United States
Environmental
Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park, NC 27711
EPA-450/4-90-00 6D
APRIL 1990
AIR
«EPA
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
-------
EPA-450/4-90-006D
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
By
Ralph E. Morris
Thomas C. Myers
Marianne C. Causley
LuAnn Gardner
Edward L. Carr
Systems Applications, Inc.
101 Lucas Valley Road
San Rafael, CA 94903
EPA Project Officers:
Gerald L. Gipson, Atmospheric Research and Exposure Assessment Laboratory
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
1 INTRODUCTION 1
Use of the Urban Airshed Model 1
The "Five Cities" UAM Study 2
2 DESCRIPTION OF THE CB-IV VERSION OF THE URBAN AIRSHED
MODEL 4
Use of the Smolarkiewicz Algorithm to Solve the Advection
Equation 5
Use of the CB-IV to Solve Photochemistry 6
3 MODELING EPISODE AND DOMAIN SELECTION 9
Episode Selection 9
Modeling Domain Selection 12
4 PREPARATION OF MODEL INPUTS 14
Routine Data Available for Atlanta 14
Preparation of UAM Inputs Files 17
5 DEVELOPMENT OF A PLANR BASE CASE FOR ATLANTA 23
Diagnostic Run 1 24
Diagnostic Run 2 24
6 APPLICATION OF THE UAM TO ATLANTA FOR EMISSION
REDUCTION SCENARIOS 36
Tracer Simulation 36
UAM Modeling Results 37
Comparison with a Previous Study 40
Discussion 42
References 44
89059r2
-------
Appendix A: Emissions Data Used in the Application of the UAM to Atlanta
Appendix B: Hourly Wind Fields for UAM Layer 1 Used in Diagnostic Run 1
Appendix C: Hourly Predicted Ozone Concentrations (pphm) for Diagnostic Run 1
Appendix D: Hourly Wind Fields for UAM Layer Used in Diagnostic Run 2
(Base Case Simulation with Biogenics)
Appendix E: Hourly Predicted Ozone Concentrations (pphm) for Diagnostic Run 2
Appendix F: 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 on
4 June 1984
Appendix G: Sensitivity Analysis of the Calculations of Ozone Concentrations
and Effects of Emission Controls on Ozone Concentrations to Wind
Speed Specification
89059r2
-------
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 2k, 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 al., 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 sour-
ces. 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
B90S9r2 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
retirement 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 qual-
ity 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 emis-
sions 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:
89059r2 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
compare results with those obtained with EKMA; 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). This report presents the
PLANR use of the UAM for Atlanta and analyzes the effects of biogenic emissions
on the anthropogenic VOC emission reductions required to meet attainment of the
ozone NAAQS. Recently, the EKMA model was applied to Atlanta for 4 June 1984 to
estimate the effects of biogenic emissions on VOC emission controls needed to bring
Atlanta into attainment of the ozone NAAQS (Chameides et al., 1988). One of the
purposes of this study is to repeat this analysis using a more comprehensive model,
the UAM(CB-IV).
89059r2 2
-------
DESCRIPTION OF THE CB-IV VERSION OF THE
URBAN AIRSHED MODEL
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 various physical and chemical processes that take place in the atmosphere.
The basis of the UAM is the atmospheric diffusion or species continuity equation.
This equation represents a mass balance in which all of the relevant emissions, trans-
port, diffusion, chemical reaction, and removal processes are expressed in mathe-
matical terms. Based on the grid concept, the model is generally employed to simu-
late an 8- to 72-hour period during which episodic meteorological conditions persist.
Because the model can resolve both spatial and temporal features of the concentra-
tion field, it is well suited to the analysis of future control strategies and their
effects on air quality in various parts of the modeling region. Before the model is
used for such an analysis, its ability to replicate measurements from an historical
ozone episode is tested. Model inputs are prepared from observed meteorological,
emission, and air quality data for a particular day or days. Once the model inputs
have been adjusted within the range of their uncertainty so that the model performs
within prescribed levels, the emission inventory can be changed to represent assump-
tions about future emission scenarios. The model is then re-run with the forecasted
emissions, and the resulting hourly ozone patterns are what the model predicts is
likely to occur under meteorological conditions similar to the historical episode.
The UAM is the only air quality model recommended by the EPA for photochemical
or reactive pollutant modeling applications involving entire urban areas (EPA,
1986). The EPA guidelines refer to the 1978-1980 version of the UAM; the formula-
tion of that version is discussed by Ames and others (1985a,b). Many improvements
to the UAM have been made over the last 10 years. The two most significant are:
Incorporation of the latest version of the Carbon-Bond Mechanism, the CB-IV
(Gery, Whitten, and Killus, 1988)
Use of the Smolarkiewicz algorithm for advection (Smolarkiewicz, 1983)
89059r2 2
-------
USE OF THE SMOLARKIEWICZ ALGORITHM
TO SOLVE THE ADVECTION EQUATION
Grid-based air quality simulation models require a numerical approximation of the
horizontal advection terms in the species conservation equations. The 1978-1980
version of the Urban Airshed Model (UAM) used a variant of the Sharp and Smooth
Transport Algorithm (SHASTA) originally formulated by Boris and Book (1973).
Since 1977 there have been many comparative studies of advection schemes. Exam-
ples of such studies relevant to the UAM are those of Zalesak (1970), Schere (1983),
Chock and Dunker (1983), Chock (1985), Smolarkiewicz (1983), and Yamertino and
Scire (1985). In each of these studies an idealized scalar function (a cone, block,
ellipse, or cosine wave) representative of a concentration distribution is advected by
a rotating wind field (constant angular velocity). The rotating wind field provides a
range of Courant numbers, depending on the radial distance from the center of the
domain. The degree to which the attributes of the idealized function (total mass,
peak value, mean value, and gradients) are preserved indicates the accuracy of the
scheme.
The above studies showed that a number of advection schemes were more accurate
than SHASTA, as measured by the idealized tests. For the UAM, a number of spec-
ific requirements constrained the selection of alternative to SHASTA. First, it is
important that the scheme be positive definite, i.e., that it not result in negative
concentrations. Second, the scheme should use forward time differencing to mini-
mize storage requirements and to insure compatibility with the chemical mechanism
numerical solution scheme. Third, the ability of an advection scheme to represent
the magnitudes and locations of peak concentrations is of major importance in regu-
latory applications. Fourth, to handle complex airflows, the scheme should display
relatively uniform accuracy over a wide range of Courant numbers (i.e., wind
speeds).
Further review indicated that the advection scheme developed by Smolarkiewicz
(1983) represented the best combination of accuracy and economy. The
Smolarkiewicz scheme is conceptually similar to SHASTA in that a highly diffusive
transport step is followed by an anti-diffusive correction step. The transport step is
essentially the well-known "upstream" finite-difference scheme. The correction step
involves a second exercise of the upstream finite difference scheme, substituting the
anti-diffusive velocity for the actual velocity. The Smolarkiewicz scheme is positive
definite and forward in time, and can be used in either a time-split or multidimen-
sional mode. Smolarkiewicz demonstrated that the scheme was superior in both
accuracy and economy to the multidimensional generalization of SHASTA formulated
by Zalesak (1979). In applications to the Los Angeles area and Kern County, Califor-
nia, the UAM with the Smolarkiewicz algorithm produced results that were more
accurate than those produced in earlier applications using SHASTA (Hogo, Mahoney,
and Yocke, 1988; Whitten et al., 1985).
890S9r2 2
-------
USE OF THE CB-IV TO SOLVE PHOTOCHEMISTRY
The latest version of the Carbon-Bond Mechanism (CBM-IV) was recently implemen-
ted in the UAM (Gery, Whitten, and Killus, 1988). Whenever a new chemical kinetics
mechanisms is merged into a complex air quality simulation model, the predictive
capabilities and solution speed of the new computer code require optimization and
evaluation. This process is even more important now because the recent gas-phase
chemical kinetics mechanisms (CAL, RADM, and the CBM-IV) are larger than pre-
vious mechanisms, and therefore require significantly more computing time. The
CBM-IV reactions are shown in Table 1. There are some minor differences between
the original CB-IV and the version implemented in the UAM. For the examination of
ethanol (ETOH) blended fuels, ETOH was added as a species. The ethanol reaction is
ETOH + OH * ALD2 + HO2 with a rate constant of 4,300 1/ppm-min. Both numerical
and chemical improvements were made to the CB-IV implementation in the UAM so
that computational speed could be increased and solution uncertainty diminished.
The implementation of the CB-IV in the UAM used a modified Crank-Nicholson
algorithm for the simultaneous solution of the differential equations that represent
the chemical changes for each species. This numerical integration scheme produced
results that were within a few percent of those obtained with the previous (Gear)
algorithm over a wide range of atmospheric conditions (Morris et al., 1989a,c).
39059r2 2
-------
TABLE 1. The Carbon Bond Mechanism-IV.*
Number
1)
2)
3)
4)
a)
6)
7)
6)
9)
10)
11)
12)
13)
H)
15)
16)
17)
IB)
19)
20)
21)
22)
23)
24)
2b)
26)
27)
28)
29)
30)
31)
32)
33)
34)
35)
36)
37)
38)
39)
40)
41)
«)
43)
44)
45)
46)
47)
4b)
49)
bO)
51)
Reaction*
Reaction Rate Data
03
0
0
0
03
03
03
N03
N03
N03
NO
NO
OH
OH
HONO
OH
OH
H02
H02
OH
H02
H02
OH.
OH
FORM
FORM
FORM
AL02
AL02
ALD2
C203
C203
C203
C203
+ H20
N02
0
NO
N02
N02
NO
N02
03
03
01D
01D + H20 -
OH
H02
N03
NO
N02
N02
N205 + H20
N205
NO
N02
NO
HONO
HONO
HONO
N02
HN03
NO
N02
PNA
PNA
H02
H02
H202
H202
CO
OH
FORM
FORM
0
N03
0
OH
N03
ALD2
NO
N02
PAN
C203
H02
..-->
....>
....>
,...>
--->
....>
H20
>
-.-->
.....>
-hU->
.....>
.....>
.....>
.....>
.....>
.....>
.....>
.....>
.....>
..._.>
.....>
.....>
.....>
....->
.....>
.....>
.....>
._...>
...->
....>
..... >
h\2->
._..>
....>
...->
....>
....>
hv6->
,..._>
,....>
...._>
,....>
,...->
OH
NO + 0
N02
NO
N03
N02
N03
0
010
0
2.000H
H02
OH
0.89N02 + 0.890 + 0.11NO
2.00N02
NO + N02
2.C
N03 + N02
2.00HONO
HONO
OH + NO
N02
NO + N02
HN03
N03
PNA
H02
N02
+ N02
+ N02
H202
2.000H
H02
H02 + CO
CO + 2.00H02
CO
OH + H02 + CO
C203 + OH
CO
FORM
+ HN03
FORM + X02
N02 + X02
PAN
C203 + N02
2.00FORM +
0.79FORM +
0.790H
X02 + FORM + H02
2.00H02
H02
2.1
0.79X02
2.00H02
0.79H02 +
Pre-factor
(pom'""
mi n
Temp. Factor Rate Constant @ 298K
exp((-E/R)/T) K298 .(ppm-nmin-l)
8.383 E+04
2.643 E+03
1.375 E+04
2.303 £+02
3.233 E+02
1.760 E+02
5.300 E-02
*EXP( 1175/T)
*EXP(- 1370/T)
*EXP{ 687/T)
*EXP( 602/T)
*EXP(- 2450/T)
1.147
3.260
2.344
2.100
3.390
1.909
3.660
7.849
1.900
2.110
2.600
1.600
6.554
1.975
9.770
1.500
1.537
7.600
5.482
1.640
2.876
1.909
8.739
7.690
2.550
4.720
3.220
1.500
E+05
E+03
E+01
E+01
E+04
E+01
E+02
E-06
E+16
E-05
E-ll
E+02
E-01
E+03
E-05
E+03
E+03
E+02
E+15
E+03
E+01
E-10
E-01
E+03
E+02
E+04
*EXP(
*EXP{-
*EXP(-
*EXP(
*EXP(-
*EXP(
*EXP(-
*EXP(
390/T)
940/T)
580/T)
250/T)
1230/T)
256/T)
10897/T)
530/T)
*EXP( 806/T)
*EXP( 713/T)
*EXP( 1000/T)
*EXP( 240/T)
*EXP( 749/T)
*EXP(-10121/T)
*EXP( 380/T)
*EXP( 1150/T)
*EXP( 5800/T)
*EXP(- 187/T)
4.302 E+04 *EXP(- 1550/T)
9.300 E-01
1.739 E+04 *EXP{- 986/T)
1.037 E+04 *EXP( 250/T)
3.700
7.915 E+03 *EXP( 250/T)
1.180 E-04 *EXP( 5500/T)
5.616 E+18 *EXP(-14000/T)
3.700 E+03
9.600 E+03
6.521 E+03 *EXP(- 1710/T)
see notes
4.323 E+06
2.664 E+01
1.375 E+04
2.309 E+03
2.438 E+03
4.731 E-02
5.300 £-02*14
see notes
4.246 E+05
3.260
1.000 E+02
2.999"
3.390 E+Olxkj
4.416 E+04
5.901 E-01
1.853 E+03
1.900 E-06
2.776
1.539 £-04
1.600 E-ll
9.799 E+03
1.975 E-Olxki
9.770 E+03
1.500 E-05
1.682 E+04
2.179 E+02
1.227 E+04
2.025 E+03
5.115
6.833 E+03
4.144 E+03
2.181 E-01
2.550 E-01*k39
2.520 E+03
3.220 E+02
1.500 E+04
see notes
see notes
2.370 E+02
9.300 E-01
6.360 E+02
2.400 E+04
3.700
see notes
1.831 E+04
1.223 E+04
2.220 E-02
3.700 E+03
9.600 E+03
2.100 E+01
As currently implemented in the DAM (CB-IV), ethanol and methanol have been
added to the CB-IV and can be treated as explicit species.
(Continued)
87048
88097
-------
TABLE 1. Concluded.
Reaction Rate Data
Number
52)
S>3)
54)
55)
58)
59)
60)
61)
62)
63)
64)
6t>)
66)
67)
68)
69)
70)
71)
72)
73)
74)
75)
76)
77)
7B)
79)
80)
81)
PAR * OH
ROR
ROR
ROK 4 N02
0 * OLE
OH 4 OLE
03 4 OLE
N03
0
OH
03
OH
T02
OH
CRES
CKO
OPEN
OPEN
OH
OH
03
4 OLE
* ETH
* ETH
4 ETH
4 TOL
4 NO
T02
* CRES
4 N03
4 N02
OPEN
4 OH
4 03
4 XYL
OH 4 MGLY
MGLY
0 4 ISOP
4 ISOP
4 ISOP
N03 4 ISOP
X02 4 NO
X02 4 X02
X02N 4 NO
Reaction4
-h\2->
-hv2->
0.87X02 4 0.13X02N + 0.11H02
0.11ALD2 4 0.76ROR - 0.11PAR
1.10ALD2 4 0.96X02 4 0.94H02
U.U4X02N 4 0.02ROR - 2.10PAR
H02
0.63ALD2
0.30CO
0.22PAK
FORM
H02
0.50ALD2
0.44H02
- PAR
0.91X02
0.09X02N
FORM
1.70H02
X02
0.22ALD2
FORM
0.08X02
0.56T02
0.90N02
CRES
0.40CRO
0.300PEN
CRO
C203
X02
C203
0.03ALD2
O.U3X02
0.76H02
0.70H02
0.80MGLY
X02
C203
0.60H02
O.SOX02
0.90PAR
X02
0.40MGLY
0.20ALD2
FORM
0.20MGLY
0.44H02
X02N
N02
0.38H02
0.20FORM
0.200H
ALD2
PAR
0.74FORM
0.22X02
FORM
N02
0.70X02
0.300H
1.56FORM
0.42CO
0.36CRES
0.90H02
H02
0.60X02
4 HN03
H02
2.00CO
FORM
0.62C203
0.69CO
0.20MGLY
0.50X02
1.10PAR
C203
H02
0.80AL02
0.50CO
4 FORM
4 0.20C203
4 0.13X02N
4 0.40ALD2
4 0.10PAR
4 0.100H
Pre-factor
(ppm'nmin-1)
Temp. Factor Rate Constant
exp((-E/R)/T) k298 (ppirf n
4 0.28X02 4
4 0.02X02N 4
4 X02 4
4 0.33CO 4
4 0.100H
4 ALD2 4
- PAR
4 CO 4
4 H02 4
4 0.12H02
4 0.44H02 4
4 0.900PEN
4 0.60H02 4
4 CO
4 2.00H02 4
4 0.70FORM 4
4 0.080H 4
4 0.20CRES 4
4 0.30T02
4 CO
4 0.550LE 4
4 0.45ETH 4
4 0.67H02 4
4 l.OOETH 4
4 0.55ETH 4
4 0.06CO 4
1.203 £403
6.250 E416
9.545 E404
2.200 E404
1.756 E404
7.740 £403
2.104 £401
1.135 £401
1.540 £404
3.000 £403
1.856 £401
3.106 £403
1.200 £404
2.500 £402
6.100 £404
3.250 £404
2.000 £404
9.040
4.400 £404
8.030 E-02
2.453 £404
2.600 £404
9.640
2.700 £404
1.420 £405
1.800 £-02
4.700 £402
1.200 £404
2.550 £401
1.000 £403
EXP(- 8000/T)
*EXP(-
*EXP(
324/T)
504/T)
*EXP(- 2105/T)
*EXP(- 792/T)
*EXP(
*EXP(-
411/T)
2633/T)
*EXP( 322/T)
*EXP(-
*EXP(
500/T)
116/T)
*EXP( 1300/T)
1.203 £403
1.371 £405
9.545 £4Q4
2.200 £404
5.920 E+03
4.200 £404
1.800 E-02
1.135 £401
1.080 £403
1.192 £404
2.700 E-03
9.150 £403
1.200 £404
2.500 £402
6.100 £404
3.250 £404
2.000 £404
9.040 x
4.400 £404
1.500 E-02
3.620 £404
2.600 £404
9.640 x
2.700 £404
1.420 £405
1.800 E-02
4.700 £402
1.200 £404
2.000 £403
1.000 £+03
298K
S70-.8r
88097
-------
3 MODELING EPISODE AND DOMAIN SELECTION
EPISODE SELECTION
Table 2 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 pertain to Atlanta and one (3/18/84) to Columbus (note that on other days
Columbus may have an exceedence of the ozone standard but the ozone concentra-
tions are less than recorded in Atlanta). The most striking feature of the occurrence
of elevated ozone concentrations in 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 concentra-
tion in Atlanta (201 ppb on 31 July 1987). On most days during this period, most
ozone monitors in Georgia, including Columbus, are recording high ozone concentra-
tions. 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 Atlan-
ta during July and August, 1987, may be attributable to regional-scale in addition to
urban-scale ozone formation.
For Atlanta, the city specific analysis will analyze the effects of biogenic emissions
on urban ozone formation and on the effects of anthropogenic VOC emissions strate-
gies 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. It is
highly desirable that this day also be chosen for the city-specific UAM analysis as
long as there are no other specific reasons, such as excessive wind shear or other
unusual meteorological phenomena, that would preclude its selection. On 4 June
1984, the maximum daily ozone concentration was 147 at Conyers, Georgia. For the
two day period of 3 to 4 June 1984, winds are predominantly from the northwest
sector with some southwesterlies late on 4 June. Maximum daily ozone concentra-
tions on 4 June are 78 ppb (DLLS) upwind of Atlanta and 130 (DKLB) and 147 (CNYR)
ppb at distances of, respectively, 15 and 40 km downwind of Atlanta (see Figure 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 in the next section, does not contain any unusual meteorological
89059r2 2
-------
TABLE 2. Highest ozone days in Atlanta, 1984 - 1987.
Rank
1
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.
168.
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 t or
130890002
130890002
130970002
132470001
130890002
130890002
130890002
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. MLK Columbus Conyers
DKLB SWTR MUM 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
8/ 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.
163
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?.
13?.
86.
80.
137.
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.
89059
10
-------
86.0
35.0
34.5
34.0 -
33.5
33.0
85.5
85.0
84.5
84.0
83.5
83.0
82.5
32 5 -
35.(
34.£
34.C
33.5
33.0
32.5
86.0
85.5
85.0
64.5
84.0
83.5
83.0
82.5
Figure 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.
89059
11
-------
conditions it appears to be an appropriate day for the city-specific biogenic analysis
for Atlanta 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 model-
ing episode will be initialized on 3 June 1984 and terminated the evening of 4 June
1984. Thus, UAM modeling inputs will be prepared for 3 and 4 June 1984 with the
actual starting time on 3 June to be determined from an inert tracer simulation.
It should be noted that other days listed in Table 2 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 day
closer in time to current, 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 1 (201 ppb observed on 31 July 1987) appears not to be a prototypical event
since it is almost 20 percent higher than any of the other high ozone events over the
four year period.
MODELING DOMAIN SELECTION
The modeling domain for the 3-4 June 1984 UAM modeling episode should include a
fairly large fetch upwind of the city of Atlanta in order to account for sufficient
amounts of biogenic emissions upwind of the urban core of Atlanta. The modeling
domain should also include enough area downwind of Atlanta in order to capture the
maximum ozone concentrations that will be formed. Grid spacing should be suffici-
ently small in order to resolve the anthropogenic emission distribution in Atlanta. To
determine the UAM 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 Dekalb were analyzed to determine the
mean flow conditions that existed on 3-4 June 1984.
Meteorological Conditions on 3-4 June 1984
The days of 3 and 4 June 1984 saw the passage of a high pressure system. The 500
mb height contours at 0700 EST on 4 June 1984 shows the approaching high pressure
ridge towards Atlanta. The 500 mb 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 aloft the winds were in the same direction, but stronger.
As the 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
morning 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
89059r2 2
12
-------
conditions during the modeling period were clear, hot and humid. Maximum temper-
atures were in the upper 80's and dew point temperatures in the low 60's.
Modeling Domain Definition
Figure 1 gives a region-wide perspective of the modeling domain for Atlanta. As
pictured in Figure 1, this modeling domain consists of a 40 by 40 array of 4 km
square grid cells. The proposed modeling domain origin is located at UTM coordin-
ates 660 km easting, 3665 km northing in zone 16 and extends 160 km in the east and
north direction. As stated in the Atlanta UAM modeling protocol (SAI, 1988) 5
vertical layer structure is to be used in the UAM: two below the mixing height and
three above. The region top will be based on the maximum mixing height occurring
during the modeling episode.
In order to accommodate the city-specific analysis of the effects of biogenic emis-
sions on urban ozone formation in Atlanta, there is a region 75 to 100 km wide
upwind of Atlanta. This will allow for a 4 to 12 hour loading of biogenic emissions
into the atmosphere upwind from the outskirts of Atlanta.
89059r2 2
13
-------
it PREPARATION OF MODEL INPUTS
An important requirement of the PLANR use of the UAM is that the procedures used
to prepare model inputs should be consistent with traditional applications of the
model. Procedures to be followed in preparing UAM inputs with limited data availa-
bility must be flexible and clearly stated. When there is a lack of data, there must
be recommended procedures to be used. However, flexibility is an important com-
ponent of the construction of UAM inputs when limited data are available. These
procedures are currently being refined and evaluated to determine the optimum
methodologies for generating UAM inputs with limited data availability.
The preprocessor programs supplied with the 1978-1980 version of the UAM generally
rely on an intensive measurement program; data are interpolated from these mea-
surements to obtain gridded fields of input parameters required by the UAM. Over
the last 10 years, many of these programs have become outdated. More recent
applications of the UAM have prepared input files using only routinely available
(although fairly dense) meteorological and air quality data (Hogo, Mahoney, and
Yocke, 1988). However, the procedures used for input preparation have been devel-
oped on a case-by-case basis and are tailored to data availability. In this section we
discuss the initial application of the PLANR procedures for preparing meteoro-
logical, air quality, and terrain inputs for the UAM in which UAM modeling inputs
are prepared using only routinely available data.
ROUTINE DATA AVAILABLE FOR ATLANTA
On June 3 and 4, 1984 there were six surface meteorological observation sites opera-
ting in and around the city of Atlanta. Table 3 lists the six surface sites along with
the locations and meteorological variables available at each site. There were no
upper-air observation sites located within the UAM modeling domain. Thus, we made
use of upper-air observations from five sites that surround the UAM modeling domain
for generating wind and mixing height inputs. These five upper-air sites, their loca-
tions, and the distance to the city of Atlanta are given in Table 4. 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 as listed in Table 5. There were no air quality data available near
the boundaries of the modeling domain for use in prescribing boundary conditions.
89059r2 2
14
-------
TABLE 3. 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
729.9^7
749.724
772.248
752.780
UTMY Variables Measured
3726.148 WS, WD, T, TD, P
3755.498 WS, WD, T, TD, P
3740.709 WS, WD, T, TD
3752.306 WS, WD, T, TD
3719.869 WS, WD
3730.991 WS, WD, T, TD
89053 3
15
-------
TABLE 4. 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'
5V
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 5. 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
89053 3
16
-------
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. The Athens Georgia upper-air site was
felt to be most representative of upper-air conditions in Atlanta because its proxim-
ity to Atlanta. For three of the surface meteorological observation sites that
recorded temperature (Fulton County Airport, Dobbins Air Force Base, and South
Dekalb monitor) 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 observations
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,^00 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,
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-
890S9r2 2
17
-------
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
corresponding to evaluation runs number 1 and 2 and a sensitivity test reported in
Appendix G. As will be discussed later in the section on the evaluation of the UAM,
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. These scalars are the NO? photolysis rate constant, the concentration
of water vapor, the temperature gradient above and below the inversion base, the
atmospheric pressure, and the exposure class. The N©2 photolysis rates were calcu-
lated for the CB-IV mechanism using procedures described by Schere and Demerjian
(1977) and actinic flux data collected by Bass and co-workers (1980) (see Gery,
Whitten, and Killus, 1988). 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 atmo-
spheric pressure value of 1 atmosphere 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-
89059r2 2
18
-------
sity: a value of 2 at night and daytime values of either 0 (one hour day/night transi-
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 June 3, 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 to "clean values" as fol-
lows:
VOC 25 ppbc (using EKMA default speciation)
ISOP 0.001 ppb
NOX 1 ppb (split 3/4 N02, 1/4 NO)
40 ppb
200 ppb
O,
CO
89059r2 2
19
-------
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 Atlanta,
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. As will be dis-
cussed in the section on the evaluation of the UAM, 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 underestimate background concentrations because of the bio-
genie emissions which increase background VOC and ISOP (isoprene) concentrations
above the clean background 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 evaluation simulation, a 1 ppb
ISOP boundary condition was prescribed and VOC boundary conditions were set at 40
ppbC, where 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 evaluation 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.
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 several step process which includes updating mobile source emis-
sions from MOBILE-3 to MOBILE-4 based emission factors, applying summer week-
89059r2 2
20
-------
day and episodic temperature adjustments, gridding of data using a known surrogate
distribution, and speciation of the VOC and NOX emissions into the CB-IV mechanism
species.
The NAPAP motor vehicle emission categories were split into exhaust and evapora-
tive emissions based on splitting factors using 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 episodic temperature condi-
tions on 4 June 1984 to obtain episodic evaporative and exhaust emissions. Estimate
of running loss emissions (which are not included in the NAPAP inventory) were then
obtained by applying the ratio of the running loss to the exhaust emission factors
from the MOBILE-4 to the mobile exhaust VOC emission rate.
The county-wide area source emissions were then mapped to the 40 by 40 4 km
modeling domain by assigning each area source Source Classification Code (SCO
category to a known surrogate distribution. Known surrogate distributions include:
agriculture, urban, rural, and water from the national Geographical Information Ser-
vices 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 4 June 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 resultant
anthropogenic and biogenic emission rates and their spatial distribution are given in
Appendix A. Of particular note is that biogenic VOC emissions account for approxi-
mately 55 percent of the VOC emissions within the modeling domain.
89059r2 2
21
-------
PTSOURCE
This file contains the point source information, including the stack height, tempera-
ture and flow rate, the plume rise, the grid cell into which the emissions are emitted,
and the emissions rates for NO, NC^, 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 sour-
ces 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 apply-
ing the summer weekday 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. As shown in Appendix A, point source NOX
(including electrical utilities) emissions account for approximately 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 the
digitization of a standard 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, 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 mechan-
ism (see Morris et al., 1989a).
SIMCONTROL
This file contains the simulation control information, such as the time of the simula-
tion, file option information, default information, and information on integration and
chemistry time steps.
89059r2 2
22
-------
5 DEVELOPMENT OF A PLANR BASE CASE FOR ATLANTA
One of the key components of the PLANR use of the UAM is a limitation on the
number of diagnostic simulations used to arrive at a base case. This is achieved by
relaxing the strict model performance standards expected of the UAM in the past.
Although the goal is to achieve a statisfactory level of performance with as few
diagnostic simulations as possible, the model must show some skill in predicting
ozone observations in order to have confidence that the model will respond properly
to changes in emissions.
The minimal performance goal in the past was to have the predicted regional maxi-
mum 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 per-
formance cannot be expected to always be as good in the past. However, there
should be some minimal expectations of model performance since incorrect charac-
terization 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. However, model evaluation should always
include discussions on whether the right answer is being obtained for the right rea-
son. In addition, it may be useful to discuss approaches that may improve UAM per-
formance but which also may deviate slightly from the demonstration of the PLANR
use of the UAM. The PLANR use of the UAM for Atlanta discussed here involved
three diagnostic simulations before an adequate base case was obtained that satisfied
the minimal performance goal.
89059r2 2
23
-------
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 3une 4 upper-air 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. Appendix B contains the wind fields
generated by the DWM using the routine data from Atlanta and the procedures dis-
cussed above.
Hourly average predicted ozone concentrations for evaluation run 1 are contained in
Appendix C. A comparison of predicted versus observed hourly ozone concentrations
for the three monitors within the UAM modeling domain is given in Figure 2. As
seen in Appendix C and Figure 2, the regional maximum predicted ozone concentra-
tion is 13.54 pphm, within 8 percent of the observed maximum of 14.7 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 evaluation 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 maximum daily observed ozone con-
centration 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 3une
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
89059r2 2
24
-------
160
160
- 120
4160 160
i i i i i I i i i i i I i i i i i I i i i i i
- CNYR 03
OBSERVED CO
PREDICTED -
a
OBSERVED CO
PREDICTED
nmmmmmm
- 40
12
TIME: (HOURS)
18
12
TIME (HOURS)
18
24
160
120
m
Q- or\
0. 80
3
12
18
1 I I I I I I I I I I I I I II I I I I I I T
- DLLS 03
OBSERVED CO
PREDICTED
m
mmmm
160
120
80
40
6 12 18
TIME (HOURS)
24
FIGURE 2. Observed and predicted ozone concentrations (ppb)
for the Atlanta evaluation run 1.
89059
SYSTEMS APPLICATIONS. INC.
25
-------
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 2) 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 obser-
ved ozone concentration at DLLS is mainly due to entrainment of ozone concentra-
tions aloft as the mixing height rises rather than to local production of ozone 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 NOV boundary conditions (40 and 1 ppb, respectively) for diagnostic run 2.
yv
An examination of the observed winds, the predicted wind fields in diagnostic run 1
(Appendix B), and patterns of predicted hourly ozone concentrations (Appendix C)
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 elevated 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 consulted. The
CNYR wind observation site is located near a grove of trees and may not be repre-
sentative of mesoscale wind flows, and the Atlanta Hartsfield Airport wind observa-
tion site (which was inadvertently left out of the original analysis) was better sited
(D. Kemmerick, personal communication, 1989). Thus, the CNYR wind measurement
was eliminated and the Atlanta Hartsfield Airport winds added for the next evalua-
tion 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 of the diagnostic run 2.
An examination of the spatial distribution of emissions (Appendix A) and the predic-
ted hourly ozone concentrations from diagnostic run 1 (Appendix C) 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
89059r2 2
26
-------
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 (1) a higher VOC value was used (40
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 are shown in Appendix D. The resulting predicted
hourly ozone concentrations for evaluation run 2 are given in Appendix E. The pre-
dicted regional maximum ozone concentration is 13.2 pphm, within approximately 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,
at the location of the peak observed ozone, the maximum observation is replicated to
within less than 30 percent. In addition, the daily maximum observed ozone concen-
trations 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 4 and 5 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
concentration for the point, a one-cell search, and a two-cell search matches 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 matches 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.
89059r2 2
27
-------
160
120
0
12
18
I I I I | I I I 1 T [ I I I I I \ r I 1 I
u DKLB
OBSERVED [H
PREDICTED
24 0
160 160
J2
TIME (HOURS)
18
120 120-
40
24
160
- 120
i l i i i i l i l
- CNYR
- 40
6 12 18
TIME (HOURS)
24'
160
120
12
18
24
i i i r T | i
- DLLS
CD
80
40
OBSERVED [JJ
PREDICTED -
6 12 18
TIME (HOURS)
160
120
80
40
24
FIGURE 3. Observed and predicted ozone concentrations (ppb) for the Atlanta
evaluation run 2.
SYSTEMS APPLICATIONS. INC.
89059
28
-------
160
160
- 120
24 0
160 160
i t i i I i i i i i I i i i i i | i i i i i
- CNYR
D PREDICTED -
i i i i i i i i i i i i i i i i i i i i i i i
- DKLB
OBSERVED [JJ
PREDICTED
12
TIME (HOURS)
12 18
TIME (HOURS)
24-
160
120
12
18
24
i i i i i i r
- DLLS
CD
80
4O
OBSERVED
PREDICTED
6 12 18
TIME (HOURS)
160
120
80
24
FIGURE 4. Observed and nearest-neighbor predicted (one-cell search) ozone concentrations
(ppb) for the Atlanta evaluation run 2.
SYSTEMS APPLICATIONS, INC.
89059
29
-------
160
120
12
18
1 I I I I | I I I I I I I I I I I | II I I T
OBSERVED CD
PREDICTED
24 0
160 160
12
18
160
- 120
i i i i i | i i i
- CNYR
- 40
12
TIME (HOURS)
18
6 12 18
TIME (HOURS)
24
160
12
18
6 12 18
TIME (HOURS)
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
OBSERVED CD
PREDICTED
24
FIGURE 5. Observed and nearest-neighbor predicted (two-cell search) ozone concentrations
(ppb) for the Atlanta evaluation run 2.
SYSTEMS APPLICATIONS, INC.
89059
30
-------
Figure 6 shows a scatter plot and model performance statistics between predicted
and observed hourly ozone concentrations for evaluation Run 2. Also contained in
Figure 6 are analyses of the residuals of the predicted and observed ozone concentra-
tions. As indicated in Figure 6 and the time series plots in Figure 3, diagnostic run 2
tends to overpredict the nighttime ozone concentrations and underpredict the peak
daytime values. As a result, the bias between predicted and observed 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% less than the observed peak (14.7 pphm).
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.
89059r2 2
31
-------
120.00
90.00
a.
CL
60.00
cc
0.
30.00
I 1 i
I I I I I I I I I I I I I I I I 1 T
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
SKEhNESS
KURTOSIS
OTHER MEASURES
MEDIAN
UPPER QUARTILE
LOWER QUARTILE
MINIMUM VALUE
MAXIMUM VALUE
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
FIGURE 6a. Scatterplot and model performance statistics for hourly ozone concentrations
and evaluation run 2 (N = 60).
89059
32
-------
0.20 -
-40.00 -20.00 0.00
RESIDUAL (OBS-PRED)
THE BINSIZE EOUALS 10.000
20.00
40.00
RESIDUAL ANALYSIS
AVERAGE -0.48384
STANDARD DEVIATION 25.16143
SKEWNESS 0.20968
KURTOSIS -1.08769
OTHER MEASURES
MEDIAN -0.23000
UPPER QUARTILE 18.80000
LOWER QUARTILE -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 6b. Residual analysis of observed minus predicted hourly ozone
concentrations for evaluation run 2 (N = 60).
89059
33
-------
77.3
58.0
40.9
31.8
20.7
12.7
£ -13.7
J> -21.7
m
Q
- -32.7
1 -41.9
£ -59.0
-78.2
- n
- n
D
n
a a a
a
°
m
a a
a
a
a
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
5.00
HOUR
10.00
15.00
20.00
THE LINEAR MODEL PARAMETERS
THE CORRELATION 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 6c. Plot of residuals versus time of day for
evaluation run 2 (N = 60) .
89059
34
-------
In addition, there is a very large uncertainty in the biogenic emission rates and
speciation as well as uncertainties in the chemistry of biogenic species. Another
possible explanation for the underprediction of the observed ozone peak at Conyers is
that the 1985 NAPAP inventory underestimates 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. 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 G.
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 G), 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.
89059r2 2
35
-------
APPLICATION OF THE UAM TO ATLANTA FOR
EMISSION REDUCTION SCENARIOS
The UAM was exercised for several different emission scenarios using the UAM
modeling inputs corresponding to evaluation run 2. Two base case emission scenarios
are used: with biogenic emissions and without biogenic emissions. Different across
the board anthropogenic VOC emission reduction scenarios are examined in order to
determine what level of anthropogenic VOC emission reductions are required in order
to reduce the peak observed ozone concentration of U.7 pphm to below the NAAQS
ozone standard of 12 pphm. The following emission scenarios were used:
Base case (all anthropogenic emissions) with biogenic emissions
30% reduction in anthropogenic VOC emissions
60% reduction in anthropogenic VOC emissions
90% reduction in anthropogenic VOC emissions
Base Case (all anthropogenic emissions) without biogenic emissions
30% reduction in anthropogenic emissions
60% reduction in anthropogenic emissions
90% reduction in anthropogenic emissions
TRACER SIMULATION
Before conducting a UAM modeling analysis, it is useful to perform a weighted tra-
cer simulation for the two principle ozone precursors, VOC and NOX, in order to
obtain a rough estimate of contributions of initial concentrations, boundary condi-
tions, and emissions to predicted concentration levels. In a weighted tracer simula-
tion, the UAM is run in an inert mode (i.e., no chemistry or deposition) with several
different "colored" tracers (species) that represent the major contributers to VOC
and NOX concentrations within the modeling domain. In the weighted tracer simula-
tion for Atlanta different colored tracers were used for initial concentrations, lat-
eral boundary conditions (four colors, one for each face), top boundary conditions,
anthropogenic area source emissions, biogenic area source emissions, and point
source emissions.
Appendix F displays the results of this tracer simulation at 0800, 1200, and 1600 LST
on it June 1984. By 0800 initial concentrations of NOX (Appendix F-la) and VOC
(Appendix F-2a) are almost completely advected out of the region and only affect
89059r2 2
36
-------
the extreme southeastern portion of the modeling domain. By noon (Appendixes F-3
and F-4), initial concentrations have no influence in the modeling domain. At noon,
NOX concentrations are dominated by emissions, where, downwind of Atlanta (i.e.,
the location of the maximum predicted ozone concentration) area source NOX con-
tributes over 70 percent, point sources contribute about 20 percent, and boundary
NOX contributes approximately 10 percent. Boundary VOC contributes a higher per-
centage to the total tracer VOC downwind of Atlanta than seen for NOX where the
percentage contribution of boundary, anthropogenic emissions, and biogenic emis-
sions at 12 noon are approximately 30, 50, and 20 percent. The relative contributions
to total NOX and VOC tracer at 1600 LSI is approximately the same as for 12 noon
(Appendixes F-5 and F-6).
In regions of the UAM modeling domain away from the influences of emissions from
the city of Atlanta, biogenic and boundary VOC dominate the total VOC tracer con-
centration. Because of the presence of several large power plants within the UAM
modeling domain (see Appendixes F-3d and F-5d) the point source NOX dominates the
NOX tracer concentrations downwind of the power plants and the boundary NOX con-
centration does not contribute significantly to NO tracer concentrations in the
interior of the UAM modeling domain.
UAM MODELING RESULTS
The predicted regional maximum ozone concentration for the base case with and
without biogenic emissions are, respectively, 13.22 and 12.28 pphm. The modeling
objective is to compare the levels of anthropogenic VOC control (with and without
biogenics) required to effect similar percentage reductions in the regional peak
ozone changes. Since the base case peaks are dissimilar, the simulations will address
the level of reductions required to reduce regional peak ozone by 18.4 percent
(equivalent to reducing the observed peak from 14.7 to 12.0 pphm).
Table 6 lists the predicted regional maximum ozone concentrations, percent reduc-
tion of the maximum ozone concentration, and the maximum predicted ozone con-
centration normalized to the observed peak value for each of the emission scen-
arios. The results from Table 6 are graphically shown in Figure 7. According to
Figure 7, to reduce the regional maximum ozone concentration 18.4 percent (i.e.,
reduce the maximum observation from 14.7 to 12.0 pphm), a 100 and 62 percent con-
trol of anthropogenic VOC emissions is required for the cases with and without bio-
genic emissions, respectively. (Since the UAM underpredicts peak ozone (13.2 pphm)
relative to the observed peak (14.7 pphm), model calculations for attainment were
adjusted by normalizing the predicted peak ozone to the observed peak.) If one
considers the ozone NAAQS as 12.4 pphm, then an 87 and 52 percent control of
anthropogenic VOC emissions are required to reduce the predicted regional maximum
ozone concentration by 15.6 percent for the, respectively, scenarios with and without
biogenic emissions. In either case, the model calculations indicate that the inclusion
of biogenic emissions in VOC control strategies for Atlanta on this day results in an
89059r2 2
-------
TABLE 6. Regional maximum ozone concentrations (pphm) predicted by the
UAM for the different emission scenarios.
Maximum Ozone
Concentration
pphm
With Biogenics
0* VOC Reduction 13.22
30% VOC Reduction 12.68
60% VOC Reduction 11.91
90* VOC Reduction 11.08
Without Biogenics
0% VOC Reduction 12.28
30* VOC Reduction 11.16
60% VOC Reduction 10.09
90* VOC Reduction 8.98
Percent
Reductions
from Base Case
Maximum Ozone
Normalized to
Peak Observation
(pphm)
Peak Observed
Ozone NAAQS*
14.7
12.0
0.0
18.4
14.7
12.0
0.0
4.1
9.9
16.2
0.0
9.1
17.8
26.9
14.7
14.1
13.2
12.3
14.7
13.4
12.1
10.7
Technically, the ozone NAAQS is 0.12 ppm rounded. Thus, an ozone
concentration of 12.4 pphm (15.6* reduction from observed peak) is
considered attainment.
69059rl 3
38
-------
30
25
-S 1 20
II
80
I
E
i
.,.
15
10
Attainment (0.12 ppm)
Attainment (0.124 ppm)
10 20 30 40 50 60 70 80
Percent Reductions of Anthropogenic VOC Emissions
90 100
EEE 89059
FIGURE 7. Relationship between percent reduction of anthropogenic VOC
emissions to percent reduction of the regional maximum ozone concentration
with and without biogenic emissions.
39
-------
additional 35 to 40 percent control on anthropogenic VOC emissions over the case
when biogenic emissions are not included.
COMPARISON WITH A PREVIOUS STUDY
As noted previously, a similar analysis of effects of biogenic emissions on VOC con-
trol strategies for Atlanta on this day was carried out by Chameides and co-workers
using the EKMA model (Chameides et al., 1988). Over the 15-hour EKMA modeling
period they estimated biogenic emissions to be 30 kg/km for isoprene, and a total
biogenic VOC emission rate of 50 kg/km . They also estimated daily total biogenic
emission rate of 65 kg/km . In their EKMA analysis, biogenic emission rates were
varied from zero to 50 kg/km and were assumed to be isoprene. The EPA has sum-
marized biogenic VOC flux rates for 12 other biogenic VOC inventories (EPA,
1984). These values for the other biogenic inventories generally range from 780 to
2,540 vg/m -h, with one extreme value at 8,890 yg/m -h. The biogenic emission rates
of 30, 50, and 65 kg/km reported by Chameides and co-workers translate into bio-
genic flux rates of 2,000, 3,333, and 2,708 yg/m-h. (Note that the 65 kg/km2 is a
daily total.) In our study we used a 24-hour average biogenic emission flux for the
UAM modeling domain of 2,197 ng/m -h. Thus our study and the work of Chameides
and co-workers used similar estimates of biogenic emission flux rates that are on the
high end, but well within the range of biogenic fluxes of other studies (EPA, 1984).
Due to the large amount of foliage in the Southeast, it is expected that biogenic
emission fluxes for the Atlanta region are on the high end of continental flux esti-
mates.
However, there are some differences in the emissions used in this study and those
used in the work of Chameidies and co-workers. These differences include the
speciation of the biogenic emissions into CB-IV species and the total anthropogenic
VOC and NOX emissions used. Chameides and co-workers assumed that all of the
biogenic emissions were isoprene in their EKMA analysis, whereas our study assumed
that isoprene made up only 19 percent by weight of the 24-hour total biogenic emis-
sions. (Note that isoprene contributed about 25 percent of the total biogenics for the
15 daylight hours.) The less reactive CB-IV species PAR made up the largest portion
of the biogenic emissions in this study (65 percent by weight), with the remainder of
the biogenic emissions assumed to be the CB-IV species OLE. Since isoprene is much
more reactive than PAR, Chameides and co-workers used a more reactive biogenic
emissions inventory than was used in our study.
Chameides and co-workers estimated that without biogenic emissions, a 30 percent
reduction of anthropogenic emissions is required in order to reduce the maximum
observed ozone concentration of 14.7 to 12.0 pphm. In contrast, we estimated that a
62 percent control of anthropogenic VOC emissions is required to reduce the maxi-
mum observed value from 14.7 to 12.0 pphm without biogenics. There may be sev-
eral reasons for the differences in VOC emission control requirements predicted by
the UAM and EKMA.
890S9r2 2
40
-------
(1) A lower base case maximum daily ozone concentration (12.28 pphm) was
used in UAM than was used in the EKMA (14.7 pphm). Thus, a larger per-
centage of the UAM-predicted maximum daily ozone concentration is due
to background concentrations, which will not be affected by emission con-
trol strategies.
(2) The UAM was exercised with boundary conditions that reflected the pre-
sence of upwind emission sources (anthropogenic and biogenic). Thus, again,
a larger fraction of the UAM-predicted maximum daily ozone concentration
is due to background, which will not be affected by anthropogenic VOC
reductions.
(3) Due to changes in the reactivity and chemistry of the atmosphere in the
different emission reduction scenarios, the regional maximum ozone con-
centration predicted by the UAM does not occur at the same location in
each scenario. If changes in emissions causes the maximum ozone concen-
tration to lie outside of the EKMA trajectory, then the maximum observed
ozone cannot be simulated by EKMA, and EKMA will overstate the ozone
reductions due to the emission reductions.
(4) The observed vertical wind profile at Athens, Georgia and the UAM layer
1 and 2 wind fields both exhibit some wind shear in the mixed-layer. This
wind shear, which cannot be simulated by EKMA, results in a larger portion
of the predicted maximum ozone concentration being attributable to back-
ground conditions rather than to urban emissions from Atlanta.
(5) The UAM modeling analysis included the emissions from several large power
plants in the region (50 percent of the NOX inventory) that were not
accounted for in the EKMA analysis. These NOX emissions combined with
the background VOC concentrations may produce ozone concentrations that
will not be affected by the anthropogenic VOC emission reductions.
(6) The anthropogenic emission inventory used in this study most probably
underestimates actual anthropogenic VOC emissions. The annual NAPAP
inventory, which is probably an underestimate to begin with, is adjusted to a
typical summer weekday based on average summer conditions. Ozone epi-
sodes tend to be highly correlated with hotter than normal temperatures,
although 4 June 1984 was not an excessively hot day. Except for mobile
sources, which were adjusted to episodic temperature conditions based on
MOBILE-4, VOC emissions (e.g., solvent use and other evaporative emis-
sions) were not adjusted to the higher emission rates expected on a hotter
than normal day.
(7) The wind speeds may be too high in the UAM simulation, causing excessive
dilution of the urban plume.
89059r2 2
41
-------
Of particular note in these differences is the underprediction of the observed peak
ozone concentration (14.7 pphm) by the UAM in evaluation run 2 (13.2 pphm). To
investigate this effect, a sensitivity test reduced the observed wind speeds at some
of the surface sites. The results of this sensitivity test are discussed in Appendix G.
Despite the differences between our study and the study reported by Chameides and
co-workers (1988), both studies conclude that by including biogenic emissions in an
attainment demonstration for Atlanta for June 4 1984, an additional 35 to 40 percent
anthropogenic emission reduction over the case without biogenics is required to meet
attainment of the ozone NAAQS. Chameides and co-workers report that the anthro-
pogenic VOC emission control requirements increase from 30 to 70 percent, whereas
we report that an increase from 62 to 100 percent is needed. However, because the
UAM underpredicts the peak observed ozone concentration, and for the reasons cited
above, it is probable that the UAM is overstating the VOC emission control require-
ments.
DISCUSSION
There were two main objectives of this study: (1) the demonstration of the PLANR
use of the UAM for the city of Atlanta, and (2) the analysis of the effects of biogenic
emissions on VOC control strategies. Using only routinely available data, it has be
shown that adequate UAM modeling inputs can be created using objective techniques
and a minimal amount of diagnostic simulations. It has also been demonstrated that
biogenic emissions do affect VOC control strategies for ozone attainment. However,
in a sense these two objectives are not entirely consistent with each other, since
more accurate results of the effects of biogenic emissions on ozone attainment
demonstration would most likely be obtained if the UAM base case simulation exhibi-
ted better model performance. Improved UAM modeling inputs could be generated
either by using data from an intensive measurement network or through modifying
the existing inputs by an experienced UAM model user. However, neither of these
approaches is consistent with the demonstration of the PLANR use of the UAM.
Because of this conflict of objectives, additional uncertainties are introduced into
the analysis of the effects of biogenic. These uncertainties are addressed somewhat
in the sensitivity analysis discussed in Appendix G.
As noted previously, several other sources of uncertainity in the analysis need to be
recognized when interpreting the results. Foremost among these are the uncertain-
ties in the biogenic and anthropogenic emission inventories. These uncertainties may
be biased in such a fashion that the calculated effects of biogenic emissions presen-
ted here possibly are overestimates of the actual effects. This is because the bio-
genic VOC emissions may be overestimates of actual emissions, and the anthropo-
genic VOC emissions are most likely underestimates of actual emissions. Recent
analysis of biogenic emissions have indicated that 15 to 35 percent of the biogenic
emissions are reacted away before leaving the forest canopy, thus those emissions
89059r2 2
42
-------
cannot contribute to the formation of ozone. The 1985 NAPAP inventory only
included major VOC emission sources; smaller sources, such as small evaporative
sources whose annual emission rate was lower than the cutoff, were not included in
the analysis. Many of these smaller sources have peak emission rate during the
summer. The uncertainties in the emission inventories contribute substantially to
the uncertainties in this analysis. It is expected that use of a better quality anthro-
pogenic emission inventory, representative of actual emission rates, will not only les-
sen the uncertainties in the analysis of the effects of biogenic emissions, but will
also improve model performance of the base case.
Even with the uncertainties in the modeling analysis presented here, two important
conclusions can be drawn concerning attainment of the ozone standard for the city of
Atlanta. First, biogenic emissions do contribute to some extent to VOC concentra-
tions in the Atlanta region and therefore may lessen the effects of ozone attainment
strategies aimed solely at reducing VOC emissions. Although it is suspected that the
results presented here are overestimates (i.e., the actual effects of biogenic emis-
sions may be less than presented here), biogenic emissions should be recognized in
future modeling studies. The second major conclusion is that a high-quality emission
inventory that represents actual emission rates is needed to properly calculate the
effects of emission reduction strategies.
89059r2 2
43
-------
References
Ames, J., T. C. Myers, L. E. Reid, D. C. Whitney, S. H. Golding, S. R. Hayes, and
S. D. Reynolds. 1985a. SAI Airshed Model Operations Manuals. Volume I
User's Manual. U.S. Environmental Protection Agency (EPA-600/8-85-007a).
Ames, 3., S. R. Hayes, T. C. Myers, and D. C. Whitney. 1985b. SAI Airshed Model
Operations Manuals. Volume IISystem's Manual. U.S. Environmental Protec-
tion Agency (EPA-600/8-85-007b).
Bass, A. M., L. C. Glasgow, C. Miller, 3. P. Jesson, and D. L. Filken. 1980. Planet
Space Sci., 28:675.
Boris, 3. P., and D. L. Book. 1973. Flux-corrected transport: I. SHASTA, a fluid
transport algorithm that works. 3. Comp. Phys., 11:38-69.
Burton, C. S. 1988. Comments on "Ozone Air Quality Models." 3. Air Pollut. Con-
trol Assoc., 38(9): 1119-1128.
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.
Chock, D. P., and A. M. Dunker. 1983. A comparison of numerical methods for solv-
ing the advection eolation. Atmos. Environ., 17:11-24.
Chock, D. P. 1985. A comparison of numerical methods for solving the advection
equationII. Atmos. Environ., 19:571-586.
Douglas, S., and R. Kessler. 1988. "User's Guide to the Diagnostic Wind Model. Ver-
sion 1.0." Systems Applications, Inc., San Rafael, California.
EPA. 1984. Air Quality Criteria for Ozone and Other Photochemical Oxidants
Draft. Volume II. U.S. Environmental Protection Agency (EPA-600/8-84-
020A).
EPA. 1986. Guideline on Air Quality Models (Revised). U.S. Environmental Pro-
tection Agency (EPA-450/2-78-027R).
89059 5
44
-------
Federal Register. 1987. "State Implementation Plans; Approval of Post-1987 Ozone
and Carbon Monoxide Plan Revisions for Areas Not Attaining the National
Ambient Air Quality Standards; Notice." Federal Register, Vol. 52, No. 226
(November 24, 1987).
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., D. R. Souten, T. W. Tesche, L. R. Chinkin, H. Hogo, and M. C. Dudik.
1986. "Evaluation and Applications of the PARIS Photochemical Model in the
South Central Coast Air Basin." Volume I. Systems Applications, Inc., San
Rafael, California (SYSAPP-86/065).
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 June 1985." Systems Applications, Inc., San Rafael, Califor-
nia (SYSAPP-88/138).
Morris, R. E., M. W. Gery, M. K. Liu, G. E. Moore, C. Daly, and S. M. Greenfield.
1989b. "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, H. Hogo, L. R. Chinkin, L. A. Gardner, and R. G.
3ohnson. 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).
O'Brien, 3. 3. 1970. A note on the vertical structure of the eddy exchange coef-
ficient in the planetary boundary layer. 3. Atmos. Sci., 27:1213-1215.
OTA. 1988a. "Urban Ozone and the Clean Air Act: Problems and Proposals for
Change." Office of Technology Assessment, Washington, D.C.
OTA. 1988b. "Ozone and the Clean Air Act: Summary of OTA Workshop with State
and Local Air Pollution Control Agency Officials." Office of Technology
Assessment, Washington, D.C.
OTA. 1988c. "Ozone and the Clean Air Act: A Summary of OTA Workshops on
Congressional Options to Address Nonattainment of the Ozone Standard."
Office of Technology Assessment, Washington, D.C.
89059 5
45
-------
Rao, S. T. 1987. "Application of the Urban Airshed Model to the New York Metro-
politan Area." U.S. Environmental Protection Agency, Research Triangle Park,
North Carolina (EPA-450/4-87-011).
SAI. 1988. "Protocol Document for Urban Airshed and EKMA Modeling in the
Atlanta Metropolitan Area." Systems Applications, Inc., San Rafael, California
(SYSAPP-88/157).
Schere, K. L. 1983. An evaluation of several numerical advection schemes. Atmos.
Environ., 17:1897-1907.
Schere, K. L., and K. L. Demerjian. 1977. "Calculation of Selected Photolytic Rate
Constants over a Diurnal Range. A Computer Algorithm." U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina (EPA-600/4-77-
015).
Science. 1988. Rural and urban ozone. Editorial in Science, 241(4873): 1569.
Seinfeld, J. H. 1988a. Ozone air quality models. A critical review. 3. Air Pollut.
Control Assoc., 38(5):616.
Seinfeld, 3. H. 1988b. Closing remarks. 3. Air Pollut. Control Assoc.. 38(8);1136-
1137.
Smolarkiewicz, P. K. 1983. A simple positive definite advection scheme with small
implicit diffusion. Monthly Weather Review, 111:479-486.
Whitten, G. Z., J. P. Kiilus, and H. Hogo. 1980. "Modeling of Simulated Photo-
chemical Smog with Kinetic Mechanisms. Volume 1. Final Report." Systems
Applications, Inc., San Rafael, California (EPA-600/3-80-028a).
Whitten, G. Z., T. C. Meyers, C. Daly, L. R. Chinkin, S. D. Reynolds, N. M. Yonkow,
and B. Austin. 1985. "Application of the Urban Airshed Model to Kern
County." Systems Applications, Inc., San Rafael, California (SYSAPP-85/200).
Yamartino, R. 3., and 3. S. Scire. 1984. "ADOM/TADAP Model Development Pro-
gram. Volume 3. The Transport and Diffusion Modules." Environmental
Research & Technology, Inc., Concord, Massachusetts (P-B980-210).
Yocke, M. A., R. E. Morris, H. Hogo, L. R. Chinkin, and L. A. Mahoney. 1985.
"Analysis of the Air Quality Impacts of the San Miguel Project. Volume I."
Systems Applications, Inc., San Rafael, California (SYSAPP-85/127).
Zalesak, S. T. 1979. Fully multi-dimensional flux-corrected transport algorithms for
fluids. 3. Comput. Phys., 31:335-362.
89059 5
-------
Appendix A
EMISSIONS DATA USED IN THE APPLICATION
OF THE UAM TO ATLANTA
89059
-------
O)
Q.
c
O
II
II
II
Q. II
OO II
I II
II
II
II
>.-«oooooooo.-'Oooorocr>
in
in
re
TJ
t.
O
10
o
01
c
o
i-
o
IS) II
II
II
II
II
II
II
II
1=1 II
<_> II
II
II
II
OOOOr1.- C
>.-'OOOOOOOOOOOOOTCO
TOOOOOOOOOOOOOO1CSJ
>riOOOOO>-iOOOOOOOOOr-.OOOOPJCTl
OOOOiiOOOOO«TOOr-»OOOCDOOOOOOOOOOOOOOOOOOOOOOOOOr~»OOOO *T o
II
z II
h- II
II
II
II
OCDOOOC3OOOOOOOtMlOO«3T-.O)ir>O<*)OOUlOC\jromOOir>O>iCMCOtMOOOOOOr-4OOOOOPO
1I t< VO T< r^ r-H^^fO ri tIIO O01
II
II
a> ii
E II
ro ii
Z II
II
a>
U
S-
II
II
II
O II
II
II
-------
II
II
II lv.,-iOOOOa>OrOCOOOOOOOI-» r-i O i-<
O_ II COOOOOOCMOCTITOOOOOOI^ «T O *T
O1 II .................
I II TOlOOOOr-tOOOOOOOOOO ro O CO
ii CM in r»- r-v
II CM CSJ
II
in ii
in ii
i i ii meooooo^TTioi loooooo' ' o r->
co o ii .... .............
tO II OTOOOOCMOO'-'OOOOOOin OO O OD
>> II 1 I ID V- 1 * I
o 01
-» XII r-~OOOOOf-«CM«TrOOOOOOOCJ> r«. O l-»
>l O II .................
10 z ii cMr-oooocnooaioooooor^ o o o
"O II i «CM « * lOiniT)
^ Hi* ro r^* r*-.
II
c
o
-M II
^^ II
ii inkooooo" 'inoi^roooooocM < i o
>, II rOroOOOOf^-OI'-'OOOOOOOOOO 1-1 O r-H
S-OII ................. «
O LJU cooOOOO^or^CMCMOOOOOOr*. r-. o r**
otiitom ^-IT CMT ^r
CU II i-H r- 1 f <
4-> II CM CM
10
o
II
0) II
O II cli lOOOOCTvOll"^- 'OOOOOOCM m o IT)
t- LJU in«TOOOOOO«TO^inOOOOOOin ii fOi-i n r^ o
II O
X II t-l --1 «M
to ii
Ol II rir-.OOOOO»CMP-.OOOOOOOO CTl o O»
t- on nr~.oooocMoo'-'O>ooooooT eo o oo
X I II CMrOOOOOTLOr-O^OOOOOO'-' ri r-I CO
II fO < » r-* *T t*^, ^^
i Ii
03 II
a>
a.
c
o
in
i
oo
:*:
<_)
00=3
i^o:
<->(
hLU LU
OO I LU LU Z UJ 3 OO O LU
OOLU (_> >XUJZ: OOOLU i ID
- >- LU ^ 11^ O*(-> LU LU O Z O
|_|__| O^LU ^i Q. h-h-OO LU > +
ii =3rx_> o h-i LU zo.«t- LU-LUi_i_JO i_) oo u>
O> II (_) _JOOO <: «S LU h- L3 X LU Q_ LU i
EII >->-QCC«:>-.Q£ZOOO!:Q:_J-oooo: z z z
(OH >^OLUCO *4Q.LJLJ*i I* OLL. LU LU LU
II LU uj O I Zu-osxi-Ol ^uj z: <£ (_> o u O
on oc Q.ZLUQ.OZO
(.11 LU LUCCOOOOO QCLUOf
il I tOOOl => I O
II O Z
-------
tr> ii
ir> ii
«i Z II OOOO>'Oi'OOOOOOOOOOOOOOOOO.-'OOl^OJOOOl'
i'OOOOOOOCDOOOrOi
a:
o
>>
IO
a
c
o
o
a*
OJ
II
II
II <
I II
> II
X II
II
II
II
>r-iOOOr-.-iOOOOOOOOOT-iO«TT-iOOOOOOOOOOOOCMO
in.'
jr^i->oooooooooocsjocncsj
II
II
II OOOOCOiiinr
o
oo
OO P
oooocooooooooooooioro
CD
Q.
c
o
UJ O
to»-
UJ U~ Q£ OO
OO UJ< OO
oooouz uj
UJ ^ LJ
OLD3E Z O
»i o
OO UJ OO I Z i Q£ QtQ
Q-Z
0£ «S I 00 00 Q£
100
_IUJ
ZJOOOO
<_>ooujo:
II UJOO UJ
QCUOOQ-
tD O UJ ^
hOOOOOOO
-
ItI
LJ =3:
-------
in
in
* i
^
CO
>^
ia
0
O
lf_
~~-
_^
>>
o
C71
CD
M
IO
0
a>
u
i-
o
co
>^
JD
T-,
at
en
s-
z
1
ra
o
1
0>
a.
^j
«/>
c
o
z
Q£
CO
U-
_l
X
_J
I
cc
^
a.
UJ
p
0
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
It
II
II
II
It
II
»
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
»
II
II
II
II
P*» t t O O O O CM **~ ' PO ^O O O O O O O ^"
Or-.OOOOCMO'-'COOOOOOO.-t
ooooooooooooooooo
CMr-.OOOO03CTlCM*TOOOOOOCM
^HCMOCSOOOrnO'-'OOOOOOr-i
CMOOOOOOOOOOOOOOO^-*
r^-inOOOOCvii^CM^TOOOOOOO
^rcTioooocrvcMO^-oooooor«-
r-«CvlOOOOOOOOOOOOOOO
r~* iO ^D ^D CD ^D CTl i^J *~* C^J ^D ^3 C^ CD ^D ^D r*»
CMlOOOCDOOi-«OOCMOOOOOOai
COiOOCiOOCMTOCMOOOOOOin
CM
r*» IO O O O O *T *T ro CO O O O O O O CM
r-.!^OOOOCMr-i,--
CO
CM
r4
in
f^
CM
in
VO
1 1
en
01
CO
f»v
CM
to
^
(_>
tors I
^ac z
C_)h LU LO
=3 Z LU O
LJ CO
a: z z
(/) I LULUZLU Z3 (/I O LU
COLU LJ >XLUZ OLOLU L3
->-LU > I I => O-CJ LU LL. O TC O
n=>r3O o i i LU z a. - LU«£ U-U- >-UJ _I_JQ LJ CO <->
OJ II LJ -JOCO ->-Q;Q;Icezcoc£o£_i> ocoo; > z z z
(aii^^oujca ilO-LJLj*iit' OLU LU LU LU
zii- O.
u ii a: O.ZLUCL.OZO
S- II LU LUCCCOOCO GfLUQt:
a ii z LUH_J~Z 3C u 3:
OH t COLJOI3 I O H-
CO II CD Z . Z
'CMninvoT^eoeriCficncftO o o o
i_)nr-r~-r^r^cococooococococococococoai i t i
-------
LT>
LO
T3
S-
o
c
O
S-
o
en
00 Q£
=J LU
o z:
z. -x. s.
o o
~~» o
i ta
ocozooo
=D Z U-IZ
Q^ot^fl:
OZ^ll
ot:.) oo
Q. U. t_) I LU
J LU<: LJ
on
en:
C£.
rj
loo
_1LU
3OOOO
ILUOZ
-------
LO
LO
,__,
T
CO
>^
It)
o
o
4-
*~-
m
T3
**^,
in
O
.U
*_*
>>
S
0
CT>
01
4_>
>a
ai
CJ
i-
o
i/i
>^
JD
a
QJ
01
S-
QJ
^
r
«3
O
1
OOOOOOOOO
OOOOOOOOOOOOOOOOO
ooooooooooooooooo
o o o o o o o o o o o o o o o o o
ooooooooooooooooo
ooooooooooooooooo
o o o o o o o o o o o o o o o o ^
ooooooooooooooooo
ooooooooooooooooo
^. o O O O O CM i*D P*- ^" O O O O O O CO
OTTOOOO'CMCMr^OOOOOOO
-lOOOOOOOOt-iOOOOOOO
OHOOOOO- ivOCOTOOOOOOCM
q-lOOOOOCMCMOUIOOOOOOO
r-iOOOOOOOOOOOOOOOr-H
CO
«T
O
O LO
IO VO
CM CM
O
O
CO
10
O
o
CO
to
r-H O
r^ o r--.
O
o
o
o>
10
cvj
CD
o
O.Z UJQ.
L>jeCOOOOOOOO»-
-------
N' ' o°
' cP cP N
O N0^ <& -7
10
20
ATLANTA 1985 base
NOx
665
Area source and low-level point source NOX emissions (kg/day)
-------
3705
665
ATLANTA POIKT SOURCE EMISSIONS
NOx (kg/day)
total for region
-------
.j.4.1. (:::L . i. .i.l.i. j . . i . i . f.- :::i. .,. j. t .
3705
gt^MVi^iiilyMB
ATLANTA MOTOR VEHICLE EMISSIONS
NOx (kg/day)
total for region
-------
3705
10
20
30
665
ATLANTA MOTOR VEHICLE EMISSIONS
NOx (kg/day)
Limited Access Roads
-------
10
ATLANTA AREA SOURCE EMISSIONS (NO MV)
NOx (kg/day)
total for region
-------
N°
0
0
10
ATLANTA 1985 base
RHC
Anthropogenic area source and low-level point source plus biogenic
VDC emissions (kg/day).
-------
825
- 3785
3745
3705
£665
ATLANTA 198S base anthropogenic only
RHC
-------
10-
10
20
30
65
:::::::: 2±j:l±jn±!
'*
ATLANfTA POINT SOURCE EMISSIONS
-------
3705
10
3665
ATLANTA AREA SOURCE EMISSIONS (NO MV)
TOG (kg/day)
-------
10
0
10
20
30
65
ATLANfTA MOTOR VEHICLE EMISSIONS
TOG (kg/day)
total for region
-------
^705
10
ATLANTA 3°
ROG
4
-------
10
20
30
ATLANTA MOTOR VEHICLE EMISSIONS
TOG (kg/day)
Limited Access Roads
-------
0}
10
ATLANTA 3°
JSOP
665
- \ \ \
Tbtal biogenic ISOP emissions (kg/day) for the Atlanta UM1 nodeling domain
-------
700
740
780
82
§825
^785
10
PAR
Tbtal biogenic PAR endssions (kg/c3ay) for the Atlanta UAM modeling domain
-------
- 3745
3705
0
10
ATLANTA 3°
OLE
5
665
>.
\
Total biogenic OLE emissions (kg/day) for the Atlanta UAM modeling domain
-------
Appendix B
HOURLY WIND FIELDS FOR UAM LAYER 1
USED IN DIAGNOSTIC RUN 1
89059
-------
Wind Speed (m/s)
0 5 10
0
10
665
Atlanta PLANE Winds
Layer 1 at hour 10 on 84156
-------
Wind Speed (m/s)
0 5 10
665
Atlanta PLANR Winds
Layer 1 at hour 11 on 84156
-------
Wind Speed (m/s)
0 5 10
I I i I i I ii I i I
825
^785
10
20
30
Atlanta PLANE Winds
Layer 1 at hour 12 on 84156
-------
Wind Speed (m/s)
0 5 10
r i i I I i
825
-3785
^13745
B705
0
10
20
30
665
Atlanta PLANR Winds
Layer 1 at hour 13 on 84156
-------
Wind Speed (m/s)
0 5 10
705
10
665
Atlanta PLANR Winds
Layer 1 at hour 14 on 84156
-------
Wind Speed (m/s)
0 5 10
»-»--, p - . . . . .,!».,,«
»-* »/ 4 . i /. . .
-*f*^ / - y
665
Atlanta PLANR Winds
Layer 1 at hour 15 on 84156
-------
Wind Speed (m/s)
0 5 10
^*s*s*=^Zi>^]ie j»d. 1m JV ^
10-
0
0
10
665
Atlanta PLANR Winds
Layer 1 at hour 16 on 84156
-------
Wind Speed (m/s)
0 5 10
I i i i i I i i i i I
0
10
20
30
Atlanta PLANR Winds
Layer 1 at hour 17 on 84156
-------
Wind Speed (m/s)
0 5 10
I i i I i I i i i i I
10
3705
10
665
Atlanta PLANE Winds
Layer 1 at hour 18 on 84156
-------
Wind Speed (m/a)
0 5 10
3705
665
Atlanta PLANR Winds
Layer 1 at hour 19 on 84156
-------
Wind Speed (m/s)
0 5 10
I i i i i I i i i i I
825
T 3785
f 3745
3705
10
20
30
665
Atlanta PLANR Winds
Layer 1 at hour 20 on 84156
-------
Appendix C
HOURLY PREDICTED OZONE CONCENTRATIONS
(pphm) FOR DIAGNOSTIC RUN 1
890S9
-------
1S3M
E
si
CL
CL
1SV3
oa
c
3
C
3
o
~D
C
o
"-M
D
L.
-^^
C
D
O
O
0)
c
o
N
o
1S3M
-------
1S3M
E
.c
Q.
0.
03
0)
3
C
o
"o
"o
LJ
C
o
o
CJ
IV
o
N
o
1S3M
-------
in
2! S
CO
1S3M
Q.
a
oo
c
3
c
3
v_
C
O
"o
"5
LJ
o
o
0)
o
N
o
1S3M
-------
1SV3
1S3M
1SV3
in
in
0)
E
P
"i i TVi.. ft 1 '> -i't. t L .r I rl
E
.c
Q.
Q.
00
O)
(1)
c
C
D
i_
C
o
_
o
>
LLJ
C
O
C
-------
1S3M
E
.c
a
a
oo
en
0)
3
c
3
L.
C
O
"5
c
o
v^
O
"c
o
CJ
CD
C
o
N
o
1S3M
-------
1SVG
8
1S3M
Q.
Q.
1SV3
00
CD
0)
c
3
O
'-4-^
D
c
O
c
0>
O
O
O
0)
O
N
O
1S3M
-------
1SV3
E
.c
1S3M
1SV3
00
CD
0)
c
D
C
3
C
o
^
o
"o
LJ
C
O
C
(1)
O
o
0)
c
o
N
o
1S3M
-------
1SV3
1S3M
E
Si
Q_
Q.
isva
00
9)
3
=«=
c
C
o
O
LU
C
0
c
V
2
o
u
C
o
N
O
-------
iSVG
1S3M
E
.c
CL
Q.
1SV3
00
0)
c
C
3
v_
C
o
+*
D
"6
LJ
c
o
c
V
2
0
o
c
o
N
O
1S3M
-------
1SV3
in in
0)
3
=«=
C
C
_0
"5
"o
UJ
L.
o
CJ
c
o
N
o
1S3M
-------
Appendix D
HOURLY WIND FIELDS FOR UAM LAYER 1
USED IN DIAGNOSTIC RUN 2
(BASE CASE SIMULATION WITH BIOGENICS)
69059
-------
Wind Speed (m/s)
0 5 10
10-
0
10
20
30
665
Atlanta PLANR Winds
Layer 1 at hour 10 on 84156
-------
Wind Speed (m/s)
0 5 10
I I I I I I i i i I I
^" _ ^ '^^^y », ^. _^ ^~ m~'K'~'^^ ^ fc 71
i J N -^-i * *- * ,r *>"
10-
0
10
20
30
665
Atlanta PLANR Winds
Layer 1 at hour 11 on 84156
-------
Wind Speed (m/s)
0 5 10
I I i i I I i i i i I
*.»».,-V»...^^^^fe
g=»J»-J».*-<<..»J»'i|J'-^J*S*S=«g'tS*g
20-
10-
665
Atlanta PLANR Winds
Layer 1 at hour 12 on 84156
-------
Wind Speed (m/s)
0 5 10
10-
0
10
665
Atlanta PLANR Winds
Layer 1 at hour 13 on 84156
-------
Wind Speed (m/s)
0 5 10
^705
10
665
Atlanta PLANR Winds
Layer 1 at hour 14 on 84156
-------
Wind Speed (m/s)
0 5 10
10-
0
665
Atlanta PLANR Winds
Layer 1 at hour 15 on 84156
-------
Wind Speed
0 5
10
20
30
Atlanta PLANR Winds
Layer 1 at hour 16 on 84156
(m/s)
10
705
665
-------
Wind Speed (m/s)
0 5 10
825
tft/ffft
4f^
i
0
0
10
Atlanta PLANR Winds
Layer 1 at hour 17 on 84156
-------
Wind Speed (m/s)
0 5 10
825
785
3745
3705
0
10
20
30
665
Atlanta PLANR Winds
Layer 1 at hour 18 on 84156
-------
Wind Speed (m/s)
0 5 10
825
WMma
w:v:fr
WVt///'
WVM/A
/
USAS.VM.
//!///
LjSl'/SA
sMWWZ/W/MVWfa
^^W^»»»K%3(SJiJ
m&/tiWM/vmwmt&tfm
Y/7/
0
665
Atlanta PLANR Winds
Layer 1 at hour 19 on 84156
-------
Wind Speed (m/s)
0 5 10
i i i i I
825
.,-. /. .1 A ,[7,1/1 A A ,i ,r///,r;* f'' £,££;--,C
^/ / x; # /; ^//7^^^w!/
t r fts*xWxs^/tffi/,//M/$f/
tfft'>^ss/%fflSMSSMM
ry//sa//m
m
m
10
20
30
665
Atlanta PLANR Winds
Layer 1 at hour 20 on 84156
-------
Appendix E
HOURLY PREDICTED OZONE CONCENTRATIONS
(pphm) FOR DIAGNOSTIC RUN 2
(BASE CASE WITH BIOGENIC EMISSIONS)
89059
-------
1SV3
tf)
" 0
m ,-;
t i
11 ii
| J
^
| £
I i
I:E
2 ^
X
r-
cc
0
z
i-
(/)
J
8
CM
1
O
O
0)
E
i
in in in in in m in
CM O CD CO 'f CM O
oo c r** f*** i"** P1** r***
OrO ro »O ro ro >O ro
03
O
O
00
0
03
O
CO
O
|V
O
CM
rv
O
O
o
0
to
f f 1 | 1 t t j t t 1 | 1 f i j I 1 1 | 1 1 1 j
: f
V
\
: \
: H
"~
cC^Sfi^^^^.
~ ^ . Q ^ ' ^
_ 1 Q g^ Q ^^"^
\5^1a₯"" m A
) QiC^f" y^
~ CD >>^
-------
*. o
I s
in
cs
oo
z o
J!
8
o
(O
too
in
o
oo
10
in
03
r-
in
to
in
in
CM
r-
in
o
in
03
to
tO
I I I 1 I
I 1 I I I I I 1 I 1 I i I I I j I 1 I
o
CM
1S3M
o
fO
00
E
IT
Q.
Q.
in
X °°
9 oto
O CM
03
11 II
I 8
>!
jl
2 2
?0
o: *
O i^
z
CO
o
O
tO
8
IOO
m
o
03
tn
03
in
to
r--
to
1SV3
in
to
in
CM
to
in
o
r--
10
m
03
to
ii r
in
-------
1SV3
*
II
§
o
aximum
2
CO
to
II
1>
3
"5
inimum
2
I
fe
O
z
10
o
o
CO
1
8
a>
.E
in
CM
oo
oo
03
o
§
0
03
"
O
tO
r-
o
r-
o
CM
0
o
o
_
CD ^<
O CD . ""
Q 0
a
^ " I
-N.
V) ~*
^ ^
Q **"
;
*-
t i i i i i l i i i i i i i i i l i i i i i t i t i i i ( i i i t i i i ~
i^-
O
o
0
1^1
O O O O
ro °1
If
in
CM
oo
£ a
K *
O r-
8
too
m
o
ao
rO
in
oo
to
in
(O
r-
tO
r-
tO
m
CM
to
in
o
r-
in
n
(O
to
I 1 I I i I I 1 l 1 l l l 1 I i I i i I t I i i l l l I i ! L 1 I til i~
o
CM
CD
03
0)
3
Cfl
O
"w
w
LT3
CO
CD
Q.
a
(D
O
N
O
1S3M
-------
ro
'
I s
2 2
CO
in
tsi
00
00
CM
03
in
o
oo
rO
in
oo
ro
in
to
1SV3
ro
m
cs
f-
rO
in
o
rO
m
oo
O
O
o
§
o
rQ
i
.
:
-
»
-
"
***
»«..
~
1
-*
,n
"*
-
T.
~
"t
(DO
minininininmin
oootOTfrMOooto
oor^r^r^r^f^tD^o
rO rO rO lO rO rO ro rO
1 1 | 1 1 1 J 1 t 1 j 1 i 1 | 1 1 1 | 1 1 1 | 1 I 1 | 1 1 1 _
^^"~\ I
"
*
«w
a: :' '
o
O '
j< . Q ^ '
a. Q ^ Q i,. . -
5§sgf* J -
CD m< * ~
o pa . .-
Q O *"
o: , - -
^ ' **-
^
to ;
w v's .> , ^
^j * %^i^^ s
-" ' Q I;1 ' \ I
0 !/:'-:
i i i i i i i i 1 t i i t i t i i i 1 i i i i i i \ t \ I i i i \ i i i } f
2
0
o
o o o o
1S3M
00
OJ
w
o
"oo
oo
"E
-------
1SV3
^s
CM
00
OiO
CM
00
o = o
is
If
.i I
S:E
2 S
in in
CM o
oo
OlO
CM
00
rO
in
00
m
(O
to
1SV3
in
10
in
CM
in
o
ro
m
03
to
in
to
oo i i i i i t i |iii i i i i { TII | n i \ r\ i
in
o
o
O)
o
o
03
z
o
1 1 1 ! 1 S
O
rO
t (J I i
O
N
1S3M
00
OJ
w
c
o
"w
"E
0)
C?)
a
a
aj
c
o
N
O
-------
1SV3
8
in
(SI
oo
in
09
en ,_:
2 2
o
o: *
O r>
in
to
10
in
to
m
CM
in
o
r-
to
o
o
[XI
o
o
i i' i
i r |F i f
in
03
to
to
in
to
(O
i i
'o
^
o
iO
i i 11111 i i i i 11 11 i i i i i i i 11111 1111 11 i i 11 t r
too
O
rO
O
(N
1S3M
E
x:
a
a.
to
q
to
ii
ii
s :§
2 2
in
CM
oo
oo
CM
in
o
in
ao
to
1SV3
in
in
CM
10
in m
o oo
i~» to
to to
co i t 1 1 ) t t
tt *
O r»
z
O
O
o
to
too
IT) r
t i
i i i i i i i i i i i i i i
o o
IO (M
1S3M
i i i t i I
in
-------
s a
o£ c$
*~ *- CM
11 II
J 5
||
il §
22 ^
o: <*
O rv
in
o
\f>
00
10
iO
1SV3
in
iO
in
CM
r-
in in
o 03
r- 10
IO fO
in
(O
§
o
03
ib to
5=
J t I
§l~i I I i i 1 I t I I t } I t I t I I i I I I t I I > I » I I I t I < I I.I 1 I I o
too
O
to
O
CM
1S3M
E
r:
a
CL
i^
r-
2 2
in in
CM o
00 03
on 10
in
oo
h-
ro
in
to
r-
ro
1SV3
in
*
h-
tO
m
CM
rO
in
o
m
CO
to
>o
03 i i T fill } ii T j i I i { I i i I I 7 ! I I i I I I
I
I
o
o
to
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 i i i i~
(DO
O
IO
o
fM
1S3M
OD
O)
a;
3
c
o
0)
10
0
_o
in
00
CD
E
.c
CL
(D
C
o
N
O
-------
CD
ID ,_
CN P-;
*~ CN
11 II
I I
>!
I!
$'t
2 2
in in
CN o
oo CD
oto to
CN
in
GO
10
m
CO
1SV3
in
r^
10
m
CM
in
o
03
-------
CM
^ CD
to r-_
*~ rO
« II
I 1
in
CM
oo
OfO
CM
03
IT)
O
CD
in
ao
r--
in
(£>
1SV3
in
lO
in
csi
in
o
rO
in
CD
(D
7; o
I!
S:E
2 5
o
cc <«
O l^
z
o
o
03
8
a>
E
I I ( I I i I I 1 I I I I I t J t I ! I I I I I i I I I I
O
00
S 1 t i i i t i i i I i i i i M i i i ) i i i i i t i i i I i i i i
too
o
(N
1S3M
E
.c
a
a
T o
K) CM
||
in
CN
OtO
in
o
in
CD
1SV3
m
s
in
CM
in
o
10
CD
(O
in
CO
ID
O
O
8
<0
o
00
E "
4 I I I I 1 I j J i I I I I I I t I 1 I I I I j J I I j I I f _
«o
\
i i i
I 1 I i
8
1S3M
00
O)
a;
3
I/)
C
O
oo
O)
a
a
a>
c
o
N
O
-------
in
^ -o
rw in
"" 10
in
o
oo
I 3
s P
II
2 2
in
_J
o
o
o
i i T i r
in
oo
in
to
(O
1SV3
in
m
CM
p-
rO
in
o
IO
in
oo
>o
in
g Pi. M t 1.1 i r i I i
i t i 1 i t i i i i t i i 1 i i i
i i i i
O
(N
1S3M
E
_c
Q.
a
CN
r-;
10
in in
CM o
oo oo
OfO fO
CM
in
03
m
to
1SV3
in
iO
CM
pv
iO
in
o
in
03
to
iO rO
in
to
(D
00 I i I I I » I I » J i^ f I i I I I I I I I i I { ' ' I ' ' ' _
O
o
o
I
o
8
I i I i I < I 1 I I I I I i I i I i I j I 1 I t I 1 I I I i I 1 1 i I I I i f
(OO
O
rO
O
CM
1S3M
00
CT)
0)
3
tfl
O
E
(D
D
_Q
ID
00
(D
E
.c
a
a.
0)
c
o
N
O
-------
Appendix F
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 ON 4 JUNE 1984
F-la: 0800 Initial NOX tracer contibution
F-lb: 0800 Boundary NOX tracer contribution
F-ic: 0800 Area source NOX tracer contribution
F-ld: 0800 Point source NOX tracer contribution
F-2a: 0800 Initial VOC tracer contribution
F-2b: 0800 Boundary VOC tracer contribution
F-2c: 0800 Anthropogenic VOC tracer contribution
F-2d: 0800 Biogenic VOC tracer contribution
F-3a: 1200 Initial NOX tracer contribution
F-3b: 1200 Boundary NOX tracer contribution
F-3c: 1200 Area source NOX tracer contribution
F-3d: 1200 Point source NOX tracer contribution
F-4a: 1200 Initial VOC tracer contribution
F-4b: 1200 Boundary VOC tracer contribution
P-^c: 1200 Anthropogenic VOC tracer contribution
F-4d: 1200 Biogenic VOC tracer contribution
F-5a: 1600 Initial NOX tracer contibution
F-5b: 1600 Boundary NOX tracer contribution
F-5c: 1600 Area source NOX tracer contribution
F-5d: 1600 Point source NOX tracer contribution
F-6a: 1600 Initial VOC tracer contribution
F-6b: 1600 Boundary VOC tracer contribution
F-6c: 1600 Anthropogenic VOC tracer contribution
F-6d: 1600 Biogenic VOC tracer contribution
89059
-------
1SV3
in
N
00
00
CM,
0}
8
s
m in
03
D
X
O
_D
U-'
'c
C
O
O
8
15
.Q
C
O
O
2
C
CD
O
L_
CD
Q_
O
(O
lilt
l i t
1 1 I I I
t I
i i
1S3AA
LJ
o:
o
Lu
-------
oo
en
1SV3
in
CM
in
in
00
rv
in
-------
00
CD
1 I I I I I 1 I I I I I I ! I j I I
t i i i i ! I i r i i i
0
c
D
CO
Ld
O
O
00
-»-'
o
l_
(U
O
D
x
O
o
s_
0)
Q-
o
UJ
a:
D
O
-------
in
CM
oo
00
o
8
o
s
o
O 1*) *"^rt ~T
I 1 1 I I I 1 1 1 1 1 t t 1 1 1 1 t 1 I I t 1 1 1 I 1 1 1 1
V^ "^:
^^-^ ~
~
~
~
^~
CC
Z
o
o
b m
< o ^
CL o 2 Q
^K r ^
fy fS^ >
CD
2 CD
0
K
I
to
-
I
Q
-
_
i l I I I I 1 l 1 1 I i I I t 1 1 1 1 I I i I I 1 1 t i 1 1 t 1 i 1 I 1 I I "
"* ^>
^r
H-"
C/l
bJ
0
o
00
o +->
K) O
L
CD
(j
D
i_
-t~f
O
0
>
0 5 ."2
O *'
in -^
'o
c
o
U '
^
JD
o 4=
*~ c
o
c
CD
O
CD
Q-
r-»
O O O O 03
K) CM *- C\J
1S3M iL
LJ
z>
0
-------
00
CD
t I I I I 1 i 1 t t t i i I I 1
0)
c
CO
LJ
O
O
00
CD
(J
O
O
O
>
X
1_
D
TD
C
c
o
c
o
o
-t-*
c
(1)
(J
1_
O)
CL
1S3M
LJ
o:
D
O
-------
00
0)
0)
c
m
M
oo
s
00
m
00
8
in
m
o
fv
ID
O
o
03
O
15
o
CM i-
1S3M
O
O
00
(J
D
O
O
>
o
'c
0>
c^
o
d
o
c
D
*4
O
c
o
c
o
o
-^
c
o
-------
in
in
s
rv
K)
r-
K)
m
CM
in
o
o
o
00
o
0}
o
s
o
O
CM
O
o
O
03
0)
c
Ld
O
O
CO
o
D
O
O
>
O
'c
0)
0>
o
jQ
C
o
O
u
l_
CD
0_
o
C\J
UJ
o
u_
-------
a
08
03
O
O
03
O
s
o
to
r-.
$
r-
o
CM
o
o
r-
O
8
O
-
__.
_
-
_
"
~
-
:
_
_
_
_
~
~
t
*?
OO
05
1SV3 "~
ininininminmin ^
OODtD^CNOOOfO C
»O **> 'O »O »O i*> K) ^o ^
< 1 1 I 1 1 1 1 1 1 1 1 1 t 1 i 1 t 1 1 1 1 1 1 1 1 1 1 1 1
^^\^ -
«
_
-
z
o
o
H CD
t t i I i ! t i i t i f i t i I i i i i i t i t I
V
Tj-
1
00
LJ
O
O
0 -^
rj (3J
t_
0)
o
D
i^.
-t-^
X
0
~z.
o ^ "o
O **"*
V) *F"
s
o
c
o
"D
-D
o i
*~ c
o
(J
c
0
-------
1SV3
1S3W
00
0)
c
GO
bJ
O
O
CN
0>
O
D
X
O
C
D
O
C
O
c
O
O
c
-------
I I I I I f I I I f I I I I I I I I I I I ! 1 I I I I I I ! I t I ILL 111
oo
CD
0)
c
3
-3-
CO
o
o
1S3AA
o
0
x
O
o
o
o
^
D
c
o
C
o
o
-^
c
d>
o
o
LJ
o
L.
-------
1SV3
m
CM
00
on
CM
03
in
oo
ID
(O
in
CM
P5
in
o
m
03
m
i i i r i i
O
O
03
O
s
o
o
CM
I
o
(D
o
K)
o
CM
v>
I 1 1 I 1 1 I I 1 I I t I t I I 1 1 I I ) I I I 1 I I I I I 1 I I t I
o o o
lO CM T-
1S3M
oo
O)
c
-
00
LJ
o
o
(M
o
L_
a>
Q_
-o
co
t
LJ
a:
O
-------
00
CD
1SV3 "~
SIO ^
tQ C
0003^»f^lN»^^»
O
r-
0
R
o
0
0
oo
i*l D
o;
o
D
L_
-f-1
0
p
05 ^5
O "*~'
(rt "^
4
0
c
o
o £
* c
o
o
c
(D
0
d>
Q_
f^ rr*
<00 0 0 O O 4?
* IO N »- |
1S3M u-
LU
O
Z
-------
t
00
O)
1SV3
0)
c
13
I l I l l l I l l 1 I t l I I f l l I t l t f l i l i l t I i i i i t t i
1S3AA
o
o
CM
O)
0
O
O
O
o
TD
C
D
O
JD
C
o
c
o
o
c
CL>
(J
L_
d>
CL
Ld
a:
o
-------
00
cr>
1SV3
in
CM
08
CM
00
o
IV
O
CM
s
s
oo
m
(D
in
CM
in
o
rv
m
oo
CM »-
1S3M
I/)
LJ
O
o
rj
o
D
O
O
>
o
'c
a>
a-
o
Q.
o
c
o
c
o
'-+-*
D
.0
'l_
-I-'
c
o
o
c
a>
u
uL
Z)
o
-------
00
01
in in
CM O
00 03
o**"> to
CM
oo
m
oo
in
CM
1S3M
I t I i i i i i I I i i
CD
c
5
3-
I~
Ld
O
o
4'
o
O
O
u
O
o
'c
0)
o
'r>
c
O
^~>
Z5
.D
'i_
-*->
C
O
O
-*'
c
(D
O
v_
0)
Q.
-o
^~
I
a:
D
O
-------
in
CNJ
00
cx>
03
0
8
O
O K> IO K) K) K> K) K>Q ^
i i 1 1 i 1 1 1 1 I 1 1 t 1 1 1 1 i 1 i I 1 1 1 1 I 1 1 1 1
_
"
.^
cc
z
o
o
H CD
j< O 5-
D- Q ^2 °
^^ m -
cc c?^ $.
CD m<
g 8
Ct
h~ «.
(/) ~*
-
in
Q
~
...
t 1 t I i 1 1 1 1 4 1 1 1 1 t 1 t 1 t 1 I 1 t 1 t t t 1 1 t i 1 t t t 1 1 1
*
'T
I"
LJ
0
O
<£>
0 +-
O O
t
o
o
^
-1*
X
o
z
o ^ ^5
o ~^^
v> (-
4
o
c
o
_Q
o i
*~ C
o
o
c
CL
«i
O O O O "">
lO CM «- 1
1S3M U-
LJ
K
13
-------
00
in
CM
08
8
IV
K)
in
tv
K)
m
o
fv
Sin
o
'--
0>
CL
LT>
. I
Ld
o;
o
-------
=1-
00
O)
1SV3
I I I I I I I I I I I
I t I I I i i I ! I I I t I 1 ! I I I I I I I I I I 1 I 1 I I 1 I
1S3M
0)
c
Z3
~5
to
LJ
O
O
(0
O
O
D
X
O
O
O
en
O
0)
c
O
It-'
Z5
.a
c
O
O
u
i_
LJ
cr
z>
o
-------
00
O)
I ] I I t 1 I I 1 I I 1 I ]
I I I 1 I 1 i I I I I I
c
ZJ
>
co
LJ
O
O
CD
O
O
X
O
0)
O
V-
3
O
O
CL
c
O
c
O
O
-t-*
c
0)
O
1S3M
LJ
ct
D
O
-------
oo
01
0)
c
13
1SVG
in
CM
ID
in in
oo
-------
Jo"
in
CM
08
03
O
8
0
S
o
JO
0
***
0
CM
o
o
(V
O
03
OOO<0*CMO03
O -t-1
10 O
L_
0
o
\^
-t~f
0
O
o^ "o
^J -^^
v> -^
s_
0
c
o
'^
JD
0 i
^~ c
o
o
Q
o>
o
Q.
/-\
O O O O CM »- lO
1S3M ul
LJ
ce
z>
0
-------
oo
Of)
m
s
ID
in
8
in
oo
c
3
I/)
O
O
<£>
o>
u
D
O
O
>
o
'c
(U
0>
o
CL
O
C
o
c
o
C
o
c
0)
o
cr
o
o
u_
-------
OO
in m
CM O
8
fv
1SV3
in
CD
c
in
CM
in
col I l l ] f l l [ f
<0
in in
CO
O
O
O
O
>
O
c
o
C
O
o
-*-*
c
CD
O
LJ
tr
O
-------
00
CD
in
CM
8
00
CM
10
o
§
o
00
r-.
o
-
1S3M
1 1 I
>>
^
O
TD
C
Z3
O
.Q
C
O
'^
13
JD
C
o
o
C
o
0)
Q_
TJ
I
LL.
Ld
O
-------
Appendix G
SENSITIVITY ANALYSIS OF THE URBAN AIRSHED MODEL
89059r2 6
-------
Appendix G
SENSITIVITY ANALYSIS OF THE URBAN AIRSHED MODEL
Because of the limited air quality and meteorological monitoring data in the vicinity
of Atlanta there is some question whether the base case simulated all of the perti-
nent physical, meteorological, and chemical processes that led to the high ozone epi-
sode of 4 June 1984. The only direct indication of whether the model correctly simu-
lated the episode is its ability to predict the historical hourly ozone concentrations
at the three ozone monitoring sites. Diagnostic run 2 barely passed the model per-
formance goal used in this study (predicted region-wide maximum ozone must be
within 20 percent and in the general location of the observed peak, and at the loca-
tion of the observed peak the predicted daily maximum ozone concentration must be
within 30 percent of the observed peak). The descrepancies between the predicted
and observed ozone concentrations could be explained by any of several factors,
including underestimation of the anthropogenic emissions inventory, use of too low
boundary conditions, and excessive wind shear and wind speeds, both of which tended
to dilute the urban plume. Thus it was decided to perform a sensitivity test that
examined the sensitivity of the predicted ozone concentrations to less dilution of the
urban plume to see if model performance could be improved. This was accomplished
by reducing the wind speeds at all FAA wind observations sites by 50 percent.
This wind speed reduction is justified because the UAM requires hourly averaged
wind speeds and wind directions, whereas the FAA meteorological sites report
instantaneous wind observations. When the wind varies throughout the hour, the vec-
tor averaging of the hourly average wind speeds results in a lower value than the one
reported at an FAA site. In addition, the FAA wind observations are used primarily
to advise pilots of adverse wind conditions at the airport. Thus, during slow wind
conditions, which are typical during an ozone episode, the FAA observer may report
higher wind gusts that might affect flight operations. The hourly average wind speed
is of no interest to the pilot if gusts exist that may affect flight operations. This
bias of FAA wind speeds was first noticed while developing UAM inputs for the
California South Coast Air Basin, where several hourly integrated and FAA wind
observation sites are located near each other. A brief analysis of these sites is
reported next.
89059r2 6
-------
COMPARISON OF HOURLY AVERAGED AND
FAA WIND SPEED OBSERVATIONS
During the development of the California South Coast Air Basin (SOCAB) Air Quality
Management Plan four three-day episodes of high ozone days were extensively
studied. A systematic bias was seen between the one-hour vector-averaged observa-
tions at SCAQMD (South Coast Air Quality Management District) surface wind speed
monitors and observations of one-minute averaged surface wind speed at nearby
NWS/FAA (National Weather Service/ Federal Aviation Administration) sites. To
estimate the extent of the bias, seven NWS/FAA stations were compared with
SCAQMD wind monitors located nearby (i.e., within approximately one UAM grid
cell). Table G-l shows the station pairs and the distance between the NWS/FAA and
SCAQMD sites.
The study periods were the high ozone episodes of 5-7 June, 12-14 August, 21-23
August, and 26-28 August 1985. A total of 1920 collocated data points were collec-
ted. 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 lowest value that FAA/NWS
wind monitoring stations can measure.
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 hourly average
wind speeds were a little over half of the FAA instantaneous observations). The dif-
ferences in the median values were also 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 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 a similar bias, ranging
from 45 to 56 percent. However, two pairs showed a significantly smaller bias; the
Upland site and Ontario Airport pairs showed a bias of only 28.8 percent, and the San
Bernadino site and Norton Air Force Base pair showed a bias of only 2.4 percent. In
an attempt to see if the agreement was caused by low nighttime wind speeds at
Norton Air Force Base, the wind observations during the daytime period were analy-
zed and 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; for
example, incorrect conversion from knots to m/s could account for the small bias.
We believe that the one-minute average wind speeds reported from FAA/NWS sta-
tions are correct. As a result of the SCAQMD study, which may be true for other
locations also, it was concluded that FAA/NWS surface wind observations may have a
positive bias of as much as a factor of 2.
89059r2 6
-------
This preliminary analysis is by no means complete or statistically roboust. However,
it does have important implications for the development of wind fields for air quality
simulations models, such as the UAM. It also helps explain some of the UAM's ten-
dency to underpredict peak ozone concentrations 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 one-minute wind
speeds with the hourly average wind speed may give some indication of the extent of
the bias.
MODEL PERFORMANCE EVALUATION
The UAM(CB-IV) was exercised with the exact same inputs as in diagnostic run 2
except that the observed wind speeds from FAA observations were reduced by 50
percent. The resultant predicted ozone concentrations at each of the ozone monitor-
ing sites are shown in Figure G-l. Similar time series plots of predicted and obser-
ved ozone concentrations using a one-cell and two-cell search are shown in Figures
G-2 and G-3. Scatterplots, residual analysis plots, and model performance statistics
for predicted and observed hourly ozone concentrations and the reduced wind speed
sensitivity test are shown in Figure G-4. The isopleths of hourly ozone concentra-
tions for the reduced wind speed sensitivity test are given in Figure G-5. Note that
these five figures can be compared to Figures 3 through 6 in the main body of the
report and Appendix E to obtain a comparison of model performance.
The predicted region-wide maximum ozone concentration in the 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 percent (DKLB) and 33
(DLLS) percent. The performance of the sensitivity test is superior to that of diag-
nostic run 2, as shown in the scatterplots of hourly predicted and observed ozone
concentrations (Figures G-4, and Figure 6 in the main body of the report).
VOC EMISSION REDUCTION SCENARIOS
Three across-the-board anthropogenic VOC emission reduction scenarios (30, 60, and
90 percent) were simulated using the inputs from the reduced wind speed sensitivity
test. Table G-2 shows the effects of these reduction scenarios, with and without
biogenic emissions, on the predicted region-wide maximum ozone concentrations.
The information in Table G-2 is graphically presented in Figure G-7. The reduced
wind speed sensitivity test estimated that anthropogenic VOC emission reductions of
62 percent are required to meet attainment of the ozone NAAQS when biogenic
emissions are included, and 53 percent when biogenic emissions are not included.
Thus, compared to diagnostic run 2, the percent increase in anthropogenic VOC
89059r2 6
-------
emission reductions required to reach attainment of the ozone NAAQS when biogenic
emissions are included is 9 percent higher, and 38 percent higher when biogenic
emissions are not included.
These differences of 9 and 38 percent can be explained because the sensitivity test
results in less dilution of the urban plume, so that the ozone peak is generated mainly
by the urban anthropogenic emissions. These differences illustrate the uncertainty in
the model calculations, be they due to uncertainties in emissions (anthropogenic and
biogenic), boundary conditions, or meteorology. Thus the emphasis in the findings
concerning biogenic emissions reported here and elsewhere (e.g., Chameides et al.,
1988) should be on the directions rather than the absolute magnitudes of the
calculations. However, the reduced wind speed sensitivity test does not change the
basic resultbiogenic emissions in Atlanta do increase the level of reduction in
anthropogenic VOC emissions required to meet attainment of the ozone standard.
89059r2 6
-------
TABLE G-1. Surface meteorological observation sites and distance between
collocated observations pairs.
NWS/FAA
Site Name
Bur bank
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
89059r2 6
-------
TABLE G-2. Regional maximum ozone concentrations (pphm) predicted by the
UAM for the base cases (with and without biogenic emissions) and the
different emission scenarios using the reduced wind speed sensitivity
test modeling inputs.
Maximum Ozone
Concentration
(pphm)
With Biogenics
Q% VOC Reduction
30* VOC Reduction
60* VOC Reduction
90* VOC Reduction
Without Biogenics
14.70
13.17
12.09
11.24
Percent
Reductions
from Base Case
Maximum Ozone
Normalized to
Peak Observation
(pphm)
Peak Observed
Ozone NAAQS*
14.7
12.0
0.0
18.4
14.7
12.0
0.0
10.4
17.8
23.5
14.7
13.2
12.1
11.2
0* VOC Reduction
30* VOC Reduction
60* VOC Reduction
90* VOC Reduction
12.93
11.83
10.09
8.82
0.0
8.5
22.0
31.8
14.7
13.5
11.5
10.0
Technically, the ozone NAAQS is 0.12 ppm rounded. Thus, an ozone
concentration of 12.4 pphm (15.6* reduction from observed peak) is
considered attainment.
890S9r2 6
-------
160
12 18 24
160
- 120
CbBSEPVED n
D PREDICTED -
6 12 18
TIME (HOURS)
160
6 12 18
TIME (HOURS)
160
- 120
24
SYSTEMS APPLICATIONS, INC.
FIGURE G-l. Observed and predicted ozone concentrations (ppb)
for Atlanta reduced wind speed sensitivity test.
-------
160
12
18
12
18
24
I ' '
D PREDICTED -
160
120
12
TIME (HOURS)
12
TIME (HOURS)
160
120 -
TIi 160
120
12
TIME (HOURS)
STSTEMS APPLICATIONS. INC.
FIGURE G-2. Observed and nearest-neighbor predicted (one-cell
search) ozone concentration (ppb) for the Atlanta reduced wind
speed sensitivity test.
-------
160
150
- 120
12
TIME (HOURS)
24 0
6 12 18
TIME (HOURS)
160
12
18
12
TIME (HOURS)
160
- 120
SYSTEMS APPLICATIONS, INC.
FIGURE G-3. Observed and nearest-neighbor predicted (two-cell
search) ozone concentration (ppb) for the Atlanta reduced
wind speed sensitivity test.
-------
120.00
90.00
a
a.
Q.
o
LU
OL
a-
60.00
30.00
I r 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
SKEWNESS
KURTOS1S
OTHER MEASURES
MEDIAN
UPPER OUARTILE
LOWER OUARTILE
MINIMUM VALUE
MAXIMUM VALUE
50.39999
47.63355
0.51320
-1.22432
40.00000
88.00000
5.00000
5.00000
147.00000
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
LOW 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 G-4a. Scatterplot and model performance statistics for hourly
ozone concentrations and Atlanta reduced wind speeds sensitivity
test (N = 60) .
-------
0.20 -
-32.00 -16.00 0.00
RESIDUAL (DBS-PRED)
THE BINS1ZE EQUALS 8.000
16.00
32.00
RESIDUAL ANALYSIS
AVERAGE -6.28719
STANDARD DEVIATION 22.19390
SKEWNESS 0.10702
KURTOSIS -1.19285
OTHER MEASURES
MEDIAN -6.30000
UPPER QUARTILE 9.57001
LOWER QUARTILE -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 G-4b. Residual analysis plot and model perforrtance
statistics for hourly ozone concentrations and Atlanta reduced
wind speeds sensitivity test.
-------
62.3
45.3
30.2
22.2
12.4
5.3
_ -6.3
o
LU
£ -17.9
J> -25.0
OQ
O
- -34.7
< -42.8
£ -57.9
-74.9
_ n
-JO.
a o n
a a
a a a a ° a n
a
m m °
n
a
n
i i i i 1 i i i i
i i i i 1 i i i i
i i i i
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 G-4c. Plot of residuals versus time of day for hourly
ozone concentrations and Atlanta reduced wind speeds
sensitivity test.
-------
IN O
"0 o
m X
I s
X =
o .E
2 2
m
CM
00
in
o
in
oo
r-
if)
to
iSV3
in
in
CM
in
o
r-.
in
03
<£>
in
o
o
CM
o
o
1 f I | I I I ) J 1 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 i { i i r i i i i i ~
(DO
o
CM
1S3M
^r
oo
9
o
8
inm
in
in
1SV3
in
in
in
in
(O
E
i l l | l t i } t I l | i t i j i l l | l l t j l l i | l l I
1 f 1 1 1 f 1 1 1 ! 1 1 1 ! 1 1 1 t 1 1 1 1 1 1 1 t 1 ! 1 5 1 I 1 i 1 1 1 1 l~
"?
O
CM
I
0)
CO
s
.8
m
oo
9
§
s
in
fa
1S3M
-------
00
m
in
CM
00
I 5
I!
x =
o .E
2 2
f o
tr <*
O f»
z
CO
o
o
I*)
o
(O
(oo
in
o
CD
K)
in
oo
m
(O
?
^
f 1:1 } I i I I I I I Ml I I I It t I 1 I t I I I 1 I t.'L I I > I t I t l"
(OO
O
(0
O
'(/)
w
'E
(D
8
o
XI
^ .
00 rrt
C7) (D
^1
^ s
Q. ,Q
Q. O
OJ
c m
o i
N O
O
-------
J.SV3
5 IN
" n
5 5
>2
§ E
.1 1
0 1
2 2
h
"
-
~ K
o
"^ _^^
o ^^"^^-^^
"~ t 5/5 ^l
f Q% g ' J
"" LL^^^ , /£**
fl_y i^tfj^**^p
ty (jfi? >
I § . m^ .
0 °o:
*** t" .
5
10
00
il'
-
»-'
~
1 1 I 1 1 i 1 1 1 1 1 I 1 1 I ! 1 1 I 1 1 1 1 t 1 i t t 1 1 1
too o o o
« tO (N T-
in in
03 <0
to
-------
CN
h-
in in
(M O
oo oo
Oio to
IT)
GO
to
1SV3
in
*
r^
rO
m
CM
to
in in
o 03
r~ to
10 to
in
to
to
|*V
2 5
I IT I I I I { I 1 I I I I I | I I i| I 1 » { I I I I I I T^
9, ~l 1 I < 1 I I > I t t I I 1 I I I i I I I 1 I t I I I t 1 I I 1 I I I. i I i r
tOO
-t
O
to
o
CM
1S3M
£
.c
a
a
1SV3
^.
^"
* 2
..
11 II
1 s
O *n
> >
| |
x *~
o .E
2 2
1-
Q
z
(_
CO
^
8
^
8
to
ID
J
in
(M
S"°
00
o
§
0
00
**"
0
to
r~-
o
rv
o
r-
o
^
o
00
to
0
to
<
*"
»
*w»
^
Mw
*
""
***
nt+
r-
*
p»
** '
"^**
>**
*
**
I
I
too
*
ininminmintnin
OOOtO^CSIOOStO
QO r^ r^ r^* r^* ^ to to
tO lO to lO lO to tO tOj-j
1 I | I 1 J | t t 1 i 1 i 1 j t 1 1 1 1 1 1 j 1 1 1 | 1 1 I _
M*^
:
^
."-
: w
«
K ^
o
.
o
IIQ^O' :
°" Q^^"' X ' -
: ^^^~ > ^
CD CD< _
§ CD
0
a -
$ " ""
01 ~
"i:-., -
- , (ft -;-><'' -" -
r " 1 *>i : ^
~! » ;: %
a ,:;: : - -
1 '- ' 8:'.:
'^ "%"-."**
^ :! 0 I:'- -"
. , -
5 1 1 1 f 1 » 1 1 1 1 II t t 1.1 1 1 1 1 1 I: I . 1. 1 V 1 i II 1 i 1 t 1 1 l~
^
s
o
rs|
o
O O O O
tO c\l «-
1S3M
03
O)
(1)
3
(0
c
o
E
0)
(fl
D
m
00
en
l;g
£8
c
o
N
O
H
b
-------
in
o
in
oo
10
1SV3
K)
in
CM
I'-
ll
in
o
in
>
I §
i I
._ c
x =
o .=
Q;
O
Z
10
8
Qi
1
?
I
p
00
0
§
Q
00
^
o
(O
r-
o
rv
o
CM
r-
o
o
o
03
(O
o
to
I 1 1 | 1 1 t J i t 1 | 1 1 i j I 1 1 | I 1 1 j 1 1 1 | 1 1 1 _
~
** "
- «
»**
* \ "
»
* ^ "'
"* °
*** - w.
~: 0 _
** -* ^ JTJn\ ""
" " ' CL * ^^"^r""*^) s -
1 iiS^ffi«Ax) /
"^ <^S^ *Li*^ ' "71
00 ^<~ ^^ 1
P°i. ; :
- f :
*" - "i /i ; -
"" ' - lf\ \ A ^*^ V^ ***
: = ^' IJ <\ ~
- ': O \ : ; ~
-«* 1 ^^ -Hi
\ / ;;
_ ^. ,_ , . . . . ^ .,
i i i f i i t 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~
*
o
rO
O
CN
O
r*
i^
-------
1SV3
II
x =
o .5
2 2
in
CM
oo
OtO
00
in
o
to
K)
in
oo
t-.
tO
in
to
fO
in
10
in
CM
to
in
o
r-.
10
m
03
to
to
a:
O
10
O
O
CM
o
to
too
t I
I I 1 j I
M. > t t i i > i I i i..i. i i i i i i > i i i t i i i t t t r M
o
(N
1S3M
in
to
E
_c
Q.
Q.
n N
" N
" ll
I s
>l
II
II
in
CM
03
OtO
CM
03
.1 .* J « 1 1 f I I t _
too
i i t i t i i j i i i r.i.t. t t r i i > i i.t i i-i i t i I \ t i t i > i i T
ooo
to CN »-
1S3M
03
o;
3
(0
O
"10
w
£
0)
0)
w
o
in
00
CD
Q.
Q.
(1)
O
N
O
B
U
-------
CO
« II
I s
o
in
CM
oo
oo
CM
in
o
1C
to
in
oo
r--
to
1SV3
in
r-
rO
K)
to
1ft
o
10
m in
03 I } I i I 1 I i r I I ! I 1 I i I I I i l'
o
cvj
1S3M
£
JI
a
a
(D
in
rig S
o r^ Oro
T- CM
03
s
I!
o S
2 2
in
o
oo
to
in
oo
iO
in
i t i i
too
o
ro
o
CM
1S3M
rf
03
O)
(D
3
(0
o
'55
w
"E
0)
S
o
oo
O)
a
a
0)
c
o
N
O
§
in
6
En
-------
j |
9. 1 I I I I I I 1 I I I I I i 1 t I I I I I I I t I 1 I t I. i .1 t I i L t 1 I f~
E
x:
a
d
to ,_
CM a>
11 II
I S
>S
If
x ;=
O .E
in
CM
00
OiO
S P
a: *
O P^
j °
£N
o r*
o
m
T 8
^ P*
<£>
in
o
ao
in
oo
ro
m
(O
1SV3
in
Tj-
r>
rO
m
CM
in
o
r>
rO
in
ao
(O
rO
i i i i i i i i I
1 i ii i i i i n ! i i i t i i i i i"
too
o
CM
1S3M
00
3
w
c
o
'w
w
"E
0)
0)
°
N
O
b
-------
in in
CM o
oo ID
m
oo
m
(O
1SV3
in
CM CM
0)
3 o
in
esi
ro
in
o
m
oo
to
in
to
v
I
I o
C 00
2 r*
in
_i
o
o
ID
O
O
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
gl~l i I t I i I t I { I t I 1 I t t I I } I 1 I t t 1 1 1 1 I I 1 I i I I I i l"
too O
o
cc
O
2 °
CM
O I*-
O
o
o
to
I I | I
i 1 I | l I I j t t I | I i f j f l Tj l I (
in
to
to
m I 111 I i I I I i I 1 I I I I I t I I I i I 11 i I i I j I I I
O 1
No
(/)
-------
J.SV3
o m
Ss J
in
o
(D
in
oo
in
CO
in
CM
r-
r- CM
CO
in
o
r--
fO
m
CD
to
in
to
II
1 s
7; o
If
So
o: *
O r«
c/i
_i
o
o
o
CM
8
O>
O
03
Q)
in
o
to
in ir
oo tc
to tc
(O K
I i
s o
II
o
o
I
o
i I i i l j I i i | i t t | i i i | i i i j i i i i i i t
I .1.1 i i r i t i I i t t.t > t I I it t i i ; I i i i .1 Tit til ill T
03
O)
0)
D
5
10
O
'(/)
w
"E
0)
(P
w
D
.0
ID
00
CT)
E
.c
a
Q.
Oj
C
O
N
O
m
N«^
E
too
1S3M
-------
35
30
25
20
.s 3
15
.2 c
la
II
10
Without
Biogenics
Attainment (0.12 ppm)
Attainment (0.124 ppm)
10 20 30 40 50 60 70 80
Percent Reductions of Anthropogenic VOC Emissions
90
100
FIGURE G-6. Relationship between percent reduction of anthropogenic VOC
emissions to percent reduction of the maximum ozone concentration at the
location of regional maximum for the base case scenarios with and without
biogenic emissions for Atlanta reduced wind speed sensitivity test UAM
modeling inputs.
EEE 89059
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO. 2.
EPA 450/4-90-006 D
4. TITLE AND SUBTITLE
URBAN AIRSHED MODEL STUDY OF FIVE CITIES -
Application of the Model to Atlanta and Eva!
Effects of Biogenic Emissions on Emission Co
7. AUTHOR(S)
Ralph E. Morris, Thomas C. Myers, Marianne C
LuAnn Gardner, Edward L. Carr
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 Planning and Standan
Research Triangle Park, NC 27711
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
Low-Cost April 1990
uation of thc6-pERFORM|NGORGAN|2AT|°NCODE
ntrol Strategies
8. PERFORMING ORGANIZATION REPORT NO.
. Causley,
10. PROGRAM ELEMENT NO.
11. CONTRACT /GRANT NO.
13. TYPE OF REPORT AND PERIOD COVERED
IS 14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This document presents Urban Airshed Modeling results showing sensitivity of peak
ozone to manmade hydrocarbon emissions reductions for two cases - inclusion and
exclusion of biogenic emissions.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
Ozone
Urban Airshed Model
Photochemistry
Biogenics
18. DISTRIBUTION STATEMENT
b.lDENTIFIERS/OPEN ENDED TERMS C. COS AT I Field/Group
19. SECURITY CLASS (Tilts Report) 21. NO. OF PAGES
152
20. SECURITY CLASS (This page) 22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLETE
------- |