# A \
pro"^
Evaluation of Prognostic Meteorological Data in
AERMOD Overwater Applications
-------
-------
EPA-454/R-23-010
October 2023
Evaluation of Prognostic Meteorological Data in AERMOD Overwater Applications
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Air Quality Assessment Division
Research Triangle Park, NC
-------
Preface
This document provides an evaluation of the performance of prognostic data in AERMOD via
the AERMET pass-through option and the COARE processing option in marine boundary layer
environments added to AERMET as part of the 2023 revisions to the Guideline on Air Quality
Models. The purpose of the document is to provide results to compare results between passing
certain meteorological data from WRF through AERMET to AERMOD and using the COARE
algorithms in AERMET for the same data. Also, results are presented to show the difference, if
any, between using multiple vertical levels of data or a single vertical level of data. Included in
this document are descriptions of the inputs, comparison of meteorological output from
AERMET using standard AERMET processing and the COARE algorithms, and comparison of
AERMOD results between both approaches.
11
-------
Acknowledgements
This report was developed as part of the 2023 proposal of The Guideline on Air Quality Models,
Appendix W with input from the meteorological data workgroup comprised of staff from EPA's
Office of Air Quality Planning and Standards and Region 10. WRF a processing for the
evaluations were processed by General Dynamics Information Technology.
111
-------
Contents
1.0 Introduction 1
2.0 Methodology 2
2.1 Study Areas 3
2.1.1 Cameron, LA 3
2.1.2 Carpinteria, CA 7
2.1.3 Pismo Beach, CA 10
2.1.4 Ventura, CA 14
2.2 AERMET configurations 16
2.2.1 WRF simulations 16
2.2.2 AERMET-COARE configurations 23
2.3 Meteorological data evaluation 26
2.4 AERMOD evaluation 27
3.0 Results 28
3.1 Meteorological data comparisons 28
3.2 AERMOD results 36
4.0 Summary and Conclusions 37
5.0 References 39
iv
-------
Figures
Figure 1. Study areas for COARE to AERMET testing 3
Figure 2. Cameron, LA study area 4
Figure 3. Carpinteria, CA study area 7
Figure 4. Pismo Beach, CA study area 11
Figure 5. Ventura, CA study area 14
Figure 6. Cameron, LA WRF domains. The large outer box is the 12-km domain, the white box
is the 4-km domain, and the red box is the 1.33 km domain 18
Figure 7. Carpinteria, CA WRF domains. The large outer box is the 12-km domain, the white
box is the 4-km domain, and the red box is the 1.33 km domain 19
Figure 8. Pismo Beach, CA WRF domains. The large outer box is the 12-km domain, the white
box is the 4-km domain, and the red box is the 1.33 km domain 20
Figure 9. Ventura, CA WRF domains. The large outer box is the 12-km domain, the white box
is the 4-km domain, and the red box is the 1.33 km domain 21
v
-------
Tables
Table 1. Cameron measured meteorological data 5
Table 2. Cameron source and receptor data 6
Table 3. Carpinteria measured meteorological data 9
Table 4. Carpinteria source parameters data 10
Table 5. Pismo Beach measured meteorological data 12
Table 6. Pismo Beach release heights and receptor distances 13
Table 7. Ventura measured meteorological data 15
Table 8. Ventura receptor distances 16
Table 9. Time periods modeled for each location 22
Table 10. AERMET-COARE configurations for prognostic data 25
Table 11. AERMET-COARE configuration names for prognostic data 26
Table 12. Mean and fractional biases (Prognostic - observations) for each study area for wind
speed, air temperature, sea surface temperature, air-sea temperature difference, and relative
humidity 29
Table 13. Mean and fractional biases for each study area for surface friction velocity (u*). An X
indicates that a particular scenario is not valid for a particular case study 30
Table 14. Mean and fractional biases for each study area for convective mixing height. An X
indicates that a particular scenario is not valid for a particular case study 31
Table 15. Mean and fractional biases for each study area for mechanical mixing height. An X
indicates that a particular scenario is not valid for a particular case study 32
Table 16. Mean and fractional biases for each study area for mixing height used in AERMOD.
An X indicates that a particular scenario is not valid for a particular case study 33
Table 17. Mean and fractional biases for each study area for surface roughness (z0). An X
indicates that a particular scenario is not valid for a particular case study 34
Table 18. Mean and fractional biases for each study area for Monin-Obukhov length. An X
indicates that a particular scenario is not valid for a particular case study 35
Table 19. Mean and fractional biases for each study area for absolute Monin-Obukhov length.
An X indicates that a particular scenario is not valid for a particular case study 36
vi
-------
Table 20. AERMOD Robust Highest Concentration for measured meteorological data and
prognostic data. Observed RHC value are in parenthesis with the study area name. An X
indicates that a particular scenario is not valid for a particular case study 37
vii
-------
1.0 Introduction
In recent years, applications of AERMOD (U.S. EPA, 2023a) in marine boundary layer
environments, i.e., overwater applications, have increased. Calculations of boundary layer
parameters for the marine boundary layer present special challenges as the marine boundary
layer can be very different from the boundary layer over land. For example, convective
conditions can occur in the overnight hours in the marine boundary layer while typically over
land, stable conditions occur at night. Also, surface roughness in the marine environment is a
function of wave height and wind speed and less static with time than surface roughness over
land.
While the Offshore and Coastal Dispersion Model (OCD) (DiCristofaro and Hanna, 1989) is the
preferred model for overwater applications, there are applications where the use of AERMOD is
applicable. These include applications that utilize features of AERMOD not included in OCD
(e.g., NO2 chemistry). Such use of AERMOD would require consultation with the Regional
Office and appropriate reviewing authority to ensure that platform downwash and shoreline
fumigation are adequately considered in the modeling demonstration.
For the reasons stated above, a standalone pre-processor to AERMOD, called AERCOARE
(U.S. EPA, 2012a) was developed to use the Coupled Ocean Atmosphere Response Experiment
(COARE) bulk-flux algorithms (Fairall et al., 2003) to bypass AERMET and calculate the
boundary layer parameters for input to AERMOD for the marine boundary layer. AERCOARE
can process either measurements from water-based sites such as buoys or prognostic data
processed via the Mesoscale Model Interface program (MMMIF) (Ramboll, 2023).
AERCOARE was developed in response of a need for overwater meteorology for an AERMOD
application in an Arctic Ice Free Environment (U.S. EPA, 201 la) and that the boundary layer
calculations in AERMET (U.S. EPA, 2023b) are more suited for land-based data.
To better facilitate the use of the COARE algorithms for AERMOD, EPA included the COARE
algorithms into AERMET version 23132 (U.S. EPA, 2023b) as part of the 2023 proposed
updates to the Guideline on Air Quality Models (U.S. EPA, 2023c), thus eliminating the need
for a standalone pre-processor and ensures the algorithms are updated as part of routine
AERMET updates. The evaluation of the implementation of COARE into AERMET is
presented in U. S. EPA (2023 d) and the results of the evaluation indicated that COARE was
implemented into AERMET with no issues.
With the implementation of COARE into AERMET, there are now two options for processing
prognostic data in AERMET for overwater applications. The first option was incorporated into
MMTF 4.0 (Ramboll, 2023) and AERMET 22112. With this option, MMTF outputs an optional
data flag with the DATA keyword for the PROG pathway of AERMET (U.S. EPA, 2023b) to
1
-------
let AERMET know if the data is overland, "OL" or overwater, "OW." When the data are
overwater, AERMET will use the MMIF output surface friction velocity (u*), Monin-Obukhov
length, convective velocity scale (w*), potential temperature lapse rate, sensible heat flux,
hourly surface roughness, and cloud cover, instead of calculating the first five variables, and
using monthly surface roughness. This "pass-through" is used instead of AERMET calculating
the variables as done for land-based data as the equations used in AERMET are more suitable
for land-based applications. With the second option, COARE, AERMET will calculate the
variables listed above using the COARE algorithms, as done with the AERCOARE processor
using the standard input variables (wind, temperature, etc) in addition to three new variables
added to the MMIF output for AERMET: sea surface temperature and measurement depth and
longwave downward radiation. See the AERMET User's Guide (U.S. EPA, 2023b) for details
on the COARE processing in AERMET.
This report details the evaluation process to determine the differences between the two options
in AERMET for prognostic data in the marine boundary layer environment, the pass through of
key variables or COARE calculations. Also, this report will determine the differences between
using one level of data as generally done with AERCOARE output from MMIF or multi-level
data often generated for most applications of AERMET involving prognostic data. This report
will not attempt to determine if prognostic data performs better or worse than observed
meteorological data but the results based on observations are shared as a benchmark for
comparisons of COARE vs. AERMET pass through and the number of levels.
The comparisons presented in this report do not include the warm layer or cool skin options
available for COARE. These options have been included in AERMET but have not been
evaluated as the data necessary for these options are not available in the datasets used in this
evaluation. Section 2 discusses the methodology of the case studies used for the evaluations.
There are four case studies used to evaluate the incorporation of COARE into AERMET: 1)
Cameron, LA; 2) Carpinteria, CA, 3) Pismo Beach, CA, and 4) Ventura, CA. This report
includes comparisons of meteorological data output from AERMET using the pass-through
option, COARE processing, as well as comparisons of the use of single vs. multi-level data.
The evaluations also include comparison of AERMOD results using both methods of processing
and levels of data. Section 2.0 describes the methodology of the evaluations, Section 3.0
discusses the results of the evaluations, and Section 4.0 is the summary and conclusion of the
evaluation.
2.0 Methodology
Following is the methodology of the evaluation of prognostic data with AERMET "pass-
through" and COARE processing. Section 2.1 describes the study areas, Section 2.2 describes
2
-------
the AERMET configurations, Section 2.3 describes the meteorological data evaluation and
Section 2.4 describes the AERMOD evaluation.
2.1 Study Areas
Four case study areas were considered for evaluation (Figure 1) as noted in Section 1.0. Each
study area is detailed below and more information about each can be found in U.S. EPA
(2012b).
( \ h 5—J \LY
_ /
Pismo Beach
h— -jrfl
Carpinteria
|
Ventura
>
Cameron
Figure 1. Study areas for COARE to AERMET testing.
2.1.1 Cameron. LA
The Cameron case study consisted of 26 tracer releases from field studies in July 1981 and
February 1982. Tracer was released from both a boat and a low-profile platform at a height of
13 m. Receptors were located in flat terrain near the shoreline with transport distances ranging
3
-------
from 4 to 10 km (U.S. EPA, 2012b). Error! Reference source not found, shows the general
study area. The meteorological data for Cameron is shown in Table 1. Note, for all hours, the
station pressure was set to 1000 mb and wind direction was assumed to be 270° because
AERMOD would be run in screening mode. The data set contains both very stable and fairly
unstable conditions. There are several hours of stable lapse rates accompanied by unstable air-sea
temperature differences. For example, on February 15, 1982, hour 1700, the air-sea temperature
difference is -0.8 °C, while the virtual potential temperature lapse rate is 0.06 °C/m (extreme
stability "G" in OCD). Over 10 m, this virtual potential temperature lapse rate would result in at
least an air-sea temperature difference of +0.5 °C. The data was adjusted for the AERCOARE
evaluations by adjusting the air-sea temperature difference to be at least as stable as indicated by
the virtual potential temperature lapse rate. The sea temperature was adjusted so the air-sea
temperature difference matched the measured potential temperature lapse rate (U.S. EPA, 2012b)
UTM East (km) Zone 15N, Datum: NAS-C
Figure 2. Cameron, LA study area
4
-------
Table 1. Cameron measured meteorological data.
Date
Hour
(LST)
Wind
ht
(m)
Wind
speed
(m/s)
Temperature/RH
height (m)
RH
(%)
Air
temperature
(°C)
Sea
temperature
(°C)
ae
(°)
Mixing
height
(m)
7/20/81
14
10
4.6
10
63
29.25
31.95
6.4
800
7/20/81
15
10
4.8
10
64
29.45
32.05
4.9
800
7/23/81
17
10
4.3
18
73
30.45
31.85
4.7
225
7/23/81
18
10
5.1
18
74
30.55
31.75
4.7
225
7/27/81
20
10
2.1
18
82
27.05
31.45
999
400
7/27/81
22
10
4.5
18
82
26.85
31.35
999
450
7/29/81
16
10
4.6
18
69
29.85
32.05
9.6
420
7/29/81
17
10
5
18
68
29.85
31.85
6.4
430
7/29/81
19
10
5
18
68
29.95
31.65
9.6
450
2/15/82
16
10
5.7
10
89
14.25
13.75
999
200
2/15/82
17
10
5.6
10
88
13.95
13.45
999
200
2/15/82
20
10
5.9
10
87
14.25
13.75
999
200
2/17/82
14
10
3.3
10
93
15.65
13.55
2.5
200
2/17/82
15
18
3.7
18
93
14.95
14.05
7.6
200
2/17/82
16
18
4.3
18
93
14.85
14.25
3.9
200
2/17/82
17
18
3.5
18
93
14.55
14.19
3.8
200
2/17/82
18
18
3.5
18
93
14.25
13.89
2.1
200
2/22/82
14
18
5.2
18
75
17.45
16.15
2.7
100
2/22/82
16
18
4.7
18
76
17.45
16.55
2.4
100
2/22/82
17
18
4.5
18
76
17.75
16.95
2.8
100
2/23/82
14
18
4.8
18
84
18.35
14.65
0.6
50
2/23/82
17
18
6.2
18
88
18.05
15.75
3.2
80
2/24/82
15
18
3.7
18
49
19.95
14.95
2.7
50
2/24/82
16
18
3.7
18
50
19.75
15.15
3.2
50
2/24/82
17
18
3.5
18
50
19.75
15.05
3.3
50
2/24/82
19
18
4.1
18
52
17.55
14.85
2.6
50
5
-------
Table 2 shows the Cameron source and receptor data for AERMOD. Release heights for
releases was 13.0 m. AERMOD was run in screening mode with westerly winds with the
source location at (0,0). Receptor coordinates are (X,0) where X is the downwind distance of
the peak observed concentration.
Table 2. Cameron source and receptor data.
Release
number
Date
Hour
(LST)
Building ht
(m)
Building width
(m)
Receptor distance
(m)
1
7/20/81
14
0.0
0.0
7180
2
7/20/81
15
0.0
0.0
7400
3
7/23/81
17
0.0
0.0
8930
4
7/23/81
18
0.0
0.0
8710
5
7/27/81
20
0.0
0.0
7020
6
7/27/81
22
0.0
0.0
7859
7
7/29/81
16
0.0
0.0
7820
8
7/29/81
17
0.0
0.0
9780
9
7/29/81
19
0.0
0.0
9950
10
2/15/82
16
7.0
20.0
4834
11
2/15/82
17
7.0
20.0
5762
12
2/15/82
20
7.0
20.0
4526
13
2/17/82
14
0.0
0.0
7000
14
2/17/82
15
0.0
0.0
6985
15
2/17/82
16
0.0
0.0
7400
16
2/17/82
17
0.0
0.0
7260
17
2/17/82
18
0.0
0.0
6950
18
2/22/82
14
0.0
0.0
7095
19
2/22/82
16
0.0
0.0
7070
20
2/22/82
17
0.0
0.0
6955
21
2/23/82
14
0.0
0.0
7769
22
2/23/82
17
0.0
0.0
7245
23
2/24/82
15
7.0
20.0
5669
24
2/24/82
16
7.0
20.0
5669
25
2/24/82
17
7.0
20.0
6023
26
2/24/82
19
7.0
20.0
4786
6
-------
2.1.2 Carpinteria. CA
The Carpinteria tracer study was conducted in September and October 1985. Studies were
conducted to examine offshore impacts caused by both interaction with complex terrain and
shoreline fumigation. The current analysis only evaluated the complex terrain data set as the
AERCOARE-AERMOD approach currently cannot simulate shoreline fumigation.
270 271 272 273 274
UTM East (km) Zone 11N, Datum: NAS-C
100
Snow/Ice
-90
"85 Tundra
-80
"'5 Barren
-70
-65 Wetland
60
-55 Water
-50
-45 Forest
40
35 Range
-30
-25 Agriculture
-20
-15 Llrban/Built-Up
10
Land Use
sampler Locations:
X - Complex Terrain
X - Fumigation
Tracer Releases:
~ - Complex Terrain
~ — Fumigation
shows the land use and terrain for the Carpinteria field study. The shoreline receptors are located
on a 20 m to 30 m high bluff within 0.8 km to 1.5 km of the offshore tethersonde release. Two
tracers were released with heights varying from 18 m to 61 m. The tethersonde was well above
the anchor boat and downwash was not considered in the simulations.
7
-------
¦too
Snow/Ice
-90
"85 Tundra
-80
" 75 Barren
.-70
"65 Wetland
-60
-55 Water
J-SO
-45 Forest
-40
• 35 Range
-30
-25 Agnculture
-20
. 15 Urban/Buill-Up
-10
land Use
Sampler Locations:
X - Complex Terrain
X - Fumigation
Tracer Releases:
A - Complex Terrain
A - Fumigation
268 269 270 271 272 273 274 275 276
UTM East (km) Zone 11N. Datum: NAS-C
Figure 3. Carpinteria, CA study area.
Table 3 displays the meteorological data used in the current simulations and previous evaluations
of OCD and CALPUFF. The winds were very light for most of the releases, especially
considering the wind measurement heights were from 30 m to 49 m. Note that the air
temperature and relative humidity measuring height was 9 m for all hours, station pressure was
1,000 mb for all hours, and the mixing height was 500 m for all hours. The combined influences
of low wind speeds and the air-sea temperature differences in Table 3 result in cases with
unstable to very stable stratifications. Unlike the Cameron data set, the
virtual potential temperature lapse rates do not contradict the gradient inferred from the air
temperature difference measurements. One suspect aspect of the data is the constant mixed layer
height of 500 m for the entire data set. In cases where plumes are not trapped under a strong
inversion, CALPUFF and OCD are less sensitive to the mixing height than AERMOD. Thus,
uncertainty in the boundary layer height in this experiment may not have been important to the
original investigators.
Table 4 lists the source release parameters used for the AERCOARE simulations of the
Carpinteria data set. Unlike the other databases, actual wind directions, source locations and
receptor sites were used in the analysis to consider the effects of terrain elevation on the model
predictions. Receptor elevations and scale heights for AERMOD were calculated with AERMAP
8
-------
(Version 11103) (EPA, 201 lb) using 1/3 arc-second terrain data from the National Elevation
Data (NED) set. The peak predicted concentration was compared to the peak measured
concentration for each release.
Table 3. Carpinteria measured meteorological data.
Date
Hour
(LST)
Wind ht
(m)
Wind
speed
(m/s)
Wind
direction
(°)
RH
(%)
Air
temperature
(°C)
Sea
temperature
(°C)
oe(°)
9/19/85
9
30
1.3
259.7
78.8
16.3
17.4
26.8
9/19/85
10
30
1.3
235.4
79
16.8
17.6
28.4
9/19/85
11
30
2.6
214.1
80.1
17
17.7
24.4
9/19/85
12
30
3.1
252.9
80.1
17.1
17.8
32.9
9/22/85
9
30
1
220.8
70.6
17.4
16.9
32.1
9/22/85
10
30
1.2
251.1
81
17
16.7
17.4
9/22/85
11
30
2.4
253.8
92.1
16.4
15.4
8
9/22/85
11
30
2.4
230
92.1
16.4
15.4
8
9/22/85
12
30
2.8
248.4
91.1
16.3
15.2
17.4
9
-------
9/22/85
12
30
2.8
237.7
91.1
16.3
15.2
17.4
9/25/85
10
24
1
163.8
60.3
21.2
18.4
41.7
9/25/85
11
46
1.6
163.8
69.9
21
18.7
9.9
9/25/85
12
46
1
165.6
90.3
20.9
18.8
26.1
9/25/85
13
46
1
175
90.4
21.4
18.7
18.4
9/26/85
12
49
3.8
262
83.5
18.7
19.4
10.9
9/26/85
13
49
4
262.2
81
18.8
19.8
11.8
9/28/85
10
24
5.4
155.8
85.1
18.1
18.7
8.9
9/28/85
10
24
5.4
155.8
85.1
18.1
18.7
8.9
9/28/85
11
24
3.2
174.7
84.1
18
18.8
10.9
9/28/85
11
24
3.2
177
84.1
18
18.8
10.9
9/28/85
13
24
1.5
234.5
82.5
18.3
18.9
10.9
9/28/85
13
24
1.5
229.5
82.5
18.3
18.9
10.9
9/28/85
14
24
2.1
215
81.7
18.5
18.8
11.8
9/28/85
14
24
2.1
215
81.7
18.5
18.8
11.8
9/29/85
11
30
3.4
243.7
86
18.2
18.5
18.4
9/29/85
12
30
3.1
238.9
87.8
18.1
18.5
5
9/29/85
12
30
3.1
232.7
87.8
18.1
18.5
5
10/1/85
9
61
2
215.5
92.1
16.5
17.4
19.2
10/3/85
10
61
1
164.6
89
26.3
24.2
12.8
10/3/85
11
61
1.8
215.5
95.9
24.8
21.4
32.9
10/4/85
12
76
1.7
216.9
70.3
21.6
18.3
14.7
10/4/85
9
76
2.6
231.2
71.9
21.7
18.4
11.8
10/4/85
10
76
1.7
186.4
76.4
21.3
18
13.7
10/5/85
11
91
1.3
171.3
66.8
20.9
20.2
28.4
10/5/85
11
91
1.5
208.2
64.8
21.3
20.6
19.2
10/5/85
12
91
1
195.2
62.7
21.5
20.8
28.4
Table 4. Carpinteria source parameters data.
Release
number
Date
Hour
(LST)
Release
type
Release
ht (m)
UTM
East
(m)
UTM
North (m)
1
9/19/85
9
SF6
30.5
270,343
3,806,910
2
9/19/85
10
SF6
30.5
270,343
3,806,910
3
9/19/85
11
SF6
30.5
270,343
3,806,910
4
9/19/85
12
SF6
30.5
270,343
3,806,910
5
9/22/85
9
SF6
18.3
270,133
3,806,520
6
9/22/85
10
SF6
18.3
270,133
3,806,520
7
9/22/85
11
SF6
18.3
270,133
3,806,520
8
9/22/85
11
Freon
36.6
270,133
3,806,520
9
9/22/85
12
SF6
18.3
270,133
3,806,520
10
-------
10
9/22/85
12
Freon
36.6
270,133
3,806,520
11
9/25/85
10
SF6
24.4
271,024
3,806,660
12
9/25/85
11
SF6
24.4
271,024
3,806,660
13
9/25/85
12
SF6
24.4
271,024
3,806,660
14
9/25/85
13
SF6
24.4
271,024
3,806,660
15
9/26/85
12
Freon
24.4
269,524
3,807,330
16
9/26/85
13
Freon
24.4
269,524
3,807,330
17
9/28/85
10
SF6
24.4
271,289
3,806,340
18
9/28/85
10
Freon
42.7
271,289
3,806,340
19
9/28/85
11
SF6
24.4
271,289
3,806,340
20
9/28/85
11
Freon
42.7
271,289
3,806,340
21
9/28/85
13
SF6
24.4
270,133
3,806,520
22
9/28/85
13
Freon
39.6
270,133
3,806,520
23
9/28/85
14
SF6
24.4
270,133
3,806,520
24
9/28/85
14
Freon
39.6
270,133
3,806,520
25
9/29/85
11
SF6
30.5
270,133
3,806,520
26
9/29/85
12
SF6
30.5
270,133
3,806,520
27
9/29/85
12
Freon
61
270,133
3,806,520
2.1.3 Pismo Beach. CA
The Pismo Beach experiment was conducted during December 1981 and June 1982. A depiction
of land use, release point locations and receptor sites are shown in Figure 4 based on U.S. EPA
(2012b). Tracer was released from a boat mast height of 13.1 m to 13.6 m above the water. Peak
concentrations occurred near the shoreline at sampling distances from 6 km to 8 km away. The
Pismo Beach evaluation database consists of 31 samples.
11
-------
CO
<
z
Q
-------
Table 5. Pismo Beach measured meteorological data.
Date
Hour
(LST)
Wind
speed
(m/s)
RH (%)
Air
temperature
(°C)
Sea
temperature
(°C)
-------
with the highest observed concentration, a constant westerly wind was assumed, and predictions
were obtained at a single receptor located the correct distance east of the release point.
Table 6. Pismo Beach release heights and receptor distances.
Release
number
Date
Hour
(LST)
Release
ht (m)
Receptor
distance
(m)
1
12/8/81
15
13.1
6730
2
12/8/81
16
13.1
6506
3
12/11/81
14
13.1
6422
4
12/11/81
15
13.1
6509
5
12/11/81
17
13.1
6619
6
12/11/81
19
13.1
7316
7
12/13/81
14
13.1
6516
8
12/13/81
15
13.1
6372
9
12/13/81
17
13.1
6870
10
12/14/81
13
13.1
6378
11
12/14/81
15
13.1
6378
12
12/14/81
17
13.1
6526
13
12/15/81
13
13.1
6944
14
12/15/81
14
13.1
6697
15
12/15/81
19
13.1
8312
16
6/21/82
15
13.6
6532
17
6/21/82
16
13.6
6589
18
6/21/82
17
13.6
6748
19
6/21/82
18
13.6
6532
20
6/22/82
15
13.6
6125
21
6/22/82
16
13.6
6214
22
6/22/82
19
13.6
6054
23
6/24/82
13
13.6
6244
24
6/24/82
15
13.6
6244
25
6/25/82
12
13.6
6406
26
6/25/82
13
13.6
6377
27
6/25/82
15
13.6
6406
28
6/25/82
16
13.6
6435
29
6/25/82
17
13.6
6455
30
6/27/82
16
13.6
6630
31
6/27/82
18
13.6
6579
14
-------
2.1.4 Ventura. CA
The Ventura experiment was conducted during September 1980 and January 1981. Land use,
release point locations and receptor sites are shown in Figure 5 based on the files from the
CALPUFF evaluation archives. The tracer was released from a boat mast height of 8.1 m above
the water. Peak concentrations occurred along the closet arc of receptors in Figure 5 at sampling
distances from 7 km to 11 km away. The Ventura evaluation database consists of 17 samples.
VENTURA, CA
379ft-
3796-
0 3794-
(0
<
z
1 3702-
W
o
5 3790-
3788-
3786-
3784-
3782-
3780-
-
Hi
/ 1
m
/
'H
|
4
*
n
©
u
u
—
y
y.
'li
—or
X
X
<
X
X'
A
M.
A
X
x>
X .
xx
X
X
*
L x
X
X
y x
>
<
X
X
X
X
284 286 288 290 292 294 296 298 300 302 304
UTM East (km) Zone 11N, Datum NAS-C
Figure 5. Ventura, CA study area.
r 100
-95
Snow/Ice
-90
-85
Tundra
-80
-75
Barren
70
-65
Wetland
-60
-55
Water
-50
-45
Forest
-40
-35
Range
-30
-25
Agnculture
-20
-15
Urban'Built-Up
- 10
Land Use
X Sampler Locations
~ Tracer Releases
The Ventura meteorological data used in the current analysis are shown in Table 7. Note for all
hours the station pressure was 1000 mb, wind measurement height was 20.5 m, air
temperature/relative humidity measurement height was 7.0 m, and wind direction was 270°
because AERMOD was run in screening mode. The OCD and CALPUFF model evaluation data
set stabilities ranged from moderately unstable to slightly stable. As with the Pismo Beach data,
there are several hours of stable lapse rates accompanied by unstable air-sea temperature
differences. For example, on September 29, 1980, hour 1400, the air-sea temperature difference
is -0.8 °C, while the virtual potential temperature lapse rate is 0.03 °C/m. These contradictory
15
-------
data were resolved using the same methodology as in the Pismo Beach and Cameron datasets
and the revised estimates are highlighted in gray in Table 7.
Table 7. Ventura measured meteorological data.
Date
Hour
(LST)
Wind
speed
(m/s)
RH (%)
Air
temperature
(°C)
Sea
temperature
(°C)
°e (°)
Mixing
height
(m)
9/24/80
16
4.1
72
15.15
17.25
8
400
9/24/80
18
6.2
78
14.85
16.85
6.5
400
9/24/80
19
6.9
77
14.85
16.95
6
400
9/27/80
14
6.3
80
14.85
16.75
4.7
400
9/27/80
19
6.1
80
15.85
16.85
3.6
400
9/28/80
18
3.1
80
16.85
16.85
4.4
250
9/29/80
14
3.3
76
15.55
15.44
5
100
9/29/80
16
5.1
76
16.15
16.04
3.9
100
9/29/80
18
5.2
76
16.05
15.94
5.2
50
1/6/81
16
4
60
17.15
15.55
21.5
50
1/6/81
17
5.1
58
17.45
15.75
13.1
50
1/6/81
18
4.9
60
17.25
15.45
9.4
50
1/9/81
15
4.7
87
14.45
15.35
3.4
100
1/9/81
16
4.6
85
14.85
15.35
4.8
100
1/9/81
18
4.9
87
15.05
15.35
3.1
100
1/13/81
15
5.8
65
16.95
15.55
11.6
50
1/13/81
17
4.2
84
15.85
15.45
8.5
50
Table 8 shows the source and receptor characteristics used in the Ventura tracer simulations. The
boat releases assumed a release height of 8.1 m, building height of 7 m and a width (and length)
of 20 m. Downwind receptor distances were varied to match the downwind distances of the
measurement site with the highest observed concentration for each period.
16
-------
Table 8. Ventura receptor distances.
Release
number
Date
Hour (LST)
Receptor distance (m)
1
9/24/80
16
9291
2
9/24/80
18
9211
3
9/24/80
19
10799
4
9/27/80
14
9123
5
9/27/80
19
9123
6
9/28/80
18
9145
7
9/29/80
14
8085
8
9/29/80
16
7854
9
9/29/80
18
7854
10
1/6/81
16
7463
11
1/6/81
17
7416
12
1/6/81
18
7463
13
1/9/81
15
7956
14
1/9/81
16
7749
15
1/9/81
18
7704
16
1/13/81
15
7705
17
1/13/81
17
6914
2.2 AERMET configurations
2.2.1 WRF simulations
WRF version 4.4.2 was applied over multiple near-shore locations in Louisiana and California.
The time periods modeled for each location are indicated in Table 9 below. These simulations
were conducted using nested domains of 12-km, 4-km, and 1.33-km and utilizing a 354ayer
vertical resolution. These WRF domains encompass the entire dispersion modeling domain and
are shown for each location in Figure 6 through Figure 9. The ERA-Interim 6-hourly reanalysis
dataset was used for initialization. All WRF simulations utilized the physics options outlined
below:
• Microphysics: Thompson
17
-------
• Planetary Boundary Layer: UW
• Cumulus: Kain-Fritsch
• Radiation: RRTMG
• Land Surface Model: NOAH
• Surface Layer: Eta
An effort was made to select model options and domains like work conducted during the
development of AERCOARE (U.S. EPA, 2015). That report outlines extensive model
performance evaluation and is the basis for the options selected here.
18
-------
WRF Domain - Cameron
Figure 6. Cameron, LA WRF domains. The large outer box is the 12-km domain, the white box
is the 4-km domain, and the red box is the 1.33 km domain.
19
-------
WRF Domain - Carpinteria
Figure 7. Carpinteria, CA WRF domains. The large outer box is the 12-km domain, the white
box is the 4-km domain, and the red box is the 1.33 km domain.
20
-------
WRF Domain - Pismo
Figure 8. Pismo Beach, CA WRF domains. The large outer box is the 12-krn domain, the white
box is the 4-km domain, and the red box is the 1.33 km domain.
21
-------
WRF Domain - Ventura
Figure 9. Ventura, CA WRF domains. The large outer box is the 12-km domain, the white box
is the 4-km domain, and the red box is the 1.33 km domain.
22
-------
Table 9. Time periods modeled for each location.
Location
Period
Cameron, LA
Period 1: 7/15/1981 - 7/31/1981
Period 2: 2/10/1982-2/25/1982
Carpinteria, CA
Period 1: 9/15/1985 -9/30/1985
Pismo Beach, CA
Period 1: 12/5/1981 - 12/20/1981
Period 2: 6/15/1982-6/30/1982
Ventura, CA
Period 1: 9/15/1980-9/30/1980
Period 2: 1/1/1981 - 1/15/1981
2.2.1.1 MMIF output
Once WRF simulations were completed, the 1.3 km WRF output was processed in MMIF to
generate data formatted for input to AERMET. Locations for extraction were based on the
release point locations shown in Figure 2 through
23
-------
VENTURA. CA
'£M.
mt
ln?c
!Av=
W
1
X
X
X
X
wtm
n
je
X
"X*
J
&
X
*x
X
*
2*1
=^r
x-x-
284 286 288 290 292 294 296 298 300 302 304
UTM East (Km) Zone 1 IN, Datum NAS-C
r 100
-95
Snow/Ice
-90
-85
Tundra
-80
-75
Barren
70
-65
Wetland
-60
-55
Water
-50
-45
Forest
-40
-35
Range
-30
-25
Agriculture
-20
-15
UrbanBuilt-Up
-10
Land Use
X Sampler Locations
~ Tracer Releases
Figure 5.
Files generated for AERMET input would be processed in AERMET using both the pass-
through option and the COARE processing option. These files contained all of the variables
output by MMIF for overwater grid cells for AERMET processing in MM IF (Ramboll, 2023).
Winds, temperature, and relative humidity were output at several levels in meters: 12.5, 37.5,
62.5, 87.5, 112.5, 137.5, 162.5, 187.5, 225, 275, 325, 375, 425, 75, 550, 650, 750, 850, 950,
1250, 1750, 2250, 2750, 3250, 3750, 4250, and 4750. Additionally, 2 m temperature was
output.
Files generated for AERCOARE output were processed in AERMET using the COARE option
and included winds at 10 m and temperature and relative humidity at 2m. See the MMIF user's
guide (Ramboll, 2023) for AERCOARE formatted output. In addition to winds, temperature,
and relative humidity, sea surface temperature, pressure, downward solar radiation, downward
longwave radiation, precipitation, total sky cover, mixing height, vertical potential temperature
gradient above the PEL, and depth of sea surface temperature measurement. See the MMIF
user's guide (Ramboll, 2023) for AERCOARE formatted output.
24
-------
2.2.2 AERMET-COARE configurations
The following scenarios were run for the COARE processing and are shown in Table 10. An 'X'
in the cell for a scenario and location indicates that scenario was run for the case location.
These scenarios are analogous to the scenarios used in the original AERCOARE work (U.S.
EPA, 2012b). In addition to the prognostic data, the observed data were also processed in
AERMET and AERMOD for the following scenarios as part of the evaluation of COARE
implementation into AERMET. Details about the observed data processing are in U.S. EPA
(2023 d). Table 11 shows the names of the AERMET/AERMOD simulations associated with
both the observed and prognostic data scenarios processed with COARE. Prognostic data run in
AERMET with the pass-through option is designated as AERMET PASS. For all COARE
processing, surface roughness values were calculated based on surface friction velocity (option
ZO_U* in AERMET).
• Scenario l1:
o Reset absolute value of Monin-Obukhov length to 5 m if absolute value of
Monin-Obukhov length is less than 5 m. Retain original sign (+ or -) of Monin-
Obukhov length
o Use observed mixing height for convective mixing height and calculate
mechanical mixing height without smoothing; Reset mechanical mixing height to
25 m if less than 25 m.
• Scenario la:
o Reset absolute value of Monin-Obukhov length to 5 m if absolute value of
Monin-Obukhov length is less than 5 m. Retain original sign (+ or -) of Monin-
Obukhov length
o Use observed mixing height for convective mixing height and calculate
mechanical mixing height without smoothing; Reset mechanical mixing height to
1 m if less than 1 m.
• Scenario lb:
1 The observed meteorological data for Scenario 1 and la-lc contains turbulence data for use in AERMOD
while the prognostic data does not contain turbulence. The observed meteorological data also has a Scenario 2,
where the observed turbulence data is not used AERMOD but the other meteorological parameters are the same as
Scenario 1. See U.S. EPA (2023d) for details.
25
-------
o Reset absolute value of Monin-Obukhov length to 5 m if absolute value of
Monin-Obukhov length is less than 5 m. Retain original sign (+ or -) of Monin-
Obukhov length
o Use observed mixing height for convective mixing height and calculate
mechanical mixing height without smoothing; Reset mechanical mixing height to
5 m if less than 5 m.
• Scenario lc:
o Reset absolute value of Monin-Obukhov length to 5 m if absolute value of
Monin-Obukhov length is less than 5 m. Retain original sign (+ or -) of Monin-
Obukhov length
o Use observed mixing height for convective mixing height and calculate
mechanical mixing height without smoothing; Reset mechanical mixing height to
15 m if less than 15 m.
• Scenario 2:
o Reset absolute value of Monin-Obukhov length to 1 m if absolute value of
Monin-Obukhov length is less than 1 m. Retain original sign (+ or -) of Monin-
Obukhov length
o Use observed mixing height for convective and mechanical mixing heights;
Reset mechanical mixing height to 1 m if less than 1 m.
• Scenario 3:
o Reset absolute value of Monin-Obukhov length to 5 m if absolute value of
Monin-Obukhov length is less than 5 m. Retain original sign (+ or -) of Monin-
Obukhov length
o Use observed mixing height for convective and mechanical mixing heights;
Reset mechanical mixing height to 1 m if less than 1 m.
26
-------
Table 10. AERMET-COARE configurations for prognostic data.
Scenario
Cameron
Carpinteria
Pismo Beach
Ventura
1
X
X
X
X
la
X
X
lb
X
X
lc
X
X
2
X
X
X
X
3
X
X
X
X
For the comparisons using prognostic data, AERMET was run with a wind speed threshold of
0.3 m/s instead of the recommended value of 0 m/s in U.S. EPA (2023 e). This was done
because the same prognostic data was used in the COARE implementation evaluation in U.S.
EPA (2023d) and AERCOARE does not contain the reset of winds below 2m x Ovmin where
Ovmin=0,2 m/s to 21/2 x Ovmin that AERMET does.
27
-------
Table 11. AERMET-COARE configuration names for prognostic data.
Scenario
Observed data
Prognostic data processed with
COARE
AERMET multi-
level
AERMET single
level
1
OBSl
COAREMLl
COARESL1
la
OBS1A
COAREML1A
COARESL1A
lb
OBS1B
COAREML1B
COARESL1B
lc
OB S_1C
COAREML1C
COARESL1C
2
OB S 3
COAREML2
COARESL2
3
OBS4
COAREML3
COARESL3
2.3 Meteorological data evaluation
Meteorological data comparisons between the observed data and prognostic data will
encompass mean bias and fractional bias calculations for several key variables, wind speed,
reference air temperature, sea surface temperature, air-sea temperature differences, relative
humidity, Monin-Obuklov length, mixing height, surface friction velocity, and surface
roughness. Fractional bias is calculated as:
FB = 2
lOB+PRl
(1)
Where PR is the prognostic value and OB is the observed value. Negative (positive)
values indicate that the prognostic data underpredicts (overpredicts) compared to observations.
The mean bias is the essentially the numerator of equation 1.
28
-------
For mixing height, convective and mechanical mixing heights are compared separately.
Also, for each hour and scenario, the maximum of the two heights is used to represent the
mixing height for a given hour and scenario. This is done because AERMOD uses the
maximum of the two heights for each hour. Carpinteria was the only case study that used actual
wind directions, so wind direction difference statistics were calculated. For the wind direction
difference statistics, a difference called displacement, which is the difference in the U and V
vectors of the modeled and observed winds and was used. This was used in the assessment of
the 2011 12km WRF simulations over the U.S. (US EPA, 2014). The displacement can be
calculated as:
D = abs{(JJM -U0 + VM- Vq) x (1 km/1000 m) x (3600 s/hr) x lhr) (2)
Where D is the displacement in km, Um and Vm are the u and v components respectively
of the prognostic wind vector and Uo and Vo are the u and v components of the observed wind
vector.
2.4 AERMOD evaluation
Except for Carpinteria, all AERMOD runs were run in screening mode, i.e., the receptor was
assumed to be on the plume centerline and the AERMOD SCREEN model option used. For
those screening mode cases using measured data, the wind direction was set to 270°, or westerly
winds. Carpinteria AERMOD runs reflected actual source-receptor distances and orientation.
All AERMOD runs were with version 22112. A test statistic called Robust Highest
Concentrations (RHC) (U.S. EPA, 1992) was calculated and compared as well for each study
area. The RHC is calculated as:
RHC = X(JV) + [X - X(JV)] x In (3)
Where X(N) is the Nth largest value, X is the average of N-l values, and N is the number of
values exceeding the threshold value, in this case 10.
29
-------
3.0 Results
3.1 Meteorological data comparisons
Mean biases and fractional biases of wind speed, temperature, sea surface temperature,
air-sea temperature differences, and relative humidity are shown in Table 12. These variables
represent input variables and are independent of the scenarios discussed in Section 2.2.2. Table
13 through Table 19 show the mean and fractional biases for surface friction velocity (u*),
convective mixing height, mechanical mixing height, mixing height used in AERMOD
(maximum of the convective or mechanical mixing height for an hour), surface roughness
length used in AERMOD2, Monin-Obukhov length, and absolute Monin-Obukhov length
respectively. Both actual Monin-Obukhov length and absolute Monin-Obukhov length are
presented because there are hours for each study area where the sign of Monin-Obukhov length
can be opposite between the observed and prognostic meteorological data, skewing the bias
statistics. Results in Table 12 show that the prognostic data and observed data are generally in
fairly good agreement, fraction biases less than 2.0. Wind speeds tend to be less than 1 m/s in
differences and air and sea temperatures tend to be within 2 degrees. Relative humidity also
tends to be in relatively good agreement. For Carpinteria, wind displacement ranged from 0.39
km to 28 km with a mean of 11 km. For the calculated variables in Table 13 through Table 19
there are differences and fractional biases tend to be generally acceptable. Fractional biases for
Monin-Obukhov length for Ventura (Table 18) tend to have large fractional biases but that is
due to the hours where the signs of Monin-Obukhov length are opposites between observed data
and prognostic data. Based on Table 19, the magnitudes of the Monin-Obukhov lengths are
comparable between observed and prognostic data.
2 When AERMOD reads the surface meteorological file, it will reset surface roughness values < 0.0001 m
to 0.0001 m. The surface roughness values compared in the table are based on those reset values.
30
-------
Table 12. Mean and fractional biases (Prognostic - observations) for each study area for wind speed, air temperature, sea surface
temperature, air-sea temperature difference, and relative humidity.
Variable
Study area
Cameron
Carpinteria
Pismo Beach
Ventura
Mean Bias
Fractional
Bias
Mean Bias
Fractional
Bias
Mean Bias
Fractional
Bias
Mean Bias
Fractional
Bias
Wind speed
0.79
0.13
1.80
0.44
1.48
0.29
-0.49
-0.11
Air temperature
1.88
0.006
0.15
0.0005
0.16
0.0005
0.1
0.0003
Sea Surface Temperature
2.17
0.008
-0.34
-0.001
1.61
0.005
-0.66
-0.002
Air-sea surface temperature
difference
-0.29
-1.63
0.49
-1.67
-1.45
-1.37
0.76
-1.54
Relative humidity
12.46
0.17
0.59
0.009
6.11
0.07
12.47
0.16
31
-------
Table 13. Mean and fractional biases for each study area for surface friction velocity (u*). An X
indicates that a particular scenario is not valid for a particular case study.
Scenario
Cameron
Car]
jinteria
Pismo Beach
Ventura
MB
FB
MB
FB
MB
FB
MB
FB
AERMET-OBS 1
0.04
0.28
0.06
0.68
0.08
0.54
-0.02
-0.15
AERMET-OBS 1A
X
X
0.06
0.68
0.08
0.54
X
X
AERMET-OBS IB
X
X
0.06
0.68
0.08
0.54
X
X
AERMET-OBS 1C
X
X
0.06
0.68
0.08
0.54
X
X
AERMET-OBS 3
0.04
0.30
0.06
0.72
0.08
0.56
-0.02
-0.15
AERMET-OBS 4
0.04
0.28
0.06
0.68
0.08
0.54
-0.02
-0.15
COARE 1-OBS 1
0.04
0.28
0.06
0.60
0.09
0.57
-0.02
-0.13
COARE 1A-OBS 1A
X
X
0.06
0.60
0.09
0.57
X
X
COARE 1B-OBS IB
X
X
0.06
0.60
0.09
0.57
X
X
COARE 1C-OBS 1C
X
X
0.06
0.60
0.09
0.57
X
X
COARE 3-OBS 3
0.04
0.30
0.06
0.58
0.09
0.59
-0.02
-0.13
COARE 4-OBS 4
0.04
0.28
0.06
0.60
0.09
0.57
-0.02
-0.13
COARE 1 1-OBS 1
0.05
0.30
0.07
0.63
0.09
0.59
-0.01
-0.10
COARE 1 1A-OBS 1A
X
X
0.07
0.63
0.09
0.59
X
X
COARE 1 1B-OBS IB
X
X
0.07
0.63
0.09
0.59
X
X
COARE 1 1C-OBS 1C
X
X
0.07
0.63
0.09
0.59
X
X
COARE 1 3-OBS 3
0.05
0.32
0.06
0.62
0.09
0.61
-0.01
-0.10
COARE 1 4-OBS 4
0.05
0.30
0.07
0.63
0.09
0.59
-0.01
-0.10
32
-------
Table 14. Mean and fractional biases for each study area for convective mixing height. An X
indicates that a particular scenario is not valid for a particular case study.
Scenario
Cameron
Carpinteria
Pismo Beach
Ventura
MB
FB
MB
FB
MB
FB
MB
FB
AERMET-OBS 1
-228.56
-0.69
-357.31
-1.13
-335.40
-0.92
-73.00
-1.17
AERMET-OBS 1A
X
X
-357.31
-1.13
-335.40
-0.92
X
X
AERMET-OBS IB
X
X
-357.31
-1.13
-335.40
-0.92
X
X
AERMET-OBS 1C
X
X
-357.31
-1.13
-335.40
-0.92
X
X
AERMET-OBS 3
-228.56
-0.69
-357.31
-1.13
-335.40
-0.92
-73.00
-1.17
AERMET-OBS 4
-228.56
-0.69
-357.31
-1.13
-335.40
-0.92
-73.00
-1.17
COARE 1-OBS 1
-228.56
-0.69
-357.31
-1.13
-335.40
-0.92
-69.67
-1.08
COARE 1A-OBS 1A
X
X
-357.31
-1.13
-335.40
-0.92
X
X
COARE 1B-OBS IB
X
X
-357.31
-1.13
-335.40
-0.92
X
X
COARE 1C-OBS 1C
X
X
-357.31
-1.13
-335.40
-0.92
X
X
COARE 3-OBS 3
-228.56
-0.69
-357.31
-1.13
-335.40
-0.92
-73.00
-1.17
COARE 4-OBS 4
-228.56
-0.69
-357.31
-1.13
-335.40
-0.92
-73.00
-1.17
COARE 1 1-OBS 1
-228.56
-0.69
-357.31
-1.13
-335.40
-0.92
-69.67
-1.08
COARE 1 1A-OBS 1A
X
X
-357.31
-1.13
-335.40
-0.92
X
X
COARE 1 1B-OBS IB
X
X
-357.31
-1.13
-335.40
-0.92
X
X
COARE 1 1C-OBS 1C
X
X
-357.31
-1.13
-335.40
-0.92
X
X
COARE 1 3-OBS 3
-228.56
-0.69
-357.31
-1.13
-335.40
-0.92
-73.00
-1.17
COARE 1 4-OBS 4
-228.56
-0.69
-357.31
-1.13
-335.40
-0.92
-73.00
-1.17
33
-------
Table 15. Mean and fractional biases for each study area for mechanical mixing height. An X
indicates that a particular scenario is not valid for a particular case study.
Scenario
Cameron
Carpinteria
Pismo Beach
Ventura
MB
FB
MB
FB
MB
FB
MB
FB
AERMET-OBS 1
17.96
0.21
51.96
0.25
21.82
0.19
-105.41
-1.08
AERMET-OBS 1A
X
X
57.78
0.72
23.43
0.28
X
X
AERMET-OBS IB
X
X
57.19
0.58
23.43
0.28
X
X
AERMET-OBS 1C
X
X
55.30
0.40
22.96
0.24
X
X
AERMET-OBS 3
-112.69
-0.32
-390.33
-1.37
-181.54
-0.39
-139.82
-0.92
AERMET-OBS 4
-112.69
-0.32
-390.33
-1.37
-181.54
-0.39
-139.82
-0.92
COARE 1-OBS 1
63.12
0.41
82.37
0.46
129.57
0.71
-27.06
-0.19
COARE 1A-OBS 1A
X
X
85.37
0.76
131.18
0.76
X
X
COARE 1B-OBS IB
X
X
84.78
0.66
131.18
0.76
X
X
COARE 1C-OBS 1C
X
X
83.00
0.50
130.71
0.75
X
X
COARE 3-OBS 3
-121.77
-0.60
-411.59
-1.44
-269.61
-0.85
-149.88
-1.10
COARE 4-OBS 4
-121.77
-0.60
-411.59
-1.44
-269.61
-0.85
-149.88
-1.10
COARE 1 1-OBS 1
68.73
0.43
86.56
0.48
139.61
0.74
-20.88
-0.14
COARE 1 1A-OBS 1A
X
X
90.41
0.80
141.21
0.79
X
X
COARE 1 1B-OBS IB
X
X
89.81
0.71
141.21
0.79
X
X
COARE 1 1C-OBS 1C
X
X
88.00
0.55
140.75
0.77
X
X
COARE 1 3-OBS 3
-107.92
-0.52
-411.59
-1.44
-269.61
-0.85
-149.88
-1.10
COARE 1 4-OBS 4
-107.92
-0.52
-411.59
-1.44
-269.61
-0.85
-149.88
-1.10
34
-------
Table 16. Mean and fractional biases for each study area for mixing height used in AERMOD.
An X indicates that a particular scenario is not valid for a particular case study.
Scenario
Cameron
Carpinteria
Pismo Beach
Ventura
MB
FB
MB
FB
MB
FB
MB
FB
AERMET-OBS 1
-57.38
0.0006
-133.93
-0.49
-13.71
0.10
-105.41
-1.08
AERMET-OBS 1A
X
X
-128.11
-0.01
-12.11
0.19
X
X
AERMET-OBS IB
X
X
-128.70
-0.16
-12.11
0.19
X
X
AERMET-OBS 1C
X
X
-130.59
-0.34
-12.57
0.15
X
X
AERMET-OBS 3
-81.81
-0.21
-370.07
-1.27
-172.36
-0.36
-139.82
-0.92
AERMET-OBS 4
-81.81
-0.21
-370.07
-1.27
-172.36
-0.36
-139.82
-0.92
COARE 1-OBS 1
-14.04
0.19
-101.70
-0.23
87.04
0.59
-27.06
-0.19
COARE 1A-OBS 1A
X
X
-98.70
0.06
88.64
0.65
X
X
COARE 1B-OBS IB
X
X
-99.30
-0.03
88.64
0.65
X
X
COARE 1C-OBS 1C
X
X
-101.07
-0.20
88.18
0.63
X
X
COARE 3-OBS 3
-121.77
-0.60
-411.59
-1.44
-269.61
-0.85
-149.88
-1.10
COARE 4-OBS 4
-121.77
-0.60
-411.59
-1.44
-269.61
-0.85
-149.88
-1.10
COARE 1 1-OBS 1
-8.92
0.21
-97.67
-0.22
94.89
0.62
-20.88
-0.14
COARE 1 1A-OBS 1A
X
X
-93.81
0.10
96.50
0.67
X
X
COARE 1 1B-OBS IB
X
X
-94.41
0.02
96.50
0.67
X
X
COARE 1 1C-OBS 1C
X
X
-96.22
-0.14
96.04
0.66
X
X
COARE 1 3-OBS 3
-107.92
-0.52
-411.59
-1.44
-269.61
-0.85
-149.88
-1.10
COARE 1 4-OBS 4
-107.92
-0.52
-411.59
-1.44
-269.61
-0.85
-149.88
-1.10
35
-------
Table 17. Mean and fractional biases for each study area for surface roughness (z0). An X
indicates that a particular scenario is not valid for a particular case study.
Scenario
Cameron
Carpinteria
Pismo Beach
Ventura
MB
FB
MB
FB
MB
FB
MB
FB
AERMET-OBS 1
0.00001
0.08
-0.00006
-0.15
0.00003
0.19
0.00
0.00
AERMET-OBS 1A
X
X
-0.00006
-0.15
0.00003
0.19
X
X
AERMET-OBS IB
X
X
-0.00006
-0.15
0.00003
0.19
X
X
AERMET-OBS 1C
X
X
-0.00006
-0.15
0.00003
0.19
X
X
AERMET-OBS 3
0.00001
0.08
-0.00006
-0.15
0.00003
0.19
0.00
0.00
AERMET-OBS 4
0.00001
0.08
-0.00006
-0.15
0.00003
0.19
0.00
0.00
COARE 1-OBS 1
0.00
0.00
-0.00006
-0.20
0.00
0.00
0.00
0.00
COARE 1A-OBS 1A
X
X
-0.00006
-0.20
0.00
0.00
X
X
COARE 1B-OBS IB
X
X
-0.00006
-0.20
0.00
0.00
X
X
COARE 1C-OBS 1C
X
X
-0.00006
-0.20
0.00
0.00
X
X
COARE 3-OBS 3
0.00
0.00
-0.00006
-0.20
0.00
0.00
0.00
0.00
COARE 4-OBS 4
0.00
0.00
-0.00006
-0.20
0.00
0.00
0.00
0.00
COARE 1 1-OBS 1
0.00
0.00
-0.00006
-0.20
0.00
0.00
0.00
0.00
COARE 1 1A-OBS 1A
X
X
-0.00006
-0.20
0.00
0.00
X
X
COARE 1 1B-OBS IB
X
X
-0.00006
-0.20
0.00
0.00
X
X
COARE 1 1C-OBS 1C
X
X
-0.00006
-0.20
0.00
0.00
X
X
COARE 1 3-OBS 3
0.00
0.00
-0.00006
-0.20
0.00
0.00
0.00
0.00
COARE 1 4-OBS 4
0.00
0.00
-0.00006
-0.20
0.00
0.00
0.00
0.00
36
-------
Table 18. Mean and fractional biases for each study area for Monin-Obukhov length. An X
indicates that a particular scenario is not valid for a particular case study.
Scenario
Cameron
Carpinteria
Pismo Beach
Ventura
MB
FB
MB
FB
MB
FB
MB
FB
AERMET-OBS 1
-80.94
2.00
51.59
0.91
-287.24
0.90
112.41
3.78
AERMET-OBS 1A
X
X
51.59
0.91
-287.24
0.90
X
X
AERMET-OBS IB
X
X
51.59
0.91
-287.24
0.90
X
X
AERMET-OBS 1C
X
X
51.59
0.91
-287.24
0.90
X
X
AERMET-OBS 3
-80.90
2.02
52.19
1.20
-286.88
0.92
112.41
3.78
AERMET-OBS 4
-80.94
2.00
51.59
0.91
-287.24
0.90
112.41
3.78
COARE 1-OBS 1
-158.83
1.83
-49.16
1.01
-55.81
1.78
134.18
1.83
COARE 1A-OBS 1A
X
X
-49.16
1.01
-55.81
1.78
X
X
COARE 1B-OBS IB
X
X
-49.16
1.01
-55.81
1.78
X
X
COARE 1C-OBS 1C
X
X
-49.16
1.01
-55.81
1.78
X
X
COARE 3-OBS 3
-158.80
1.84
-49.65
0.99
-55.46
1.80
134.18
1.83
COARE 4-OBS 4
-158.83
1.83
-49.16
1.01
-55.81
1.78
134.18
1.83
COARE 1 1-OBS 1
-161.72
1.83
-50.86
1.01
-680.56
1.76
140.17
1.63
COARE 1 1A-OBS 1A
X
X
-50.86
1.01
-680.56
1.76
X
X
COARE 1 1B-OBS IB
X
X
-50.86
1.01
-680.56
1.76
X
X
COARE 1 1C-OBS 1C
X
X
-50.86
1.01
-680.56
1.76
X
X
COARE 1 3-OBS 3
-161.68
1.84
-51.20
1.03
-680.21
1.78
140.17
1.63
COARE 1 4-OBS 4
-161.72
1.83
-50.86
1.01
-680.56
1.76
140.17
1.63
37
-------
Table 19. Mean and fractional biases for each study area for absolute Monin-Obukhov length.
An X indicates that a particular scenario is not valid for a particular case study.
Scenario
Cameron
Carpinteria
Pismo Beach
Ventura
MB
FB
MB
FB
MB
FB
MB
FB
AERMET-OBS 1
105.66
1.16
277.70
0.64
229.38
1.20
38.65
0.19
AERMET-OBS 1A
X
X
277.70
0.64
229.38
1.20
X
X
AERMET-OBS IB
X
X
277.70
0.64
229.38
1.20
X
X
AERMET-OBS 1C
X
X
277.70
0.64
229.38
1.20
X
X
AERMET-OBS 3
105.93
1.18
278.60
0.93
229.73
1.22
38.65
0.19
AERMET-OBS 4
105.66
1.16
277.70
0.64
229.38
1.20
38.65
0.19
COARE 1-OBS 1
178.52
1.36
415.23
0.82
608.97
1.33
74.38
0.40
COARE 1A-OBS 1A
X
X
415.23
0.82
608.97
1.33
X
X
COARE 1B-OBS IB
X
X
415.23
0.82
608.97
1.33
X
X
COARE 1C-OBS 1C
X
X
415.23
0.82
608.97
1.33
X
X
COARE 3-OBS 3
178.79
1.37
415.04
0.80
609.33
1.35
74.38
0.40
COARE 4-OBS 4
178.52
1.36
415.23
0.82
608.97
1.33
74.38
0.40
COARE 1 1-OBS 1
186.72
1.38
436.44
0.83
632.64
1.34
82.02
0.45
COARE 1 1A-OBS 1A
X
X
436.44
0.83
632.64
1.34
X
X
COARE 1 1B-OBS IB
X
X
436.44
0.83
632.64
1.34
X
X
COARE 1 1C-OBS 1C
X
X
436.44
0.83
632.64
1.34
X
X
COARE 1 3-OBS 3
186.99
1.39
436.39
0.85
632.99
1.36
82.02
0.45
COARE 1 4-OBS 4
186.72
1.38
436.44
0.83
632.64
1.34
82.02
0.45
3.2 AERMOD results
Table 20 lists the lists the modeled Robust Highest Concentration (RHC) for each scenario and
case study area based on prognostic data as well as the observed meteorology (for comparison)
for each of the study areas. Except for Cameron, the prognostic data output tended to
overpredict when compared to the observed RHC and did not agree as well with the observed
RHC as did the observed meteorological data, which is not unexpected. The treatment of the
prognostic data with COARE vs. the AERMET pass through tended to also agree better with the
observed RHC. Results between the number of levels used with the COARE treatment tended
to be mixed, with Carpinteria presenting more difference between RHC values for the multi-
level and single-level RHC values, while the other study areas showed little difference between
multi-level and single-level RHC values.
38
-------
Table 20. AERMOD Robust Highest Concentration for measured meteorological data and
prognostic data. Observed RHC value are in parenthesis with the study area name. An X
indicates that a particular scenario is not valid for a particular case study.
Scenario
Cameron
(40.8)
Carpinteria
(142.9)
Pismo Beach
(9.0)
Ventura
(4.3)
OBS 1 (with turbulence)
49.9
148.5
35.0
5.7
OBS 2 (no turbulence)
51.2
323.1
55.1
19.8
OBS 1A
X
373.2
36.1
X
OBS IB
X
263.5
36.1
X
OBS 1C
X
221.1
36.2
X
OBS 3
39.9
469.1
18.3
7.8
OBS 4
43.7
307.5
20.5
7.8
AERMET PASS
16.4
298.4
25.3
32.3
COARE ML 1
17.9
302.9
16.9
20.0
COARE ML 1A
X
311.2
16.9
X
COARE ML IB
X
311.2
16.9
X
COARE ML 1C
X
311.2
16.9
X
COARE ML 2
31.0
326.2
32.8
52.0
COARE ML 3
31.0
301.9
32.8
52.0
COARE SL 1
17.1
245.4
15.0
18.2
COARE SL 1A
X
247.3
15.0
X
COARE SL IB
X
247.3
15.0
X
COARE SL 1C
X
247.3
15.0
X
COARE SL 2
30.8
417.1
31.7
50.2
COARE SL 3
30.8
243.7
31.7
50.2
4.0 Summary and Conclusions
COARE algorithms were incorporated into AERMET version 23132 to allow processing of
measured or prognostic meteorological data to calculate representative boundary layer
parameters for the marine boundary layer environment. Four case studies were used to assess
the differences between treating prognostic data with the COARE algorithms in AERMET vs a
pass through of variables from MMIF to AERMET. Also, results between single-level and
multi-level data using COARE were assessed. The results generally showed that treatment of
prognostic data with COARE in AERMET agreed better with observations than the pass-
through. Results were mixed between comparisons of the single-level and multi-level data in
COARE. For guidance on the use of COARE algorithms with prognostic data see the
39
-------
AERMET User's Guide (U.S. EPA, 2023b) and MMIF guidance for AERMOD applications
(U.S. EPA, 2023e).
40
-------
5.0 References
DiCristofaro, D.C. and S.R. Hanna, 1989. OCD: The Offshore and Coastal Dispersion Model,
Version 4. Volume I: User's Guide, and Volume II: Appendices. Sigma Research
Corporation, Westford, MA. (NTIS Nos. PB 93-144384 and PB 93-144392).
Fairall, C.W., E.F. Bradley, J.E. Hare, A.A. Grachev, and J.B. Edson, 2003: "Bulk
Parameterization of Air-Sea Fluxes: Updates and Verification for the COARE
Algorithm." J. Climate, 16, 571-591.
Ramboll, 2023: The Mesoscale Model Interface Program (MMIF) Version 4.1 User's Manual.
U.S. EPA, 1992: Protocol for Determining the Best Performing Model, EPA-454/R-92-025.
U.S. Environmental Protection Agency, Research Triangle Park, NC.
U. S. EPA, 2011a: Model Clearinghouse Review AERMOD-COARE as an Alternative Model in
an Arctic Ice Free Environment. George Bridgers Memorandum dated May 6, 2011,
Office of Air Quality Planning and Standards, Research Triangle Park, North Carolina,
27711.
https://gaftp.epa.gov/Air/aqmg/SCRAM/mchisrs/R1444 Bridgers 6 May 11 AMERO
D-COARE.pdf
U.S. EPA, 201 lb: User's Guide for the AERMOD Terrain Preprocessor (AERMAP). EPA-
454/B-03-003. U.S. Environmental Protection Agency, Research Triangle Park, North
Carolina.
U.S. EPA, 2012a: User's Manual AERCO ARE Version 1.0. EPA-910/R-12-008. U.S. EPA,
Region 10, Seattle, WA.
U.S. EPA, 2012b: Evaluation of the Combined AERCO ARE AERMOD Modeling Approach for
Offshore Sources. EPA-910/R-12-007. U.S. EPA, Region 10, Seattle, WA.
U.S. EPA, 2014: Meteorological Model Performance for Annual 2011 WRF v3.4 Simulation.
http://www.epa.gov/ttn/scram/reports/MET_TSD_2011_final_ll-26-14.pdf
U.S. EPA, 2015: Combined WRF MMIF AERCO ARE AERMOD Overwater Modeling Approach
for Offshore Emission Sources: Volume 3 - Analysis of AERMOD Performance Using
Weather Research and Forecasting Model Predicted Meteorology and Measured
Meteorology in the Arctic. EPA-910/R-15-001c. U.S. EPA, Region 10, Seattle, WA.
41
-------
U.S. EPA, 2023a: User's Guide for the AMS/EPA Regulatory Model (AERMOD). EPA-454/B-
23-008. U.S. Environmental Protection Agency, Research Triangle Park, North Carolina.
U. S. EPA, 2023b: User's Guide for the AERMOD Meteorological Preprocessor (AERMET).
EPA-454/B-23-005. U.S. Environmental Protection Agency, Research Triangle Park,
North Carolina.
U.S. EPA, 2023c: Guideline on Air Quality Models. 40 CFR Part 51 Appendix W.
U.S. EPA, 2023 d: Evaluation of the Implementation of the Coupled Ocean Atmosphere
Response Experiment (COARE) Algorithms into AERMETfor Marine Boundary Layer
Environments. EPA-454/R-23-008. U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina.
U. S. EPA, 2023 e: Guidance on the Use of the Mesoscale Model Interface Program (MMIF) for
AERMOD Applications. EPA-454/B-23-006. U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina.
42
-------
United States Office of Air Quality Planning and Standards Publication No. EPA-454/R-23-010
Environmental Protection Air Quality Assessment Division October 2023
Agency Research Triangle Park, NC
------- |