciation (reactivity) of the biogenic emissions, and the chemistry of biogenic
species. A further possible reason for the underprediction is that the wind speeds are
too high. There has been some discussion that wind speeds reported at FAA sites
(instantaneous 1-minute averages) tend to be higher than the hourly integrated
values. Thus, as a sensitivity simulation, all observed wind speeds at FAA sites were
reduced 50 percent and new UAM wind fields were created. The results of this
sensitivity test are discussed in Appendix D.
8910Hrl it
54
-------
Although implementation of some of the above changes to the boundary conditions
and wind fields would probably improve model performance, the use of those proce-
dures may not be consistent with some of the objectives of this study, namely the
demonstration of the PLANR use of the UAM. Thus, since model performance from
diagnostic run 2 satisfies the model performance goal, it was deemed a suitable base
case. Nevertheless we recommend that when the UAM(CB-IV) and associated pre-
processor programs are delivered to the state of Georgia, they should carry out
sensitivity studies like those discussed above (and presented in Appendix D), which
may improve the UAM inputs data bases and model performance. This exercise will
also help them gain experience with the causes and effects of UAM inputs on predic-
ted ozone concentrations.
8910
-------
H APPLICATION OF THE UAM TO THE
DALLAS-FORT WORTH METROPLEX AREA
In this section we discuss the PLANR application of the UAM to the Dallas-Fort
Worth metroplex area. As discussed in Section 2, the PLANR application of the
UAM involves the following activities: episode selection, meteorological characteri-
zation, modeling domain definition, emissions preparation, preparation of the UAM
modeling inputs, and a limited number of diagnostic simulations designed to obtain a
base case with statisfactory model performance. These topics are discussed in the
following sections.
EPISODE SELECTION
Ozone and meteorological data from the Dallas-Fort Worth area for 1984 through
1987, and parts of 1988 at some stations, were analyzed to determine an ozone epi-
sode suitable for demonstrating the PLANR use of the UAM. From 106 days of
ozone exceedences in this period, three candidate days of high observed ozone con-
centrations were identified as possible modeling days. A memorandum was prepared
for the Texas Air Control Board (TACB), EPA Region VI, and EPA Office of Air
Quality and Planning Standards (OAQPS) on 10 May 1989 that documented the
meteorological conditions, observed ozone concentrations, and appropriateness for
using each of the candidate days for UAM modeling. This memorandum is presented
in Appendix E. The three candidate UAM modeling days and their daily maximum
observed ozone concentrations are as follows:
3une 25, 1984 Dallas 150 ppb, Fort Worth 150 ppb
August 31, 1985 Dallas 170 ppb, Fort Worth 130 ppb
August 1, 1986 Dallas 170 ppb, Fort Worth 120 ppb
Based on the information provided in the memorandum in Appendix E, 31 August
1985 was selected for UAM modeling by the TACB, EPA OAQPS, and EPA Region
.VI. The episode on 31 August 1985 recorded the second highest observed ozone (170
ppb) in the period studied, contained exceedences in both Dallas and Fort Worth, and
had meteorological conditions that are fairly typical of those that produce ozone
exceedences in the Dallas-Fort Worth area.
8910«»rl S
56
-------
The winds on 31 August 1985 were light and variable. Observed daily maximum
ozone concentrations in Fort Worth on 30 August 1985 were quite high (140 ppb). In
order to simulate the high ozone concentrations in Dallas on 31 August it was neces-
sary to model August 30 and 29 as well. Thus, UAM modeling inputs were prepared
for the three-day period 29-31 August 1985. The exact starting time on 29 August
was determined after performing an inert tracer simulation, which indicates how
long it takes to flush the initial conditions out of the modeling domain.
METEOROLOGICAL CHARACTERIZATION
The meteorological conditions of 31 August 1985 were characterized by examining
daily weather maps, observed surface and upper-air meteorological data, and
observed ozone concentrations. Data from the twice-daily upper-air soundings were
used to estimate the heights of the 950, 900, and 850 mb constant pressure sur-
faces. Some of the analyses for 1900 August 30, 0700 August 31, and 1900 August 31
are given in Figures 4-1 through 4-3.
On 30 August a high-pressure system is the dominant feature at both 950 and 900 mb,
centered west of the Dallas-Fort Worth area (Figure 4-1). This high-pressure per-
sisted and continued to influence the region through 1 September (Figures 4-2 and
4-3). The surface winds were light and variable for the three-day period spanning 30
August through 1 September 1985. At 0700 LST on 31 August, surface winds were
light and from the northwest and, as indicated in Figure 4-2, winds aloft had a
southerly component. By 1900 LST the surface winds and winds aloft were light and
from the south (Figure 4-3). The daily maximum temperature increased steadily dur-
ing the episode with maximum surface values of 96, 100, and 103 °F measured on
August 29, 30, and 31, respectively. No frontal systems passed through the area dur-
ing this period. Under the influences of a high-pressure system a region may experi-
ence subsidence (large-scale downward motion) that can limit vertical mixing. Thus,
the combination of high pressure, light winds, high temperature, and limited vertical
mixing allowed ozone and ozone precursors to build up during the three-day period.
The distribution of observed daily maximum ozone concentrations for 29 August
through 1 September 1985 in the Dallas-Fort Worth area are displayed in Figure
4-4. On 29 August the observed daily maximum ozone concentrations range from 5
to 9 pphm. By 30 August high ozone concentrations are observed in Fort Worth (14
pphm) while in Dallas the daily maximum ozone concentrations remain fairly low (5
to 7 pphm). On August 31 there are very high observed daily maximum ozone con-
centrations in Dallas and north of Dallas (17 pphm), and fairly high observed ozone
concentrations in Fort Worth (13 pphm). Daily maximum ozone concentrations are
still fairly high on 1 September (8 to 13 pphm), however there is only one exceedence
of the ozone NAAQS and that occurs in Fort Worth. It appears that the light winds
from the east on 30 August combined emissions from Dallas and Fort Worth to cause
exceedences of the ozone standard in Fort Worth while Dallas remained fairly
8910»»rl ;
57
-------
£
45 765 885 1005 1125 1245 1365 1485 160;
10-
3384
264
FIGURE 4-la. 950 nb constant pressure heists at 1800 on 30 August 1985
(H denotes estimated center of high pressure).
58
-------
1005 1125 1345
1365 1485 1605__
I i | i I i I i I I I j 4^24
= 4104
1216
IAMAR
t i M / i i ; I l I l l i l i IVL I i l l I i M l i I i l l I i l l l l I i M l I I i I l I l l i M l i I l I M M M \K.ff( i I II l I l 1 _T
4-lh. 900 nb constant pressure heights at 1800 on 30 August 1985
(H denotes estimated center of high pressure).
59
-------
8
765 885 1005 1125 1245 1365 1485 160?
J4%24
= 4104
- 3984
10 20 30 40 50 60 70
3384
264
FIGURE 4-2a. 950 nb constant pressure heights at 0.600 on 31 August 1985
(H denotes estimated center of high pressure).
60
-------
fi45 765 885 1005
OJ l i I i l I M| i I i l , I I i ! | i i i i i i i i i | i i;
1125 1245
"i i i | i i i i i i i i i i i
__
i i i i | i i ii i i ii i | i i . M M r j 4224
~1 I I I I / ! I I I I I 1 I I I 1VLI I I I I I I I I I I I
101
0
0 10 20
264
FIGURE 4-2b. 850 nb constant pressure heists at 0600 on 31 August 1985
(H denotes estimated center of high pressure).
61
-------
645765 885 1005 1125 1245
1365 1485
I I i ui I I i , | M i i i M I j
__
Uu 1 1 1 ; M i ;ii ii
20-
101
- 3744
- 3624
3504
3384
\t~lllll/lllllllllll I^L I I I I I I I I I I I I I I I I I I I I I I I I I M I I I I I I I I I I I I I I I I I I I I \K,W I I I I I I I I r
10 20 30 40 50 60 70
264
FIGURE 4-3a. 950 nb constant pressure heights at 1800 on 31 August 1985
(H denotes estiirated center of high pressure).
62
-------
fi45 765 885 1005 1135 1245 1365 1485 160
1512
/AMAR
~l I I I I / I I i I I I I I I I lU.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 I I I I I I I I I I I I I I I I \Kl&( I I II II I I
264
FIGURE 4-3b. 850 r±> constant pressure heights at 1800 on 31 August 1985
(H denotes estimated center of high pressure).
63
-------
375C
3700 ^
3650 J
3600-
3550 -
600
650
700
750
600 850
•* i i i I i i i i i i i i i I 3'
350
750
3700
- 3650
- 3600
- 3550
500
FIGURE 4-4a. Observed daily rraxinum ozone concentrations (ppb) in the
Dallas-Fort Worth area, 29 August 1985.
64
-------
375'
3650 =f
i i i I i i i i i i i I i I 3750
FIGURE 4-4b. 30 August 1985.
65
-------
375
3700-
3650 J
i i i i i i i i i i i i i i 3750
500
FIGURE 4-4c. 31 August 1985.
66
-------
375§
3700-
3650
i i i I i i i i i i i I i | 3750
3600::
FIGURE 4-4d. 1 Septenber 1985,
-------
clean. As the surface winds shifted from the east to the northwest to the south the
elevated cloud of ozone moved east to Dallas and then north of the city to cause the
high ozone concentrations in Dallas on 31 August.
Winds were still fairly calm on 1 September, yet the only ozone exceedence was 13
pphm recorded in Fort Worth. Although observed daily maximum ozone concentra-
tions were fairly high on 1 September (8 to 13 pphm), one might expect higher ozone
on this day because of the calm conditions and high observed ozone concentrations on
previous days. Possible explanations for this are that high ozone concentrations
occurred but the monitoring network did not pick them up, or the lower emissions on
Saturday (31 August) and Sunday (1 September) did not emit enough ozone precursors
to cause a high exceedence despite the indications that meteorological conditions
were favorable for ozone formation. The exceedences recorded on 31 August (Satur-
day) are most likely a combination of the emissions on Friday (30 August) and Satur-
day.
MODELING DOMAIN DEFINITION
Because of the stagnant meteorological conditions of 29-31 August 1985 it is diffi-
cult to determine how large a modeling domain is required to contain the predicted
ozone peaks. The meteorological data indicate that there will be a sloshing of emis-
sions from the Dallas-Fort Worth area to the west on 30 August and back to the
northeast on 31 August. Thus an overly large modeling domain was selected, with an
origin at UTM zone 14 coordinates of 600 km easting and 3560 km northing, spanning
180 km in the east-west direction and 160 km in the north-south direction (see Figure
4-5). As stated in the Dallas-Fort Worth modeling protocol, and to be consistent
with other PLANR applications of the UAM, a horizontal grid spacing of 4 km was
used and the vertical structure consisted of two layers below and three layers above
the mixing height.
Meteorological and emission inputs were prepared for the entire modeling domain.
However, after performing inert tracer simulations and examining the emissions dis-
tribution, the computational grid was reduced by defining new boundaries in the UAM
BOUNDARY input file. This reduction in the computational grid is to reduce
unnecessary computations.
ROUTINE DATA AVAILABLE FOR DALLAS-FORT WORTH
Surface meteorological data (hourly average wind speed, wind direction and tempera-
ture) collected at Continuous Air Monitoring Stations (CAMS) were obtained from
EPA Region VI. These eight surface CAMS sites are listed in Table 4-1. Additional
surface meteorological data from FAA sites at the Dallas-Fort Worth airport, Red-
bird, Love Field, Addison, and Meacham were also available.
8910Vrl 5
68
-------
45 765 865 1005 1125 1245 1365 1485 160g>
10-
&24
= 4104
UAM Modeling Domain
~i I I I II I I I I I I I l I I ll.1 II I I I I I I I I I I I I I I M I I I I I I II I I I I I I I I I I I I I I I M I I
10
3384
264
FIGURE 4-5. Locations of upper-air meteorological observation sites and
UAM modeling domain for the Dallas-Fort Worth metroplex area.
69
-------
TABLE 4-1. Surface meteorological observation sites at the continuous
air monitoring stations (CAMS).
Site
ID Site Code
Site Location
MOCK 451310044H01
NTRA 451310045F01
HNTN 451310069H01
BNVW 451310055H01
DNTN 451420054H01
ROSS 451880002F01
PECN 451880003F01
KLLR 455070003F01
1500 West Mockingbird, Dallas
12532 Nuestra Drive, Dallas
1415 Hinton Drive, Dallas
10501 Bonnie View, Dallas
Quarter Mile Road off Highway 121, Denton County
Ross Avenue, between Long and 34th Street, Fort Worth
TOO N. Pecan, Fort Worth
FAA site off Alta Vista Road, Fort Worth
89104 3
70
-------
Upper-air measurements at Longview, Stephenville, Midland, and Amarillo, Texas,
and Oklahoma City and Fort Sill, Oklahoma, were also obtained (see Table 4-2). The
locations of these upper-air sites and their relationship with the modeling domain is
shown in Figure 4-4.
There were eight air quality monitor sites in operation during 29-31 August 1985.
The locations and parameters measured at these eight sites are given in Table 4-3.
One site was in Fort Worth (ROSS) and another just north of Fort Worth (KLLR). The
remaining sites were in or around Dallas: one was located south of the city (BNVW),
four were roughly within the city (MOCK, NSTA, ILLI, and NRTH), and one was north
of the city (DNTN). There were no air quality monitors located near the modeling
domain boundaries from which boundary conditions could be based.
EMISSIONS PREPARATION
Two separate emission files are required by the UAM: an area source emission file
(EMISSION) whose emissions are injected into the lowest layer of the UAM; and a
point source emission file (PTSOURC) whose emissions are injected into the appro-
priate vertical layers depending on plume rise calculations. The 1985 base case
emissions data files were based on the 1985 NAPAP anthropogenic emission inven-
tory (Zimmerman et al., 1989) and biogenic emissions estimates obtained from Wash-
ington State University (Lamb et al., 1986, 1987).
Anthropogenic Emissions
The anthropogenic emission inventory for the 1985 base case was created from the
1985 NAPAP point and area source anthropogenic emission data. The procedures for
converting the 1985 NAPAP anthropogenic emissions into the UAM format were the
same as those used for the Atlanta application and followed the methodology given in
Figure 2-8 using software documented in "User's Manual for Preparing Emission Files
for use in the Urban Airshed Model" (SAI, 1989).
Biogenic Emissions
The biogenic emissions inputs for the UAM application to Dallas-Fort Worth were
prepared using different data sources and techniques than used for the Atlanta appli-
cation. Biogenic emissions for Atlanta were originally generated for the EPA
Regional Oxidant Model (ROM) and were obtained from the EPA. However, similar
estimates for Dallas were not available because the ROM southeast modeling domain
did not include Dallas. Thus the WSU monthly average, county-wide biogenic emis-
sions data were used in the Dallas application.
8910trl 5
71
-------
.c
c.
Q
^IP
40
£-
0
[L,
1
CO
CO
Q
CM
O
^j
40
•vH
C
o
Q)
40
C
0)
0)
•*•>
01
•c
o
•H
40
<0
*
0)
n
JO
o
i-H
co
0
•H
0
W
i— 1
O
u
o
0>
40
i
t.
1-1
cd
1
(^
&
o.
•
CM
1
Ed
J
CD
£5
.C
40
O O
40
0) 40 t.
O (* O
coo.
S tu £-
40 I -H
CO CO
-------
c
0
iH
43
CO
i.
0>
Q.
O
c
•**
CO
0)
£.
ca
43
t.
Q
*
43
t*
0
bu
cd
•H
.-H
£
Q)
JC
C
•H
ra
t.
•5
^H
c
^,
4J
iH
.-H
CO
3
•rH ON
tt> 43
o n
cd 3
If
CO
CO »-
1 CM
tO
1x3 C
•J "»H
CQ tt
«S 3
H -0
0}
Q) 0)
43 J_
0) 3
E CO
cd (d
cd 2:
cu
•—•
^ Ji
«— «*v*
^k
Q)
O
^P (f**N
s i
MI? JC
X
c
o
iH
43
D.
t.
O
CO
tO
CO
to
43 Q
— H I-H
CO
o
o
•k
0^
z
•t.
o
z
•k
CO
o
:;
oo
CO
CO
VO
CO
^—
vO
ON
ON
vO
CO
cd
rt
Q
»
•o
t.
2
to
3
o
0
03
o
O
in
^*
^^
i
o1'
^F
•t
0
Z
«h
CO
o
oo
o
in
a-
^O
CO
CO
^^
CO
ST
0
CO
cd
i-H
cd
f^
••
§:
•H
C.
a
cd
^J
CO
0)
58
CM
CO
in
CM
gj
vO
vO
s:
43
£^
£
43
S.
O
•t
•o
g
OS
cd
43
CO
**
5
rH
CM
CM
O
to
43
•H
CO
4
^2
(X*
OS
f*J
3
o
CJ
in
CM
CM
ON
CM
vO
CO
^
O
CO
VO
O
CO
cd
i-H
-H
cd
0
•.
c
cd
CO
43
cd
^k|
(0
t*
u
-------
Spatial Allocation
The monthly average biogenic emissions data obtained from WSU consisted of
county-wide emissions of alpha-pinene, isoprene, and other NMHC by six land use
categories. The biogenic emissions were allocated to the modeling grid cells based
on the distribution of a surrogate. An examination of the available surrogates (popu-
lation, roadways, area, and the 11 CIS land cover categories) identified only two
appropriate surrogates that matched the WSU land use categories: alpha-pinene
emissions from scrub and coniferous forest. Thus the county-wide alpha-pinene
emissions were allocatec tc the UAM modeling grid cells based on these two land use
categories, whereas the isoprene and other NMHC biogenic emissions were allocated
to the modeling grid based on fractional area covered.
Speciation
The speciation of the biogenic emissions followed the recommendations of Hogo and
Gery (1988). Isoprene is an explict species (ISOP) in the UAM(CB-IV) and was
assigned to the CB-IV TSOP category. Alpha-pinene was split into the CB-IV species
as 0.5 OLE, 6 PAR, and 1.5 ALD2. The catgory of other NMHC in the WSU biogenic
emission estimates were assumed to be monoterpenes and were split into 1 OLE and
SPAR.
Temporal Allocation
The diurnal variation of the biogenic emissions was based on the analysis of biogenic
emission rates reported by Tingey and co-workers (1978a,b). As illustrated in Figure
4-6a, the diurnal variation of the alpha-pinene and other NMHC biogenic emissions
was based on measured variations for five monoterpenes for slash pine in Tampa,
Florida. The diurnal variation for isoprene emissions was based on emission rates
from oak leaves assuming shaded leaves with one-half ambient sunlight (curve B in
Figure 4-6b).
Results
Figure 4-7 displays the spatial distribution of the emissions used in the Dallas-Fort
Worth diagnostic simulations. The emission totals for the entire modeling domain
are summarized in Table 4-4.
PREPARATION OF UAM INPUT FILES
The following paragraphs describe how the UAM input files were prepared for the
PLANR application of the UAM to Dallas-Fort Worth. A definition of the
parameters contained in each file is discussed in Section 2.
8910<»rl &
74
-------
2400
0600 1200
Time of Day-Hr
1800
2400
FIGURE 4-6a. Estimated diurnal emissions of five monoterpenes for slash
pine in Tampa, Florida, for an average of summer days. (Adapted from
Tingey et al. 1978b)
10
600
900
1200
Time of Day
1500
1800
FIGURE 4-6b. Estimated diurnal emissions of isoprene
for oak leaves in Tampa, Florida, for an average of
summer days. A - sunlit leaves; B - shaded leaves (one-half
ambient sunlight); C - shaded leaves (one-fourth ambient
sunlight). (Adapted from Tingey et al., 1978b)
EEE89KM
75
-------
*r v
00°
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 1 I I I I I I I I I
- 3600
30
DALLAS '85 Area Weekday
NOx
Total: 280130 (Kg/day)
FIGURE 4-7a. 1985 emissions for Dallas-Fort Worth.
40
3560
76
-------
«?°
X X
i i T i ii i iii i i 'i"i '1 3720
i i i i i i i ii i r ; -i 'i
= 3640
- 3600
3560
DALLAS — '85 Area Weekend
NOx
Total: 237061 (Kg/day)
FIGURE 4-7b.
77
-------
*
cP
*F -7
3720
- 3680
= 3640
- 3600
40
3560
DALLAS — '85 Motor Vehicle Weekday
NOx
Total: 346986 (Kg/day)
FIGURE 4-7c.
78
-------
i i i I i i i i r i i
1 I I I I I1 1 I1 TT I I'll I l"l T I
3720
- 3680
= 3640
- 3600
30
DALLAS — '85 Motor Vehicle Weekend
NOx
Total: 260239 (Kg/day)
40
3560
FIGURE 4-7d.
-------
00
640
680
720
I I I I
30
20
10
I I T I I I ] T I I
o°
3680
3640
3600
0
10
20
30
40
3560
DALLAS -— '85 Low Level Points Weekday
NOx
Total: 18153 (Kg/day)
FIGURE 4-7e.
80
-------
00
30
20
10
0
640 680
i i i i ii i i i i i i i i i I i i ii i
x<\.
720
i I i l l
760
i l i i l i
0__
3720
3680
3640
3600
i i l i i i i I i i i l i i i
0
10
20
30
40
3560
DALLAS — '85 Low Level Points Weekend
NOx
Total: 17661 (Kg/day)
FIGURE 4-7f.
81
-------
640
680
720
760
3720
- 3680
- 3640
- 3600
30
40
3560
* \
FIGURE 4-7g.
Dallas 85 Base Elevated Point Sources
NOx (kg/day)
Total = 162761.90
82
-------
3720
- 3680
= 3640
- 3600
10
40
3560
DALLAS — '85 Area Weekday
RHC
Total: 400017 (Kg/day)
FIGURE 4-7h.
83
-------
N^ ^ /N Q0°
r\ r\ sV
i i I i i II i i i I i i I i i i i i ii i i i i i i • i' i 13720
i i i i i i i i i] i i
- 3600
3560
DALLAS — '85 Area Weekend
RHC
Total: 221166 (Kg/day)
FIGURE 4-7i.
84
-------
3720
- 3680
^ 3640
- 3600
0
3560
DALLAS — '85 Motor Vehicle Weekday
RHC
Total: 831840 (Kg/day)
FIGURE 4-7j.
85
-------
i l | i I l I l
l l l I I l I l I I I I I I I I I I l I I I I ! I l
I I i >^i \ I '•[~*~~tt i I I i I
3720
- 3680
30
DALLAS — '85 Motor Vehicle Weekend
RHC
Total: 623882 (Kg/day)
40
= 3640
- 3600
3560
FIGURE 4-7k.
86
-------
600 640
40i i i i i i i i i i | i i i i
30
20
10
0
0
680 720
i i i i I i i ii i i i i i i I i i
760
i i H i i i I i
.
i i
3720
3680
3640
I I I I I I I I IX I I I I I I I
3600
10
20
30
40
3560
DALLAS — '85 Low Level Points Weekday
RHC
Total: 23393 (Kg/day)
FIGURE 4-71.
87
-------
00
640
680
720
760
30
20
10
0
Jr\ K I \r—r^ll I I I I
3720
0
10
20
30
40
DALLAS — '85 Low Level Points Weekend
RHC
Total: 10429 (Kg/day)
3680
3640
3600
3560
FIGURE 4-7m.
-------
00 640 680 730
• i i i i i i i i |i i i r i i i i i i | i i ii i i i i i i | i i i i
30
20
10
0,
760
T—r~i—i—r~r~i—r~i—i 3720
3680
3640
3600
10
20
30
40
3560
Dallas 85 Base Elevated Point Sources
THC (kg/day)
Total = 45399.44
FIGURE 4-7n.
89
-------
TABLE 4-4. Summary of emission totals (tons per day) for the
Dallas-Fort Worth modeling domain.
Weekday Weekend
VOC NOX CO VOC N0y CO
Motor vehicle 926 430 3174 734 344 2701
Other anthropogenic 454 439 193 190 382 184
Total anthropogenic 1380 869 3367 924 726 2885
Biogenic 631 0 0 631 0 0
Total anthropogenic 2011 869 3367 1555 726 2885
and biogenic
8910H 3
90
-------
DIFFBREAK
This file contains the daytime mixing height or nighttime inversion height for each
column of cells at the beginning and end of each hour of the simulation. Hourly mix-
ing heights were estimated at several surface meteorological sites through use of the
hourly surface measurements of temperature and the twice-daily upper-air observa-
tions from a representative upper-air site. The Longview, Texas upper-air site was
felt to be most representative of upper-air conditions in Dallas-Fort Worth because
it is closer to Dallas, the location of the observed maximum ozone concentration on
31 August (note that the Stephenville, Texas upper-air site is closer to Fort Worth).
However, both the Longview and Stephenville sites yielded similar maximum mixing
heights on the days selected for modeling. The diurnal variation in mixing heights at
the eight CAMS surface meteorological observation sites (Table 4-1) was calculated
using the RAMMET meteorological processor. These hourly values were then input
into the standard UAM mixing height interpolation program* DFSNBK (Ames et al.,
1985b) using the 1/r interpolation option to produce the spatially varying hourly
fields of mixing heights. The maximum mixing height in diagnostic run 1 on 31
August was approximately 3,100 m AGL. Because it was believed that mixing
heights lower than 3,100 m AGL actually existed in Dallas-Fort Worth on 31 August,
diagnostic run 2 used a maximum mixing height of 2,000 m AGL. This is discussed
later in this chapter.
REGIONTOP
This file contains the height of each column of cells at the beginning and end of each
hour of the simulation. If this height is greater than the mixing height, the cell or
cells above the mixing height are assumed to be within an inversion. For the Dallas-
Fort Worth application a constant 3,000 m AGL region top was initially used. How-
ever, in diagnostic run 2, with the reduced maximum mixing height, a region top of
2,000 m AGL was specified.
WIND
This file contains the x and y components of the wind velocity for every grid cell for
each hour of the simulation. There are two steps in creating the wind fields for
Dallas-Fort Worth: (1) application of the Diagnostic Wind Model (DWM) (Morris et
al., 1988, 1989a; Douglas and Kessler, 1988) using 15 vertical levels and data from
the eight surface CAMS and two upper-air meteorological observation sites (Long-
view and Stephenville, Texas); and (2) vertical interpolation of the 15-layer hourly
wind fields into the five-layer UAM configuration used for the Dallas-Fort Worth
application (see Morris et al., 1989a).
8910»frl 5
91
-------
The DWM creates an hourly 15-layer wind field by first creating a wind field by
adjusting a domain-mean wind (based on upper-air observations) for terrain effects
(kinematic effects, thermodynamically generated slope flows, blocking, and deflec-
tion). Then the observational information is added to the wind field by weighting the
initial wind field heavily away from the observations and the observed winds heavily
in the vicinity of the observations.
The 15-layer wind fields created by the DWM are then vertically averaged into the
UAM five layer configuration. Since the surface observations provide a more
detailed representation of the air flows in the vicinity of Dallas-Fort Worth and their
presence is only felt within the lowest layer of the DWM, the first UAM vertical
layer was assumed to have the same wind field as the first DWM layer. The DWM
first layer is also weighted 50 percent in the second UAM vertical layer. The wind
fields for the remainder of the UAM vertical layers were obtained from the DWM
wind fields using weighted averaging. The resultant five-layer wind field was then
modified using the procedure suggested by O'Brien (1970), which minimizes the verti-
cal velocity out of the top of the region. In this manner, the boundary concentration
assumed to exist above the region top (TOPCONC) does not greatly influence con-
centrations within the modeling domain.
For the application of the UAM to Dallas-Fort Worth, the hourly varying domain-
mean wind input to the DWM was defined by averaging the winds from the two
twice-daily upper-air soundings and then linearly interpolating these average sound-
ings in time to the hour in question. Surface wind observations from the eight CAMS
sites (Table
-------
(TGRADABOVE) the inversion were based on the twice-daily upper-air soundings
taken at Longview, Texas, and hourly surface temperatures observed at the NWS sur-
face site in Dallas. Values for TGRADBELOW from the surface to the mixing height
from the Longview evening sounding (1800 LSI) are -0.0091, -0.0107, and -0.0113
K/m for August 29 through 31, respectively. Based on these measurements, values
for TGRADBELOW were -0.0110 K/m for exposure classe 3, -0.0105 for exposure
class 2, -0.01 K/m for exposure class i, and -0.098 for exposure class 0.
The Longview morning sounding (0600 LST) measured values for TGRADBELOW of
0.012, 0.008, and 0.017 K/m on the mornings of August 29 to 31, respectively, assum-
ing that the top of the surface based inversion was 250 m AGL. Observed tempera-
tures at 250 m on August 29 through 31 were, respectively, 297.8, 297.4, and 301.3
K. Nighttime hourly values for TGRADBELOW were then calculated for 29-31
August based on the surface temperature at the Dallas NWS surface site and the
measured value at 250 m from the 0600 LST sounding.
For predawn hours, the temperature gradient from 250 m to the original region top
(3,000 m) from the 0600 sounding at Longview was used, resulting in values for
TGRADABOVE of -0.0061, -0.0055, and -0.0062 K/m for August 29 through 31,
respectively. For daytime hours a temperature at the mixing height was calculated
for each hour of the day using the surface measurement at the Dallas NWS site and
the value for TGRADBELOW. Then the hourly temperature gradient between the
mixing height and the region top was estimated using the calculated value at the
mixing height and the measured temperature from the 0600 sounding at Longview at
3,000 m AGL (281.8, 284.1, and 283.4 K for August 29 through 31, respectively)
AIRQUALITY
This file contains the initial concentrations of each species for each grid cell at the
start of the simulation. Since the UAM was initiated at 12 noon on 29 August 1985,
most of the material from the initial conditions was advected out of the region by
the morning of 31 August; thus, initial concentrations do not influence ozone forma-
tion on 31 August. In the majority of the region the initial concentrations were
assigned "clean values" as follows:
VOC 25 ppbc (using EKMA default speciation)
ISOP 0.001 ppb
NOX 1 ppb (split 3/4 NO2, 1/4 NO)
O3 40 ppb
CO 200 ppb
Air quality measurements for ozone, NO, NO2» and CO were available at up to eight
sites in the Dallas-Fort Worth region (Table 4-3). Initial conditions were set accord-
ing to the measured air quality data in the neighborhood of the sites.
8910trl 5
93
-------
BOUNDARY
This file contains the location of the modeling region boundaries. It also contains the
concentration of each species that is used as the boundary condition along each
boundary segment at each vertical level. For the application of the UAM to Dallas-
Fort Worth the minimum of a one-cell buffer of unsimulated cells was used in the
tracer simulation, resulting in a simulation region of 172 km by 152 km. The "clean"
values listed previously were specified for boundary c^nditicr-s. . lp.e tracer simula-
tion showed that the stagnant conditions of the episode resulted in little transport of
pollutants away from the urban areas. Therefore for the photochemical diagnostic
simulations 3-, 4-, 7-, and 1-cell buffer zones were specified for respectively the
north, south, east, and west boundaries, resulting in a 148 km by 124 km computa-
tional modeling domain.
TOPCONC
This file contains the concentration of each species for the area above the modeling
region. Since the wind fields are processed to eliminate the vertical velocity through
the'region top, the model results are not sensitive to TOPCONC. Accordingly, the
clean concentration values listed previously were used.
TEMPERATUR
This file contains the hourly temperature for each surface layer grid cell. Hourly
spatial varying temperatures were obtained by using 1/r interpolation from the eight
CAMS surface meteorological observations.
TERRAIN
This file contains the value of the surface roughness and deposition factor for each
grid square. Each 4 km by k km grid cell was assigned to a land-use category based
on a digitized USGS map of the modeling region. The land-use categories were then
converted to roughness lengths and vegetation factors according to data published by
Argonne National Laboratory (see Morris et al., 1989a).
CHEMPARAM
This file contains information regarding the chemical species to be simulated,
including reaction rate constants, upper and lower bounds, activation energy, and
reference temperature. Reaction rate constants correspond to those given in the
94
-------
report documenting the CB-IV mechanism (Gery, Whitten, and Killus, 1988), except
that reactions for methanol (MEOH) and ethanol (ETOH) have also been added to the
mechanism (see Morris et al., 1989a).
TRACER SIMULATION
In order to determine the starting time and the computational modeling region for
the UAM simulation, an inert tracer simulation was performed using an initial con-
centration field of all 100 and zero boundary conditions and emissions (Appendix F).
Similarly, an inert boundary condition tracer simulation (100 on boundaries, zero
initial concentrations and emissions) and emission tracer (constant emission rate with
zero initial and boundary conditions) were performed (not shown). The tracer simula-
tion was initiated at noon on 29 August. Note that these tracer simulations differ
substantially from the weighted tracer simulation performed for Atlanta (Appendix
A). In the tracer simulation for Atlanta the inert species were used to represent
VOC and NOX pollutant fluxes, whereas the inert tracer simulations performed for
Dallas-Fort Worth were used only to determine how much of the initial conditions
are present and gross flow aspects of the modeling inputs.
After 12 hours of simulation (0100 on 30 August) the initial concentrations still
dominate about half of the region in the upper-northwest sector of the modeling
domain. By noon on 30 August the initial condition tracer ranges from 5 to 30 in the
upper northwestern portion of the region. At midnight on 30 August, after 36 hours
of simulation, the initial condition tracer is still in the northwestern portion of the
modeling domain with concentrations as high as 11. By 0800 on 31 August the maxi-
mum initial condition tracer is reduced to 4 and located in the northwestern corner
of the modeling domain, well away from Fort Worth and Dallas.
The inert initial concentration tracer analysis indicates that if a starting time of
noon on 29 August is used, the initial concentrations will not have any influence on
the predicted concentrations on 31 August but will influence concentrations in the
upper-northwest sector of the modeling domain on 30 August. The initial concentra-
tions will also have a little influence on predicted concentrations in Fort Worth on
30 August but will not affect concentrations in Dallas on that day.
The inert initial, boundary, and emission tracer simulations also provided insight on
how the computational modeling domain could be reduced without affecting the pre-
dicted concentrations. Because pollutants tend to slosh to the west on 30 August,
the western boundary condition was not adjusted downward. It appears that pollu-
tants travel somewhat eastward on 31 August, but then turn more toward the north-
east. Since the wind speeds are so low on 31 August, pollutants are not expected to
travel too far eastward from Dallas; thus eight grid cells (32 km) from the eastern
boundary were eliminated from the computational grid. Finally, five grid cells from
the southern boundary and four grid cells from the northern boundary were also
eliminated in the computational grid.
8910trl 5
95
-------
EVALUATION SIMULATIONS
Two diagnostic simulations were performed before an adequate base case was
obtained. As discussed above, the modeling period began at noon on 29 August 1985
and ended at 2000 on 31 August. All meteorological inputs were based on the six
CAMS surface sites and the two upper-air meteorological observation sites. The
first 12 hours of the simulation on 29 August was used to initialize the model and the
model was evaluated against hourly average ozone, NO, NO2» and CO concentrations
at the surface air quality monitoring sites on August 30 and 31.
Diagnostic Run I
For the first diagnostic simulation the observed meteorological data were used "as
is" in the wind field model and other UAM preprocessors. The maximum mixing
height on 31 August exceeded 3,000 m AGL. As noted in the tracer simulation, there
is still significant amounts of initial conditions within the modeling domain on 30
August (see Appendix F). Since the model was initialized with "clean" concentra-
tions, it is expected that the model may underpredict the peak observed ozone con-
centrations on 30 August.
30 August 1985
A comparison of predicted and observed hourly average ozone concentrations for
diagnostic run 1 at the air quality monitoring sites are given in Figure 4-8. (Isopleths
of hourly average ozone concentrations for diagnostic run 1 are shown in Appendix
G.) As seen in Figure 4-4b, the observed peak ozone concentration on 30 August is
14 pphm and occurs in Fort Worth (ROSS). The predicted region-wide maximum
ozone concentration on 30 August is 12.4 pphm (within 11 percent) and occurs
approximately 40 km west of ROSS (Appendix G). Predicted daily maximum ozone
concentrations in Fort Worth range from approximately 7 to 10 pphm on 30 August.
At the location of the observed peak ozone concentration on 30 August (ROSS) the
model predicts the observed peak (14 pphm) to within only 43 percent (7.9 pphm) At
the other ozone site in Fort Worth operating on this day (KLLR) the observed maxi-
mum ozone concentration for 30 August (11 pphm) is replicated by the model to
within 7 percent (10.2 pphm).
In the Dallas area the observed maximum ozone concentrations on 30 August range
from 5 to 7 pphm (Figure 4-4b) while maximum model predictions range from 7 to 10
pphm (Appendix G). It appears that for 30 August in Dallas the model overpredicts
the observed peak ozone concentrations (MOCK, NSTR, ILLI), while to the north
(DNTN) and south (BNVW) the model replicates the observations quite well.
8910"+rl S
96
-------
200
12
16 24
30 36
42 46
150
m
100
eo
O
50
i i i i
MOCK
i I I I I I
OBSERVED O
m
CD
I I I I/I I I I I I I ! I I I I I I I I I
200
150
100
50
6 12 16
AUGUST 30. 1985
24 30 36 42
TIME (HOURS) AUGUST 31. 1985
46
200
12
18 24
30 36
42 48.
150
m
100
50
I I I 1 I 1 i I I I I I I I 1 I I I I i I I ] I I I 1 I ( I ! 1 I T ] i I I I I | I I 1 I
NSTR
OBSERVED D
I I I I I I I I J I I I I J 1 I I I
200
150
100
50
6 12 16 24 30 36 42 46
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
FIGURE 4-8. Tine series of predicted and observed hourly ozone
concentrations (.ppb) for the Dallas-Fort Worth diagnostic run 1.
' SYSTEMS APPLICATIONS, INC.
89101+
Q7
-------
200
12
16
24
30
36
48
150
200
ILLI
OBSERVED O
- 150
- 100
6 12 16 24 30 36 42 46
AUGUST 30, 1985 TIME (HOURS) AUGUST 31, 1985
200
12
18
24
30
36
42
46
150
CD
eu 100
50
NRTH
OBSERVED D
I I I I
I I I I I I I I I I I I I
200
150
100
50
6 12 16 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31, 1985
48
FIGURE 4-8 continued.
SYSTEMS APPLICATIONS., INC.
89104
98
-------
200
150
en
100
50
0
BNVW
OBSERVED
12
18
24
30
36
42
43
200
150
100
0 6 12 16 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31. 1985
43
200
200
6 12 16 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31, 1985
- 150
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
DNTN
OBSERVED d
i i i i I i' i i i i I i i i i i I i i i i i I i i i i i I i
- 100
46
FIGURE 4-8 continued.
SYSTEMS APPLICATIONS. INC.
891
-------
200
200
- 150
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
- 100
6 12 16
AUGUST 30. 1985
24 30 36 42
TIME (HOURS) AUGUST 31. 1985
200
12
16 24
30 36
42 48.
150
CD
CL 100
n
o
50
PECN
OBSERVED
200
150
100
6 12 16
AUGUST 30. 1985
24
TIME (HOURS)
30 36 42
AUGUST 31, 1985
FIGURE 4-8 continued.
SYSTEMS APPLICATIONS, INC.
8910%
100
-------
200
200
- 150
- 100
6 12 18
AUGUST 30. 1985
24
TIME (HOURS)
30 36 42
AUGUST 31. 1985
48
FIGURE 4-8 concluded.
SYSTEMS APPLICATIONS, INC.
-------
Figure 4-9 presents a scatterplot along with model performance statistics for hourly
ozone concentrations on 30 August. Over all hours and sites the model predictions
have a bias of about 12 percent (-0.54 pphm) tending toward overprediction and a
gross (average absolute error) of about 43 percent (2.0 pphm). The correlation coef-
ficient between predicted and observed hourly ozone concentrations on 30 August for
diagnostic run 1 is 0.64.
For diagnostic run 1 on 30 August the model predicts higher concentrations in Fort
Worth than in Dallas. This ozone distribution is verified by the observations. The
model underpredicts the observed peak ozone concentrations in Fort Worth and
..lightly overpredicts the observed daily maximum ozone concentrations in Dallas.
31 August 1983
On 31 August the observed peak (17 pphm) occurred at two sites: one in Dallas
(MOCK) and another north of Dallas (DNTN). As seen in Appendix G, the predicted
region-wide maximum ozone concentrations for diagnostic run 1 was 14.5 pphm
(within 15 percent of the peak observation) and occurred in north Dallas between the
two monitors that recorded the maximum observed value. At the location of one of
the observed peaks in Dallas (MOCK) the model underpredicts the peak (17 pphm) by
approximately 56 percent (7.5 pphm), whereas at the site north of Dallas (DNTN) the
observed peak (17 pphm) is underpredicted by about 35 percent (11.1 pphm). At the
only other site in Dallas that recorded a violation of the ozone NAAQS (14 pphm) on
31 August (NSTR) the model underpredicts by 36 percent (8.9 pphm).
Both of the sites in Fort Worth also recorded violations of the ozone NAAQS on 31
August (13 pphm). At these two sites the predicted maximum ozone concentrations
on 31 August were within 39 (7.9 pphm at ROSS) and 22 (10.2 pphm at KLLR) percent
of the peak observation.
Figure 4-10 displays a scatterplot and model performance statistics for predicted and
observed hourly ozone concentrations on 31 August 1985 and diagnostic run 1. Over
all hours the model has a bias of 18 percent tending toward underprediction (1.3
pphm) and a gross error of 33 percent (2.3 pphm). The hourly ozone predictions fol-
low the observations well, as evidenced by the lack of excessive scatter in the scat-
terplot (Figure 4-10a) and the fairly high correlation coefficient (0.83).
NO, NO2, and CO
Hourly observations of NO and NO2 were available at four of the air quality monitor-
ing sites: MOCK, NSTR, BNVW, and ROSS. CO observations were also available at
four sites: MOCK, ILLI, NRTH, and ROSS. Historically, the performance of air
quality simulation models for primary species, such as NO, NO2> and CO, has been
much worse than for secondary species, such as ozone. This is because the effects of
102
-------
160.00
120.00
o
a
a.
tc
a.
60.00
40.00
40.00 80.00 120.00
OBSERVED (ppb)
160.00
MOMENTS OF THE PROBABILITY DENSITY
OBSERVED
AVERAGE
STANDARD DEVIATION
SKEWNESS
KURTOSIS
OTHER MEASURES
MEDIAN
UPPER OUARTILE
LOWER OUARTILE
MINIMUM VALUE
MAXIMUM VALUE
46.96721
31.01291
0.82584
0.17512
40.00000
60.00000
20.00000
10.00000
140.00000
FUNCTION
PREDICTED
52.43745
25.12773
-0.13819
-0.60323
53.49000
69.10001
31.96000
0.01000
104.00000
SKILL OF PREDICTION PARAMETERS
CORRELATION COEFFICIENT OF PREDICTED
VERSUS OBSERVED 0.635
THE BOUNDS OF THE CORRELATION AT THE
CONFIDENCE LEVEL OF 0.050 ARE
LOW BOUND 0.515 HIGH BOUND 0.730
RATIO OF OVER TO UNDER PREDICTIONS 1.711
PERCENT OF OVER PREDICTIONS
GREATER THAN 200 PERCENT OF THE
OBSERVED 19.672
PERCENT OF UNDER PREDICTIONS
LESS THAN 50 PERCENT OF THE
OBSERVED 6.557
FIGURE 4-9a. Scatterplot and model performance statistics for predicted
and observed hourly ozone concentrations (ppb) on 30 August 1985 in
Dallas-Fort Worth for diagnostic run 1 (N=122).
8910%
103
-------
0.20 -
-55.00 -33.00 -11.00 11.00
RESIDUAL (OBS-FRED)
33.00 55.00
THE B1NS1ZE EQUALS 11.000
RESIDUAL ANALYSIS
AVERAGE -5.47053
STANDARD DEVIATION 24.57855
SKEHNESS 0.41158
KURTOS3S 0.30432
OTHER MEASURES
MEDIAN -8.36000
UPPER OUARTJLE 8.22000
LOWER OUARTILE -21.04000
MINIMUM VALUE -70.71001
MAXIMUM VALUE 69.17000
BIAS CONFIDENCE INTERVAL
AT THE 0.0500 LEVEL
LOWER BOUND -11.4614
UPPER BOUND 0.5203
STD RESIDUAL CONFIDENCE INTERVAL
AT THE 0.0500 LEVEL
LOWER BOUND 495.3563
UPPER BOUND 755.9985
THE MEASURES OF GROSS ERROR
THE ROOT MEAN SQUARE ERROR IS 25.08
THE AVERAGE ABSOLUTE ERROR IS 20.23
VARIOUS MEASURES OF RELATIVE VARIABILITY
OBSERVATION COEFFICIENT OF VARIATION
0.6603
RESIDUAL COEFFICIENT OF VARIATION
0.5233
RATIO OF RESIDUAL TO OBSERVED ST. DEV.
0.7925
FIGURE 4-9b. Residual analysis and nodel performance statistics for
predicted and observed hourly ozone concentrations (ppb) on 30 August 1985
in Dallas-Fort Worth for diagnostic run 1 (N=122) .
104
-------
160.00
120.00
o
a
a.
cr
a.
80.00
40.00
I I I I I I I T
] I I I I I I I I I
I I I I I I I I
40.00 80.00 120.00
OBSERVED (ppb)
160.00
MOMENTS OF THE PROBABILITY DENSITY FUNCTION
OBSERVED
AVERAGE
STANDARD DEVIATION
SKEWNESS
KURTOSIS
OTHER MEASURES
MEDIAN
UPPER OUARTILE
LOWER OUARTILE
MINIMUM VALUE
MAXIMUM VALUE
70.60869
44.49103
0.38833
-0.79089
70.00000
100.00000
30.00000
10.00000
170.00000
PREDICTED
57.81913
24.27184
0.15263
-0.49933
56.25000
73.1100
.37.64000
1.63000
111.399
SKILL OF PREDICTION PARAMETERS
CORRELATION' COEFFICIENT OF PREDICTED
VERSUS OBSERVED O.B26
THE BOUNDS OF THE CORRELATION AT THE
CONFIDENCE LEVEL OF 0.050 ARE
LOW BOUND 0.761 HIGH BOUND 0.876
RATIO OF OVER TO UNDER PREDICTIONS 0.438
PERCENT OF OVER PREDICTIONS
GREATER THAN 200 PERCENT OF THE
OBSERVED 9.565
PERCENT OF UNDER PREDICTIONS
LESS THAN 50 PERCENT OF THE
OBSERVED 3.478
FIGURE 4-10a. Scatterplot and model performance statistics for
predicted and observed hourly ozone concentrations (ppb) on
31 August 1985 in Dallas-Port Worth for diagnostic run 1 (N=115)
105
-------
0.40 -
-80.00 -48.00 -16.00
RESIDUAL 10BS-PRED)
16.00 4B.OO 60.00
THE BINSIZE EQUALS 36.000
RESIDUAL ANALYSIS
AVERAGE 12.78924
STANDARD DEVIATION 27.91764
SKEWNESS 0.32808
KURTOSIS 0.56592
OTHER MEASURES
MEDIAN 12.89000
UPPER OUARTILE 26.61000
LOWER OUARTILE -6.91000
MINIMUM VALUE -59.66000
MAXIMUM VALUE 98.00000
BIAS CONFIDENCE INTERVAL
AT THE 0.0500 LEVEL
LOWER BOUND 4.9507
UPPER BOUND 20.6277
STD RESIDUAL CONFIDENCE INTERVAL
AT THE 0.0500 LEVEL
LOWER BOUND 635.5250
UPPER BOUND 982.3846
THE MEASURES OF GROSS ERROR
THE ROOT MEAN SQUARE ERROR IS 30.60
THE AVERAGE ABSOLUTE ERROR IS 23.61
VARIOUS MEASURES OF RELATIVE VARIABILITY
OBSERVATION COEFFICIENT OF VARIATION
0.6301
RESIDUAL COEFFICIENT OF VARIATION
0.3954
RATIO OF RESIDUAL TO OBSERVED ST. DEV.
0.6275
FIGURE 4-10b. Residual analysis and model performance statistics for
predicted and observed hourly ozone concentrations (ppb) on 31 August 1985
in Dallas-Port Worth for diagnostic run 1 (N=115).
106
-------
subgrid-scale variability (due to local emission sources) is much more pronounced in
primary species. However, it is always useful to compare model predictions with
observations of these primary species in order to determine whether the model is
getting the right answer for the right reason.
As seen in Figure 4-11, the model tends to underpredict NO observations at all four
sites. At the two sites in Dallas (MOCK and NSTR) the model predicts a primary NO
peak at around 0300 and a secondary peak at 0700, whereas the observations only
support an NO peak at 0700. The NO peak at 0700 is most likely due to increased
emissions for that period. The model-predicted peak at 0300 may be due to a wind
shift. At the monitor located south of Dalii. ',!?M'",V) the observations are at the
lower threshold of the NO instrument (1 pphm) or zero. The model predictions for
NO at BNVW are also very low. In Fort Worth (ROSS) the model does an excellent
job replicating the pattern of the observed NO concentrations, however it appears to
underpredict the peak NO observations by over a factor of two.
Diagnostic run 1 displays better agreement with hourly N©2 observations than for
NO as evidenced in Figure 4-12. The diurnal variations in the hourly observed N©2
concentrations are replicated by the model quite well.
As is usually the case for a grid model like the UAM, the model has difficulty in
reproducing the scatter in the hourly CO concentrations (Figure 4-13). In general it
appears that the model underpredicts the observed CO concentrations during most
hours and sites. The model does pick up some of the trends of the observations such
as the peak around midnight; however, the scatter in the observations makes it diffi-
cult to evaluate model performance for CO concentrations.
Summary
The peak ozone observations on 30-31 August are replicated by the predicted region-
wide maximum on these days to within 11 and 15 percent and is in the general
vicinity of the peak observation. However, there is much spatial variation in the
predicted ozone concentrations, resulting in predicted daily maximum ozone concen-
trations at the location of the peak observations that are only within 43 percent on
31 August and 56 and 35 percent (two sites) on 31 August. The model also tends to
underpredict primary emitted species (NO, NO2» and CO)
Diagnostic run 1 statisfies the primary model performance goal, that the predicted
region-wide maximum ozone concentration be within 20 percent and in the general
vicinity of the peak observation. However, it does not statisfy the secondary goal
because the predicted daily maximum ozone concentration at the location of the
observed region-wide maximum is only within 35 percent rather than the goal of 30
percent. The model predicts a high amount of spatial variation in the ozone concen-
trations. Thus a slight displacement of the ozone cloud, which is very likely under
stagnant conditions, will result in poor model performance. Because the model
8910trl 5
107
-------
100
80
60
m
CL
a,
6 12
i i i i i I i i i i i I i i i
MOCK
OBSERVED D
18
24
30
36
42
46
6 12 18 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31, 1985
100
80
60
40
20
48
100
80
6
12
18
24
30
36
42
48
60
m
a,
a.
z 40
i i i i i |
NSTR
OBSERVED
i i r i i i i
nan
100
80
60
40
20
6 12 ' 18 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
46
FIGURE 4-11. Tine series of predicted and observed hourly NO
concentrations (ppb) in Dallas-Fort Worth for diagnostic run 1.
SYSTEMS APPLICATIONS, INC.
89104
108
-------
(
i on
1 \J w
80
60
s
CL,
e.
0 40
20
1 6 12 18 24 30 36 42 4
iiiijiiii | 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 -
ILLl
OBSERVED D
~ —
'_
-
- -
1 1 1 1 IK t 1 1 ! 1 |^"1 "t"^l 1 ' < 1 t 1 1 1 i 1 1 1 1 1 I -J^l 1 1 i ! 1 1 1 r*«-L | 1 I 1 1 1 1
3 6 12 18 24 30 36 42 4
< f\r\
1UU
80
60
40
20
f\
6°
AUGUST 30, 1985 TIME (HOURS) AUGUST 31, 1985
100
80
60
12
18
24
30
36
42
48
CL*
55 40
20
i i i i i | ii[ii[[iiii|i i T iijii i i i | i i r i i | i i i i i [ i i i i
NRTH
OBSERVED E
6 12 18
AUGUST 30, 1985
80
60
40
20
24 30 36 42
TIME (HOURS) AUGUST 31, 1985
FIGURE 4-11 continued.
SYSTEMS APPLICATIONS, INC.
89104
-------
100
80
12
16
24
30
36
42
46
60
m
a,
a,
i 40
20
I I i I I | I I I I I I I I I I I | I I I I I I I I I I I | I I I I I I I I I I I I I I I I T
BNVW
OBSERVED D
ncnncicn
6 12 16
AUGUST 30. 1985
24
TIME (HOURS)
a a
100
80
60
40
20
30 36 42
AUGUST 31. 1985
46
100
60
6
12
16
24
30
36
42
46
60
05
CL
CL
2 40
20
DNTN
OBSERVED Q
i i i i I i i i
100
80
60
40
20
6 12 16 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31. 1985
46
FIGURE 4-11 continued.
SYSTEMS APPLICATIONS, INC.
8910U
110
-------
100
12
18
24
30
36
I I I I
I I 1 I i I ] (
ROSS
OBSERVED
D
60
60
-ED
m
a,
D D
CD
40
20
: 46
I I I I I l100
80
60
40
20
[DDDDDDEIQII1CDED
CDCDCD-
6 12 16 24
AUGUST 30, 1985 TIME (HOURS)
30 36 42
AUGUST 31, 1985
46°
100
12
16
24
30
36
42
46
l T i T 1 I
PECN
OBSERVED
r i i i T r i i ^ i i i i i \ i \ i T i | i i i i r i i i i i i
m
o.
a,
60
60
40
20
i l i J_
l I _ixi i rvJ.
6 12 16 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31, 1985
100
80
60
40
20
46
FIGURE 4-11 continued.
SYSTEMS APPLICATIONS. INC.
-------
100
80
60
m
a.
a.
6 12
T~T I I I | i i i i i
KLLR
OBSERVED D
24
30
36
48
i i i i r | i "i ~i T n—i—r~i—r
\ i I i i i | i i i i r
z 40
20
I I i I I I I I
6 12 18 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31. 1985
100
80
60
40
20
46°
FIOTRE 4-11 concluded.
SYSTEMS APPLICATIONS, INC.
89104
IT)
-------
300
6
12
16 24
30 36
42 46
200
m
o
z
100
I I I I I I I I I I I I I I
- MOCK
- OBSERVED D
I I I I I 4-"-!-1 I I I I I I I I I
6 12 16 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
300
200
100
46
300
12
16
24
30 36
42 48
200
s
O
z
100
IIIII|IIIII|ITIII[|ITII|1I I I I | I I I I T I I I I T I
- NSTR
- OBSERVED CD
nm,
nun
6 12 18 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
300
200
100
48
FIGURE 4-12. Tine series of predicted and observed hourly N02
concentrations (ppb) in Dallas-Fort Worth for diagnostic run 1.
SYSTEMS APPLICATIONS. INC.
891 0*f
-------
300
6 12 16
! I I I I | i I I I I | I I I I I |
ILLI
- OBSERVED O
24
30 36
42 48
200
m
c_
O
100
T i i i r
i i i
300
200
100
6 12 16 24 30 36 42 48
AUGUST 30. 1985 TIME (HOURS) AUGUST 31, 1985
300
12 18
24
30 36
42 48
i i i i i r ' ' ' ' ' I '' ' ' ' I '
- NRTH
- OBSERVED D
200
m
c-
1
100
I I I I I I I I I I I I I I I I I I I I I I I I
300
200
100
6 12 18 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31. 1985
48
FIGURE 4-12 continued.
SYSTEMS APPLICATIONS. INC.
8910^
114
-------
300
0
6
12
16 24
30 36
42
48
\- BNVW
- OBSERVED Q
200
£
O
100
COD
6 12 16
AUGUST 30. 1985
24
TIME (HOURS)
30 36 42
AUGUST 31, 1985
300
200
100
48
300
6
12
16 24
30 36
42
200
m
e-
o
100
48.
- DNTN
- OBSERVED
I T I 1 I I I I JT I I I I | I T I I I I I I I
i I i I I i i
Q
300
200
100
6 12 16 24 30 36 42
AUGUST 30, 1985 TME (HOURS) AUGUST 31, 1985
48
FIGURE 4-12 continued.
SYSTEMS APPLICATIONS, INC.
-------
300
12
16 24
30 36
42 46
200
ca
c.
a.
i I I I T | 1 1 I I r
ROSS
OBSERVED CD
100-
300
200
100
0
0 6 12 16 24 30 36 42 46
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
300
12
16 24
30 36
42
i i i i i i i i i i i | i r T
- PECN
- OBSERVED CD
200
m
a
a
o
100
48300
200
100
6 12 18 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
FIGURE 4-12 continued.
SYSTEMS APPLICATIONS. INC.
8910<+
116
-------
300
12
18
24
30
36
42
4-6
200
ea
o
100
I I I I I I i 1 1
- KLLR
- OBSERVED 0
6 12 16 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
300
200
100
48
FIOJRE 4-12 concluded.
SYSTEMS APPLICATIONS. INC.
8910«*
-------
CD
2000
12
18
24
30 36
46
_ 1 I I I I | I I l-M
* MOCK
- DESERVED E
1500
CD
1000
500
CD
CD
CD
n>
CD :
CD
CD
CD CD
CD
CD
CD
CD
2000
1500
1000
500
6 12 IB 24 30 ' 36 42 46
AUGUST 30. 1985 TIME (HOURS) AUGUST 31, 1985
2000
6
12
18
24
30 36
42 46
1500
CD
1000
500
i i i i | i i i i i
NSTR
OBSERVED CD
2000
1500
1000
500
6 12 18 24 30 36 42 46
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
FIGURE 4-13. Time series of predicted and observed hourly CO
concentrations (ppb) in Dallas-Fort Worth for diagnostic run 1.
SYSTEMS APPLICATIONS. INC.
89104
118
-------
-S B-
2000
- ILLI
- OBSERVED
1500
Q
1000
c
u
500
12 16 24 30 36
I i | i i i i i | i i I I I | I I I I I | I I I I I | I I
46
a nn
2000
1500
1000
500
6 12 16 24 30 36 ' 42
AUGUST 30, 1985 Q TIME (HOURS) AUGUST 31, 1985
m
46
2000
2000
- 1500
_ i i i i i | i i i i i | i i i i i | i i i i i | i i i i i | i i i i i I i i i i i I i i i i i _
; OBSERVED CD
- 1000
- 500
6 12 16 24 30 36 42
AUGUST 30, 1985 TME (HOURS) AUGUST 31. 1985
46
FIGURE 4-13 continued.
SYSTEMS APPLICATIONS, INC.
89104
-------
OD
o
O
PHAA
1500
1000
500
n
D 6 12 18 24 30 36 42
_ 1 1 1 1 1 | 1 1 1 1 1 | 1 1 ! 1 1 | 1 1 1 1 1 | 1 1 I 1 1 | 1 1 1 1 1 | 1 1 1 1 ! | 1 1
I BNYW
I OBSERVED 0
—
—
-
- i i i i i i i i i i i i -i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i
4
—
-
—
—
—
-
6
1500
1000
500
n
6 12 18 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31, 1985
46
2000
12
18
24
30
36
42
46
1500
CO
1000
8
500
^iI1II|IIIIIIiIT
- DNTN-
I OBSERVED E
2000
1500
1000
500
6 12 18 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31. 1985
.FIGURE 4-13 continued.
48
SYSTEMS APPLICATIONS, INC.
-------
200C
1500
Q
1000
500
12
I ROSS
- OBSERVED
16
CD
LJLLJ
24
30
36
42
48
P]
CD SO ID
ED
O " I I I I I I ' I I I I I I I I I I I I I ' I I I I I I I I I I I I I I I I I I I - I I I I I I I I Q
2000
1500
1000
500
6 12 18 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31. 1985
48
2000
12
18
24
30
36
42
48
1500
n
fc 1000
8
500
= PECN
- OBSERVED El
-I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I ~
2000
1500
1000
500
6 12 18 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
FIGURE 4-13 continued.
SYSTEMS APPLICATIONS, INC.
121
-------
2000
12
16
24
30
36
42
46
~I I I I I I I I I I I I I I I I | I I I I 1 I I I T
= KLLR
- OBSERVED D
1500
CD
1000
o
o
500
~ i I i I l I l I l I l I l I i l i I l I l l l I l l l I i I l l l I l I i l i I l I l i l I l ~
2000
1500
1000
500
6 12 16 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31. 1985
46
FIGURE 4-13 concluded,
SYSTEMS APPLICATIONS, INC.
89101*
122
-------
underpredicts the peak ozone observations as well as observations of primary pollu-
tants (NO, NO2> and CO), it appears that the ozone concentrations and precursors
may be excessively diluted.
Diagnostic Run 2
As indicated above, it was suspected that diagnostic run 1 underpredicted the peak
observations due to excessive dilution. Several modeling inputs can be adjusted to
lower the amount of dilution of the ozone and ozone precursors: smaller horizontal
grid spacing, lower mixing height, slower wind speeds, elimination of any wind shear,
and increases in emission levels. Because of cost and logistical limitations it is not
possible to reduce the horizontal grid spacing. In the past there has been some justi-
fication for lowering wind speeds when instantaneous FAA wind observations are
used (see Appendix D). However, for this application only hourly average surface
wind observations were used. The elimination of wind shear in the mixed layer may
produce higher ozone concentrations, but would involve alterations to the wind input
and wind field preparation procedure that may not be desirable. Although it is sus-
pected that the amount of actual VOC emissions is underestimated in the emission
inventory, the inventory is the current best estimate and any modifications would be
arbitrary. However, there is quite a bit of justification for lowering the maximum
daily mixing height.
Calculated maximum mixing heights at both the Longview and Stephenville upper-air
sites on August 31 exceeded 3,000 m AGL. Both of these soundings indicated super-
adiabatic conditions near the surface. The use of a mixing height algorithm that uses
an hourly surface temperature and a dry adiabatic during super-adiabatic conditions
(e.g., Kelley, 1981) may result in an overestimate of the actual mixing height. Ozone
episodes are usually associated with hot, dry, stagnant conditions with limited mix-
ing. Due to the presence of a high-pressure system over the area during the episode,
it is expected that subsidence may exist over the region, resulting in limited vertical
mixing. Thus, for diagnostic run 2 the maximum mixing height was limited to the
summer afternoon climatological average of 2,000 m AGL as reported by Holzworth
(1976). All other UAM modeling inputs were as in diagnostic run 1.
30 August 1985
On 30 August the predicted region-wide maximum ozone concentration for diagnostic
run 2 is 13.2 pphm (Figure 4-14 and Appendix H) which is within 6 percent of the
peak observation (14 pphm) for this day. As shown in the comparison of isopleths of
predicted daily maximum ozone concentrations with observations (Figure 4-14), the
predicted region-wide maximum occurred approximately 40 km east of the location
of the peak observation. The predicted maximum concentration at the location of
the peak observation (ROSS) is within only 48 percent (6.2 pphm) (Figure 4-15). As in
the previous diagnostic simulation, there is considerable spatial variability in the
8910»trl S
123
-------
1SV3
r' i I i" i i i r i i 'i i
in
•H
o S
•U o
SI
° T3
§i
O |2
I "^
§ P
•1^
g a
u o
•H U-l
£ ^
^^
'o -u
0)
s
1S3M
124
-------
200
12
13
24
30 36
42
46
150
ca
100
CD
o
50
I I I I I | I I I I I | I I 1 1 I I I T
MOCK
OBSERVED CD
i i
6 12 16
AUGUST 30, 1985
24 30 36 42
TIME (HOURS) AUGUST 31, 1985
200
150
100
50
46
200
150
m
fc 100
50
6
12
16 24
30 36
42 46,
I I Tl 1 | I I I I I ] I I I I I \ I I I I I
NSTR
OBSERVED TD
200
150
100
50
0
6 12 18 24 30 36 42 46
AUGUST 30. 1985 TIME (HOURS) AUGUST 31, 1985
FIGURE 4-15. Tine series of predicted and observed hourly ozone
concentrations (ppb) for the Dallas-Fort Worth diagnostic run 2.
SYSTEMS APPLICATIONS, INC.
-------
200
200
- 350
l I I I l l I l I i I I l ! l i
- 100
6 12 18 24 30 36 42 48
AUGUST 30, 1985 TIME (HOURS) AUGUST 31. 1985
200
6
12 18
24
30 36 42 48
150
co
fc 100
50
I 1 I i | I I I I 1 | 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 T
NRTH
OBSERVED CD
J I I I I
i I i i i
200
150
100
50
6 12 18
AUGUST 30, 1985
24
TIME (HOURS)
30 36 42
AUGUST 31. 1985
48
FIGURE 4-15. Continued.
SYSTEMS APPLICATIONS. INC.
-------
200
150
ca
100
50
i i i i i I
BNVW
OBSERVED
12
16
24
30
36
48200
150
100
50
6 12 16 24 30 36 42 48
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
o
200
200
- 150
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 | !
- 100
6 12 16 24 30 36 42
AUGUST 30. 1985 TME (HOURS) AUGUST 31. 1985
46
FIGURE 4-15. Continued.
SYSTEMS APPLICATIONS. INC.
-------
200
12
18
24
30
36 42
150
100
en
O
50
I 1 i I |
ROSS
OBSERVED
48200
CD
CD
CD CD
CD
CD
lit
6 12 16
AUGUST 30. 1985
24
TIME (HOURS)
30 36 42
AUGUST 31, 1985
150
100
50
48
200
6
12 18
24
30 36
42 48
150
OB
£ 100
50
PECN
OBSERVED D
i i i i
200
150
100
50
0 6 12 16 24 30 36 42 46
AUGUST 30, 1985 TIME (HOURS) AUGUST 31. 1985
Continued 4-15. Continued.
SYSTEMS APPLICATIONS. INC.
128
-------
200
200
- 150
- 100
6 12 16 24
AUGUST 30, 1985 TIME (HOURS)
30 36 42
AUGUST 31. 1985
FIGURE 4-15. Concluded.
SYSTEMS APPLICATIONS. INC.
-------
predicted ozone concentration fields in diagnostic run 2, as shown in the isopleth
plots in Appendix H. Around the time of the peak observation at ROSS (1500 - 1600)
there is a cloud of predicted ozone concentrations to the north of the ozone monitor
in excess of 10 pphm (i.e., within 30 percent). The spatial displacement of the pre-
dicted concentration peaks is discussed below. As in diagnostic run 1, the daily
maximum ozone concentrations in Dallas predicted in run 2 (7 to 10 pphm) are higher
than the observed (5 to 7 pphm).
A scatterplot and statistical model performance measures that indicate how well
diagnostic run 2 simulates hourly ozone concentrations on 30 August is given in
Figure 4-16. Over all hours and sites the model has a bias toward overprediction of
about 6 percent (0.3 pphm) and a gross error of approximately 43 percent (2.0
pphm). Model performance statistics are comparable between diagnostic runs 1
(Figure 4-9) and 2 (Figure 4-16) on 30 August. For 30 August there does not seem to
be much improvement in model performance in diagnostic run 2, although the obser-
ved peak is reproduced better in diagnostic run 2 (within 6 percent) than in diagnostic
run 1 (within 11 percent).
31 August 1983
The predicted region-wide maximum ozone concentration in diagnostic run 2 on 31
August was 16.4 pphm (Figure 4-17 and Appendix H), within 4 percent of the peak
observation (17 pphm). The predicted region-wide maximum is located between the
two ozone monitors that recorded the highest observed ozone concentration of the
episode, approximately 8 km to the southwest of DNTN and 13 km to the northwest
of MOCK (Figure 4-17). At the location of the peak observation, the predicted
maximum ozone concentration at MOCK is 7.2 pphm (within 58 percent) and at
DNTN 13.6 pphm (within 20 percent) (see Figure 4-15). As seen in Figure 4-17 and
Appendix H, a very sharp predicted ozone gradient runs through the middle of Dallas;
in a distance of only 14 km (3-4 grid cells) the ozone concentrations increase by more
than a factor of two from 7 to 16 pphm. Diagnostic run 2 underpredicts the observed
peak ozone concentrations at ozone monitors located near the center of Dallas
(MOCK and NSTR), but does a fairly good job of replicating the observed peaks and
diurnal variation at monitors located south (ILLI and BNVW) and north (DNTN) of the
center of the city (Figure 4-15).
The model reproduces the observed peaks in Fort Worth (13 pphm) to within 33 per-
cent at ROSS (8.8 pphm) and 10 percent at KLLR (11.7 pphm). Again there is con-
siderable spatial variation in the predictions, and there are predicted ozone concen-
trations similar to the observed peaks within a few grid cells of the location of the
observed peaks.
Figure 4-18 displays a scatterplot and model performance statistics for diagnostic
run 2 predicted and observed hourly ozone concentrations on 31 August. Over all
hours and sites the model underpredicts the observed hourly ozone concentrations on
8910
-------
160.00 -
120.
£2
Q.
Q.
o
UJ
LJ
OL
a.
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
60.00 -
40.00 -
40.00 80.00 120.00
OBSERVED (ppb)
160.00
MOMENTS OF THE PROBABILITY DENSITY
OBSERVED
AVERAGE
STANDARD DEVIATION
SKEWNESS
KURTOSJS
OTHER MEASURES
MEDIAN
UPPER OUARTILE
LOWER OUARTILE
MINIMUM VALUE
MAXIMUM VALUE
46.96721
31.01291
0.82584.
0.17512
40.00000
60.00000
20.00000
10.00000
140.00000
FUNCTION
PREDICTED
49.96617
26.66011
-0.06659
-0.89927
49.79000
70.14000
28.68000
0.01000
102.20000
SKILL OF PREDICTION PARAMETERS
CORRELATION COEFFICIENT OF PREDICTED
VERSUS OBSERVED 0.617
THE BOUNDS OF THE CORRELATION AT THE
CONFIDENCE LEVEL OF 0.050 ARE
LOW BOUND 0.493 HIGH BOUND 0.716
RATIO OF OVER TO UNDER PREDICTIONS 1.490
PERCENT OF OVER PREDICTIONS
GREATER THAN 200 PERCENT OF THE
OBSERVED 17.213
PERCENT OF UNDER PREDICTIONS
LESS THAN 50 PERCENT OF THE
OBSERVED 10.656
FIGURE 4-16a. Scatterplot and nodel perfonrance statistics for predicted
and observed hourly ozone concentrations (ppb) on 30 August 1985 in
Dallas-Fort Worth for diagnostic run 2 (N=122)„
131
-------
0.20 -
-55.00 -33.00 -U.OO
RESIDUAL IOBS-PRED)
11.00 33.00 55.00
THE BINS1ZE EQUALS 11.000
RESIDUAL ANALYSIS
AVERAGE -2.99921
STANDARD DEVIATION 25.53730
SKEWNESS 0.29124
KURTOS1S 0.06662
OTHER MEASURES
MEDIAN -4.48000
UPPER OUART1LE 10.58000
LOWER OUARTILE -20.48000
MINIMUM VALUE -68.64000
MAXIMUM VALUE 71.53000
BIAS CONFIDENCE INTERVAL
AT THE 0.0500 LEVEL
LOWER BOUND -9.1375
UPPER BOUND 3.:390
STD RESIDUAL CONFIDENCE INTERVAL
AT THE 0.0500 LEVEL
LOWER BOUND 534.7552
UPPER BOUND 616.1261
THE MEASURES OF GROSS ERROR
THE ROOT MEAN SQUARE ERROR IS 25.61
THE AVERAGE ABSOLUTE ERROR IS 20.36
VARIOUS MEASURES OF RELATIVE VARIABILITY
OBSERVATION COEFFICIENT OF VARIATION
0.6603
RESIDUAL COEFFICIENT OF VARIATION
0.5437
RATIO OF RESIDUAL TO OBSERVED ST. DEV.
0.8234
FIGURE 4-16b. Residual analysis and nodel performance statistics for
predicted and observed hourly ozone concentrations (ppb) on
30 August 1985 in Dallas-Fort Worth for diagnostic run 2 (N=122).
132
-------
*— (N
JS3M
133
-------
160.00
120.00
£>
a
o
UJ
Q.
60.00
40.00
i \ I I I I I I T
I I I I I I
I I I I I I
40.00 80.00 120.00
OBSERVED (ppb)
160.00
MOMENTS OF THE PROBABILITY DENSITY
OBSERVED
AVERAGE
STANDARD DEVIATION
SKEWNESS
KURTOSIS
OTHER MEASURES
MEDIAN
UPPER QUARTILE
LOWER OUART3LE
MINIMUM VALUE
MAXIMUM VALUE
70.60869
44.49103
0.3BB33
-0.79069
70.00000
100.00000
30.00000
10.00000
170.00000
FUNCTION
PREDICTED
57.71497
29.65752
0.51526
-0.04070
53.30000
72.10001
34.67000
0.01000
135.69990
SKILL OF PREDICTION PARAMETERS
CORRELATION COEFFICIENT OF PREDICTED
VERSUS OBSERVED 0.763
THE BOUNDS OF THE CORRELATION AT THE
CONFIDENCE LEVEL OF 0.050 ARE
LOW BOUND 0.674 HIGH BOUND 0.630
RATIO OF OVER TO UNDER PREDICTIONS 0.420
PERCENT OF OVER PREDICTIONS
GREATER THAN 200 PERCENT OF THE
OBSERVED 9.565
PERCENT OF UNDER PREDICTIONS
LESS THAN 50 PERCENT OF THE
OBSERVED 9.565
FIGURE 4-18a. Scatterplot and nodel performance statistics for
predicted and observed hourly ozone concentrations (ppb) on
31 August 1985 in Dallas-Fort Worth for diagnostic run 2 (N=115)
134
-------
0.40 -
-90.00 -54.00 -18.00
RESIDUAL IOBS-PRED)
THE BINSIZE EQUALS 18.000
18.00 54.00 90.00
RESIDUAL ANALYSIS
AVERAGE 12.89339
STANDARD DEVIATION 29.04443
SKEWNESS 0.74048
KURTOSIS 1.61677
OTHER MEASURES
MEDIAN 10.41000
UPPER OUARTILE 23.65000
LOWER OUARTILE -3.30000
MINIMUM VALUE -57.83000
MAXIMUM VALUE 112.53990
BIAS CONFIDENCE INTERVAL
AT THE 0.0500 LEVEL
LOWER BOUND 4.6064
UPPER BOUND 21.1804
STD RESIDUAL CONFIDENCE INTERVAL
AT THE 0.0500 LEVEL
LOWER BOUND 687.8611
UPPER BOUND 1063.2840
THE MEASURES OF GROSS ERROR
THE ROOT MEAN SQUARE ERROR IS 31.66
THE AVERAGE ABSOLUTE ERROR IS 22.85
VARIOUS MEASURES OF RELATIVE VARIABILITY
OBSERVATION COEFFICIENT OF VARIATION
0.6301
RESIDUAL COEFFICIENT OF VARIATION
0.4113
RATIO OF RESIDUAL TO OBSERVED ST. DEV.
0.6528
FIGURE 4-18b. Residual analysis and model performance statistics for
predicted and observed hourly ozone concentrations (ppb) on
31 August 1985 in Dallas-Fort Worth for diagnostic run 2 (N=115) .
135
-------
average by about 18 percent (1.3 pphm). The gross error is about 32 percent, slightly
lower than was seen in diagnostic run 1 (33 percent). In general, model performance
statistics for hourly ozone concentrations for diagnostic runs 1 and 2 are
comparable. However, diagnostic run 2 does a much better job of replicating the
peak observed ozone concentration on 31 August (within 4 percent) than does
diagnostic run 1 (within 15 percent).
NO, NO2, and CO
Time j-vies of prac'ictec and observed hourly NO, NO2> and CO concentrations for
diagnostic run 2 are shown in Figures 4-19, 4-20, and 4-21. In general, the remarks
made in the analysis of diagnostic run 1 hold for run 2 also, except diagnostic run 2
does a better job reproducing the peak observations of these species. For example
the peak NO observation at ROSS at midnight on 30 August (8 pphm) is reproduced to
within 42 percent (4.7 pphm) in diagnostic run 2 (Figure 4-19) as compared to 58 per-
cent (3.4 pphm) in diagnostic run 1. It is also interesting to note that in both
diagnostic simulations the model does an excellent job reproducing the observed
hourly NO2 concentrations at ROSS (Figure 4-20)
Nearest-Neighbor Analysis
It appears that the model predictions of peak ozone concentrations are similar to the
observed peaks but slightly displaced from the location of the observed peak. This
displacement should not affect the usefulness of the UAM in an evaluation of alter-
native emission control strategies, but does affect model performance. In order to
calculate model performance measures that take into account small displacements at
the predicted concentrations, it has been suggested that model predictions in adja-
cent cells of the observation site be compared with the observation in evaluating
model performance (Seinfeld, 1988a; Burton, 1988). This is accomplished by
examining predicted concentrations in grid cells adjacent to the cell containing the
ozone monitoring (one-cell search) and matching up the prediction that is most like
the observation for that hour. Similarly, a two-cell search involves examing predic-
ted concentrations within two cells of the ozone monitor. The results of the one-cell
and two-cell searches for diagnostic run 2 are presented in, respectively, Figures
4-22 through 4-24 and Figures 4-25 through 4-27.
It appears that the model predictions of ozone, NO, and NO2 are comparable to the
observations within only one or two grid cells almost all of the time. The exceptions
are (1) the peak ozone observations at ROSS on 30 August (Figure 4-22 and 4-25); (2)
the peak ozone observations at MOCK on 31 August (Figure 4-22 and 4-25); and (3)
the high NO observations around midnight 30 August at MOCK and ROSS (Figures
4-23 and 4-26). Other then these three situations, model performance based on
nearest-neighbor analysis is quite good.
8910>frl 5
136
-------
100
6 12 16 24
AUGUST 30. 1985 TIME (HOURS)
30 36 42
AUGUST 31. 1985
100
- 60
- 60
I I I I | I I I
MOCK
OBSERVED m
100
12
16
24
30
36
42
46
i i i i I i i i i i I i i i i i I i i r i i | i i i i i | T I i i i | i i i i i | i i i i i
NSTR
OBSERVED Q
60
60
CD
0.
O,
I
I I I |"T"~'KJ__ I I I I
100
80
60
40
20
6 12 16 24 30 36 42 46
AUGUST 30. 1985 TME (HOURS) AUGUST 31. 1985
FIGURE 4-19. Time series of predicted and observed hourly NO
concentrations (ppb) in Dallas-Fort Worth for diagnostic run 2.
SYSTEMS APPLICATIONS. INC.
-------
100
80
12
16
24
30
36
42
48
60
cn
cu
a.
2 40
20
T I 1 i 1 |
ILLI
OBSERVED
I
I
6 12 18
AUGUST 30, 1985
24 30 36 42
TIME (HOURS) AUGUST 31. 1985
100
80
60
40
20
46
100
80
6
12
18
24
30
36
42
46
60
m
cu
Cu
2 40
20
\ I i 1 i |
NRTH
OBSERVED
i i i i i
.L i i i i i
100
80
60
40
20
6 12 18
AUGUST 30, 1985
24
TIME (HOURS)
30 36 42
AUGUST 31. 1985
48
FIGURE 4-19. Continued.
SYSTEMS APPLICATIONS, INC.
138
-------
100
80
12
18
24
30
36
46
1 i I i | 1 I 1 I I I I I I I I I I 1 I I T
BNVW
OBSERVED D
i i i i ] i i i i r j i i i i i i i i i i r
60
(3
C-,
c-
2 40
20
CD CD CD COD
a m
I I I I
100
80
60
40
20
6 12 18 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31, 1985
46
100
6
12
18
24
30
36
42
46
n i i i i pi r^ i
DNTN
OBSERVED O
i i i i r | \ \ \ r i | i i i i r
80
60
CD
Cu
a,
z 40
20
1 ,
I i i i I
100
80
60
40
20
6 12 IB 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
48
FIGURE 4-19. Continued.
SYSTEMS APPLICATIONS. INC.
-------
100
80 -
12
16
100
Q
- 20
6 12 18 24 30 36 42 48"
AUGUST 30, 1985 TIME (HOURS) AUGUST 31, 1985
100
80
6
12
18
24
30
36
42
48
60
O
C.
c.
2 40
20
I I i I I I I I I I | I i I I 1 I i I I I I | I T i I i I I I I I I | I 1 I I I | I I i I I
PECN
OBSERVED O
100
80
60
40
6 12 18
AUGUST 30. 1985
24 30 36 42
TIME (HOURS) AUGUST 31. 1985
FIGURE 4-19. Continued.
SYSTEMS APPLICATIONS. INC.
140
-------
100
80
12
16
24
30
36
42
48
i i i
KLLR
OBSERVED E
60
a
c-
a,
Z 40
20
100
80
60
40
20
6 12 16 24
AUGUST 30, 1985 TIME (HOURS)
30 36 42
AUGUST 31, 1985
48
FIGURE 4-19. Concluded.
SYSTEMS APPLICATIONS. INC.
-------
300
6
12 IB
24
30 36
42
II Tll|lllll|Tllll|IIIII|ll I I I | I I I I I | I I I
- MOCK
- OBSERVED O
200
o
o
100
I I I I I
48300
6 12 18 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31. 1985
200
100
48
300
6
12
18 24
30 36
42 48
T ] r i T ] i i r i i | i i i i i | i i i T^ i
- NSTR
- OBSERVED E
200
Q
C.
O
100
6 12 18 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31, 1985
300
200
100
48°
FIGURE 4-20. Tine series of predicted and observed hourly NC>2
concentrations (ppb) in Dallas-Fort Worth for diagnostic run 2.
SYSTEMS APPLICATIONS. INC.
142
-------
300
12
18
24
30
36
42
46
I I I I I I I I I
ILLI
- OBSERVSD CD
200
m
CL
a.
(M
C
z
100
300
200
100
6 12 IB 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31, 1985
46
300
12
ie
24
30
36
42
48
200
OJ
a.
100
i i r T i | i i i i i | r i i i i | i \ \ i r | i r \ i r j i i i i r | \ i i i^ i | T i i i i
- NRTH
- OBSERVED ED
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 t I I I I ! I I I I I I I
300
200
100
AUGUST 30, 1985 TIME (HOURS) AUGUST 31. 1985
FIGURE 4-20. Continued.
SYSTEMS APPLICATIONS, INC.
-------
300
6
12 16
24
30 36
42 46
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 i i i [ I I I I I I i i i i i
- KLLR
- OBSERVED Q -
200
m
a.
o
z
100
300
200
100
6 12 16 24 30 36 42 46
AUGUST 30, 1985 TIME (HOURS) AUGUST 31. 1985
FIGURE A-2 0. Cont inued.
SYSTEMS APPLICATIONS, INC.
144
-------
300
12
16
24
30 36
42 46
200
CO
A.
o
100
riiii|iiiii|iiiiT| i r i i i ] i i i i I i i i I I i I I I I i I [ I I i r
- BNVW
- OBSERVED D
300
6 12 16 24
AUGUST 30. 1985 TIME (HOURS)
30 36 42
AUGUST 31, 1985
200
100
46
300
6
12
16
24
30 36
42 46
300
200
100
DNTN
OBSERVED O
i i i i i ]i i i i i | i i i i i | i i i ii \ i i i i r
200
100
i i i
6 12 16 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
48
FIGUEE 4-20. Continued.
SYSTEMS APPLICATIONS. INC.
-------
300
12 18
24
30 36
42
46
200
CO
C\)
O
100
I I I I I I I 1 II|IIITT|III I I \ 1 I 1 T
ROSS
- OBSERVED D
300
200
100
"0 6 12 18 24 30 36 42 46
AUGUST 30, 1985 TIME (HOURS) AUGUST 31, 1985
300
12
16 24
30
36 42
46
T i i r i i i i r i i i : i i i i | i i i i i j i i i i i | i i i i i | i i i i i | i i i i i
- PECN
- OBSERVED D
200
CD
CL.
O
100
I I I I I I I I I I I I I I I I I I I I I I I I I I » I I I I I I
300
200
100
6 12 16 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31. 1985
46
FIGURE 4-20. Continued.
SYSTEMS APPLICATIONS. INC.
AC
-------
CD
12
Q
Cu
O,
16
24
30
36
42
ouuu
2000
L
1000
" n
_ i i i i i i i LM i i
- MOCK
- OBSERVED E
-
-m
-_ mm
^~**~—~*.
1 | 1 1 1 1 1 | 1 1 1 1 1 j 1 1 1 1 ! | 1 1 1 1 1 | 1 1 1 1 1 | 1 1 1 1 1
mm • \
m :
-
m ° m^
mm
° m m m :
mm
iliiiiiliiiiiliiiiiliiiiiliiU-'i'-i-Miiiii'
3000
2000
1000
6 12 16 24
AUGUST 30, 1985 TIME (HOURS)
30 36 42
AUGUST 31. 1985
46
3000
0
6
12
IB
24
30
36
42
46
T^ii|ii r iiIiir i I|rITri]iiiir] iiiir [ i i ir \p i
NSTR
- OBSERVED E
2000 -
n
O,
CM
8
1000
3000
2000
1000
6 12 16 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
46
FIGURE 4-21. Tine series of predicted and observed hourly CO
concentrations (ppb) in Dallas-Fort Worth for diagnostic run 2.
SYSTEMS APPLICATIONS. INC.
-------
3000
12
16
24
30
36
2000
o
c-
c-
1000
I I I I 1 I
ILLI
OBSERVED H
CD
CD
~ I I I I I I I I I I I I I I I rt-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
3000
2000
1000
. 1985 TIME (HOURS) AUGUST 31. 1985
3000
12
16
24
30
36
42
46
2000
CD
C-
C-
o
o
1000
1 1 1 1 1 l-p 1 l-j-l i 1 1 j 1 1 1
- NRTH CD
- OBSERVED 0
-
f n
m*
-
=n m
~ 1 Mh-' 1 l 1 l 1 l 1 1 1 1 1 l
l l | i i i i l | i l i l i | i l l l i | l l l l i | i l l l i _
CD CD
CD -
CD CD CD
CD
CD -
m :
i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i~
6 12 16
AUGUST 30, 1985
24
TIME (HOURS)
30 36 42
AUGUST 31, 1985
3000
2000
1000
46
FIGURE 4-21. Continued.
SYSTEMS APPLICATIONS, INC.
148
-------
3000
12
16
24
30
36
48
i i i i i |
- BNVW
- OBSERVED
2000
O
a,
©
o
1000
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 1 1 I I I I 1 I
3000
2000
1000
0 6 12 16 24 30 36 42 48"
AUGUST 30, 1985 TWE (HOURS) AUGUST 31. 1985
3000
12
2000
CQ
a.
a,
8
1000
Q
I I I I I ] I I I I 1
DNTN
- OBSERVED m
16
24
30
36
42
48.
I I I I I t 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 I I I I I
3000
2000
1000
0 6 12 16 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
FIGURE 4-21. Continued.
SYSTEMS APPLICATIONS. INC.
-------
3000
20001}-
a
cu
o
o
1000
6 12
I i i i g l I l l i j i l i l l
- ROSS
- OBSERVED D
13
30
CD
m
CD
CD
m
CD
CD
36
42
CD
48
a
3000
2000
1000
6 12 18 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
46
3000
12
16
24
30
36
42
i i i i i | i i i i i | i i r in | t i i i \ i r i i i | i i r
PECN
- OBSERVED Q
2000
a
cu
Cu
8
1000
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 1 I I 1 I I I I I I I I I I I I I I I
6 12 16 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
48.
I I I
3000
2000
1000
48
FIGURE 4-21. Continued.
SYSTEMS APPLICATIONS, INC.
-------
300
12
16
24
30
36
42
48
200
a
OH
t,
100
I I I I I I I I I I I I I I I I I I I I I I l] I 1 I i i | i i i t I I I 1 I 1 I I I I I T
- MOCK
- OBSERVED CD
I I I I I I I I I I I I I I I ! I I I I I I I i I I I I I I I I I I I ! I I 1 I I ! I I I
300
200
100
6 12 16 24 30 36 42 46
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
o
300
12
16
24
30
36
42
46
200
m
CM
a.
100
i i i i I i i i i i I i i T i i I \ r i i i |i i i
NSTR
- OBSERVED CD
300
200
100
6 12 16 24 30 36 42
AUGUST 30. 1985 TIME (HOURS) AUGUST 31. 1985
46
FIGUBE 4-21. Continued.
SYSTEMS APPLICATIONS. INC.
-------
3000
12
18
24
30
36
42
43
i I i I I j 1 I I I r
KLLR
OBSERVED O
2000
en
eu
a,
1000
3000
2000
1000
6 12 18 24 30 36 42
AUGUST 30, 1985 TIME (HOURS) AUGUST 31, 1985
48
FIGURE 4-21. Concluded.
SYSTEMS APPLICATIONS, INC.
152
-------
200
150
m
fc 100
o
o
50
0
6 12
i i i i i | i I I I I | i I
' MOCK
OBSERVED D
18 24
30 36
42 48
I I I I 1 1 I I T
m , o
CD
CD
200
150
100
50
0 6 12 18 24 30 36 42 48
August 30, 1985 Time (hours) August 31, 1985
0
200
12
18 24
30 36
42 48
150
m
100
n
O
50
T 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 r i i i i j i i i i i \ i riii
NSTR
OBSERVED O
I i i i
I I I I I I I I I I I I I I I I I I I
200
150
100
50
6 12 18 24 30 36 42
August 30, 1985 Time (hours) August 31. 1985
48
FIGURE 4-22. Tims series of predicted and observed hourly ozone
concentrations for the Dallas-Fort Worth diagnostic run 2 using
a one-cell search nearest neighbor analysis.
SYSTEMS APPLICATIONS, INC.
153
-------
200
150
CQ
100
6
~\IIII|
ILLI
OBSERVED
12
16
24
30
36
42
48
I I I I I I I I I I I I I
200
150
100
50
6 12 18 24
August 30, 1985 Time (hours)
30 36 42
August 31, 1985
48
200
6
12
16
24
30
36
42
48
150
03
100
50
^ I I 1 I I 1 I I I | I I 1 I I | i I I I I j I I I ii I I 1 I I T
BNVW
OBSERVED O
I I I I I
I I I I I I I I I
mm-
DDCD
200
150
100
50
12
16
24
30
36
42
48
FIGURE 4-22. Continued.
SYSTEMS APPLICATIONS, INC.
154
-------
200
12
16
24
30
36
42
48
i I I I I I I 1 I I I I 1 1 I I 1 I I I I i
DNTN
OBSERVED 0
150
100
eo
O
50
«^ . ,
i i i i I i i i i i I i i t i
200
150
100
50
6 12 18 24 30 36 42
August 30. 1985 Time (hours) August 31, 1985
48
200
12
18
24
30
36
42
48
i 1 i I I I I I I I I | 1 I I 1 I I I i I T
ROSS
OBSERVED Q
150
o
CU 100
50
i i i
I I I i I I I i I i I I I i I t I I I i I I I I I i I I I I
200
150
100
50
6 12 18 24 30 36 42
August 30, 1985 Time (hours) August 31. 1985
48
FIGURE 4-22. Continued.
SYSTEMS APPLICATIONS, INC.
-------
200
6
12
18 24
30
36 42
46
150
m
100
50
KLLR
OBSERVED D
CD
a -
a
i i i i i I i i i i i I i i i i i I i I i i i T i i i i i I i i i i i I i i i- i i I i i i i
200
150
100
50
6 12 18 24 30 36 42 48
August 30, 1985 Time (hours) August 31, 1985
FIGURE 4-22. Concluded.
SYSTEMS APPLICATIONS, INC.
-------
100
80
12
IB
24
30
36
42
46
CO
a.
a.
60
40
20
I t I II I I I I II
MOCK
OBSERVED D
T i i i i | i i i i r j i i i i i [ i i i i i ] i i i i i i i i i i r
BESHE3
i i i I i i i /1 I i i i i i I i r i i i I i i i i i I i i i i
100
80
60
40
20
0 6 12 18 24 30 36 42
August 30, 1985 Tim* (hours) August 31, 1985
48
100
12
16
24
30
I I I I | I I I I I | I I I I I I I I ! I I | I I I I I | I I I I I I I
6 12 18 24 30 36 42
August 30. 1985 Tim* (hours) August 31, 1985
- 20
48
FIGURE 4-23. Tine series of predicted and observed hourly NO concentrations
(ppb) for the Dallas-Fort Worth diagnostic run 2 using a one-cell search
nearest neighbor analysis.
SYSTEMS APPLICATIONS, INC.
-------
100
80
60
o
cu
a.
2 40
20
6
~~\ I I i |
ILLI
OBSERVED
12
18
24
30
36
42
46
I I I i I I I 1 I I I I i I i 1 I r
ililfTTIIIIilllll I T I I I
I I I
I I I I I I I I
100
80
60
20
6 12 18 24 30 36 42 48
August 30. 1985 Tim* (hours) August 31, 1985
100
80
60
6
12
18
24
30
36
42
48
a,
a,
2 40
20
i i i i i
BNW
OBSE5VED
i i i i i r i v\ i i
HI
i i i TT | i i i i r
i i i N^/ I i i i i i I i i i—i i I i i i i i 1 i i i i i 1 i i i i
i i I i i i i i
100
80
60
40
20
6 12 18 24 30 36 42
August 30, 1985 Time (hours) August 31, 1985
46
FIQJKE 4-23. Continued.
SYSTEMS APPLICATIONS, INC.
-------
100
80
12
16
24
30
36
42
46
i i i i i i i i i i i | i i i i i | i i i \ i i i i i i i i i i i i i ( i i i i i i i i i i r
DNTN
OBSERVED a
CD
a.
a,
60
40
20
100
80
60
40
20
6 12 16 24 30 36 42
August 30, 1985 Time (hours) August 31. 1985
48
100
12
16
24
100
I ! I I I | i I ! I I | I I I I I | I I I I I I I ! I I I | I
6 12 16 24
AUgUSt 30, 1985 Time (hours)
30 36 42
August 31. 1985
48
FIGURE 4-23. Concluded.
SYSTEMS APPLICATIONS. INC.
159
-------
300
12
18
24
30
36 42
48
iI I I 1 I i I I I I | i I I I 1 I I I I I I j 1 1 r
- MOCK
- OBSERVED O
200
en
a.
O
Z
100
S 12 18 24 30 36 42
August 30, 1985 Time (hours) August 31, 1985
300
200
100
48
300
200
a.
a,
o
100
12 18
24
30 36
42 48
III1|TTTTT|III II \I I I II| I I I [ III IiJIIII11I| IIIT
- NSTR
- OBSERVED CJ
300
200
100
6 12 18 24 30 36 42 48'
August 30. 1985 Time (hours) August 31, 1985
0
FIGURE 4-24. Tine series of predicted and observed hourly NC>2 concentrations
(ppb) for the Dallas-Fort Worth diagnostic run 2 using a one-cell search
nearest neighbor analysis.
SYSTEMS APPLICATIONS, INC.
-------
300
0
6
12
16 24 3036
42 48
- ILLI
r OBSERVED O
200
j 1 I I I I I 1I f I T^
m
c.
1
100
Ql I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I _
0 6 12 18 24 30 36 42 48
300
200
100
24 30 36
August 30, 1985 Time (hours) August 31, 1985
300
6
12
18 24
30 36
42 48
200
CD
O.
CM
i
100
i i i T i i i i i
BNVW
OBSERVED m
J I
6 12 18 24 30 36 42
August 30, 1985 Time (hours) August 31, 1985
300
200
100
48
FIGURE 4-24. Continued.
SYSTEMS APPLICATIONS, INC.
-------
300
12 18
24
30 36
1 I I I 1 I I I I I I I I I I I I I I I I I I I I I T
- DNTN
- OBSERVED D
200
Q
o,
O
100
42 48
T [ I I I I I
Ql I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I Q
0 6 12 18 24 30 36 42 48
300
200
100
August 30, 1985 Time (hours) August 31. 1985
300
6
12 18
24
30 36 42 48
200
m
CM
i
100
i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i
ROSS
OBSERVED O
CD
300
200
100
6 12 18 24 30 36 42 48
August 30, 1985 Time (hours) August 31, 1985
FIGURE 4-24. Concluded.
SYSTEMS APPLICATIONS, INC.
-------
200
12
18 24
30 36
42
150
CD
n
O
100
50
'.)' I I I i I I I I I I I I I I I I I I I I j I I I I ! | I I I I I I I I I I I I I I I
6 12 18 24 30 36 42
August 30, 1985 Time (hours) August 31, 1985
150
100
50
48
200
6
12 18
24
30 36
42 48
150
100
50
NSTR
OBSERVED
n
6 12 18 24
August 30, 1985 Time (hours)
30 36 42
August 31, 1985
200
150
100
50
48
FIGURE 4-25. Time series of predicted and observed hourly ozone concentrations
for the Dallas-Fort Worth diagnostic run 2 using a two-cell search nearest
neighbor analysis.
SYSTEMS APPLICATIONS, INC.
-------
200
12
18
24
30
36
42
48
150
m
1001—
I I I I I | I I I I I I I I I I I | I I I I I | I I T
ILLI
OBSERVED CD
I I I I I
200
150
100
50
0 6 12 18 24 30 36 42 48
August 30, 1985 Time (hours) August 31, 1985
200
6
12
18
24
30
36
42
150
O
100
to
O
50
\ i i i i \ i i i i i i i i i i i i r \ i i i. i i i
BNVW
OBSERVED CD
i i i i i | i i i i r
48
200
150
100
50
6 12 18 24 30 36 42
August 30, 1985 Time (hours) August 31, 1985
48
FIGURE 4-25. Continued.
SYSTEMS APPLICATIONS, INC.
-------
200
12
18
24
30
36
42
46
150
CQ
100
50
I i I I 1 I I I i i I I i I i i I ] I I i I r
DNTN
OBSERVED 0
i i i i r
i i i i I i i i i i i i i i i I i i i i i i i i i i I i i i i i i i i i i i i
200
150
100
50
0 6 12 18 24 30 36 42
August 30, 1985 Time (hours) August 31, 1985
48
200
6
12
18
24
30
36
42
48.
150
CO
100
n
o
50
i i i i i I i i i i i
ROSS
OBSERVED Q
i i i i i I t I i i i I i i i I i I i i i I i I i i i i i I i I i i i I i I i i i I i i i i
200
150
100
50
6 12 18 24 30 36 42 48
August 30, 1985 Time (hours) August 31, 1985
FIGURE 4-25. Continued.
SYSTEMS APPLICATIONS. INC.
-------
200
200
- 150
i i i i i I i i i i i I i i i i i I i i i i i | i i i i i I i i
- 100
6 12 16 24 30 36 42
August 30, 1985 Time (hours) August 31, 1985
FIGURE 4-25. Concluded.
SYSTEMS APPLICATIONS, INC.
ice
-------
100
80
12
13
24
30
36
42
46
60
ffl
O.
0,
I 40
20
I I I I I I I I T
MOCK
OBSERVED D
i i i i i
iI I I I | I I I i I I I I I I I I I I I I
i i i i i
QSBSE1
i i i i i I i i i _j i I i i i i i I i i i i i I i i i i i I i i i
100
80
60
40
20
5 12 18 24 30 36 42
August 30, 1985 Time (hours) August 31, 1985
48°
100
6
12
18
24
30
36
42
48
I I I I | I T T
NSTR
OBSERVED CD
100
80
60
40
20
6 12 18 24 30 36 42
August 30. 1985 Time (hours) August 31, 1985
48
FIGURE 4-26. Time series of predicted and observed hourly NC>2 concentrations
(ppb) for the Dallas-Fort Worth diagnostic run 2 using a two-cell search
nearest neighbor analysis.
SYSTEMS APPLICATIONS, INC.
-------
100
80
6
12
18
24
30
36
48
60
m
CL
cu
2 40
20
I I 1 I I I I I I I I I I I I 1 I I I I T
ILLI
OBSERVED Q
r i i i i i i r
i i i I i I i i
100
80
60
40
20
6 12 16 24 30 36 42 46
August 30, 1985 Time (hours) August 31, 1985
100
80
12
16
24
30
36
42
48
60
m
a.
Cu
2 40
20
BNVW
OBSERVED D
a m
i i I i i i i i I i i r"i i I i i i i i I i i i i
100
80
60
40
20
6 12 16 24 30 36 42
August 30, 1985 Time (hours) August 31, 1985
46
FIGURE 4-26. Continued.
SYSTEMS APPLICATIONS. INC.
-------
100
80
6
i i i i I |
DNTN
OBSERVED
12
18
24
30
36
42
48
60
CO
a.
a.
2 40
20
100
60
60
40
20
6 12 18 24 30 36 42 48
August 30, 1985 Tim* (hours) August 31, 1985
100
12
18
100
I I I I I I I I I I
ROSS
OBSERVED
I I I I I I I I I I I I I I I ! I I I I I
6 12 18 24 30 36 42
August 30. 1985 Time (hours) August 31, 1985
- 20
43
FIGUPE 4-26. Concluded.
SYSTEMS APPLICATIONS. INC.
-------
300
12
18
24
30 36
42 46
200
£
CM
O
100
\ I II I I I I I I I I I 1 I I I I I I
MOCK
OBSERVED D
J^sa
i I I I l H-"iJ l l I l I I I l I l
6 12 18
August 30, 1985
24
Time (hours)
30 36 42
August 31, 1985
300
200
100
46
300
12 18
24
30
36 42
46.
200
o
a,
a,
O
100
I I I I I I I 1 i I I I I I I I f 1
h NSTR
- OBSERVED O
mn-
300
200
100
6 12 18
August 30, 1985
24
Time (hours)
30 36 42
August 31, 1985
46
FIGUEE 4-27. Tine series of predicted and observed hourly ozone concentrations
for the Dallas-Fort Worth diagnostic run 2 using a two-cell search nearest
neighbor analyis.
SYSTEMS APPLICATIONS, INC.
-------
300
200
m
a.
O,
1
6 12 16
i i i i i | i i i i i | i i i i i I i
ILLI
OBSERVED D
24
30 36
42 46
100
300
200
100
Ql I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I ! I Q
0 6 12 16 24 30 36 42 48
August 30, 1985 Time (hours) August 31, 1985
300
6
12
16 24
30 36
42 46
200
eu
O
z
100
i i r i i | i i i i i
BNVYT
• OBSERVED D
6 12 18 24 30 36 42
August 30. 1985 Time (hours) August 31. 1985
300
200
100
48
FIGURE 4-27. Continued.
SYSTEMS APPLICATIONS. INC.
-------
300
6
I I T~l |
- DNTN
- OBSERVED
12 18
24
30 36
42 48
200
m
a.
i
100
300
200
100
6 12 18 24 30 36 42 48
August 30, 1985 Time (hours) August 31, 1985
300
6
12
18 24 30 36
42 48
200
£
100
i i i i I
ROSS
OBSERVED
i i i i i
mm,
300.
200
100
6 12 18 24
AUgUSt 30. 1985 Time (hours)
30 36 42
August 31, 1985
48
FIGURE 4-27. Concluded.
SYSTEMS APPLICATIONS, INC.
-------
Summary
The diagnostic run 2 for Dallas-Fort Worth does a quite good job of predicting the
observed region-wide peak ozone concentration on August 30 and 31. The predicted
region-wide maximums on August 30 and 31 are within 6 and 4 percent of the
observed peak respectively. On August 31 at the location of one of the peak
observations (DNTN) the model replicates the peak to within 20 percent. However,
at the location of the peak observed ozone on 30 August (ROSS) and at the other
location of the peak observed ozone on 31 August (MOCK) the model replicates the
peak observations to within only 48 and 58 percent, respectively. The nearest-
neighbor analysis indicates that there are slight displacements of the elevated ozone
cloud in the model predictions and that predicted concentrations of comparable
values to the peak observations are within a few grid cells of the locations of the
peak observations. Because this slight displacement of predicted ozone concentra-
tions will not affect usefulness of the UAM when evaluating alternative emission
control strategies, and the fact that diagnostic run 2 does satisfy the model perform-
ance goal (the predicted region-wide maximum is within 4 percent of the observed
peak and occurs 8 km to the southwest of the observed peak, and the predicted peak
at the location of the observed peak is within 20 percent), then diagnostic run 2 is a
suitable base case.
Because of the high degree of spatial variability in the predicted ozone concentra-
tions in diagnostic run 2, model performance could be improved, by adjusting the wind
fields so that the predicted elevated ozone clouds occur over the ozone monitors
where the highest ozone observations occur. However, this may involve several
diagnostic simulations with arbitrary adjustments to the wind fields, a process not
consistent with the PLANR use of the UAM. Besides, as noted earlier, this dis-
placement of the ozone peaks will not affect the evaluation of emission control
strategies.
8910<»rl S
173
-------
5 SUMMARY AND CONCLUSIONS
The PLANR use of the 'JAM was demonstrated for two regions: the city of Atlanta
and the Dallas-Fort worth metroplex. It was shown that UAM performance is ade-
quate (i.e., model performance satisfied the model performance goal) when only
routinely available data are used as inputs in a limited number of diagnostic simu-
lations. However, it has also been shown here and elsewhere (Morris, Myers, and
Carr, 1989) that use of routine meteorological data (hourly surface measurements
and twice-daily upper-air sites) to develop UAM inputs in a "cook book" type fashion
(i.e., running data through preprocessors without interpretation of the representa-
tiveness of the observations) generally results in less then satisfactory model per-
formance. Thus two of the key elements of the PLANR use of the UAM are (1)
characterization of the meteorological conditions that lead to the elevated ozone
episode (base on analysis of meteorological data, daily weather maps, and observed
air quality data) and (2) adjustment of the meteorological data to represent the
conceptual picture of the meteorological conditions. In this adjustment process we
have indentified the following as important considerations:
Twice-daily upper-air soundings
Surface wind speeds
Emissions
TWICE-DAILY UPPER-AIR SOUNDINGS
It is particularly difficult to develop hourly wind fields above the surface with just
routine twice-daily upper-air observations. The upper-air observations are collected
in the morning at 1200 GMT (approximately 0700 LST) and evening at 0000 GMT
(approximately 1900 LST). This makes it very difficult to characterize transport of
pollutants in the late morning and afternoon, the key period of photochemical acti-
vity. For the first diagnostic simulation we used linear vector interpolation between
the morning and afternoon soundings; for subsequent diagnostic simulations we used
more creative interpolation techniques (e.g., the techniques used to characterize
frontal passages as described below) when warranted.
8910i»rl 7
174
-------
Characterization of Fronts
When a front passes through the modeling region, as occurred in the Atlanta episode
reported here and in a St. Louis episode reported elsewhere (Morris, Myers, and Carr,
1989), it is particularly difficult to characterize the upper-level winds using the
twice-daily upper-air soundings. The time of passage of the front through a region
can be estimated by examining the surface meteorological observations to find an
hour with a wind shift, temperature change, or other meteorological phenomena.
This information can then be used to develop an interpolation scheme for defining
the hourly wind sounding from the twice-daily upper-air soundings for input into the
DWM. For example, if surface data indicate that a front passes through at 1300 LST,
then use of the morning sounding until 1300 and the afternoon sounding after 1300
will probably result in more representative wind fields than is obtained by linear
interpolation of soundings.
Mixing Height Definition
The characterization of the mixing height is also very difficult using just routine
twice-daily upper-air observations and the hourly surface temperatures. In the UAM
application to Dallas-Fort Worth reported here it was suspected that a subsidence
inversion caused by a stationary high-pressure system was responsible for the limited
mixing conditions that led to the ozone episode. Since ozone episodes generally
occur during hot, stagnant conditions with limited mixing, we used the summer
afternoon climatalogical mixing height as the maximum mixing height for the Dallas
base case.
Wind Shear
During the middle of the day, when convective activity is greatest and there is a
well-defined, well-mixed layer below the temperature inversion, not much wind shear
within the mixed-layer is expected. However, interpolating between the morning and
evening soundings, periods in which there may be significant amounts of wind shear,
may result in incorrect specification of wind shear within the mixed layer in the
middle of the day. This wind shear, in turn, would result in excessive dilution of the
urban plume and, consequently, underprediction of the peak ozone concentrations.
One approach to get around this problem is to use identical wind fields in all UAM
layers below the mixing height. For the PLANR use of the UAM this is partially
done because the lowest layer of the Diagnostic Wind Model (DWM), which uses the
surface wind observations, is used exclusively in layer one of the UAM and weighted
50 percent in the remaining layer of the UAM (layer 2) that lies below the mixing
height. It should be pointed out that wind shear may occur in the mixed layer during
the day. For example, in the Atlanta episode wind shear occurred due to the passage
of a front through the modeling domain in the afternoon.
8910«*r 7
175
-------
Recommendations
Specification of upper-level hourly wind fields using routine twice-daily upper-air
soundings may involve developing interpolation schemes based on the meteorological
characterization of the episode. Guidelines for interpolation procedures need to be
developed and tested. The collection of an afternoon upper-air sounding (1800 GMT)
would greatly reduce the uncertainty in the upper-level wind fields and mixing
heights. It is recommended that consideration be given to collecting one extra
sounding at 1800 GMT in the vicinity of cities with severe ozone attainment prob-
lems during the ozone season. At little extra cost in monkeying, *he uncertainty in
the wind fields and mixing heights can be reduced and the preparation of the UAM
meteorological inputs can be simplified.
SURFACE WIND SPEEDS
Surface meteorological measurements are usually the only hourly meteorological
measurements available for constructing the hourly varying UAM inputs. As noted in
Appendix D and in the evaluation of the PLANR use of the UAM for St. Louis
(Morris, Myers, and Carr, 1989), the instantaneous wind speeds reported at Federal
Aviation Administration or National Weather Service (FAA/NWS) sites may over-
estimate the wind speeds for the hourly average wind vectors required by the UAM
during low wind speed conditions. An analysis of wind speeds from seven "colloca-
ted" hourly average and FAA/NWS wind monitoring sites during episodic ozone condi-
tions in the South Coast Air Basin (SOCAB) of California revealed that the continu-
ous hourly average wind speeds are on average only 55 percent of their instantaneous
FAA/NWS counterpart. When examining each of the seven station pairs individually,
it was found that five of the seven continuous stations measured wind speeds that
were W to 55 percent of the FAA/NWS values, while the other two continuous sites
recorded wind speeds that were 71 and 98 percent of the instantaneous values.
Based on this apparent bias in wind speeds reported at FAA/NWS sites, a sensitivity
test for Atlanta was run in which FAA/NWS wind speeds were reduced 50 percent.
This sensitivity test resulted in much better model performance. Similar results
were reported in the PLANR application of the UAM to St. Louis (Morris, Myers, and
Carr, 1989). Note that because of the possibility of this wind speed bias and the
fairly good coverage of continuous wind measurements in Dallas-Fort Worth, the
FAA/NWS sites were not used in the development of wind fields for the Dallas-Fort
Worth application.
Further analysis of the possibility of wind speed bias at FAA/NWS stations during
episodic ozone (low wind speed) conditions needs to be performed. The wind speed
data from the seven collocated sites from the SOCAB needs to be examined in a
more statistically rigorous manner to determine if the bias is statistically signifi-
cant. Collocated instantaneous and hourly average wind data from other regions
8910"tr 7
176
-------
(e.g., the Regional Air Pollution Study and Northeast Corridor studies) also needs to
be analyzed. Finally, some guidence needs to be formulated to eliminate this bias
from the UAM analysis. For the PLANR UAM sensitivity analysis for Atlanta and
the PLANR UAM application to St. Louis the FAA/NWS wind speeds were reduced by
50 percent. This is probably on the high side of reductions, but more research is
needed to determine what is an appropriate reduction factor or whether it is appro-
priate to modify the wind speeds at all. It is recommended that a network of surface
wind monitors reporting hourly averages be set up in ozone nonattainment regions.
AIR QUALITY DATA
Air quality data for photochemical modeling is needed for initial concentrations,
boundary conditions, and model performance evaluation. The trend in UAM modeling
is to simulate multiple days (in order to eliminate the need for defining initial con-
centrations) and to simulate a large region (in order to reduce the sensitivity of the
model calculations to the specification of boundary conditions). Accordingly, for the
PLANR application of the UAM, "clean" concentration values have been used for
initial and boundary conditions in the first diagnostic simulations. Note that for the
PLANR UAM application to Atlanta the presence of biogenic emissions indicated
that higher VOC and ISOP concentrations were present, thus the boundary conditions
were adjusted accordingly. For the PLANR application of the UAM to Philadelphia,
air quality data indicated that there was a large amount of transported ozone and
ozone precursors, thus the initial and boundary conditions were adjusted accordingly.
The importance of air quality data for model performance evaluation cannot be
understated. The comparison of predicted and observed concentrations is the main
indication that the model is working properly for each application. Because of the
limited air quality data available for Atlanta (three ozone sites), it was very difficult
to determine whether the model was operating correctly, thus a sensitivity test was
also performed. For the Dallas-Fort Worth application, despite some descrepancies
at a couple of the air quality monitoring sites, the model replicated enough of the
features of the observed ozone, NO, NC^, and CO concentrations that there is more
confidence in model predictions. More air quality monitoring data is always desir-
able for model performance evaluation.
EMISSIONS
There is considerable uncertainty in the anthropogenic and biogenic emissions data.
The uncertainity may be biased toward an underprediction in the total emissions
which makes it very difficult to evaluate the model since a model's underprediction
may be due to insufficient emissions. Attempts to improve model performance by
adjusting other inputs (e.g., boundary conditions or meteorology) may result in better
model performance, but the model may be getting the right answer for the wrong
reason.
8910"*r 7
177
-------
Anthropogenic Emissions
Recent photochemical modeling simulations using the UAM(CB-IV) have produced a
disturbing trend toward underprediction of the peak ozone observations. This under-
prediction has occurred in the California South Coast Air Basin (Los Angeles) (Hogo,
Mahoney, and Yocke, 1988), Philadelphia (Haney and Burton, 1988), Atlanta, and
Dallas-Fort Worth. The old UAM(CB-II), which we know may have considerable
problems lr .<:€ cheTustry (e.g., aromatic and carbonyl chemistry), tended to over-
and underpredict about equally. However, since the improvements in chemistry were
made, it appears that the model tends to underpredict more than overpredict. In
light of recent discoveries of previously uninventoried VOC sources (e.g., running
losses, solvents) and emission reconcilation studies that have measured higher VOC
emissions than estimated in emission inventories (e.g., the SCAQS tunnel study), this
tendency toward underprediction indicates that VOCs in current emission inventories
may be underestimated.
In all of the PLANR UAM applications performed to date the peak observations are
underpredicted. Remedial actions generally are limited to adjustments in the boun-
dary conditions, wind fields, or mixing heights. Although it appears the VOC emis-
sions are underestimated, information on this underprediction is not available; thus
the emissions have not been adjusted to improve model performance. However, sen-
sitivity tests that look at the level of VOC emission increase needed to obtain the
best model performance may provide insight into the degree of the underestimate of
the VOC emissions.
In the PLANR diagnostic simulations to improve UAM performance (i.e., reduce the
level of underprediction) we have focused on adjusting wind fields and mixing depths
to better reflect the conceptual picture of the meteorological conditions of the
ozone episode. This adjustment has consisted mainly of interpolation schemes for
the upper-air data, reduction of wind speeds at FAA/NWS sites, and reduction of the
maximum mixing height. The reduction in wind speeds and mixing height will have
an effect similar to increasing the emissions when analyzing emission control strate-
gies. An increase in boundary conditions, which also would reduce an underpredic-
tion, would have a different effect than increasing emissions when analyzing emis-
sion control strategies. Until high-quality emission inventories are developed,
adjustments to improve model performance should be limited to variables that act in
the same direction as increasing emissions when analyzing emission control strate-
gies. Unless there is evidence of transported ozone or ozone precursors into the
region of interest, or unless a small modeling domain is specified, there should not be
extensive modifications to the boundary conditions to improve model performance.
8910»»r 7
178
-------
Biogenic Emissions
Current estimates of biogenic emissions are very uncertain in the amounts, reacti-
vity and composition, spatial distribution, and diurnal variations. More research is
needed to reduce the uncertainities in all areas of biogenic emission estimates. In
addition, the chemistry of biogenic emissions is not well understood. The use of
chemical mechanisms designed mainly to simulate the atmospheric chemistry of
anthropogenic emissions may incorrectly characterize the chemistry of rural bio-
genic emissions. Again, more research is required to better define the chemical
kinetics of biogenic emissions. Even with all the uncertainties in the current esti-
mates and chemistry of biogenic emissions, biogenic emissions affect the efficiency
of emission control reduction strategies designed to reduce ozone concentrations and
should be included, to the extent possible, in all future analyses.
CONCLUSIONS
It has been demonstrated that the UAM can be exercised using only routine data and
that it performs in an acceptable way. Given the current uncertainity in estimates
and reactivities of anthropogenic and biogenic emissions, it appears that obtaining
very good model performance would be fortuitous at best. However, even given the
current uncertainty in emissions and meteorological inputs, the use of the UAM by
air quality planners for air quality planning, ozone attainment demonstration, and
sensitivity analyses will provide a powerful tool for developing air quality manage-
ment plans.
8910<*r 7
179
-------
References
Ames, 3., T. C. Myers, L. E. Reid, D. C. Whitney, S. H. Golding, S. R. Hayes, and
S. D. Reynolds. 1985a. SAI Airshed M .-tel r>.y;rati-'r: Manuals. Volume I—
User's Manual. U.S. Environmental Proisc^o.i rvgcncy vtPA-600/8-85-007a).
Ames, 3., S. R. Hayes, T. C. Myers, and D. C. Whitney. 1985b. SAI Airshed Model
Operations Manuals. Volume II— System's Manual. U.S. Environmental Protec-
tion Agency (EPA-600/8-85-007b).
Chameides, W. L., R. W. Lindsay, 3. Richardson, and C. S. Kiang. 1988. The role of
biogenic hydrocarbons in urban photochemical smog: Atlanta as a case study.
Science, 241:1473-1475.
Douglas, S., and R. Kessler. 1988. "User's Guide to the Diagnostic Wind Model. Ver-
sion 1.0." Systems Applications, Inc., San Rafael, California.
Gery, M. W., G. Z. Whitten, and 3. P. Killus. 1988. "Development and Testing of the
CBM-IV for Urban and Regional Modeling." Systems Applications, Inc., San
Rafael, California (SYSAPP-88/002).
Haney, 3. L., and C. S. Burton. 1988. "UAM-CBM-IV-Calculated Effects of Emission
Reductions on Ozone Levels in the Philadelphia Metropolitan Area." Systems
Applications, Inc., San Rafael, California (SYSAPP-88/117)
Haney, 3. L., D. R. Souten, T. W. Tesche, L. R. Chinkin, H. Hogo, and M. C. Dudik.
1986. "Evaluation and Application of the PARIS Photochemical Model in the
South Central Coast Air Basin. Volume I." Systems Applications, Inc., San
Rafael, California (SYSAPP-85/065).
Hogo, H., M. W. Gery. 1988. "Guidelines for Using OZIPM-4 With CBM-IV or
Optional Mechanisms. Vo'lume 1. Description of the Ozone Isopleth Plotting
Package Version 4." Systems Applications, Inc., San Rafael, California
(SYSAPP-88/001).
Hogo, H., L. A. Mahoney, and M. A. Yocke. 1988. "Draft Air Quality Management
Plan 1988 Revision. Draft Appendix V-R. Urban Airshed Model Performance
Evaluation for 5-7 3une 1985." Systems Applications, Inc., San Rafael, Califor-
nia (SYSAPP-88/ 138).
89104 9
180
-------
Holzworth, G. C. 1972. Mixing Heights, Wind Speeds, and Potential for Urban Air
Pollution Throughout the Continuous United States. U.S. Environmental Pro-
tection Agency.
Kelley, R. F. 1981. User's Manual for Mixing Height Computer Program. U.S.
Environmental Protection Agency (EPA-450/4-81-022).
Killus, 3. P., and P. D. Gutfreund. 1984. The Development and Use of Protocols in
the Urban Airshed Model in Regulatory and Research Applications." Systems
Applications, Inc., San Rafael, California (SYSAPP-84/133).
Lamb, B., A. Guenther, D. Gay, and H. Westberg. 1987. A national inventory of bio-
genie hydrocarbon emissions. Atmos. Environ., 21(8): 1695-1706.
Lamb, B., H. Westberg, and G. Allwine. 1986. Isoprene emission fluxes determined
by an atmospheric tracer technique. Atmos. Environ., 20(1); 1-8.
Morris, R. E., T. C. Myers, and E. L. Carr. 1989. "Urban Airshed Model Study of
Five Cities: Evaluation of Base Case Model Performance for the City of St.
Louis Using Rich and Sparse Meteorological Inputs." Systems Applications,
Inc., San Rafael, California.
Morris, R. E., M. W. Gery, M. K. Liu, G. E. Moore, C. Daly, and S. M. Greenfield.
1989c. "Sensitivity of a Regional Oxidant Model to Variations in Climate
Parameters. Volume I: Results." Systems Applications, Inc., San Rafael, Cali-
fornia (SYSAPP-89/014a).
Morris, R. E., R. C. Kessler, S. G. Douglas, K. R. Styles, and G. E. Moore. 1988.
"Rocky Mountain Acid Deposition Model Assessment Acid Rain Mountain Meso-
scale Model (ARM3)." Systems Applications, Inc., San Rafael, California
(SYSAPP-88/152).
Morris, R. E., T. C. Myers, M. C. Causley, L. Gardner, and E. L. Carr. I989b.
"Urban Airshed Model Study of Five Cities. Low-Cost Application of the Model
to Atlanta and Evaluation of the Effects of Biogenic Emissions on Emission
Control Strategies." Systems Applications, Inc., San Rafael, California
(SYSAPP-89/117).
•
Morris, R. E., T. C. Myers, H. Hogo, L. R. Chinkin, L. A. Gardner, and R. G.
Johnson. 1989a. "A Low-Cost Application of the Urban Airshed Model to the
New York Metropolitan Area and the City of St. Louis." Systems Applications,
Inc., San Rafael, California (SYSAPP-89/038).
89101* 9
181
-------
Reynolds, S. D., T. W. Tesche, and L. E. Reid. 1979. "An Introduction to the SAI
Airshed Model and Its Usage." Systems Applications, Inc., San Rafael, Califor-
nia(SYSAPP-79/031).
Smolarkiewicz, P. K. 1983. A simple positive definite advection scheme with small
implicit diffusion. Monthly Weather Review, 111:479-486.
SAI. 1989. "User's Manual for Preparing Emission Files for Use in the Urban Airshed
Model." Systems Applications, Inc., San Rafael, California (SYSAPP-89/114).
Tingey, D. T., et al. 1978a. Isoprene Emissions Rates from Live Oak. U.S. Environ-
mental Protection Agency (EPA-904/9-78-004).
Tingey, D. T., et al. 1978b. Monoterpene Emission Rates from Slash Pine. U.S.
Environmental Protection Agency (EPA-904/9-78-013).
Zimmerman, D., W. Tax, M. Smith, 3. Demmy, and R. Battye. 1983. "Anthropogenic
Emissions Data for the 1985 NAPAP Inventory." Alliance Technologies
Corporation, Chapel Hill, North Carolina (EPA-600/7-88-022).
89104 9
182
-------
Appendix A
PERCENT CONTRIBUTION OF INITIAL CONCENTRATIONS, BOUNDARY
CONDITIONS (FOUR LATERAL FACES PLUS TOP BOUNDARY)
ANTHROPOGENIC AREA SOURCE EMISSIONS, POINT SOURCE EMISSIONS,
AND BIOGENIC EMISSIONS TO HOURLY TRACER, NOX, AND VOC
CONCENTRATIONS FOR ATLANTA ON 4 JUNE 1984
A-la: 0800 Initial NOX tracer contibution
A-lb: 0800 Boundary NOX tracer contribution
A-lc: 0800 Area source NOX tracer contribution
A-Id: 0800 Point source NOX tracer contribution
A-2a: 0800 Initial VOC tracer contribution
A-2b: 0800 Boundary VOC tracer contribution
A-2c: 0800 Anthropogenic VOC tracer contribution
A-2d: 0800 Biogenic VOC tracer contribution
A-3a: 1200 Initial NOX tracer contribution
A-3b: 1200 Boundary NOX tracer contribution
A-3c: 1200 Area source NOX tracer contribution
A-3d: 1200 Point source NOX tracer contribution
A-4a: 1200 Initial VOC tracer contribution
A-4b: 1200 Boundary VOC tracer contribution
A-4c: 1200 Anthropogenic VOC tracer contribution
A-4d: 1200 Biogenic VOC tracer contribution
A-5a: 1600 Initial NOX tracer contibution
A-5b: 1600 Boundary NOX tracer contribution
A-5c: 1600 Area source NOX tracer contribution
A-5d: 1600 Point source NOX tracer contribution
A-6a: 1600 Initial VOC tracer contribution
A-6b: 1600 Boundary VOC tracer contribution
A-6c: 1600 Anthropogenic VOC tracer contribution
A-6d: 1600 Biogenic VOC tracer contribution
-------
m
CM
CM
00
in
s
KJ
IT)
D
in
Q.
o
«5
i i »
8
1S3M
I I
LJ
cr:
O
-------
00
0)
in in
CM O
oo oo
in
oo
in
-------
oo
I 1 I I I I I t I I I I
c
—5
<
h-"
u
o
o
00
0)
O
!_
—'
x
O
z
0>
u
u
o
to
O
i_
O
c
o
C
o
u
<--
c
0>
u
1S3M
«£
U
•D
-------
00
in
CM
08
CM
CO
in
o
03
in in
00
O
•o
i
O
b_
-------
00
in
CM
08
00
8
o
03
O
SPi'iofiinioS* £-
i I i i < 1 i t i 1 i i i 1 i i i 1 i i i 1 i i i 1 i i i
V^ "^:
^^^^ — -
~
—
£
o
™
o ~
^ ID .
^ O ^
Q_ Q "^ C) _
^P ^ ~
CD
03 ^ •
8 8
ft:
H- _
•
_l —
O "
••
dw
I 1 1 1 1 1 1 1 1 t 1 1 i 1 1 1 1 1 !! 1 1 t 1 1 1 1 1 1 1 1 t f 1 1 1 ~
* ^>
>^
t— •
to
U
o
o
00
8 -5
1_
o
o
>
I _
o tr o
OT £
vt—
o
c
o
"3
JD
s 1
o
o
"c
U
L-
0>
Q-
n
O O O O CM »- CM
1S3M i
U
D
o
-------
00
o
in
CM
08
Sn T
8
00
o
s
o
s
o
c
O
C
o
C
O
c
a>
u
CVJ
I
o
-------
00
O)
0>
c
I I I I I I I I I I I I
f I I I I I I I I I I I
! I I 1 I I I I I I 1
LJ
o
o
oo
1S3M
o
o
:c
o
o.
o
c
0
c
o
C
o
o
•+*
c
0>
o
l_
-------
o
c
;0
m
ro
O.
0)
-i
o
O
o
rs
cr
c
r+-
o'
3
cr
o'
O
O
O
o
o
00
O
O
n
WEST
I I I i I I I I I I I I I I I I I t I I I I I I t I
01 . ". 01
t i i t i i i CD
EAST
00
4-
-------
o
c
m
5»
£ o
O
o
CD
f**-
n
O
% 0
c
/-*•
0
Z5
O
— *
^
5 2 ro
o 5 °
•2.
O
X
-i
o
o
0)
-1
Q Crf
/— *• O
O
O
n
—(
3*
A
1 1
i»
•»
—
-
-
-
.
•
-
~_
-
-
:
t
1 i
'£>
CO
WEST
0 0 0 $»
i i i i i i i | i i i i i t i i i I i i i i i i t i i | t t i i i t t i i
«
-
P I
i/)
-
^
^ ™
>0
0 C
030
^ CD
CD ^fc X-jj —
O _i
£ U >
03 5
O —
O
z
«••
^— :
i i i i i i i t i i \ i i \ \ i i i i i i i i i i i i i i i
O)
O
a
o
XI
O
O
ro
0
s
a>
0
0
0
i
CD
U tri U Ul III Oi OJ C-O
8O N> *• » CD O ro
ui u< cn ui ui cn cn
EAST
-------
in
I
in
(0
8
1SV3
§
8
I I I T I
l|lll|lll|lll|
O
R
3
o
s
o
tM
fv
O
s
§
-------
00
I I 1 I I i I r I I ii l l I l 1 I l i i i II till t 1 II I "I
CO
LJ
O
o
x
O
c
c
C
o
o
j-'
c
0)
(J
1_
0)
CL
(J
CO
1S3M
o
-------
00
Cr>
0)
c
(f)
LJ
o
o
o
O
x
O
z
(U
o
i_
2
O
en
"c
'o
CL
C
O
c
O
O
+-*
C
0>
o
u.
(1)
Q.
•o
ro
i
LJ
o:
D
O
-------
in in in
CN o so
o>*> m i*>
CM———— 1 . I i i i
031 I I I I I I I P
o
PJ
i i I i i i 1
a:
m
m
g
8
2 ) i i i i i i i i F i i i t > i i i t I t t i t i i i i t I i t i i » i i t i
1S3M
of
No
V)
co
0)
to
LJ
o
o
O
O
"o
*-*^
'c
c
o
c
o
c
-------
T
00
1SV3
i i \ i i i r i i I i
c
3
to
LJ
O
O
CM
u
o
o
o
>.
1_
0
•o
c
D
o
c
o
^•s
3
.0
'C
"c
o
o
"c
o>
o
1_
o>
CL
UJ
o:
o
-------
00
O)
in
CM
CM
i
•O
8
10
1SV3
in
8
in
i f i
a
1S3M
10
LJ
O
O
rg
o
D
O
O
a-
o
CL
o
c
o
c
o
c
o
o
*-<
c
a>
u
-------
00
en
in
CM
08
CM
03
in
o
CD
lO
in
05
r*
10
8
r^
•o
1SV3
n
r«.
o
a
»x
•O
f>
O
r*
K)
in
Q.
•o
^«
I
tr
o
LL.
-------
00
in
CM
oS
0}
o
§
o
5
0
*
0
g
o
03
O
«(
1 1 1
•
-
-
-
-
—
^^
™
_
-
~
—
-
"
—
B
_
^
—
~
-
1 1 1 1
5
*
1SV3
innininmninin
(O ^v ^^ ^^ ^^ ^v ^o ^p «^
'O ^^ 'O ^O 'O 'O 1^ PO^^ ^Kk
1 1 i 1 1 1 1 t 1 1 1 1 1 1 1 1 ] i 1 1 1 1 < 1 1 1 1 1
I
-
-
—
-
K ~
1 I
™
^ \ —
b m -
J. ^j
^f ^J ^X • ^
t n ..^ Ci
^P S :
QD
BCD
° I
CL
t ~
W
«i
tn J
o • *
•
«i
•i
t 1 1 1 1 1 I 1 t 1 1 1 1 1 1 1 1 1 1 1 f 1 1 1 t t 1 1 1 1 1 1 1 1 1
O O O (
*•
T
h-"
00
U
o
o
10
O "*"'
•O O
^^
u
O
i_
•***
X
O
•z.
o| |
V) '£
"o
c
o
JD
2 1
o
u
c
8
•
^v
0)
CL
•
•?
!•» CM *• i
1S3M <
U
o:
z>
o
L.
-------
00
in
CM
CM
in
o
s
2
1SV3
S
in
CM
m
i i j i i f i f I i i i i r
8
03
O
s
O
s
o
M
iiijiiiii i i | i
2 i i i i i i i i i I i > i t i > i t i I i t i i i i t i t t i i i i i i ii i
1S3M
j
n-
00
u
o
o
(£>
a>
o
o
O
z
>v
o
c
D
O
c
o
c
O
o
"c
a>
o
i_
0)
CL
IT)
<
LJ
o:
o
L_
-------
00
o>
c
i i I i i i I i i i I
I i i i ] i i i I i i i j i i i
i i i i t t I i i < i i i i t i I i i l i i i i i i I i i i i i i t i
1S3M
(S)
LJ
O
o
(£>
0)
O
O
X
O
o
L-
o
o
l_
o
c
o
c
o
o
"c
0)
o
-------
00
en
CM
CM
O
R
o
o
i_
a>
CL
1S3M
in
i
o:
o
-------
II
r
0?
00
o
o
o
R
o
m in l>
OB}
a o
^K
0)
o
'_
o
o
o| |
^ '?
««—
O
c
o
.0
o i
— c
o
"c
0
0_
»°
vo
IS3M <
LJ
0
-------
00
O)
m
IM
8
in
in
-------
CO
en
1SV3
CM
08
CM
in
$
o
CO
o
CM
s
in
CM
in
i
CD
C
to
LJ
O
O
O
v.
o
o
o
15
c
o
c
o
u
*-•
c
0)
u
u
(U
CL
u
V0
1S3M
o
-------
00
CD
in
CM
in
o
CO
in
oo
(M
CO
in
r*>
1SV3
m
CM
m
o
fv
s
2
ft
1S3M
0>
c
D
LJ
O
O
10
CD
O
D
o
o
X
l_
o
c
o
C
o
"3
.0
'C
^^
c
o
o
+-I
c
CP
o
•o
(O
LJ
a:
o
-------
Appendix B "
HOURLY PREDICTED OZONE CONCENTRATIONS
(pphm) FOR ATLANTA DIAGNOSTIC RUN 1
-------
1SV3
a
a
OD
01
9)
3
-3
'C
3
L.
g
^
o
o
"c
o
o
o
N
o
-------
1SV3
"i i i Vi i r i i t i t Li i i i
JLS3M
a
a
JLSV3
CO
05
0)
3
**S
C
0
"o
"6
c
o
I
o
CJ
O
N
O
-------
JLS3M
Q.
a
in
03
O)
V
3
-5
C
3
k.
O
"o
"5
LJ
C
o
I
o
o
o
N
o
-------
1S3M
Q.
Q.
1SV3
CD
O)
9)
3
-5
=«s
C
O
O
LU
C
O
I
O
O
8
O
N
O
-------
1S3M
Q.
a
00
O)
a>
3
-3
C
3
"5
LU
0
8
o
£_>
8
o
N
O
-------
1SV3
1S3M
a.
o.
1SV3
CO
O)
9)
3
-5
5*6
c
L.
O
I
o
o
"c
o
CJ
-=- »
o
N
o
-------
1SV3
1S3M
a
a
to
r
-------
1S3M
Q.
Q.
03
0)
3
C
3
i_
C
o
"5
"5
•
8
o
o
N
o
-------
a
a
03
03
a>
3
-5
C
3
i_
O
o
LU
C
.0
7
0
•*<
C
o
o
o
N
o
1S3M
-------
1SVG
888
"l I I Vl I I I t I I I C I I I I
1S3M
O.
d
1SV3
00
O5
3
C
3
^.
0
o
UJ
c
o
I
o
o
o
N
o
-------
•Appendix C
HOURLY PREDICTED OZONE CONCENTRATIONS
(pphm) FOR ATLANTA DIAGNOSTIC RUN 2
-------
«
U
Si!
2 S
CM
I
§
II
P
in
CM
00
or>
CM
K>
in
o
oo
in
oo
to
in
ID
1SV3
m
*
r-
•o
m
CXI
in
o
m
oo
(O
in
to
ID
""V
i i i i r i i [ i i T | i ii i i i
i t 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 t i i i i i i i~
1S3M
9
I 5
£ E
ii
in
CM
s
CO
in
8
in
to
m
r
h-
m
CM
in
o
fv
, i i p. i i r | I i
i j i t. T ]
m
to
m
00
a
a
•s^'
21
o
N
O
(DO
1S3M
-------
II
§
E
in
CM
oo
(O
in in
o GO
s
in
(O
1SV3
in
in
CM
\r>
o
.n
03
1C
(O
ID
t I I I l I l t l I I I I j I ) Tji I J j ) I I)
'o
*
'i 1 I I i I i I i I i 1 i ri i i t i 1 i I i i i t I 1 I I i i i 1 I i II ~
o
N
1S3M
a.
a.
O o
si
• II
Is
>°l
II
o ^
2 "S
in
CM
oS
CM
03
?o
E •*
o ^
V)
1
in
o
CO
n
in
CO
fc
m
1C
5
J.SV3
• in
*
r*
K)
in
CM
t^
o
8
i»»
10
m
oo
to
to
in
(O
i. ..i..i i i «_
o|
NS
;;l. I
(OO
8
1S3M
03
tn
c
o
"w
.!2
0)
o
.o
ao
O)
g
o
N
O
-------
1SV3
t£
f*} J~
*'
II
" II
1 -2
•* >
| E
.§ I
*X •=
i i
t-
^
z
in
i
8
in
•>
E
;=
in m in in in m in
CM O 03 (O T CM O
O O O O
* IO (M •«-
1S3M
1C ID
03 (O
*£ (O
fO **"^(~i
1 1 ' ' '-
^^
-
,—
-
—
—
^
-
-
•—
-
-
(^—
-
"•
.* _
-------
1C
*
in
PJ
oo
or>
CM
ID
O
ID
K)
in
oo
in
10
1SV3
in
in
in
o
in
oo
ID
E
D
"x
C
8)
I 1
I O
C 00
"i ^
cr
o
o
• to
to
III I i I I 1 I ijII I I I I I I IIT J I IIIIIT_
O
O
t I 1 I t I I I 1 I I I I I I I I I I I t I 1 I I I I I I I I I I I I I } l~
-------
in
co
ID
in
in
-
r 1- « /\ i
- / / -
— I / —
\ / _
V— S :
"i i i i i i i i i i i i i i i i i i i t i i i i i i i i i t i i i t i i i i r
"*
o
o
tM
o
1S3M
1SV3
a
a
CD
o>
w
o
"55
.2
0)
03
O5
a
a
0)
o
N
O
«oo
1S3M
-------
0> ^
11 II
1 §
in m
CM o
08 S
CM
in
03
m
*>
J.SV3
in
in
CM
in
o
r-
in
oo
•o,
•s P
II
x '?
o .E
I
z
o
o
CM
I
'O
CD FT r~i | iiijiii iiit] i i r~j i i t i i i r~| i i i_j •*
.^M > I 1 I 1 I I I 1 I I 1 I I 1 1 I I 1 I 1 I 1 I I I I I I I I I I I I I I l~
IB
D
.0
(D
GO
O)
&
o
N
O
1S3M
-------
CM
o>
6
I §
if
in
CM
ID
or>
tM
o
»
rO
ID
r~
fO
in
o
fM
1S3M
m
<0
,°
O
•O
in
CN
*~. • *-r
» ^ CM
in
o
s
TI i j i i- i j.
23
tfi
_i
§
I
O
ID
r^
o
rv
o
fN
r*.
o
o
r»
o
s
m
ID
1SVG
in
in
CM
r-
in
o
r-
in
m
(O
-------
00
in
^r
oo [_t i |—f T T i | i i i"T~r~riT* i I I i I 1 1 ~»~
ol
§1
o
03
1^
O
(C
h-
SI i i i i i i i i i I i i i i i t i i i } i i i i i i t i i | t
(C
i ) i i i
1S3M
a
a
GO
O)
§
W
O
"55
B
o
03
O)
CL
a.
a>
o
N
O
1S3M
-------
CM
^ <0
in
CM
08
CM
00
•R O
> >
II
II
t/i
ID
I
II
in in
O co
oo P-
rO fO
in
(0
in
K)
in in in
CM o CD
^* r^ to
I I t | I I I | I I I J I I I | I I I | 1 I I | I I
in
0)
3
o
•55
.!£
0)
O
.0
in
03
I
'*-*'
9)
O
N
O
(DO
1S3M
-------
CM
II
J!
ii
in
CM
or>
CM
V
in in
o CD
oo r-
in
ID
1SV3
in
in
o
m
oo
o
o
o
o
to
r*-
o
>D
f*
O
*
r»
o
CM
r^
o
g
o
o
CM
r*
I
in
eo
in
(O
1SV3
in
in
cs
in
o
m
oo
>s_l
IT"-
V'1^-
15 *r
rv'M
S>^
el
"8
00
O5
§
-5
o
"w
.12
0)
m
00
en
0)
o
N
o
-------
Appendix D
SENSITIVITY ANALYSIS OF THE CALCULATIONS OF OZONE
CONCENTRATIONS TO WIND SPEED SPECIFICATION
8910** 8
-------
Appendix D
SENSITIVITY ANALYSIS OF THE CALCULATIONS OF OZONE
CONCENTRATIONS TO WIND SPEED SPECIFICATION
Because of the limited air quality and meteorological monitoring data in the vicinity
of Atlanta, there is some question whether the base case UAM simulation is simula-
ting all of the pertinent physical, meteorological, and chemical processes that led to
the ozone episode of June 4, 1984. The only indication of whether the model is
correctly simulating such an episode is its ability to predict the hourly ozone concen-
trations at the three ozone monitoring sites. The base case modeling inputs from
diagnostic run 2 produced model performance that barely passed the performance
goal. There were several explanations given for the descrepancies between the
predicted and observed ozone concentrations, including underestimation of the
anthropogenic emissions inventory, use of too low boundary conditions; excessive
wind shear that tended to dilute the urban plume, and excessive wind speads that also
tended to dilute the urban plume. Thus it was decided to do a sensitivity test that
would result in less dilution of the urban plume. This was accomplished by reducing
the wind speeds at all FAA wind observations sites by 50 percent.
The justification of this wind speed reduction is based on the instantaneous nature of
the observed FAA wind speeds versus the hourly integrated wind speeds required by
the UAM and the bias in the human FAA observations versus the mechanical wind
averaging techniques. When there is wind variation throughout the hour, the vector-
averaging of the hourly average wind speeds will result in a lower value than
reported at an FAA site, which is essentially an instantaneous value taken at the end
of an hour. In addition, the human element and uses of the FAA wind speeds are
different from the wind information required by the UAM. An important use of the
FAA wind observations is to advise pilots of adverse wind conditions at the airport
for take-offs and landings. During slow wind conditions, which are typical of ozone
episodes, the FAA observer may report higher wind gusts that will affect an aircraft
8910* 8
-------
operation. The fact that the hourly average wind speed is calm is of no use to the
pilot. This bias of FAA wind speeds was noted while developing UAM inputs for the
California South Coast Air Basin where there were several hourly integrated and
FAA wind observation sites located in the proximity to each other. A brief analysis
of these sites is reported next.
COMPARISON OF HOURLY AVERAGED AND
FAA WIND SPEED OBSERVATIONS
During the development of the California South Coast Air Basin (SCAB) Air Quality
Management Plan (AQMP), in which four 3-day episodes of high ozone days were
extensively studied, a systematic bias was seen between the i-hour vector-averaged
observations from SCAQMD (South Coast Air Quality Management District) surface
wind speed monitors and nearby one-minute averaged NWS/FAA (National Weather
Service/ Federal Aviation Administration) observations of surface wind speed. To
estimate the extent of the bias, seven "collocated" station pairs of NWS/FAA
stations with SCAQMD wind monitors were studied. Table D-l shows the station
pairs and the distance between the collocated pairs.
The study periods were the ozone episodes of 5-7 3une, 12-1* August, 21-23 August,
and 26-28 August 1985. A total of 1920 collocated data points were collected. This
set was reduced by 801 data points because of either a missing station pair or a
station pair that was below the speed of 1 knot. The speed of 1 knot represents the
lowest value that FAA/NWS wind monitoring stations can measure. Thus,
comparison between SCAQMD stations reporting wind speeds as low as 0.1 mph
might introduce a bias toward the higher values being reported for FAA/NWS
stations.
The mean value for the remaining 1119 data points for the FAA/NWS stations was
3.49 m/s, while the mean for the SCAQMD stations was 1.93 m/s, suggesting that the
FAA/NWS stations were biased by approximately 45 percent (i.e., the observed
hourly average wind speeds were a Ittle over half of the FAA wind observations).
The median values were very similar, with values of 3.10 m/s for FAA/NWS stations
and 1.80 m/s for SCAQMD stations. Each day was examined to see if any particular
891 OH A
-------
day showed an extreme bias; the bias ranged from 36 percent on 13 August to 51
percent on 26 August.
Similarly, all seven stations were examined to see if any particular station may have
been the cause of the bias. Five of the seven stations showed similar bias, ranging
from 45 to 56 percent. However, two pairs showed a significantly smaller bias; the
Upland and Ontario airport showed a bias of only 28.8 percent and the San Bernadino
and Norton Air Force Base pair showed a bias of only 2.4 p«rc?nt In an attempt to
see if the agreement was caused by low wind speeds at Norton Air Force Base at
night, the wind observations were stratified into a daytime period only; we still found
that almost no bias occurred for the San Bernadino-Norton Air Force pair. The
possibility exists that the data may have been incorrectly processed or reported; if
the data was incorrectly converted from knots to m/s, this may account for the small
bias.
We believe that the wind speeds reported for 1-minute averages from FAA/NWS sta-
tions are correct; however, what is needed for the UAM is the 1-hour vector-
averaged wind speed. As a result of the SCAQMD study, which should be true for all
locations in the United States, it was concluded that FAA/NWS surface wind
observations have a positive bias of about a factor of 2 and should be reduced by
approximately 50 to 75 percent when hourly average wind speeds are desired.
This preliminary analysis is by no means complete or statistically roboust, however it
does have important implications in the development of wind fields for air quality
simulations models, such as the UAM. It also helps explain some underprediction
tendency of the UAM at some locations in the past (e.g., St. Louis and
Philadelphia). Further analysis of the SCAQMD wind data base and data from other
locations is necessary to determine the extent and frequency of this .bias. For
example, hourly average wind speeds are calculated by vector-averaging of a series
of lower-frequency wind observations. A comparison of the final one-minute wind
speed for each hour with the hourly average may give some indication of the extent
of the bias.
8910«f 8
-------
MODEL PERFORMANCE EVALUATION
The UAM(CB-IV) was exercised with the exact same inputs as in diagnostic run 2,
except that in the wind fields the observed wind speeds from FAA observations were
reduced by 50 percent. The resultant predicted ozone concentrations at each of the
ozone monitoring sites are shown in Figure D-l. Similar time series plots of
predicted and observed ozone concentrations using a one-cell and two-cell search are
given in Figure D-2 and D-3. Scatterplots, residual analysis plots, and model per-
formance statistics for predicted and observed hourly ozone concentrations and the
sensitivity test (reduced wind speed) are shown in Figure D-4. The isopleths of
hourly ozone concentrations for sensitivity test are given in Figure D-5. Note that
these five figures correspond to Figures 3-4 through 3-7 in the main body of the
report and Appendix C for diagnostic run 2.
The predicted region-wide maximum ozone concentration in the reduced wind speed
sensitivity test matches the observed peak ozone concentration exactly (14.7
pphm). At the location of the observed peak ozone the model prediction (11.4) is
within 22 percent. At other ozone monitors the model predicts the peak within 2
(DKLB) and 33 (DLLS) percent. The sensitivity test produces superior model
performance over diagnostic run 2 as exhibited in the scatterplots of hourly
predicted and observed ozone concentrations (Figure D-4, and Figure 3-7 in the main
body of the report).
8910* 8
-------
TABLE D-l. Surface meteorological observation sites and distance between
collocated observations pairs.
NWS/FAA
Site Name
Burbank
Airport
Los Angeles
International
Airport
Long Beach
Airport
El Toro
Airport
Ontario
Airport
Norton
Air Force
Base
Compton
Airport
Location
UTMX
356.48
352.48
370.88
401.60
411.68
438.80
364.40
UTM (ZonelO)
UTMy
3768.00
3744.40
3733.76
3720.16
3754.16
3756.96
3740.24
SCAQMD
Site Name
Burbank
Lennox
Long
Beach
El Toro
Upland
San Bern-
ardino
Lynwood
Location
UTMX
359.60
354.40
368.00
404.8
408.00
434.40
366.40
UTM (Zone 10)
UTMy
3766.40
3744.00
3734.40
3716.72
3758.48
3759.20
3743.20
Distance
between
Collocated
Stations
3.52 km
1.96 km
2.95 km
4.70 km
5.67 km
4.94 km
3.57 km
89104 8
-------
160
12
18
160
- 120
I I I I I | I I I I I | I I ( I I I I I I I I
• CNYR
D~ PREDICTED
6 12 18
TIME (HOURS)
160
12
6 12
TIME (HOURS)
160
- 120
18
24
SYSTEMS APPUCXnONS. INC.
FIGURE D-l. Observed and predicted ozone concentrations (ppb)
for Atlanta reduced wind speed -sensitivity test.
89104
-------
160
160
- 120
160
120 -
12
18
6 12
TIME (HOURS)
18
160
- 120
- 40
24°
SYSTEMS APPLICATIONS. INC
FIGURE D-3. Observed and nearest-neighbor predicted (two-cell
search) ozone concentration (ppb) for the Atlanta reduced
wind speed sensitivity test.
8910U
-------
120.00 -
90.00 -
o
Q.
Q.
u 60.00 -
o
UJ
K
Q.
30.00
i i i i I i i i t I i i i i I i i i
30.00 60.00 90.00
OBSERVED (ppb)
120.00
MOMENTS OF THE PROBABILITY DENSITY FUNCTION
OBSERVED
AVERAGE
STANDARD DEVIATION
SKEWNESS
KURTOS1S
OTHER MEASURES
MEDIAN
UPPER OUART1LE
LOWER OUART1LE
MINIMUM VALUE
MAXIMUM VALUE
50.39999
47.63355
0.51320
-1.22432
40.00000
86.00000
5.00000
5.00000
147.00000
PREDICTED
56.68703
34.20863
0.52057
•0.84048
46.90000
82.56000
31.13000
0.02000
126.60000
SKILL OF PREDICTION PARAMETERS
CORRELATION COEFFICIENT OF PREDICTED
VERSUS OBSERVED 0.904
THE BOUNDS OF THE CORRELATION AT THE
CONFIDENCE LEVEL OF 0.050 ARE
LOH BOUND 0.844 HIGH BOUND 0.942
RATIO OF OVER TO UNDER PREDICTIONS 1.308
PERCENT OF OVER PREDICTIONS
GREATER THAN 200 PERCENT OF THE
OBSERVED 41.667
PERCENT OF UNDER PREDICTIONS
LESS THAN 50 PERCENT OF THE
OBSERVED 1.667
FIGURE D-4a. Scatterplot and nodel perfonrance statistics for hourly
ozone concentrations and Atlanta reduced wind speeds sensitivity
test (N = 60).
89104
-------
0.20 -
-32.00 -16.00 0.00
RESIDUAL fOBS-PRED)
THE B1NS1ZE EQUALS 6.000
16.00
32.00
RESIDUAL ANALYSIS
AVERAGE -6.28719
STANDARD DEVIATION 22.19390
SKEWNESS 0.10702
KURTOS1S -1.19285
OTHER MEASURES
MEDIAN -6.30000
UPPER OUART1LE 9.57001
LOWER OUARTILE -26.13000
MINIMUM VALUE -42.56000
MAXIMUM VALUE 35.00000
BIAS CONFIDENCE INTERVAL
AT THE 0.0500 LEVEL
LOWER BOUND -18.9385
UPPER BOUND 6.3641
STD RESIDUAL CONFIDENCE INTERVAL
AT THE 0.0500 LEVEL
LOWER BOUND 373.7107
UPPER BOUND 684.3732
THE MEASURES OF GROSS ERROR
THE ROOT MEAN SQUARE ERROR IS 22.89
THE AVERAGE ABSOLUTE ERROR IS 19.52
VARIOUS MEASURES OF RELATIVE VARIABILITY
OBSERVATION COEFFICIENT OF VARIATION
0.9451
RESIDUAL COEFFICIENT OF VARIATION
0.4404
RATIO OF RESIDUAL TO OBSERVED ST. DEV.
0.4659
FIGURE D-4b. Residual analysis plot and nodel perfornance
statistics for hourly ozone concentrations and Atlanta reduced
wind speeds sensitivity test.
8910«+
-------
62.3
45.3
30.2
22.2
12.4
5.3
_ -6.3
o
LU
£ -17.9
J> -25.0
CD
? -34.7
< -42.8
£ -57.9
-74.9
O D D
CD O
n o D n D ° Q
Q
B
D B
-D
- B
B
B B
_
B
B Q
Q
B
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 I 1 1 1 1 1 1
5.00
HOUR
10.00
15.00
20.00
THE LINEAR MODEL PARAMETERS
THE CORRELATION IS 0.5829
THE LOWER BOUND IS 0.3861
THE UPPER BOUND IS 0.7290
AT THE 0.0500 PERCENT LEVEL
THE Y-X LINEAR MODEL INTERCEPT IS 11.460
THE Y-X LINEAR MODEL SLOPE IS 0.153
THE X-Y LINEAR MODEL INTERCEPT IS -29.647
THE X-Y LINEAR MODEL SLOPE IS 2.225
FIGURE D-4c. Plot of" residuals versus tine of day for hourly
ozone ounsentrations and Atlanta reduced wind speeds
sensitivity test.
8910»*
-------
m
CM
00
» 6
» II
!ht
II o
in
o
s
in
oo
r>
X •=
o .E
oo —
rv
o:
o
*~ ru
12 "
8
™ o
' g
8
n i i » i .j
in
10
r-
1SV3
in
in in
CM o
TV. r-.
in
co
ID
in
-------
o
II
«
"5
|
I
o :i
in
CM
00
05
§
-3
to
o
"55
.!2
o>
8
D
^
"5 .
£ 5
il
a .0
Q. O
v**'
! ^
o 1
^ »
o ..
1S3M
-------
in in in in
tN O 00 (O
eo eo r- r»
>O if) 10 t*5
in
in
CM
in
o
in
00
in
*• CM o
m
ao
E
a
a
o
N
O
in
J.S3M
-------
I s
1 1
ll
in
(N
oS
cs
03
in in
o GO
oo r>»
in
(O
1*5
1SV3
in
m
CM
fv
K)
in
o
in
03
ID
O
I I I I 1 I \ I | T I T } II I | J1 I | I T I (T
(O
too
i > i » i i i i i I i t i i i t i t i 1 i i i i
in
(O
(O
^f
JLS3M
Q.
a
O)
" it
I 5
l
in
m
o
m
oo
in
f*.
IN.
10
m
CM
in
O
£
«o
CD -1 -I 1 | 1 I I. J,. I I a J J. II J .J
•S o
in
«o
^o
1 J 1 4 1 j I 1 1
i i*2 2
CD
0
gl>»<
! i ! I ("t i t I t r i i i f i i :i i i
t..\..f
8
1S3M
00
O)
i
-3
in
o
"w
w
E
9
D
.O
ID
(D
O)
O
-------
1SV3
II
If
S.E
Z 2
U1
_J
o
8
o>
V
p
in m
CM O
CO CD
OO f)
00
o
o
OD
o
co
r^
o
o o -o
t | I i 1 | I I I | i i i | i
0 #$
V/' 1- . CD ^ *"
*•*- g'
£L 0 «5 °
H*^' i '
03 jfruj»J
§• ID •
0
tc ^
h» ^*
^ •
t/)
f
i- 0
1 1 1 1 1 1 1 1 f f t 1 f 1 1 1 ! 1 1 1
m in
o co
fv to
•0 O
I i i i i | i
i
V
•TV
/ ;
\ /r
1 1 ! 1 ! 1 I 1 I
in
to
(O
"o
1 i _
—
n
—
—
—
I
—
^
-
™
'
^
_
• —
. ^ ,
3
" • •"•
V
S
S
0
<00
in ^
11 i
I I
£|
J f
X •=
O ^
2 'i
in
CM
CO
or>
CM
CO
o
g
0
00
o
s
1S3M
1SV3
in
m
CM
' 1 ' 4.
in
o
' • ,1 '
in
D
ID
in
'igps^i^ii^in^^i^^^pif^ts?1^^'"
*i i 1 't i t a t T I :r-i'^'iN :t Tipir1':r'i^t^-:F;f'? ^^I'-ii-ii^f -i t' i'l-.l t
D.
Q.
00
O)
0)
3
-5
en
o
'w
i/>
*E
0>
00
2 1
X
8
0
N
O
in
IS3M
00
-------
)
/-
-------
1C
« fM
in
CM
00
o
00
m
co
in
to
1SV3
in
m
CM
in
o
in
oo
in
to
to
00
11
•s P
If
ii
tr
O
o
o
' §
I'll I
'L
g Ii i i i i t i i i t i i i i i i i i i ! i > i i i \ t r i t i i i i i i i i i~
(OO O
V)
o
J.S3M
Q.
Q.
1SV3
o
O Is;
:i
2
in
0%
oo » i t | i i i i i i
V)
—I
1
I
§
IN
in
o
s
m
00
m
to
in
10
in
CM
in
o
10
in in
oo to
(0 (0
I J,. i .« J. .1 i i .J < i »_
«o
00
O)
0)
3
O
'55
(0
"E
9)
m
CD
05
E
I
O
N
o
in
B *
1S3M
-------
1SV3
to
M
I/)
o
o
in
(M
oo
oo
M
00
38 o
51 §
IE "
-I o
5 c oo
22 ^
O i
ID I—I
(OO
o
00
If)
oo
in
to
r-
in
in
CM
r-
in
o
oo
<£)
in
to
' >•. I. > i .'„! ', l..J.».J
03
O5
(D
3
05
O
"w
o
X)
oo
O)
.c
a
a
§ 5
s ?
(DO
1S3M
-------
in
(Nl
oo
in
o
in
03
ID
in
in
CM
in
o
in
GO
(O
in
(O
«o
CM
oo
I 5
I E
.11
o
O
00
o
co
rv
o
to
r*-
I*
i i i I i i i i i i i i i i i i i i i i i i i i i i i I i i
o
to
too
o
CM
1S3W
I
a
a
in
II
II
If
en
_j
§
I
o
00
r«
o
3
-5
(fl
O
'5
w
o
.o
in
CD
o>
/—v .2
I
Q. O
Q. U
^x
0) .
C if)
O I
N Q
O ..
J-
o
1S3M
-------
o
*>
in m
o CD
(s. <0
I §
I I
11
'x •=•
o .E
o:
O
01
_i
o
o
o
CM
0>
__» I I I I I i I I I I I 1 I I 'I I I r JJ I 1 | I I T] TT
TD*
o
CN4
'*:'- •'
i .-i i t i i .1 i -i t i: i i . r i i .t i i I i t i i > i i t it t t r t \ -\ t » JL
®o
o
fM
1S3M
1.1 * i T i t i t j i 1.1 i
00
O)
§
w
o
"55
m
CO
Ol
&
o>
o
N
o
I
in
(D
«DO
-------
Appendix £
MEMORANDUM OF 10 MAY 1989 ON THE SELECTION
OF A UAM MODELING EPISODE FOR THE
DALLAS-FORT WORTH METROPLEX AREA
-------
MEMORANDUM
TO: Richard Scheffe, EPA/OAQPS
Les Montgomery, TACB
Jim Yarborough, EPA Region VI
FROM: Ralph Morris, Tom Myers, and Ed Carr, Systems Applications, Inc.
DATE: 10 May 1989
SUBJECT: EPA Five Cities UAM Study - Selection of a UAM modeling episode and
modeling domain for Dallas-Fort Worth, Texas.
This memorandum provides information for the selection of a UAM modeling episode
and domain for the Dallas-Fort Worth metroplex area. At this point, we would like
to solicit input from TACB, EPA/OAQPS, and EPA Region VI concerning the
definition of the final episode and modeling domain.
SELECTION OF A UAM MODELING EPISODE
A multiday ozone episode will be chosen for the Dallas-Fort Worth area based on
observed air quality and meteorological conditions from the years 1984 through
1987. The modeling episode will be selected on the basis of the following criteria:
High and widespread ozone concentrations;
No atypical meteorological conditions; and
Completeness of routine data.
The selection process consists of the following activities:
Analysis of observed elevated ozone days (ozone concentrations greater than
the ozone National Ambient Air Quality Standard (NAAQS) of 120 ppb) in the
Dallas-Fort Worth area for the years 1984 to 1987;
Definition of possible UAM modeling episodes from grouping of highest
observed ozone days;
Examination of the number of station-hours within each possible modeling
episode exceeding the ozone NAAQS, and elimination of candidate episodes
with low number of station-hours of elevated ozone concentrations;
-------
Scheffe, Montgomery, Yarborough
10 May 1989
Page Two
Analysis of meteorological conditions for the candidate ozone episodes.
Perform meteorological classification to determine most typical conditions
that lead to elevated ozone episodes in the Dallas-Fort Worth area;
Determination of routine data availability (meteorological and air quality) for
candidate modeling episodes; and
Recommendations of possible UAM modeling episodes.
Starting with 106 days of ozone exceedences from the years 1984 to 1987 (and parts
of 1988) we have identified three modeling episodes as candidates for the UAM
modeling episode. The following paragraphs describe this selection process.
Elevated Ozone Observations
Ozone monitors in the Dallas-Fort Worth area that were in operation during the
period of 1984 to 1988 are listed in Table 1. The locations of these monitors are
given in Figure 1. Also given in Table 1 is the period of operation for the 14 ozone
monitors. Many of these sites were in operation for only a few months in 1986
(FRNK, ERWN, and COIT), 1987 (DAIR), or 1988 (WTHR and FATE). However, there
were 6 monitors in continuous operation during the years 1984 to 1987 (NSTR, ILLI,
BNVW, DNTN, ROSS, KLLR, and the combined record of MOCK and HNTN which
were located close together).
For the period of record that we obtained from the. AIRS data base, which includes
1984 to 1987 and parts of 1988 for some monitors, there were 106 days of
exceedences of the ozone NAAQS. The rankings of these exceedence days, the
monitor where the highest ozone was observed, and the maximum daily ozone
concentrations at the other monitors are given in Table 2. By analyzing the top 15
percent of days with ozone exceedences (i.e. the top 16 days) we have identified 12
possible UAM modeling episodes that are listed in Table 3. Table 3 also contains the.
number of station hours for each day in which the observed ozone concentration
exceeds the NAAQS. Since we wish to pick a modeling episode with wide spread
occurrence of elevated ozone concentrations (i.e. we do not want to model an ozone
exceedence day that just occurs at one monitor and may be due to local effects) we
eliminated the days from Table 3 which have less than 15 station-hours of ozone
concentrations in excess of the ozone NAAQS. This results in the six candidate UAM
modeling episodes that are listed in Table 4.
-------
Scheffe, Montgomery, Yarborough
10 May 1989
Page Three
Meteorological Characterization of Candidate Modeling Days
The meteorology for the high ozone days from Table 4 have been reviewed. The
following is a brief discussion of the meteorology on each of those days and
recommendations as to it's suitability as a modeling day.
31 July - 3 August, 1984
The dominant feature during this 4-day period (Tuesday-Friday) was the persistent
upper-level low pressure system centered along the Oklahoma-Texas border. This
persistent upper-level low suggests that elevated levels of ozone and precursors may
have been the result of transport from upwind areas. Between the upper levels (at
700 mb approximately 3000m) and the surface, a wind shear in direction of 180
degrees is present throughout the 4-day period. The surface winds are generally
light, with the direction predominately from the southeast. Recommendation; The
transport of ozone and precursors is probably significant on this day. The multiday
nature of the ozone episode would require modeling many days and a larger region,
resulting in a need for more computer resources than required for other episodes.
Therefore, this episode is probably not appropriate for demonstrating the low-cost
application of the UAM.
31 August, 1985
On Friday the 30th of August an upper-level high pressure ridge was located
approximately 800 km to the west. The upper level high pressure continued to
dominate the region on the 31st and on into the 1st of September. The surface
pattern showed light and variable winds throughout the 3 days. At 7 a.m. on 31
August, surface winds were light and from the northwest, while southerly winds were
measured aloft. By 7 p.m., the surface and winds aloft were light and from the
south. The maximum temperature climbs during this three day period from 96 to 100
to 103. No frontal passages or troughs were present seen throughout the period. It is
speculated that the 30th served as a buildup period for the high ozone values seen on
the 31st. No significant break in the meteorology occurred during this period. The
most likely reason the maximum ozone concentration was lower on the 1st was that
the precursor emissions were probably much lower on Sunday the 1st of September
resulting in much lower observed ozone values. Recommendation: This is a possible
modeling day, however, the only reservation is that the emissions for Saturday may
be more difficult to characterize than normal weekday emissions. In addition, the
Saturday emissions may be atypical and not representative of conditions that occur
during weekday periods.
-------
Scheffe, Montgomery, Yarborough
10 May 1989
Page Four
25 June, 1984
Surface high pressure dom.'hated the region on the 24th - 25th. Upper level winds
were light and vai iabie from the east Lc northeast with speeds less than 9 knots
throughout the planetary boundary layer. These nearly stagnant conditions at the
surface and aloft allowed the high levels of ozone to develop on the Monday the
25th. Recommendation; This is a possible modeling day, no unusual meteorology
occurs during this time period and ozone is locally produced. Again the only
reservations are that the emissions on Sunday the 24th (the initialization day) may be
more difficult to estimate than normal weekday emissions.
3 September, 1985
The transition of tropical Storm "Elena" to an extratropical cyclone created the
conditions necessary for the high ozone episode. The upper air flow near Dallas on
Monday the 2nd showed winds from the northeast. These winds became stronger and
more southerly on the 3rd when the extratropical cyclone regressed towards the
Texas-Louisiana border. Recommendation; This should not be used as a model day,
it does not appear to be a typical high ozone episode case because of the influence of
the tropical storm "Elena".
9 September, 1987
A fairly complex surface pattern existed on Wednesday morning the 9th of
September with a low pressure trough located to the NW near Amarillo, Texas. The
low deepened during the day which resulted in areas of widely spread precipitation
just north of Dallas the following day. Surface winds were generally light and from
the northwest while upper air winds were more westerly. Ozone generated on this
day was most likely locally produced. Recommendation; This would not be the best
choice for a modeling day, because of the previous day precipitation and the complex
flow that developed on the 9th.
1 August, 1986
The three day period (30 July to 1 August, 1986) from a Wednesday - Friday was
characterized by a surface high pressure and an upper-level high pressure system.
This resulted in three consecutive days in which the temperature reached a maximum
of 106 F. Upper-level air flow during this three day period was generally weak with
most winds under 10 knots below 700mb. Surface flow was from the southwest to
-------
Scheffe, Montgomery, Yarborough
10 May 1989
Page Five
southeast. Passage of a very weak cold front on the afternoon of the 1st and the
migration of the upper level high eastward resulted in a large increase in the
ventilation with wind speeds of 12 knots or more throughout the layer from the
surface to 1500 meters ending the stagnation episode. Recommendation; This looks
like a possible choice for an ozone modeling day because the meteorology appears to
be typical of a high ozone day and the conditions show that transport of ozone and
precursors are of minimal importance. However, the frontal passage that occurs on
1 August may be difficult to characterize the transition of the upper-level winds
with only the routine twice daily upper air observations.
Data Availability and Discussion
Routine meteorological data was available for each of the six candidate UAM
modeling episodes listed in Table 4. Figure 2 displays time series of ozone
concentrations at each of the 1* ozone monitors for the days from the six candidate
modeling episodes. The spatial distribution of the observed maximum daily ozone
concentrations for each of the candidate modeling days are given in Figure 3. All of
the candidate episodes had at least 7 ozone monitors in operation for some of the
time with the August 1, 1986 episode having up to nine ozone monitors in operation.
The meteorological analysis has identified three modeling days as most suitable:
June 25, 1984; August 31 1985; and August 1, 1986.
June 25, 1984
As seen in Figure 2f, some observed hourly ozone concentrations were missing from
the seven ozone monitors in operation on June 25, 1984. The stagnant conditions on
this day resulted in the occurrence of high observed ozone concentrations (150 ppb)
in both Dallas and Fort Worth (see Figure 3f). The 150 ppb observation in Fort Worth
occurred at noon, which indicates it may be partially due to the previous days ozone
cloud that is entrained into the mixed-layer as the mixing height rises. The
maximum daily ozone concentration (150 ppb) for Dallas occurred in the late
afternoon, thus it was most likely locally produced. The twice daily upper-air
soundings for June 25, 1984 show upper-level winds that are fairly persistent from
the east to northeast at all levels. The upwind ozone monitor (DNT2) measured a
maximum daily ozone concentration of 90 ppb which persisted throughout the day
from 9 in the morning to 6 in the afternoon. Thus, sources upwind of the Dallas-Fort
Worth metroplex area, or recirculation of emissions, may be contributing 90 ppb to
the peak observed ozone concentrations.
-------
Scheffe, Montgomery, Yarborough
10 May 1989
Page Six
August 31, 1985
As noted previously, August 31, 1985 is a Saturday at the end of a three-day ozone
episode. There are most likely differences in emissions levels and a'l-:.. ibution on this
day due to differences in weekend industrial activity and traffic patterns. As seen in
the time series (Figure 2e), the seven ozone monitors in operation have a complete
data record and the maximum daily observed ozone concentrations occur in the
afternoon. This indicates the possibility that the ozone was most likely formed from
emitted precursors on this day rather through the entrainment of concentrations
aloft that may have been formed earlier. Winds are predominantly from the south,
and the upwind ozone monitor (BNVW) observes relatively low ozone concentrations
with a peak of 80 ppb (see Figure 3e). The August 31, 1985 is mainly a Dallas ozone
episode with high maximum daily ozone concentrations (170 ppb) occurring at two
monitors (MCK3 and DNTN). Although ozone concentrations are lower in Fort
Worth, ozone exceedences are observed at both ozone monitors there (130 ppb).
August 1, 1986
During the August 1, 1986 ozone episode there are more ozone monitors in operation
than the other candidate episodes (see Figure 2i), however two of the monitors
upwind or in Dallas (HNT2 and BNVW) are missing afternoon ozone observations. The
frontal passage that occurs on this day results in a turning of the winds from the
southwest at 7 a.m. to the southeast by 7 p.m.. The westerly component in the
morning wind may result in the transport of ozone precursors from Fort Worth to
Dallas. There is an orderly buildup of maximum daily ozone concentrations going
from south to north through Dallas. Fort Worth does not observed and exceedence of
the ozone standard on this day.
-------
FIGURE 1. Location of ozone monitors in the Dallas-Fort Worth region.
-------
12
18
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.2O
0.15
0.1O
0.05
0.00
0.2O
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
k MOO
L B B
^
1— NST2
S — B
g
gj
M- MJJ
H—H—B-
1- ILL2
1- BN/W
0.15
0.10
o.osb
n
0.00
a—a—a—a—B—B
I I I
i i i
6 12 18
. JULY 31, 1984
F;igure 2a. Observed ozone concentrations (pprn) for
Dallas - Fort Worth Area
-------
12
f"
18
24
0.15
0.10
O.O5
O.OO
0.20
0.15
0.1O
0.05
0.00
0.2O
0.15
0.1O
0.05
0.00
0.20
0.15
0.1O
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
O.1O
0.05
O.OO
0.20
0.15
0.1O
0.05
0.00.
_ corr
>- DNTN
- DNT2
3. m m
-B—B
B—H—B—B-
•a—B
- ROS2
~J3—B—Eg,
- KLLR
- DAIR
j i
i i I i i i i t
i t I i i i
6 12 18
- JULY 31, 1984
Figure 2a. Observed ozone concentrations (ppm) for
Dallas - Fort Worth Area
24
-------
12
ia
0.2O
0.15
0,10
0.05
0.00
0.20
0.15
0.10
-O.O5
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
O.OO
0.2O
0.15
0.1O
0.05
0.00
0.20
0.15
0.10
0.05
0.00
I i I
r
_ EBWN
- MOO
B—B-
- NST2
- ILLJ
- ILL2
a 13 m
•e-
J I
6 12 18
AUG 1, 1984-
Figure 2b. Observed ozone concentrations (ppm) for
Dallas - Fort Worth Area
24
-------
0.20
0.15
0.10
0.05
O.OO
0.2O
0.15
O.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
O.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.1O
0.05
0.00.
18
I i ii " i i
_ car
— DNTN
— DNT2
- ROS2
r a o-
- KLLR
j
Figure 2b.
6 12 18
AUG 1, 1984-
Observed ozone concentrations (ppm) for
Dallas - Fort Worth Area
24
-------
0
0.20-
0.15-
0.10-
O.O5-
12
24
0.00
0.20 j-
0.15-
0.10-
0.05-
o.oo
0.20
0.15
0.10
0.05
O.OO
0.20
0.15
0.10
o.osb
- MCK3
-B—B—B-
- NST2
o.oo
0.20
- ILU
- ILL2
0.15
0.10
0.05
O.OO
0.2O
0.15
0.1O
0.05
O.OO
6
12
AUG 2, 1984
18
24
Figure 2c. Observed ozone concentrations (ppm) for
Dallas - Fort Worth Area
-------
12
18
0.20
0.15
0.10
O.O5
O.OO
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.1O
0.05
0.00
0.20
0.15
O.1O
0.05
0.00
0.20
I i
_ COT
- DNTN
- DNT2
- ROS2
e—s—B-
-a.
- KLLR
0.15-
0.1O-
0.05-
i i i i I i i i i i I i i i
1 i i i
6 12 18
AUG 2, 1984-
Figure 2c. Observed ozone concentrations (ppm) for
Dallas - Fort Worth Area
-------
12
0.20
0.15
0.1O
0.05
0.00
0.20
0.15
0.1O
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
O.O5
O.OO
0.20
0.15
0.10
O.05
0.00
0.20
0.15
0.10
0.05
0.00
18
~1 T"
_ ERWN
- MCK3
^g—&-^-B-
NST2
- ILL!
"g
•a n a p.
- 102
- BNAV
-H—63
i i i
j i
6
12
AU6 3, 1984-
18
24
Figure 2d.
Observed ozone concentrations (ppm) for
Dallas — Fort Worth Area
-------
0.20
0.15
0.10
0.05
O.OO
0.20
0.15
0.10
0.05
O.OO
0.20
0.15
0.1O
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.2O
0.15
0.10
0.05
0.00
0.20
III!
12
~I T
18
24
_ COT
- DNTN
- DNT2
-S—H—B—El _
- RCS2
- KLLR
0.15
O.JO
0.05 K.
0.00
0.20
- I>JR
0.15-
0.10-
0.05-
O.00_
i i i i i I i i i
i I i i
i i I
j t
i i
6 12 18
AUG 3, 1984-
Figure 2d. Observed ozone concentrations (ppm) for
Dallas - Fort Worth Area
-------
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
O.OO
0.2O
0.15
0.10
O.O5
0.00
0.20
0.15
0.1O
0.05
0.00
0.20
0.15
0.1O
0.05
O.OO
0.20
0.15
O.1O
0.05
O.OO
12
_.
18
I I
_ COT
- DNIN
- DNT2
- ROS2
- KLLR
j i i i
ll i
t I
j I
6
12
AUG 31, 1985
18
Figure 2e. Observed ozone concentrations (ppm) for
Dallas - Fort Worth Area
-------
12
0.20
0.15
0.10
0.06
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
O.JO
0.05
O.OO
0.20
0.15
0.10
O.O5
0.00
0.20
0.15
0.10
0.05
O.OO
0.20
0.15
0.1O
0.05
O.OO
0.20
0.15
0.1O
0.05
_ FRNK
T j I I T
24
_ ERWN
- MOO
- NST2
- ILLI
- ILL2
&—&T
0.00,.
I I I I I I
6 12 IB
AUG 31, 1985
Figure 2e. Observed ozone concentrations (ppm) for
Dallas - Fort Worth Area •
-------
0.2O
0.15
0.10
0.05
0.00
0.20
0.15
0.10
.0.05
O.OO
0.20
0.15
0.10
O.05
O.OO
0.20
0.15
0.10
O.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
r- FRNK!
12
1 1 T
18
_ ERtfN
- MO<3
B—B-
- NST2
j—-a—B—B-
- ILK
5—B-*
- ILL2
B—El -
0.00,
i i i
18
24
JUNE 25, 1984
Figure 2f. Observed ozone concentrations (ppm)
Dallas - Fort Worth Area
for
-------
12
18
24
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.1O
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
O.15
0.10
0.05
O.OO
0.20
0.15
0.10
0.05
0.00
- COT
r- DNTN
- DNT2
i- RCS2
- KLLR
•a—H—B-
3F-H—B
- OJR
I I 1
1 t I I
I I
6
12
JUNE 25, 1984
18
Figure 2f.
Observed ozone concentrations (ppm) for
Dallas - Fort Worth Area
-------
12
IS
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
O.OO
0.20
0.15
0.10
0.05
O.OQ
24
- MOO
-g g p &-=»<
- NST2
J—S—H—B—B-
- IUJ
- ILL2
•{DEE D~-^jn m.
-a
B
r- BN/W
m m m.
I I
6 12 18
- SEPT 3, 1985
Figure 2g. Observed ozone concentrations (ppm) for
Dallas - Fort Worth Area
24
-------
12
18
0.2O
0.15
0.10
0.05
O.OO
0.20
0.15
0.1O
0.05
O.OO
0.20
0.15
0.10
O.O5
O.OO
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
O.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
O.05
- HNT2
I 1 T
U COT
— DNTN
E—B—B—B-
- DNT2
- ROS2
•a-
-&.
m
- KLLR
m m m m n-
— DAIR
0.00
I I I I I I I I I I I I ( I I I I I ( I I t I
6 12 18
SEPT 3, 1985
Figure 2g. Observed ozone concentrations (ppm) for
Dallas - Fort Worth Area
-------
0.2O
0.15
0,10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.1O
0.05
0.00
0.20
0.15
O.MO
0.05
O.OO
0.2O
0.15
0.10
0.05
O.OO
0.20
0.15
0.1O
0.05
0.00
0.20
0.15
0.10
0.05
0.00
I I I
12
"I I T
18
24
_ EFWN
- MOO
- NST2
r n n B-
- ILL!
- ILL2
- BhVW
n n n
6
12
SEPT 9, 1987
18
Figure 2h. Observed ozone concentrations (ppm)
Dallas - Fort Worth Area •
for
-------
18
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
O.O5
0.00
0.20
0.15
0.10
0.05
O.OO
0.20
0.15
0.10
O.05
O.OO
0.20
0.15
0.10
0.05
0.00
L!*—«JB_^
_ COT
- DNTN
- DNT2
- ROS2
S—S
- KLLR
n
B
-B"
i
-B—E9—B—B
i i i i I
n ED
n -
I
I I
6
12
SEPT 9, 1987
18
24
Figure 2h. Observed ozone concentrations (ppm) for
Dallas - Fort Worth Area
-------
0.20
0.15
0.10
0.05
O.OO
0.20
0.15
0.1O
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
O.2O
0.15
0.1O
0.05
0.00
0.20
0.15
0.10
0.05
0.00
I I
12
T 1 T
18
24.
i I I r
_ EFWN
-E9—B—E&-
- MOO
- NST2
- ILL!
O G3 ED P —
- ILL2
- 8WW
Q—€1 _
i
I I I
6
12
AUG 1, 1986
Figure 2i. Observed ozone concentrations (ppm) for
Dallas - Fort Worth Area
-------
12
18
0.20
0.15
0.10
0.05
O.OO
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
0.20
0.15
O.tO
0.05
0.00
0.20
0.15
O.1O
0.05
i i i r
_ COT
B—B—B
- DNTN .
e—a
- DNT2
- ROS2
- KLLR
0.00
0.20
0.15
O.tO
0.05
0.00;
i i i
i I t i i i
I i i
i i i j i
6 12
AUG 1,"1986
Figure 2i . Observed ozone concentrations
Dallas - Fort Worth Area
18
for
-------
375C
3700-
3650
3600
3550 -
350
600 650 700 7bO
800 850
i I i i i i i i i i i I 3'
750
3700
- 3650
- 3600
- 3550
500
Figure 3a. Observed Peak Ozone Concentrations (ppb) for
Dallas - Fort Worth Area
7/31/84
-------
375C
600 650 700 750
3700-
3650 d
800 850
•ii i i i iii i iii( i | 3
750
3700
- 3650
- 3600
- 3550
600 650 700
.\l i i i . i ! , i l^~, A .11,1 ^-rH -
750 800 850
500
Figure 3b. Observed Peak Ozone Concentrations (ppb) for
Dallas - Fort Worth Area
8/ 1/84
-------
3758
600
3700-
3650 J
3600=i
3550 -
350Q
- 3550
550
600
650
700
750
800
85?
500
Figure 3c. Observed Peak Ozone Concentrations (ppb) for
Dallas - Fort Worth Area
8/ 2/84
-------
375C
600 650 700 750
3700-
3650=^
3600-
3550 -
600 650,
•^ i ii [ i 11 ii i i i i O
35
°850
;750
3700
- 3650
600 650 700
Figure 3d. Observed Peak Ozone Concentrations (ppb) for
Dallas - Fort Worth Area
8/ 3/84
-------
375'
3700-
3650 d
3600:i
3550-
750
3500^1 ' ' '
5oO
3700
600 650 700 750 800 65(
Figure 3e. Observed Peak Ozone Concentrations (ppb) for
Dallas - Fort Worth Area
8/31/85
-------
750
3700
- 3650
- 3600
3550
500
Figure 3f. Observed Peak Ozone Concentrations (ppb) for
Dallas - Fort Worth Area
6/25/84
-------
700 750
35
600 850
i I i I i i i i I I i I 3
750
3700
- 3650
- 3600
- 3550
500
Figure 3g. Observed Peak Ozone Concentrations (ppb) for
Dallas - Fort Worth Area
9/ 3/85
-------
375'
>50 600 650 700
50
3700-
3650
600 S5Q
I I I ! I I il II] O i wVJ
3700
- 3650
- 3600
- 3550
500
Figure 3h. Observed Peak Ozone Concentrations (ppb) for
Dallas - Fort Worth Area
9/ 9/87
-------
375
3700-
3650^
3600-
3550 -
i i i i i i i i i i i i i i 375C
35°§50 600
3550
650 700
750
800
500
Figure 3i. Observed Peak Ozone Concentrations (ppb) for
Dallas - Fort Worth Area
8/ 1/86
-------
oo
00
ON
8s
0> o>
OL >.
•o ~
.2 §
NO
oo
NO
oo
00
ON
I
NO
oo
NO
00
•3-
00
00
00
NO
I
•a-
00
oo
00
oo
oo
.3-
00
00
00
NO
oo
00
NO
00
00
CM
•3-
00
00
00
00
00
oo
00
00
oo
00
I
*
00
00
ON
00
•a- — —« ^ —«
m
Q.
en
N.
SO
O
oo
ON
Ol
CO
0)
O
U.
I
1
a;
.5
c
CO
I
to
O
UTM Coordinats* (Zone 14!
UTMX UTMY
ON
NO
ON
NO
0
i
U
C
"o
U
^-
(0
4)
1
lo
d Frankford Cemet
6
(A
CO
i
718.322 3680.889
"o
U
win Central Park
L.
LU
i
i
00
_ __•
NO
NO
ON
ON
ON
to
JO
"3
Q
^3
Jj
CQ
00
S
•8
o
•fj
CO
§
en
u
rf
!*£
u
00
NO
0
NO
<»
o
to
J2
Q
532 Nuestra Drive
CNI
P
Z
CNJ
NO
ON
NO
NO
ON
NO
to
JS
"c3
Q
^3
*«4
i
—
O*
04
H
Q
CO
"c"
CO
"c"
1
a
&
jij
"o
U
^^
CO
3
jatherford Agricult
«v
>
j
-2.
-2.
"3
J£
8
cd
u
CO
a
c
CO
OJ
s
•M
I
o
«*N
1
O
cJ
NO
o
00
NO
•t-»
CO
(0
H
•
c/1
p*
^2
*
0Q
BO
C
>ss Ave between Lo
e
Qi
(N
lo
i
oo
NO
r-.
o
ON
O
NO
NO
4-»
i
u
(0
H
•o
od
^J
*M<
^-
CO
««-4
O
WH
0
(U
*-•
lo
u.
1
f^^
^|>
d
00
00
NO
V
to
*
2
"HI
£
i_
<
1
s
i
g
S
tu
_J
CO
<
Q!
U.
Qi
UJ
01
^
3
.j
g
2
efi
Z
Z
Z
fc
o
u
Z
Z
u
Ci
H
&
HI
^*
uu
t^
o
ai
a!
j
N^
£
^
Q
-------
in
O —
oo
in
^i
OCSJ
•— z
oo
OOC
Ot—
OO
rou)
— X
*-^
00
r">u>
CO
Ol
e
CO
00
CD
oo —»-:
f« IOCO
OO:*:
ou
£ E
i-
O
I/I
IB
!.
O
ID
O
0)
c
o
M
O
ECO
•— oz
XO2
O
CO
inoe
oou.
O
oo
0)
CM
CD
•*—'^*COCnO>COi&cr>CftCD
ii ill
i i i I i i i i i i i I i I i I r I I I i I i I i i i i i I i I I i I I i I i I i i i i i i
i l < i i i i l i i i l l l i l < l l l i l i l i i i l l l l I i l i l i l f i i l i l i l
i I i I i I t i
O^ Ol CD CYl *~^ Cft Cft Ol Cft Ot O^ Cft PNJ CD CD CD CD CD CD CD CD CD CD CD CD CD t0 *^CD CD CD CD CD CD CD CD CD 40 CD CD CD CD CD CD CD CD
CD CD CD CD 10 CD CD CD CD CD CD CD 7 CD CD CD CD CD CD CD CD CD CD CD CD CD r*> CD CD CD CD CD CD CD CD CD CD *•* CD CD CD CD CD CD CD CD
llll^-lllllll~lllllllllllll»-« I I I I I I I I I <—i l I I I I I t I
a»o
«B
oc
oooooooooooooooooooooooooooooooooooooooooooooo
-------
in
OCSJ
Ot—
— z
in
OZ
tMO
CO
VD
OO
o —
mo
oo
<£>
OCM
CO
CO
Ol
Ol
vO
o:
oo
^
Ol
c
Oin
I OOCSI
CO
Ol
i..
o
ID
o
s-
o
ID
o
ai
c
o
at
01
CM
CO
— OX
xo>
— 'CD
CO
LH
OCM
Ol Ol O> Ol Ol O1
I I I I I I I I
O> Ol Ol Ol C^ Ol Ol Ol Ol Ol Ol Ol fNJ Lf> rSh ^ Ol Ol Ol Ol Ol Ol Ol Ol Ol Ol I"1) C2 ^^ Ol Ol Ol OlOO) Ol <**) ^» C^ Ol Ol Ol Ol Ol Ol Ol
I I I I I—> I I I I I I I 1—II— •—I I I I I I I I I I I !-«t—I I I I I I < I <—I ^- I I I I I
I I I I I I I I I I I I I I I I I I I 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 I I I I I I f I I I I I I I I t I I I I I I I I I t I I I I I I I I I I I I I I I I
I I I I I
fn ro CSICM i— n — esi cvi csj m csi c\j cvjesi e\»fn«—•esjcsji-'— rocsitsicsiesj m CM —
ee CM 2 n t- <\j M CM x 3 CM ee z e£ x esi r*t «sj CM i£—t— VO—l_li<:t— »— — _«t— >_IIS>*£*£IS>»-»I/)-.IX—IX^Xt— t— OO —I—I —I X Ul I— tOt——II——I i^
_izzLJOooo_i_it_>zzo—izz—iot-xjozzzo—ie<_izo<->zzo-j.-i-Joeozovi-J—J<
—om—•ccxzaeocc
re
OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
C I f—
-------
CO
occ
00
CO
o
OQf
CM_I
0.-J
00
CM
O
OfM
0>
c
OCM
NO
OOoo
eo
90
V
00
=-. IOCO
EOO
— OUJ
re r^
XOi
m
CO
O
00
ooe
ox
s-
o
10
o
01
c
o
rsl
O
*•»
O>
£
O)
x
CVJ
0)
(O
00
Ol
4->
o
w
O
c
o
i i i i i r i i i i i i i i i i i i i i i i i i i i i t i i i i i i i i i i i i i i i i i i
i i i i i i i i I i I i i i i > i i i i i i I i I i I I I i I i i i i i I I i i I i i i
i i i i i i i i i i 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
eecsixrot— tMCMoeeerofMath-
—it—;>!£•—i—o->_i_ii<:>—t—~
_jzzooi^o_i_iuzzo
i
i OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
aio i
-------
fSJ
un
O —
CO
^
tn
OCM
eu-
ro v>
— z
CO
OCC
Ot—
00
«T
or*>
^~
•o-
Ofsi
CO
CO
Ol
CO
O)
00 — — '
O> — —
•— • toco
S-.
O
1-
o
V)
10
IB
O
s_
O
10
0
O)
s.
en
E
3LO
ECO
"-OZ
XO2
*incr
ZOOLU
O
CO
oz
moe
COLL.
o
CO
0>
4^
o^ eft en Cf> en en en en ••* en en en en en en en en en en en en en en en en en en en en en en en en en en en en en en en en en en en r
en o^ en oi en 01 ot ^^
-------
OCM
Ol—
~- z
CMO
CO
LT>
OZ
OH-
— Z
CMO
00
vO
CO
01-
o —
rOO
er>
OCM
01
<•»;
C*
oz
ot—
roz
— I
OO
OO Ol
00
IMLT)
OO«M
r— — OS
'ioeo
— 03:
xo>
s-
o
in
ID
ID
O
L,
O
ID
O
aj
c
O
: —oa
oo
CJ
to
OtM
I I I I I I I I I I I I I I I I I I I I I • I I I I I I I I I I I I I I I I I I I I I I I I I
i i i I i I i i i l i i l i l l I i I l i l l I l i i i l i i i l i i i i I I i I
o z o o o ee o ec
ee. tsi •- ee CM «\j esj ee. CM CM *~ «M M «sj —• «sj esi CM — ro «M
lt— —I— l»— t—H-—)H-H-— »—»—»— _lt— I— *—_J4^U1
IZ-J— IZZZ—IZZ—IZZaE—IZbOlO— lOO
o — x o o o ** o o — o o o — o ac z — z ee. uj tu CD « *. «_> ee
CM
OJ
OL. I .
-------
CO
CO
O1
rv.
CO
occ
o —
00
o
OCC
csi_i
Oi-J
CO
CM
O
OfM
— I/)
OlO
root
c
OM
NO
CO —»">(
— 1000
u
o
e
3^-
eco
zcnu.
ro
oo
O
CO
ooe
ox
ID
O
Ol
c
o
0)
f
Ol
CM
o>
•' M -* co co o en eft o»"F'
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
^^ Csj-Oi ^"CD^-O»COOODODOCTlOfNJP**CsiO*-<
f^ O CM r*» O) C\j r^ OO r*^ Csj
i i i t i i i i i i i t i i i i i I i i i i i i i i i I i t t i i i i i i i i i i i i - i t i
i i i i i i i i i i i I i i i I i i I i I i i I I I i I I I i I I I I I I I I I I i I I I I
acQez_Ji^>oo
ujujoo —ZLJOC
-------
CM
in
O —
O«M
Ot—
<*>(/•>
Oi—
CO
O f»>
CO
O en en en en en en en en en
en en en en en en en en en en en en en eh
en eft en en en en en CF* eft en en CF> en ch
i i i i i i i t i i i i i <
en en en en en en eft en en en en en en en
en en en en en en cr* en en en en en eft en
i i » i i i i i i i i i i i
en en en eft en eft en en en en en en en en
en en en eft en en en en en en en en en en
i i i i i i i i i i i i i .1
co eft en eft en en eft en en en en en en en
vo en en eft en en en en en en en en en en
co en en en en en eft en en en en en en en
co en en en en en en en en en en en en en
i i i i i i i i i i i i i
f—or—»
_izzo-i^i
-------
CO
CO
en
CO
ooc
o —
r--«
ChO
^
00
O
OK
CM—I
O>—I
CSJ
o
OfSI
t— UO
o>o
CO
«r
O)
c
O(M
MO
OOoo
CO •—f*)(
i— IOCO
t-
o
u
o
E
3«—'
ECO
CO
o
CO
ooe
ox
1C
O
o>
e
o
01
£
o»
I I I I I I I I I I I I I
I I I I I I I I I I I
O) O* O) O* O) O*> O1 Oi O1O^ Ot Ot ^d ^^
O^ O^ O) CTi O1 Oi O> O> O) Oi Ol C7) Of ^
I I I I I I I I I I I I
esi <—« CM CM CM CM CM «-« CM <™ CM CM
CM
O)
JD
fO
•em i oooooooooooooo
0)O
-------
TABLE 3. Possible UAM modeling days for Dallas-Fort Worth Area.
Date
07/31/84
08/01/84
08/02/84
08/03/84
09/05/84
06/07/85
08/31/85
06/25/84
06/26/84
06/27/84
07/23/84
09/03/85
07/10/85
06/18/87
07/22/87
09/09/87
07/30/86
07/31/86
Rank
1
8
9
10
2
3
4
17
?
6
7
12
11
14
15
16
13
27
O3
180
160
160
160
170
170
170
150
7
160
160
160
160
160
160
160
160
150
Peak
Monitor
KLLR
KLLR
KLLR
MCK2
DNT2
BNVW
MCK2
ILLI
7
NST2
ROS2
DNTN
DNT2
ILLI
DNTN
BNVW
COIT
ERWN
# station-hr
>= 120 ppb
18
23
35
22
8
10
21
21
7
10
11
20
6
13
6
16
13
12
08/01/86
170
COIT
18
-------
TABLE 4. Candidate UAM modeling days for the Dallas-Fort Worth Area
Date
07/31/84
08/01/84
= 120 ppb
18
23
35
22
21
21
20
16
18
-------
CO
CO
01
eo
o>
i.
O
10
o
Oi Oi Oi Ol Oi Oi Ol Oi Oi Oi Ol Ol Ol Ol
I I I I I I I I I I I I I I
10 CO
o<
E
n
n
OOOOOOOOOOiOiOiOiOt
~ I I I I I
CD
TT
in
OCM OiOiOOOOOOOOiOlOiOiOi
f>_) I I I I I I I
CO
(O
«cr-> i OOOOOOOOOOOOOO
0»O I CM CM CM CM CM CM CM CSJ CM CNltSl CM CM CM
c i m^-u-»or».eDOiO«-
-------
Apr
INITIAL CONDITION TRACER SIMULATION
(INITIATED 1200 ON 24 AUGUST) FOR
DALLAS-FORT WORTH
-------
O)
O)
88
OrO
O 00
o
oo
<£>
n
o
u
2
(0
g
'•6
o
u
-rr
(OO
1S3M
-------
10 o
OD o
11
>°l
It
II
o
CM
o
o
o
m
(D
§
1SV3
o
(D
O
o
(O
o
oo
§
o
o
I r»
lo
s
4 i t i H » i « |:i i i | i t u
o
CM
1S3M
o
f)
o
CM
r*
JI
Q.
D.
in
03
O)
o
n
0)
o
D
T3
O
U
."g
1?
'c
-------
1SV3
I | ( I I } .1 i. l.J i I I.); I I .1 ) I I
a
a
in
oo
en
o
0)
8
o
8
•
o
o
1S3M
-------
1SV3
a
D.
CD
01
w
0)
u
D
o
'•5
o
u
1S3M
-------
Ifl
• O
^ o
" 6
11 I
I •
> °
Ji
S II
o
CM
er
• o
V)
c
o
1SV3
go o o o o o
00
0)
u
o
•o
O
(J
15
"c.
1S3M
-------
o
o
rj
1^
Oto
00
8 8
r—
-------
o
II
1
I
S
2
J.SV2
f-
oo
X)
§
§
to
o
f
O
O
rv
|
IT)
8
^
i
o
•
t=
z S •
CM
.1 f-: t^j t-?1 ?i I«!>T'r!f -f: !•:! T I t tV
-------
o
Cvl
g
o
00
(O
o
<£>
(D
1SV3
o
•i-
(O
8
to
§ s
>°l
li
.i E
SS
1
•
E
P
_J ' ' I ' >> I
T . i 1 i
i i » i
2 .
ffi
z m •
P^% «
- *X^ rf •
-------
o
3 eS
o oo
o
o
o
oo
1SV3
o o
o
*• CM
> F-.
If ?
x •= IN.
§0
8
_l I < | i I i | ) t I ) .1 i I
I | t I I | t I. I | I I l_
C rg
^ ^
•? Z
CD
s i
?
55
w
Q.
D.
m
00
O)
V)
0)
o
2
tn
o
^
'•6
o
.o
1?
"c
1S3M
-------
1SV3
_
M 0
11 II
ll
>
| E
JE=
s :=
2 2
X
8
z
(/I
_l
o
o
1
8
N
*"
* *
«
j
o o o o o o
CM O CD «O * CM
pv r*- to ^
•* *^
: >,
^ ^^ ^^^^^
"* ^^^ ^ ^|
^ ^\7^v ^ * f
M. ^"^ O \
- ' '
^i=r=nrrwi f» i-rt-f-wn^.-f^vrw-r1*
COO Q O
* n CM
§00 ^^
oo
-------
G
PREDICTED HOURLY OZONE CONCENTRATIONS (pphm)
FOR DALLAS-FORT WORTH DIAGNOSTIC RUN 1
-------
o
CM
OfO
IE
o o o
o oo to
r-
o>
V}
o
A
in
CD
a
a
•^x
0)
O
N
o
1S3M
-------
o
o
d
n
«
o
o
00
o
00
o to
O
(O
m
f I i i f | i i f l i i i I i i i I t i i I \ i i j i ii _
I
s
2
o
r^
o
t*-
o
CM
1^
o
x
z
o
o
o
°l
E £
11
v
o
3
c
o
'v>
to
"E
0)
in
oo
Q.
Q.
N.^
0)
o
N
o
1S3M
-------
in
q
to
ii
s
oo
o o
o oo
r-. to
K) f)
o
o
(O
O
o
<00
o
o
o
O
I I I I t I I I I'l I f I t I 1 1 I t I i M I I I I I t 1 I 1 I M I 1
O O ' O
K) rg »-
1S3M
1SV3
i } [ i i i | i i i | i i i | i i i | i i « L « i J j r i i_
m
oo
o
t/1
3
01
w
c
o
'w
tn
'§
o>
0)
w
o
tn
CD
s.
Q.
a
v.^
a>
o
N
O
1S3M
-------
to
in
Ol
o
CM
o
o
>. m
I 5
>°l
i E
i §
51
Z 5
o
oo
(0
o
(O
1SV3
<£)
o
CSI
o
rs/
o
(M
8 § S
N 10 -
tr>
00
w
&
c
o
'w
V)
0
•X)
in
oo
I
^^*
s
o
N
O
1S3M
-------
to
If)
II
o
CM
r--
OO
03
E
3
"><
0
3 O
o r-
2
cr
O
in
o
o>
IOO
O
o
o
00
(0
ro
o
(O
(O
K)
ro
o
o
(0
O
oo
m
o
(D
IO
I II I i 1 I I 1 t I 1 1II])I 1 II 1 I(I1l_
?
I-
3
p
o
8
OC
I'
i i i i i > i i i i i i i i i i i t i i i i i t i i t i i i i i i i i i i~ ,
o
CM
1S3M
in
00
o
n
"w
w
c
o
'w
w
"E
0)
0>
»
D
X)
in
GO
Q.
Q.
N«X
0)
o
N
o
(OO
1S3M
-------
<°. »-
I s
>l
If
o
oo
0
O
0
00
o
o
(£>
rO
1SV3
o
v
tO
O
o
to
o
CO
m
ro
o
i i i i M M i I i i i i i i i i i I i l i i r i i i t I i i t i i i i t ~
o
00
o
00
rO
o
(O
(D
1SV3
o
-------
£s
m o
" II
11
if
s:§
o
(N
r--
00
oo
1SV3
O
_! 1 I I I I I I 1 I I | I I I ) I I I | } I 1 | I I I 1 I I l_
"i t 11 i M 111 i i i i i 111 i > i i i i M t 11 f i i i i 11 i i r ,
(DO
o
-------
Ol
9
(M
10
1 s
I!
x •=
o .E
2 S
in
_i
§
00
i
8 .
r-.
o
CM
|v
OtO
00 ,
(^
15*3
oooooooo
O83
o
M
0
XI
in
(C
Q.
a
o
N
o
-------
co
o
oo
03
p^
O
§
(O
o
o
ID
CN
O
rsi
1S3M
1SV3-.
ID
00
O)
O
'
o>
OT
C
O
'w
o>
Q>
V)
o
in
co
O
N
O
1S3M
-------
o o
O IE
r-
o
CM
to
o
o
(O
rO
o
oo
m
I i I j I i i | I I f I ) I I \l I Ij f I
I 5
If
o .
o
o
o
CM
I)
£
t i i i i i i i I t i i i t i i i i I i i i i i t i ti I t i i i i i i t i~
(DO
O
o
CM
O
ro
O
CM
1S3M
1SV3
i | 1 I I j I t I | I t I ] I J i I t I 1 | I I I j I 1 l_
ID
03
05
w
3
cn
w
c
o
'w
(0
"E
0>
0)
w
o
jQ
in
oo
o
N
O
1S3M
-------
00
o
II
I
E
J
x
O
o
q
d
ii
o>
"5
o
CXI
O
O
O
CD
o
O
i i i I i i t i i i i i i t i ) i i i i i i t T -i i t i i i i i i~
(DO
O
o
N
O
1S3M
-------
•IS3M
-------
o
^ o
o q
"" 6
11 i
I 5
I!
£ II
2 2
o
(N
OK)
1C
o
o
10
o
03
G
-------
in
i O
05 o
« II
I s
I!
o
(M
fv
00
03 ,
o
to
rx
o
*
I*.
o
CSI
r-*
O
o
r-v
o:
O
o>
O
(O
<0
O
^•
(D
o
CM
to
o
o
too
8
o o o o o o o
OD *& fc i i ! i I i i i i i i i i i I i i i i i i i i f
o
r>j
1S3M
O
tsi
o
o
g
(£>
§
(O
r>
1SV3
o
(D
O
r>j
to
o
o
ID
O
oo
o
(O
in
JitJill^lM.] Ill | 1 IT J I II) I 1 1_
in
CD
o>
3
O>
w
c
o
"E
0)
in
00
I
O
N
o
1S3M
-------
110
11
oo
03
r>> i i
o o
c co
t- (O
1SV3
o o
to ^
*o
-------
i i 11 11 i i i i i 11 i I 11 i
1S3M
00
03
o
o
O
1C
(O
o
«o
1SV3
o
*
K)
O O
oo I I j I I 1 I -1 I I I II I J I I I I i I \ I i 1 » I I i l_
trt
t>
o
O
1C
(O
O
<£>
O
CM
(O
,00
v>
o
CM
in
GO
w
3
D>
D
w
c
o
'w
(0
"E
-------
o
71 rsi £
:^s c£
t\l CD
- 2 O
o o 'f
> > i-.
It ,
O
o
I f-
1 =
oo
o
V CM
E »
o o
o co
h- (O
(O
to
<0
o
CM
i i t ) i t i i i i i i
O
O
O
CO
i i r i i i i i ! i i i i M i i i i i i i i i i i i i I i i i i i i i t r
O
(DO
O
rg
1S3M
q
O
CM
o
oo
(O
o
(D
r- tO
J.SV3
o
*•
lO
O
f>J
to
§
(O
o
CO
in
o
<£>
m
10
_J 4 i .) 14 I. j 1 i I | I i 1 | I t I | t I 1 | 4 I J J 4 -I l_
in
CD
O)
K)
1
o
'w
.£
0)
V
m
(D
I
IS3M
-------
to
CSJ
03
o
CSI
oo
03
11 ii
I 5
If
II
_!lt|lll|!II]lll| Ii1[T I I I I I 1 J I 1 l_
|
in
—i
o
o
csi
O
O
0>
£
o -,
tOO
O O
o co
r- <£>
(0
K)
1SV3
o
*
to
K)
o
CSI
ID
o
o
to
o
00
o
o
m
K)
O
O
CM
O
'i 11 11 111111 11 11 ii 11 i 111 1111 11 11 111 11 11 r
O
fsi
1S3M
•*>
r-
II
It
Ii
O
CSI
N
00
03
s
to
i
. s
i- tO
s
o
o
O
CO
g
to
ro
1SV3
o
*
<£>
in
o
CM
to
to
o
CO
ID
_I I 1 I I I I J » II } I i I | I. I I | t 1 I | 4 I -I j II l_
O
K)
O
CN
m
CD
O>
to
.^>
-------
I I I ! I ! I 1 » 1 f ! I f I I I I I ! I t I I t I 1 I I I 1 I t I I I l~
CM
1S3M
1SV3
i i.( .1 i i I .1 i i I i t i { j -t i j i .1 » | » i i j i i i_
in
on
O)
o>
CO
c
o
'w
w
"E
0)
CO
o
in
oo
x:
a
a
^^
9)
O
N
O
1S3M
-------
oo
o
CM
O
OS
00
(O
°1
It
2 2
O
o
(0
•>
E
o
(0
r~
o
•*
r-
O
CM
r^
O
O
IO
o
<£>
<0
o
*
(O
o
CM
> 111 1111 111»i« i \ \ 1
(00
o
-------
J.5Y3
in
(N _
10 «fr
"~ CN
»
0000
CM O 00 tO
<£>
-------
o
(N
8
o
QD
OO
00
(O
to
(O
to
o
o
o
00
to
to
If)
to
i i i [»n | i i i { i i i | n i | i i i i i n \n r_
I S
I!
X •=
o .E
2 2
O
in
_i
o
o
o
c\l
(71
W
P
(OO
o
fM
1S3M
o
i i i i i t i 11 I i i i i t i'i i i I 11 i i i 11 i i I i i t i
f>
1SV3
in
OJ <£>
11 ii
||
ii
x •=
o .E
O
CM
(•»•
oo
r- (0
cn
8
O)
O
o
(DO
o
03
w
c
o
'w
w
"E
-------
as
o
fM
O.
GO
(C
o
to
(D
1SV3
§
CD
o
CM
(D
8
(O
o
OD
in
o
I 1 i \ -I 1 .1 .( I I I j I I !_,
(00
•»
O
CM
i t I
O
N
1S3M
w
w
o
E
0)
ID
»
O
-Q
in
a
a
s^'
0)
o
N
O
-------
Appendix H
PREDICTED HOURLY OZONE CONCENTRATIONS (pphm)
FOR DALLAS-FORT WORTH DIAGNOSTIC RUN 2
-------
m ,-,
« o
IS
I!
X •=
o .5
O
o
s
CO
o
r>
o
CM
(O
<£>
O
CD
in
10
»/>lllllllt»IJIIIIIJIJ I I 1 J I I l_
trt
8
O
O
(DO
I * I I I I I t f f
I < ( M 1 J I » > M M 4 I 1 > 1 f I >'.! 1 I I
o
o
CM
1S3M
if)
CD
05
w
*8
5 »
i|
f |
J £
o c
2 S
(X
OO
02
o
oo
(D
rj
O
(D
(O
1SV3
o
J
p «• ?•«
. •< > „ —
-!
«>?*"
» *o^
^ **«•*"•
?M:
r
^a^^
o
cv
1S3M
CM
5fc
c
D
bJ
>>-^
(0
O
I
0>
ID
W
O
JO
in
oo
o
N
O
-------
O
CM
O
O
2 5
K •=
ID
o
oo
Kj
O
to
o
rsi
10
to
(O
tO
o
CO
m
o
ID
m
to
I I I I I I I i 1 1 I I I i I I I I I I I I I I i I i jI 1 l_
cr
o
V)
_!
§
Sg
^ 6
11 II
11
I!
x •=
o .E
O
O
(DO
o
CN
CXO
00
'l I II II I I I I I I I I 1 I I > I I I 1 I I I II M I t I 1 1 1 I I 1
P
O
rsl
opo
fM
1S3M
1SV3
§000
CD <£> <•
r*« »D (D —
o
CM
m
(D
O)
o
en
-i
CM
sfc
C
3
O
UJ
N-^
en
c
o
I
'§
0)
0)
in
o
m
oo
I
>_x
a>
o
N
O
1S3M
-------
1SV3
2 <=>
T o
1 5
If
o
CM
§o o o o
OD <£> {Ti iTl
'O 'O 'O to tO
t>
o
o
100
_< i « i » i i i » i i i i II i i i i i t i I i i i! i t i r^
I /-\ 5
i t i i i i i i rl i i > i i i i i i t i i i i i
111»
o
CM
iS3M
-J
IM
in
oo
VI
3
r-.
OKI
oo
8 g
J.SV3
(D
O
I J •( ! 1_
CM
«te
C
O
"w
n
"E
0)
D
.0
m
oo
a
a
O
N
O
1S3M
-------
Ol
•*
in
o
o
O
CO
tc
o
ID
IO
J.SV3
(O
o
CM
<£>
o
o
(O
o
GO
in
rO
o
i t i i i I \ i t r \ i t < i i i i i > i t i i i 1 t i > i t t i M~
8
o
CM
o
CM
- 8
8 S
P- «O
o
(O
o
fN
1S3M
1SV3
o
*
<0
o
CM
(D
(O
it
S
I
x
o
O
to
IO
z 2
s
(O
i .1
.1. i 1.
in_
o
CM
1S3M
oo
O)
3
CD
CN
5*fc
c
3
O
UJ
V)
C'
o
"w
tr>
"E
0)
D
XI
m
oo
I
N—^
0)
o
N
o
-------
in ,j
oooooooo
C\jOOOtf>*CMOOO
f^r^tototototoin
~ ~ fO O K) K) K) K"; K)
GO
• _lli|iil|ill|liT ( l^t I ] I I I I f I I I I k T_
If
s :§
o
t>
E
o
to
to
o
•*
to
o
CNJ
1C
cr
I'
§
S
i
O
LD
CD
O>
o
o
(00
'l ! I I I i 1 1 1 I t I 1 I I I I 1 i I I I I j I ) 1 I I I 1 1 I I 1 1 I M~
1S3M
XSV3
w
3
Q)
CN
=%;
C
O
>
I/}
C
g
"w
IT;
"E
0)
0)
CO
D
X)
in
oo
S
o
N
O
1S3M
-------
1SV3
o
rj
°!
(b
E
3
"x
D
C/l
o
o
(Nt
I
8
03
o
o
100
o
o
o
DO
o
!£>
to
o
CM
I I I I I I I I I I i I j I I i
o
00
IT;
fl
t I 1
O
o
fO
"w
O &
r 00
(O ^
11 II
1 §
1 1
.1 E
x •=
II
O
IN
OO
00
8
IO
o
to
o
1SV3
rt
O
O
tO
iO
O
03
O
-------
. o
Oi
II
o
CM
or.
oc
oooooooo
OCD i 11 11 111 i 111 i 111 H i i 11 111 11
11 i i r ,
o
CM
1S3M
oo
O)
w
3
1SV3
1 I j ! 1 i | 1 4 i | I I I. | I i
CN
«*b
C
3
O
>
C
o
"w
to
"E
0)
d)
(0
D
J3
in
CO
a
a
N_X-
8
o
N
O
1S3M
-------
ai
K>
= 5
If
x •=
o .E
ui
I
§
ID
o
CM
oo
03 r—
o
pU
O
tl CM
E <°
O
o
O
03
(O
(O
K)
1SV3
IT)
r>
MM
> MS C
i i I i i t i i i t i i t i i i i i t i t t 1 t i i i i r
o
>o
O
10
CM
in
oo
O)
1S3M
I 1 I | I I I | I I I | i I I } I ! I_J
w
3
a>
tN
=«B
c
3
O
>
in
o
"(0
n
"E
0)
o
X)
in
oo
E
.c
0)
o
N
O
1S3M
-------
en
"> r-
rsi •*
*~ CXI
11 II
II
If
5 2
o
CM
o
o
03
<£>
o
O
O
to
to
o
00
rO
O
M M I I .M 1 t 1't I t M t 1 .f .1 I I t .1 I t .1 I r
(DO
8
1S3M
1SV3
I.IT \ II I |1 1 i ] II IjJ.l l|iT;l.jltl }J i l_
CO
O)
O
fD
"«
D>
c
o
C
o
'w
_w
0)
0>
D
X)
in
03
D.
Q.
0>
O
N
O
1S3M
-------
o o o o o o o
O 03 «
(N O 00
I I I I I I I I I I I 1 I I I I I 1 I I I I I II I i I I I I
M M' t. I i 'H I I I I M r » I > I M I I M 1 I I I I » .1 M .1 •>
IS)
® CD
O
(N
r«.-
oo
§ §
(O
.s
(D
i- oo
1SV3
o
i
too
i ;| i -i i | i i ,i_
o
tM
o
CM
1S3M
oo
O)
o
n
"w
3
DI
C
L.
O
LJ
>^
V)
o
"w
(0
"E
0>
o
X)
in
oo
0)
o
N
o
-------
™ c
— o
" d
11 ii
O
CM
O»0
O
O
O
00
ID
to
O
tO
to
ID
to
O
CM
to
o
o
<£>
O
00
in
o
(O
•s -2 o
i E
If
2 2
11)111
i i i i
wi
_j
o
o
CM
CM
CM
01
O
O
(DO
I > I I I I 1 1 } I ! I M 'I 'I t I I I I I I I I t t I I I I 1 t 1 I I I } I"
O
in
oo
O)
1S3W
J.SV3
__« i i j .1 -i i | i i i j i j i ,j i i j j i .1 •» j
3
D>
CM
Its
C
D
O
8
"E
Q)
(A
D
X)
in
03
0)
o
N
O
wo
-------
o>
68
*~ o
11 ii
I 5
> °
If
o
CM
oo
GO
oooooooo
OODtD-S-CSiOCOtD
r^tOCO^tOtOtOLO
OiOtOiOiOfOioiO
I I I I i I I | I i I | I I i j Ii I | I I Tj I I T
tr
O
CM
I
O
8
(N
O
O
(00
O
o
iO
O
(\
1 I M t I t 1 I t I I I i I I 1 I t I 1 I I I I I I t I 1 I i I I 1 I
O O O
»0 CN ^
1S3M
1SV3
i I I j I 1 I j I I I .J I I i | t I « j '
ID
CD
C3)
O
n
"w
3
»
rsT
c
c
o
"E
0)
0)
K
O
.o
in
GO
0)
o
N
O
1S3M
-------
co
6
Is
II
1SV3
ooooooooo
cv o (D t0touoin
2^O 'O'OfOfO'O'OfOfO
ct
i
tN
I
0>
6
o
go
_iii/iiJ|iii|iii|iii|Til|»r I j i i i_
M .1 t i » 1 i i M i .1.1 t i I i. t i j t M 11 f t n > i
o o
™ »-
1S3M
I
IT)
(D
O5
w
3
t c?
o o
" o
3 *
2 3 O
•5 B > r«
I £
Jl ?
R S
I/
o
» CM
£ «
?=
T T J. I
V\P: I
^ *"s
I >t,;
&:«-,
jf^d
V P%
y »> * «,
/££• j
//0^>\s-
:*;i >»11 r
fv
*te
c
o
1
V
0>
(0
o
ID
03
J
a
a
»w
0)
O
N
O
o
IS3M
-------
o
CM
00
tc
o
o
O
03
O
«c
tO
fO
(O
ro
O
rsi
(O
to
O
oo
_l I i I I I 1 I I I I j I I I I I I I I I I I I ) I I I t 1\_
tl
_2
"5
|
S !E
o
tO
r^
o
N
o
CM
(X
O
O
8
00
O
S
O
'!
11
J I
X •=
11
o
CM
OO
CD
§
O
00
(O
O
to
O
O
CM
O
§
O
to
to
o
CM
to
o
o
o
(O
tO
1SV3
(O
rO
O
CN
(O
fO
o-
o
(O
o
00
m
to
o
(0
m
»• I .] I
I ] 1.1 1 I ]_
o
K)
o
(M
1S3M
CN
sfe
C
3
0
c
o
"w
(/)
"E
0)
0)
(A
O
in
oo
a.
a.
0)
o
N
o
-------
O
CM
O
O
O
00
to
CD
11 II
II
II
II
03
c:
i
ti
£
O
in
m
I ' ' '-
z
w
i > i t t r < r i f j i > i t i i i i / i » i i i t r t t f 1 i i t i ) i i r ,
too
O
«M
J.S3M
CD
03
w
o>
o£
o
o
o
K)
-------
S?
J s
J!
S J
OO
00
o
o
fO
o
03
(O
rO
O
tD
<0
r>
to
0
o
rv/
(O
o o
00 <£>
m m
K) fO
_! I I j I I I | I I I I I I I J I I I } I I I
or
ie
O
§
I
O
l i I r I } I I I I M > I I I I I I t I t 1 I I 1 t I I I t I t I M M
D
O
o
oo
O)
(DO
1S3M
w
1SV3
_l I I | -4 I j | I I .1. | i 14 j: I -i 1 j i I 'i j i i .1 | .'I -I I
CNI
««=
c
0
LJ
W
O
'w
0)
0)
0
J2
m
oo
0)
O
N
O
1S3M
-------
o
CM
O>0
I 8
ii
o
ID
rv
o
•*
r-
O
CM
rv
O
o
o
o
o
a>
o
IO
tO
o
«•
(O
o
CM
ID
o
o
o
to
'£>
10
o
1 t I I I t 1 I t II l"
too
o
0)
V)
o
JO
in
03
E
-C
0)
O
N
O
-------
o
CO
(O
o
v
(O
o
rv
I s
II
o .£
03
O
00
in
r>
o
(O
_f IT | 1 I 1 | I I I | i I I | I I I | I I I j I I I j ( 1 I_
I
cc
to
_;
O
o
rj
<£>
o
too
o
*0
(M
i 1 1 i 1 1 1 1 ( j 1 1 1 1 1 M 1 1 1 1 1 1 1 1 1 1
i 1
1 r
o
(M
1S3M
ID
00
C7)
o
00
I!
Ii
o
o
OO
O
03
(£1
•O
g
(TJ
1SV3
(O
o
CM
(O
K)
8 §
(0
o
<
rsi
=«s
C
0
UJ
N— f
V)
C
o
"w
(0
E
0)
0)
w
o
m
03
Q.
a
o
N
O
-------
o
CM
OO
00
o
o
o
BD
o
<£>
J.3'Vj
o o o
O GO (O
to m in
i t i j I t I | I I I | i i i | IT t | i IT^
o
to
r*
o
*
r^
O
cxi
r^
O
o
r~v
O
10
to
o
^-
to
o
CM
S ~l I I I I I I 1 I 1 t I I I 1 I I t I I I 1 I I I I I I I 1 I 1 M 1 I I M"
(DO
o
«M
1S3M
in
oo
w
1SV3
_i I i | ill | i i i .] i ii | 1 I I j I I i j- I i >
o
&
in
c
o
"w
Vi
0)
0)
o
in
oo
I
0)
o
N
O
1S3M
-------
*
•* •*
to to
"~ CM
11 II
5- »
.2 3
It
03
1SV3
O O O O O
CM o oo to -a-
1"^ f^ \Q to tO
rO *O IO K>
O O O O
CM O 00 tO
to to m m
rO fO O K>
in
(O
I
0>
E
1 1 I |T
I II J I I I j I I I j l I I | I I 1 | I I i_
O
rO
O
O
CM
i \ i t i < i 11 \ i it \ \ i 11 i ) 11 i i 111 i 11 \ i \ 111 i i r
(DO
o
r>4
1S3M
1SV3
•O
j 1 I \ T. IT! J I I J J I 1 J_
CD
O5
w
3
r\T
c
D
LJ
tn
c
o
8
0)
ID
n
o
.o
in
CD
CL
Q.
0)
O
N
O
1S3M
-------
^
CN
1C
°!
CN
It
»
3
O
CM
oo
03
r-.
o
o
r-
o
CO
(£>
to
O
CM
1C
o
o
(O
10
O O
00 (0
m m
I
x
o .
•
8
1C
I
*>
E
O
o
(00
_l I I I I
I 1 I j i i I I I
l I I M J.'l I H 1 I t I I I 1 .1 I 1 .M .1 M » 1 1 'H t t M t .t I .1 I"
O
04
1S3M
(D
0)
w
1SXG
CN
»*b
C
L.
O
in
o
'8
1
0?
0)
M
O
X)
in
03
0)
o
N
o
1S3M
-------
O)
r: 10
J 5
£ °
If
oo
aa
oooooooo
OOOtDTCMOCOO
r^ to t^t to to to in in
rO 'O O to *O rO fO fO
in
_j
o
o
o
IN
0>
o
o
(O
o
00
1SV3
o
> i-
J E ?
s:§ r-
en
§
o o
o to
o> «.] II !„
O
(OO
1
O
«M
1S3M
O
u
o
'w
in
"E
93
9}
V)
O
ID
00
a
a
•**^
8
o
N
o
-------
WEST
O
N
O
(D
CD
Ul
cr
o
c/>
a
0)
3
w'
92
o'
3
M
m
Q
c
3
5*fe
N)
(O
C
M
-* ro ^ j»
O O O Qq>
i I 1 J J t-l I I. I ( t { | .1, I i I | I .1 1.4 i t J J t .4. i i l..| .1 1 I til §
* .^ £ ~~ "^ ^^
N3
O
~t I 1 1 ! I 1 I t I I I I t I 1 I t 1 t 1 t t 1 I I t I I I
o> 3
KJ «
O
I
KJ
O> -*
O) O
o o
S^'
°§
S1
O
3 I
•sj J.
o> £.
O C
CJ
Ul
O>
O
CD
O
O)
o
o
fs)
O
O* Crf
o* o^
$ §
EAST
O>
CD
O
u
•vl
5 §
• II
o _
S2
CO
CD
(Ji
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO. 2. \
EPA 450/4-90-006B
4. TITLE AND SUBTITLE STUDV
URBAN AIRSHED MODEL/OF FIVE CITIES - Demonstration of
Low-Cost Application of the Model to the City of Atlanta
and the Dallas-Fort Worth Metroplex Region
7.AUTHOR Morris, Thomas C. Myers, Edward L.
Carr, Marianne C. Causley
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Systems Applications, Inc.
101 Lucas Valley Road
San Rafael, CA 94903
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Office of Air Quality Plannning and Standards
Research Triangle Park, NC 27711
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
April 1990
6. PERFORMING ORGANIZATION CODE
e. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This document presents methodology for low-cost base case Urban Airshed model
application to Atlanta and Dallas-Fort Worth.
T. KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
Dzone
Jrban Airshed Model
'hotochemistry
. DISTRIBUTION STATEMENT
b.lDENTIFIERS/OPEN ENDED TERMS
19. SECURITY CLASS (This Report)
20. SECURITY CLASS (Tllispagc)
c. COSATI Field/Group
•
21. NO. OF PAGES
320
22. PRICE
'A Form 2220-1 (R«v. 4-77) PREVIOUS EDITION IS OBSOLETE
------- |