PROPERTY OF
'i3IOiV July 1934
OF
WTEOROIQGY
EVALUATION OF THE POLLUTION EPISODIC MODEL
USING THE RAPS DATA
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
KESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
EVALUATION OF THE POLLUTION EPISODIC MODEL
USING THE RAPS DATA
by
William R. Pendergrass and K. Shankar Rao
Atmospheric Turbulence and Diffusion Division
National Oceanic and Atmospheric Administration
Oak Ridge, Tennessee 37830
IAG-AD-13-F-1-707-0
Project Officer
Jack H. Shreffler
Meteorology and Assesment Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research
Laboratory, U.S. Environmental Protection Agency, and approved for publica-
tion. Approval does not signify that the contents necessarily reflect the
views and policies of the U.S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
ii
-------
ABSTRACT
The Pollution Episodic Model (PEM) is an urban-scale model capable of
predicting short-term average ground-level concentrations and deposition
fluxes of one or two gaseous or particulate pollutants at multiple recep-
tors. The two pollutants may be nonreactive, or chemically-coupled through
a first-order chemical transformation. Up to 300 isolated point sources
and 50 distributed area sources may be considered in the calculations.
Concentration and deposition flux estimates are made using the hourly mean
meteorological data. Up to a maximum of 24 hourly scenarios of meteorology
may be included in an averaging period. PEM is intended for studies of the
atmospheric transport, transformation, and deposition of pollutants in
urban areas to assess the impact of existing or new sources or source modi-
fications on air quality, and for urban planning.
This report describes an evaluation of the PEM using the St. Louis
Regional Air Pollution Study (RAPS) data. This evaluation is designed to
test the performance of the model by comparing its concentration estimates
to the measured air quality data, using appropriate statistical measures.
Twenty days, ten summer and ten winter, are selected from the RAPS data
base for the PEM evaluation. The model's performance is judged by com-
paring the calculated 12-hour average concentrations with the corresponding
observed values for five pollutant species, namely, S02, fine and coarse
sulfates, and fine and coarse total mass. A first-order chemical transfor-
mation of SC>2 to fine sulfate is considered in the calculations in addi-
tion to the direct emission and dry deposition of all five pollutants. The
model domain, covering 125 x 125 km with a 50 x 50 receptor grid, includes
286 point sources and 36 area sources in the greater St. Louis urban area.
Hourly meteorological data and detailed emission inventories for the five
pollutants are used as inputs to the model.
Statistical tests for evaluation of the model performance include
standard measures of differences and correlation between observations and
calculations paired in space and time. For each pollutant, scatterplots of
calculated concentrations and .differences versus observed concentrations
are presented; a linear regression line is determined and evaluation
statistics are tabulated. Additional plots, examining the model perfor-
mance as a function PEM evaluation days and RAMS station numbers, are
given.
The emphasis in this evaluation is on S02 and sulfate concentration
predictions. For the twenty PEM evaluation days, PEM predicted average
concentrations of SC^, and fine and coarse sulfates to within a factor of
two. The model overpredicted the average concentrations of fine and coarse
total mass by a factor of three to four over the evaluation period. This
is attributed primarily to overestimation of emission rates and incorrect
location of area sources, which dominate the fine and coarse total mass
emissions. Other possible sources of errors in the calculations are listed
and discussed.
The work described in this report was performed by NOAA's Atmospheric
Turbulence and Diffusion Division in partial fulfillment of Interagency
Agreement No. AD-13-F-1-707-0 with the U. S. Environmental Protection
Agency. This work, covering the period October 1982 to December 1983, was
completed as of February 29, 1984.
iii
-------
-------
CONTENTS
Abstract ill
Figures vi
Tables Viii
Acknowledgements ix
1. INTRODUCTION 1
2. RAPS DATA BASE 3
2.1 Emissions 3
2.2 Meteorology 7
2.3 Measured Concentrations 10
3. MODEL EVALUATION 16
3.1 PEM Runs 17
3.1.1 Receptor Grid 17
3.1.2 Emissions 17
3.1.3 Deposition Parameters 18
3.1.4 Chemical Transformation Rate 19
3.1.5 Model Calculations 19
3.2 Evaluation Statistics -. 22
4. RESULTS AND DISCUSSION 25
4.1 Sulfur Dioxide 25
4.2 Fine and Coarse Sulfates 32
4.3 Fine and Coarse Total Mass 37
5. CONCLUSIONS 56
References 61
-------
FIGURES
Number Page
1 Location of RAMS stations in St. Louis /RAPS field program .... 8
2 Receptor grid used in PEM evaluation and its relation to the
RAMS stations ................................................ 20
3 Comparison of calculated and observed S02 concentrations for
the twenty PEM evaluation days .... ......... . ................. 27
4 S02 residuals (D^ = 0^ - P^) versus observed S02 concentrations
for the twenty PEM evaluation days ......... . ................. 29
5 Comparison of the observed and calculated daily mean concentra-
tions of SC>2 (averaged over all RAMS stations) for each of the
twenty PEM evaluation days ............ . ...................... 30
6 Comparison of the observed and calculated daily mean concentra-
tions of SC-2 (averaged over all PEM evaluation days) at each of
the RAMS stations ............................................ 31
7 Comparison of calculated and observed fine sulfate concentra-
tions for the twenty PEM evaluation days ..................... 34
8 Fine sulfate residuals (Dj. - Oi - P*) versus observed fine
sulfate concentrations for the twenty PEM evaluation days .... 35
9 Comparison of the observed and calculated mean daily centrations
of fine sulfate (averaged over all reporting RAMS stations) for
each of the twenty PEM evaluation days ....................... 36
10 Comparison of the observed and calculated mean daily concentra-
tions of fine sulfate (averaged over all PEM evaluation days) at
each of the reporting RAMS stations .......................... 38
11 Comparison of calculated and observed coarse sulfate concentra-
tions for the twenty PEM evaluation days ..................... 40
12 Coarse sulfate residuals (D± = 0^ - P^) versus observed coarse
sulfate concentrations for the twenty PEM evaluation days .... 41
13 Comparison of the observed and calculated daily mean concen-
trations of coarse sulfate (averaged over all reporting RAMS
stations) for each of the twenty PEM evaluation days ......... 42
14 Comparison of the observed and calculated daily mean concentra-
tions of coarse sulfate (averaged over all PEM evaluation days)
at each of the reporting RAMS stations ....................... 43
vi
-------
FIGURES (continued)
Number Page
15 Comparison of calculated and observed fine total mass concentra-
tions for the twenty PEM evaluation days 46
16 Fine total mass residuals (D^ » Oj_ - P-^) versus observed fine total
mass concentrations for the twenty PEM evaluation days .......... 47
17 Comparison of the observed and calculated daily mean concentrations
fine total mass (averaged over all reporting RAMS stations) for
each of the twenty PEM evaluation days 48
18 Comparison of the observed and calculated mean daily concentrations
of fine total mass (averaged over all PEM evaluation days) at each
of the reporting RAMS stations 49
19 Comparison of calculated and observed coarse total mass concen-
trations for the twenty PEM evaluation days 51
20 Coarse total mass residuals (Di=Oi~Pi) versus observed coarse total
mass concentrations for the twenty PEM evaluation days ........... 52
21 Comparison of the observed and calculated mean daily concentrations
of coarse total mass (averaged over all reporting RAMS stations)
for each of the twenty PEM evaluation days 53
22 Comparison of the observed and calculated mean daily concentrations
of coarse total mass (averaged over all PEM evaluation days) at
each of the reporting RAMS stations 54
vii
-------
TABLES
Number Page
1 PEM evaluation days of RAPS data 4
2 Area source emission file data 6
3 Point source emission file data 6
4 Base size distributions of sulfate 7
5 RAMS network instrumentation and measurements by station .... 9
6 Examples of hourly meterological data input for PEM 11-12
7 Examples of 12-hour average concentrations observed at
RAMS stations 14-15
8 PEM evaluation statistics for S(>2 •• 26
9 PEM evaluation statistics for fine sulfates 33
10 PEM evaluation statistics for coarse sulfates 39
11 PEM evaluation statistics for fine total mass 45
12 PEM evaluation statistics for coarse total mass 50
13 Average total emission rates from area and point sources .... 57
14 Mean concentration residuals by 12-hour averaging period .... 59
viii
-------
ACKNOWLEDGEMENTS
This report was prepared for the Office of Research and Development,
Environmental Sciences Research Laboratory (ESRL) of the U. S. Environmental
Protection Agency to support the needs of EPA's Office of Air Quality
Planning and Standards in urban particulate modeling. This work was
accomplished under interagency agreements among the U. S. Department of
Energy, the National Oceanic and Atmospheric Administration, and the EPA.
The authors thank Dr. Jack Shreffler of ESRL for his guidance and advice
during the course of this work, and for his interest and patience. The
authors express their appreciation to the following members of the
Atmospheric Turbulence and Diffusion Division: Director Bruce B. Hicks and
Dr. Ray Hosker for useful suggestions and discussions, Martha Stevens for
adapting PEM program for this evaluation, and Mary Rogers for her expert
typing and patient revisions.
The St. Louis/RAPS emission data tapes used in this model evaluation
were provided to EPA by Professor James Brock of the-University of Texas at
Austin, who derived the sulfate emissions and particle size distributions
for area and point sources from the 1976 RAPS Emission Inventory.
ix
-------
-------
SECTION 1
INTRODUCTION
The Pollution Episodic Model (PEM) described by Rao and Stevens (1982)
is an urban-scale model capable of predicting short-term ground-level con-
centrations and deposition fluxes of one or two gaseous or particulate
reactive pollutants in an urban environment with multiple point and area
sources. It is intended for studies of the atmospheric transport, trans-
formation, and deposition of acidic, toxic, and other pollutants in urban
areas to assess the impact of existing or new sources or source modifica-
tions on air quality, and for urban planning. PEM uses the concentration
algorithms developed by Rao (1982) which explicitly account for the effects
of dry deposition, sedimentation, and a first-order chemical transfor-
mation. Rao and Stevens (1982) discussed the analytical techniques, capa-
bilities and limitations, and input/output parameters of PEM. The PEM is
based on the Texas Episodic Model (TEM) developed by the Texas Air Control
Board (1979).
This report describes an evaluation of the PEM using the St. Louis
Regional Air Pollution Study (RAPS) data. This evaluation was designed to
test the model performance by comparing the model's concentration estimates
to the measured air quality data, using appropriate statistical measures of
performance (see, e.g., Fox, 1981).
Twenty days, ten summer and ten winter, were selected from the RAPS
data base for the PEM evaluation. The model's performance was judged by
1
-------
comparing the calculated average concentrations with the corresponding
observed values for the following five pollutant species:
1. S02
2. Fine sulfate
3. Coarse sulfate
4. Fine total mass
5. Coarse total mass
In the above, the cut-off size between fine and coarse particle fractions
was 2.5 van. A first-order chemical transformation of SC-2 to fine sulfate
was considered in the calculations in addition to the direct emission and
dry deposition of all five pollutants.
-------
SECTION 2
RAPS DATA BASE
The St. Louis Regional Air Pollution Study experiment and data base
have been described in detail in other publications (e.g., Schiermeier,
1978) and will not be discussed here; only the data used in the evaluation
of PEM will be described. Twenty days, ten summer and ten winter, were
selected from the RAPS data base by the Environmental Protection Agency
(EPA) for the PEM evaluation. The selected days are listed in Table 1.
Detailed emission inventories of the RAPS region, and meteorology and con-
centration measurements corresponding to these evaluation days were
supplied by the EPA from the RAPS data base.
2.1 EMISSIONS
Hourly area and point source emission inventories for a typical winter
day (January 19, 1976) and a typical summer day (July 26, 1976) for the
St. Louis metropolitan area were supplied by the EPA on two magnetic tapes.
For both days, precipitation was absent. The first tape included only area
sources, and the second only point'sources. The emission inventories were
supplied on a numerical grid with a fixed origin at XUTM = 710 km and
YUTM = 4250 km which extended to 60 km in both x and y directions. The
size of each emission grid cell for area sources was 5x5 km, thus giving
144 emission squares in the grid. The data tapes contained information on
S02, sulfate, and total particulate mass emissions, and the particle size
-------
TABLE 1
PEM Evaluation Days of RAPS data
WINTER
Date Julian day
Dec. 12, 1975 346
Dec. 22, 1975 356
Dec. 23, 1975 357
Dec. 31, 1975 365
Jan. 22, 1976 022
Feb. 3, 1976 034
Feb. 13, 1976 044
Dec. 8, 1976 343
Dec. 10, 1976 345
Dec. 17, 1976 352
SUMMER
Date Julian day
Jun. 8, 1976 160
Jun. 22, 1976 174
Jul. 6, 1976 188
Jul. 9, 1976 191
Jul. 19, 1976 201
Jul. 22, 1976 204
Jul. 30, 1976 212
Aug. 5, 1976 218
Aug. 13, 1976 226
Aug. 19, 1976 232
-------
derived from the 1976 RAPS Emission Inventory. Table 2 shows details of
the data supplied in the hourly emission files for area sources. Table 3
shows the corresponding information for point sources.
The sulfate emissions and their particle size distributions from both
point and area sources were derived by Professor James Brock (1982,
Personal communication) of the University of Texas at Austin. Briefly, an
average conversion rate of 1.85% of SO2 emissions was used to estimate the
sulfate emission rates for both area and point sources from the known
information on S02 emissions, provided particulate emissions existed. In
the case of point sources with no particulate emissions, but relatively
large 802 emissions, the sulfate emissions were calculated on the assump-
tion that in a short period of time, the conversion of S02 to 803 occurs
and contributes to the total mass in the region of interest. However, in
the case of area sources, the 802 emissions were relatively small (3% of
total 802 emissions) and, therefore, sulfate emissions could be neglected
if there were no associated particulate emissions.
The size distributions of sulfate particle emissions from area and
point sources were more difficult to estimate. This-information plays a
critical role in the evaluation of the health and visibility effects, and
yet little has appeared in the literature on this subject. Brock (1982)
estimated approximate base size distributions of sulfate for typical winter
and summer days, as shown in Table 4. Based on the studies of Tanner
et_ al^. (1979), about 50% of the sulfate was assigned to the size range less
than 0.25 urn for the summer aerosol, while approximately 25% of the sulfate
was assigned to this size range for winter aerosol.
-------
TABLE 2
Area Source Emission File Data
1) XUTM,YUTM: southwest corner of a numerical grid in UTM coordinates
(km)
2) Length of grid square (km)
3) Total emissions of mass (g/sec)
4) Total mass size spectrum in weight frations of total mass emissions.
(PART(l)"larger than 7 microns, PART(2)=3-7 microns, PART(3)=l-3
microns, PART(4)»less than 1 micron in size)
5) Total emissions of sulfate (g/sec)
6) Total sulfate size spectrum in weight fraction of sulfate emissions
(size ranges are the same as above)
7) Emissions of SC>2 (g/sec)
TABLE 3
Point Source Emission File Data
1) XUTM coordinate (km)
2) YUTM coordinate (km)
3) RAPS stack ID
4) Stack parameters; a) stack height (m)
b) stack diameter (m)
c) stack velocity (m/sec) •
(if stack diameter is unknown, then flow
rate is given in units of m^/sec.)
d) stack temperature (°C)
5) Emissions of total mass (g/sec)
6) Total mass size spectrum: PART(l), PART(2),...., PART(7), represent
the weight fractions of total mass emissions in the size range
greater than 7, 3-7, 1-3, 0.5-1.0, 0.1-05, 0.05-0.1, and 0.01-0.05
microns, respectively.
7) Emissions of sulfate (g/sec)
8) Sulfate size spectrum: PARTS(l) through PARTS(7) represent the weight
fractions of sulfate emission rate in the same size range as above.
9) Emissions of S02 (g/sec)
-------
TABLE 4
Base Size Distributions of Sulfate
Size Class (ym) 0.01-0.05 0.05-0.1 0.1-0.5 0.5-1.0 1.0-3.0 3.0-7.0 >7
Summer Aerosol 0.13 0.36 0.24 0.18 0.08 0.01 0.0
Winter Aerosol 0.085 0.15 0.32 0.25 0.18 0.015 0.0
The base size distributions approximated as above were then used to
estimate the sulfate size distributions in the area and point sources by
relating the total particulate emissions (with associated size spectrum)
from these sources to sulfate emissions. This procedure, discussed in
detail by Brock (1982), clearly yields a gross approximation to source
sulfate size distributions, which should be improved as additional infor-
mation becomes available.
2.2 METEOROLOGY
The Regional Air Monitoring System (RAMS) used in the RAPS program con-
sisted of 25 remotely operated, automated stations controlled and polled
via telemetry by a central data acquisition system. The locations of the
RAMS stations are shown in Figure 1. These stations were installed in
approximate rings with average radii from the central urban station (101) of
5, 11, 20, and 44 km. The elevations of the stations averaged 154 m ± 23 m
above mean sea level. The instrumentation and measurements available at
each of the RAMS stations are shown in Table 5, reproduced here from
Schiermeier (1978).
-------
e
cd c
U O
60 t-(
O 4J
U (0
CU 4J
CD
T3
1-1 en
4-1 «
en ID
cu jr
3 T3
q c
>J o
a
• CD •
•U 0) O
CO fc. O
M r-t
C O
•H O OB
3
en
------- An error occurred while trying to OCR this image.
-------
Computer tapes containing hourly measurements of both rural and urban
mixing heights, stability classifications, wind speeds and directions, and
temperatures from the RAMS network were supplied by the EPA. The data were
compiled into data files suitable for input into PEM. The input files con-
tained urban mixing heights, wind speeds and directions, atmospheric stabi-
lity class and temperatures; the input winds were RAMS network resultant
winds (as suggested by the EPA). The stability classifications were
supplied in the format required by PEM (i.e., stability classes 1-7).
Examples of hourly meteorological data input files for a winter day
(January 22, 1976) and a summer day (July 22, 1976) are shown in Table 6.
Several of the selected evaluation days showed significant meteorologi-
cal changes over a time period of a few hours. For example, on Day 346
(December 12, 1975), a nearly 180 degree wind shift (from a northwest to a
southeast wind) occured over the 24 hour period. These windshifts,
occuring on nearly all winter evaluation days, affect the background con-
centrations which were added to the calculated fine particulate concentra-
tions to account for inflow across the model boundaries.
2.3 MEASURED CONCENTRATIONS
Data tapes containing the observed gas concentration values from the
RAMS network, corresponding to the twenty evaluation days, were supplied by
the EPA. Separate files containing the high volume and dichotomous sampler
data were also provided. The data files were scanned for hourly average
S02 concentrations, and 12-hour average concentrations of total mass and
total sulfur. The S02 concentrations, recorded in ppm, were multiplied by
2612.2 to convert to ug/m^. The observed gaseous total sulfur concentra-
10
-------
TABLE 6
Examples of Hourly Meteorological Data Input for PEM
Winter (Julian day 22, January 22, 1976)
Hour Wind Speed Direction Temperature Stability Mixing Depth
(m/s) (degrees) (°C) class (m)
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
3.21
3.09
3.24
2.90
2.80
2.78
2.26
1.91
1.22
1.72
2.29
1.38
0.97
1.31
2.15
2.40
2.56
2.90
3.49
3.85
3.97
3.78
3.55
3.45
277.4
276.6
263.5
261.2
258.9
256.2
266.6
273.7
286.1
10.7
50.0
83.1
121.6
177.5
192.3
169.4
146.0
122.0
121.7
121.7
128.2
142.1
151.4
162.0
-0.38
-0.51
-0.54
-0.74
-1.21
-1.60
-1.91
-2.09
-0.87
1.14
2.28
3.21
4.37
5.33
5.98
6.38
5.94
4.50
3.21
2.56
2.01
• 1.67
1.36
1.14
6
7
6
6
7
7
7
7
6
4
3
2
3
2
3
4
4 -
6
6
6
6
6
7
6
100
100
100
100
100
100
109
192
275
358
441
524
587
504
420
336
253
169
100
100
100
100
100
100
11
-------
TABLE 6 (Continued)
Examples of Hourly Meteorological Data Input for PEM
Summer (Julian day 204, July 22, 1976)
Hour Wind Speed Direction Temperature Stability Mixing Depth
(m/s) (degrees) (°C) class (m)
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
1.40
1.87
2.74
2.72
2.08
2.49
2.60
2.69
3.38
4.61
4.28
4.07
4.13
4.11
3.91
4.29
4.11
3.87
3.37
3.36
3.98
4.07
3.95
3.65
150.4
170.2
183.3
181.3
182.6
191.7
196.6
216.0
251.1
263.7
258.3
249.7
247.4
249.8
247.2
244.0
235.9
231.2
222.2
216.2
213.5
217.2
219.8
222.5
23.69
23.31
23.14
23.01
22.92
22.96
24.20
26.45
28.48
30.18
31.45
32.86
34.16
35.33
36.09
36.42
36.12
35.52
34.11
31.93
30.31
•29.29
28.37
27.41
6
5
5
5
4
4
3
3
3
3
3
3
3
4
4
4
3 -
4
4
6
6
6
5
6
100
100
100
100
121
220
319
418
517
616
715
1004
1370
1736
2103
2469
2666
2226
1786
1345
905
465
100
100
12
-------
tions were used to approximate S02 concentrations at the RAMS stations
where the latter were not measured (see Table 5). The high volume and
dichotomous sampler data contained total sulfur and total particulate mass
concentrations in ug/m^. The particulate data were further divided into
fine and coarse categories based on a cut-off size of 2.5 vtm. The total
sulfur measurements were multiplied by a factor of 3, which is ratio of the
molecular weight of 804 to the molecular weight of sulfur, to obtain the
equivalent total sulfate concentrations.
Table 5 clearly shows that concentration measurements were not made at
all of the 25 RAMS stations. The observed S02 concentrations are 1-hour
average values. The total sulfur and total mass concentrations measured by
eight out of the ten reporting RAMS stations were 12-hour average values;
only stations 103 and 105 recorded 6-hour averages. To facilitate com-
parison with the model calculations, the observed concentrations of S02,
fine and coarse sulfates, and fine and coarse total mass were converted
into 12-hour averages. This procedure gave two (12-hour average) observed
concentrations per day for each of the five pollutants. Examples of these
observed concentration values are shown in Table 7 for a typical winter day
(December 12, 1975) and a typical summer day (July 30, 1976).
To remove outliers from the concentration measurements, a mean and
standard deviation were computed, and data points greater than three stan-
dard deviations from the mean were omitted from the measured con-
centrations ; this procedure removed less than 1% of the measured
concentrations. A new mean and standard deviation were computed after
removal of outliers.
13
-------
TABLE 7
Examples of 12-hour Average Concentrations (yg/m3)
Observed at RAMS Stations
Winter (Julian day 346, December 12, 1975)
Station SO'
Fine Coarse Fine Coarse
sulfate sulfate total total
mass mass
103
103
105
105
106
106
108*
112
112
115
115
118
118
120*
122
122
124
124
62.56
15.57
113.05
57.36
•
*
38.96
45.36
43.95
49.37
•
61.14
10.24
52.79
39.50
6.53
38.12
6.53
2.373
13.278
14.265
17.349
13.389
17.304
10.959
12.885
17.220
11.304
12.228
13.530
13.356
14.481
10.551
9.912
10.359 •
9.939
1.926
1.935
1.704
2.580
1.608
2.814
1.362
1.389
2.154
1.056
0.747
1.557
0.810
1.452
1.395
0.597
1.248
0.891
41.45
52.35
52.80
62.80
40.10
56.50
32.60
36.80
60.10
35.00
37.30
44.80
46.40
46.20
28.60
29.90
33.00
31.70
10.90
17.85
26.55
25.75
17.10
15.80
15.10
12.50
16.90
10.50
6.10
24.20
9.80
10.70
7.60
2.60
13.30
6.10
14
-------
TABLE 7 (Continued)
Examples of 12-hour Average Concentrations
Observed at RAMS Stations
Summer (Julian day 212, July 30, 1976)
Station
103
103
105
105
106*
108*
112
112
115
115
118
118
120
120
122
122
124
124
+
so2
38.996
8.605
12.134
8.381
9
14.408
8.318
6.711
9.648
6.790
18.436
6.539
6.855
17.654
.
.
6.531
6.544
Fine
sulf ate
16.884
12.159
13.968
10.089
10.728
11.022
17.148
10.605
13.731
14.469
12.372
9.495
19.014
10.965
19.596
9.180
7.092 '
7.371
Coarse
sulf ate
1.218
1.371
0.555
1.452
1.053
2.997
1.074
0.222
0.303
2.538
1.020
1.857
0.615
1.128
0.612
0.690
0.618
Fine
total
mass
45.40
32.25
25.10
22.45
25.00
28.05
40.20
27.40
23.05
27.30
27.20
20.30-
43.40
28.50
47.80
22.10
16.50
18.30
Coarse
total
mass
35.15
27.75
11.55
7.05
30.60
21.80
29.10
21.30
.
.
14.90
14.40
13.80
12.20
17.10
25.20
4.80
10.00
+ S02 observations were available at more stations than shown in this
Table.
* Only one set of 12-hour average observed concentrations were available
at these stations on this day due to missing or incomplete data.
15
-------
SECTION 3
MODEL EVALUATION
The details of the PEM computer runs, input parameters, and statistical
procedures of the model evaluation are discussed in this section.
3.1 PEM RONS
PEM concentration predictions were evaluated against the measured con-
centrations for five pollutants:
1) S02
2) Fine particulate sulfate
3) Coarse particulate sulfate
4) Fine particulate total mass
5) Coarse particulate total mass
These five quantities were calculated in three model runs utilizing dif-
ferent sets of input data. These runs can be summarized as follows:
Pollutant Pollutant
Run species-1 species-2 Note
I S02 Fine sulfate Chemical transformation
from S02 to fine sulfate
II Fine non-sulfate Coarse non-sulfate No chemical transformation
mass mass
III Coarse sulfate No chemical transformation
or decay
As shown above, it was assumed that S02 chemically transforms into fine
sulfate at a constant rate and there is no contribution to coarse sulfate
16
-------
concentrations from this transformation. Predictions from the three runs
were combined to obtain concentrations of S02, fine and coarse particulate
sulfates, and fine and coarse particulate total mass.
3.1.1 Receptor Grid
The storage capabilities of PEM were fully utilized by using a 50 x 50
receptor grid with a fine grid cell size of 2.5 x 2.5 km. The southwest
corner of the grid was set at XOTM = 681.25 and YUTM = 4231.25 km, and the
modeling domain covers 125 x 125 km to encompass the majority of the point
sources. These included the Union Electric Meramac Generating Station and
the National Lead smelting operation which were large contributors to the
anthropogenic emissions in the St. Louis area.
3.1.2 Emissions
The emissions for the three PEM runs were obtained as follows. We
defined
Ql = S02 emission rate
Q2 » Fine sulfate emission rate (size < 3 pm)
Q3 = Coarse sulfate emission rate (size > 3 pm) •
Q4 =• Fine total mass emission rate (size < 3 pm)
Q5 =• Coarse total mass emission rate (size > 3 ym)
These five emission rates were readily obtained from the area and point
source emission data files shown in Tables 2 and 3. Then, the emissions
for PEM Run II were calculated as
Q-6 = Q4 ~ Q2 = fine non-sulfate mass emission rate
Q7 "* Q5 ~ Q3 = coarse non-sulfate mass emission rate
17
-------
All hourly emission data were summed and averaged to obtain a 12-hour
average emission data file appropriate for input to the PEM runs. To
reduce run costs, the S02 inventory was scanned and point sources with
emissions less than 1 g/s were eliminated in PEM Run I.
The area source emission inventories consisted of 144 area sources,
each a 5 km grid square. Since the maximum number of area sources in PEM is
restricted to 50, the 5 km emission grid squares were merged to give 36
area sources, each a 10 km grid square. This enabled inclusion of all area
source emission data and satisfied the requirement that the area source
grid size be an integer multiple of the calculation (or receptor) grid size
(2.5 km for this evaluation).
3.1.3 Deposition Parameters
Rao (1982) discussed the specification of the deposition and gravita-
tional settling velocities (V^ and W, respectively) in PEM. These parame-
ters were varied depending on the pollutant in each model run, as follows:
Pollutant V
-------
3.1.4 Chemical Transformation Rate
The oxidation rates of SC>2 reported (in percent per hour) for the urban
St. Louis region are as follows:
Range Average Study
5.3 - 32 16 Breeding e£ ad. (1976)
5 5 White et_ aJU (1976)
10 - 14 12 Alkezweeny and Powell (1977)
8-11.5 9.8 Alkezweeny (1978)
0-4 2 Forrest et_ al. (1979).
These studies were conducted both in summer and fall seasons, and the above
values represent general daytime averages.
A chemical transformation rate of S02 to fine particulate sulfate of 5%
per hour was used in PEM Run I. This was a conservative estimate based on
studies within St. Louis region quoted above. This value was held constant
throughout the model runs regardless of the meteorological and other con-
ditions.
3.1.5 Model Calculations
Because PEM uses a fixed calculation and receptor grid system, an
array of receptors was needed to allow comparison with the RAMS network
stations. The grid system, shown in Figure 2, was designed such that PEM
receptors either matched or formed a grid around the actual RAMS network
stations. For point comparisons with the RAMS network stations, the four
19
-------
e
•* «ma
2
g
12S
120
i?a
141
Hi
•
In
l°«
lus
111
f
,122
11".
?02
L U L
IDS 1
113
OH
,03
lou
."-
12 j
ID!
Li'5
117
•
IS
123
M1.25 Iff 25 70125 711 23 721.25 731.23 74125 731.23 7«123 77125 711 23 7»l 25 M1.2S
XUTM (km)
Figure 2. Receptor grid used in PEM evaluation and its relation to the
RAMS stations.
20
-------
receptors in the grid squares around the RAMS station were summed and their
average assigned to the RAMS station location.
The number of point sources in this evaluation were 286 in winter and
275 in summer, thus nearly utilizing the maximum capacity of the model of
300 point sources. For point source calculations in this evaluation, a
modification was made to the PEM program such that concentrations were
calculated only for the receptors surrounding each RAMS station, and not at
the rest of the receptors. This required calculation of only 84 out of a
total of 2500 receptors. Use of this calculation scheme resulted in a reduc-
tion by a factor of ten in run costs with no loss of capabilities. Default
option values in PEM were used for the input parameters for the stack-tip
downwash (option in effect), and atmospheric potential temperature gra-
dients (0.02 and 0.035 °C/m for E and F stability classes, respectively).
The inversion penetration factor (see Rao and Stevens, 1982) was specified
as 1.
The area source calculations did not include the modification to the
program discussed above. For each of the 36 area sources used in this eva-
luation, the contributions to the concentrations in the five affected grid
squares immediately downwind of the source were calculated, as discussed by
Rao and Stevens (1982).
The concentrations calculated by PEM Runs I, II, and III were combined
to obtain the concentrations of the five pollutants. The calculated fine
sulfate and fine total mass concentrations, resulting only from the contri-
butions of the point and area sources to the receptors, were added to their
respective background concentrations. The lowest observed fine sulfate and
fine total mass concentrations for the 12-hour averaging period were used
21
-------
as the background concentrations. If the RAMS stations reporting these
lowest concentrations were not located upwind in the receptor grid for the
12-hour averaging period, then the second lowest concentrations were used.
The background concentrations of SC>2, coarse sulfate, and coarse par-
ticulate mass were assumed to be zero, since an analysis of the RAPS data
by Dr. Jack Shreffler (1983, Personal communication) of the EPA showed
that there was no significant regional inflow of these species across the
model boundaries.
3.2 EVALUATION STATISTICS
The model performance was evaluated by using several statistical
measures. The statistical approach to model evaluation has been reviewed
at the recent American Meteorological Society (AMS) Workshop (Fox, 1981).
Two general measures of performance were used here: a) measures of dif-
ference which include the bias, variance, gross variability or root mean
squared error (RMSE), and average absolute gross error; b) measures of
correlation paired in space and time. The measured and predicted con-
centrations were analyzed and plotted with a standard SAS statistical and
data-handling package (Ray, 1982), Release 82.3.
In the discussion that follows, Oi refers to observed concentrations
(i = 1,2, , N), and P± refers to the corresponding concentrations cal-
culated by the model at the same location for the same time period; N is the
total number of observations. Standard means are computed as
N
0 - TT I 0, (la)
l N
3- I P!
1=1
22
-------
(a) Measures of Difference
Residuals are based'on the difference between observed and calculated
concentrations such that
Di - <>± ~ pi (2)
A negative residual indicates model overprediction and vice versa.
The bias 15 of the concentration difference is defined as
- - - 1 ?
D-0-P-irlD1 (3)
The average absolute gross error is defined as
N
W-^ I l°il <4>
The estimated variance of the concentration difference is calculated
from
Sd- 3 FT
where S
-------
Pearson's correlation coefficient, R, was computed as
I (oi - o)- I (PI - P)
I (o± - o)2- I (P± - P)
where all sums were calculated over i= 1,2, — ,N.
Scatter diagrams of the differences D^ versus the observed concentra-
tions 0^ were also plotted to show the model performance. Additional plots
were generated to examine the model behavior as a function of evaluation
days and RAMS station locations.
24
-------
SECTION 4
RESULTS AND DISCUSSION
The Pollution Episodic Model was evaluated for the five pollutant spe-
cies: S02, fine and coarse sulfates, and fine and coarse total mass. The
evaluation results comparing the model's concentration estimates to the
measured air quality data are presented and discussed in this section.
4.1 Sulfur Dioxide
Figure 3 shows a comparison of the calculated and observed 12-hour
average S02 concentrations for the twenty PEM evaluation days. This scat-
terplot is a composite of case-by-case comparisons for all RAMS stations.
A linear regression line, computed by the method of least squares, is also
shown in this figure. The statistics for this plot are given in Table 8.
The ratio of the means, P/0, is 1.24 and the ratio of the corresponding
standard deviations is 1.12. This agreement between the observed
and calculated means and standard deviations suggests reasonable ability of
PEM to predict SO- concentrations averaged over a large data base from dif-
ferent stations and seasons. The correlation coefficient, however, is only
0.23 over the compared range (6.5 - 250 pg/m3) of concentrations. This
suggests a large degree of randomness in the individual case-by-case com-
parisons of SO- concentrations. No attempt was made to improve the corre-
lation coefficients in this evaluation by removing the outliers, or by
considering a shorter range of concentrations for comparison.
25
-------
TABLE 8
PEM Evaluation Statistics for S02
Variable
°i
Pi
Di
|D±l
Mean
(ug/m3)
54.3
67.1
-12.8
48.5
Standard
deviation
(ug/m3)
50.0
56.1
66.1
46.6
RMSE - 67.3 yg/m3
N = 612
Linear Regression
Slope 0.255
Intercept 53.237
Pearson's R 0.227
26
-------
S02 CONCENTRATIONS (iJg/m3)
LEGEND: A =* 1 OBS, B = 2 OBS, ETC.
250-
200-
Q
W
H
-------
The differences D^ between observed and calculated 862 concentrations
are plotted in Figure 4 against the observed concentrations. There is a
clear bias for PEM to overpredict observed concentrations less than 75 ug/n
and underpredict observed concentrations greater than about 125 ug/m-\ The
bias "D over the entire evaluated range of SO concentrations is -12.8 pg/rn-^.
Thus, PEM is conservative with a tendency to slightly overpredict the
average SO concentrations. The average absolute gross error |D| is
48.5 ug/m^ which is less than the mean of observed concentrations (see
Table 8). Therefore, on the average, PEM predictions are within a factor
of two of the observed S02 concentrations.
Figure 5 shows a comparison of the calculated and observed daily mean
concentrations of SO- (averaged over all RAMS stations) for each of the
twenty PEM evaluation days. There is no discernible difference in the
model performance over the ten winter days and the ten summer days. The
calculated daily concentrations are generally within a factor of two of the
corresponding observed values, except for three days early in summer when
the model overpredicted by a factor of three or more.
Figure 6 shows a comparison of the calculated and observed daily mean
concentrations of S02 (averaged over all 20 PEM-evaluation days) at each of
the RAMS stations. The agreement is generally within about a factor of
two; PEM tends to overpredict at center-city receptors (e.g., at stations
101-103). Figures 5 and 6 together show the day-to-day and station-to-
station variation of the model's performance for daily mean S02 con-
centrations.
28
-------
S02
M
U
a*
250-
200-
150-
100-
50-
0-
-50-
-100-
-150-
-200-
-250-
LEGEND: A = 1 OBS, B = 2 OBS, ETC.
A AA
AA A
A A A
AAAABA A A
BB A BAG AAAA A
ABAA CAAABAAC B A A A AA
B ACBAAAA AAA ABAA AAAB A A
AA B
A AAA
ADACDBGDCCCBACDCA AB AABBA AB
ZSQLOLGDFEAECDB---A-BCAA-AABA A—A-A—
ZILHFBCGAFDABAABAB CACAA AA A
EDCFGI BDAABBABBAA AAB AA B C
GAEAB AABBAAAAABABABB B B AAA
A
A
A A
A
A
AAA
ABBBBAA BBB AABA
BA AAA BAD A AA
BDAAA BAAAAA AAB
A AAA
A A
A AA
A
A
50
100
150
200
250
OBSERVED CONCENTRATION (yg/m3)
Figure 4. S02 residuals (D^ = 0^ - P^) versus observed S02 concentra-
tions for the twenty PEM evaluation days.
29
-------
125-
0 OBSERVED
P CALCULATED
0
100-
y^N
**s
bp
>«/
§ 75-
H
K
H
W
CJ
? sn-
p
P 0
P 0
0 P
P
P
*
0 0
0
25-
P P
0
P
PP 0
P
P
0 P
0 0 00
00
0
PO
P
0-
315
365
50
100 150
*
JULIAN DAY
200
250
300
350
NOTE; OBS (0) of Day 357 coincides with CALC (P) of Day 356.
OBS (0) of Day 345 coincides with CALC (P) of Day 345.
Figure 5. Comparison of the observed and calculated daily mean concentrations
of S02 (averaged over all RAMS stations) for each of the
twenty PEM evaluation days.
30
-------
0 OBSERVED
P CALCULATED
125-
100-
/_^
en
e
N— x
z 75-
o
H
H
W
a 50-
8
evi
O
en
25-
0-
P P
P
0
P P 0
P 00
0 OP
0
P
P P 0
OOP
0 0 0 P P
0 0 P 0
0 P 0
P P P
0 P P 0
0 OP
P P
0 P 0
100
NOTE:
Figure 6.
105
110
115
120
125
RAMS STATION NUMBER
CALC (P) value of 143.2 ug/m3 at Station No. 104 is outside the
range of the plot and not shown.
OBS (0) coincides with CALC (P) at Station No. 120.
Comparison of the observed and calculated daily mean concentrations
of S02 (averaged over all PEM evaluation days) at each of the
RAMS stations.
31
-------
4.2 FINE AND COARSE SULFATES
The scatterplot of calculated versus observed 12-hour average fine
sulfate concentrations is shown in Figure 7. This is a composite plot for
all reporting RAMS stations for the twenty evaluation days. The linear
regression line is also shown in this figure. The statistics for this eva-
luation are given in Table 9.
The ratio of the means of calculated and observed values of fine
sulfate concentrations, P/0, is 1.1 and the ratio of the corresponding
standard deviations is 1.2. The correlation coefficient is 0.41 over the
compared range (1-30 yg/m3) of concentrations. The fine sulfate residuals
Di are plotted against the observed concentrations in Figure 8. The model
tends to overpredict 0^ < 18 yg/m^ and underpredict 0^ > 20 yg/m^. The
bias D over the entire range of concentrations is -1.0 pg/m3, i.e., the
model is slightly conservative. The average absolute gross error JD | is
4.8 yg/m3 which is much less than the mean of observed concentrations (see
Table 9). Therefore, averaged over the entire data base, PEM calculations
of fine sulfate concentrations are within a factor of two of the
corresponding observed values.
Figure 9 shows a comparison of the calculated and observed daily mean
concentrations of fine sulfate (averaged over all reporting RAMS stations)
for each of the twenty PEM evaluation days. The model tends to slightly
overpredict in the winter and underpredict in the summer. This may be due
to the seasonal variability of the chemical transformation rate (kt) which
was not considered in this evaluation (a constant rate, k^ - 5% per hour,
was used for all evaluation days regardless of the season). The agreement
is even better when comparing station-to-station variations of the
32
-------
TABLE 9
PEM Evaluation Statistics for Fine Sulfates
Standard
Mean deviation
Variable (ug/m3) (yig/m3)
Pi
12.6
13.6
-1.0
4.8
5.9
7.0
7.1
5.3
RMSE =7.2 yg/m3
N = 280
Linear Regression
Slope 0.479
Intercept 7.551 ug/m3
Pearson's R 0.405
33
-------
o
W
H
U
-CAB AB A
A A A^rlSCBABA A ABAB B A
A A ^^AB ADBBAB AB A
A A A A A A A
AAA ABA ABA AAA
A A A BAB A
A
AA
AA
A
AA
A A
A AA A
A A
A A
A ACABDA AAB BA
ABA BAA. A
A A A CA A AAB
C B BAA A
AAAA A AAAA A
A
A
A
B
10
20
OBSERVED
30
Figure 7. Comparison of calculated and observed fine sulfate concentrations
for the twenty PEM evaluation days. The solid line shows the linear
regression fit.
34
-------
FINE SULFATE
LEGEND: A - 1 OBS, B - 2 OBS, ETC.
60
o
M
W
Ed
30-
20-
10-
0-
-10-
-20-
-30-
0
AAA CB
A
ABCB
A
CC
A
B A B A
AA AAB AABAB
ABBACAA A CB AAB
-B-A-A-F-CBABCBAABBDA-BDE-GBCCBAD-AA—
BAA
BA
BC CBA BC
AAAA A B
A A A A A A
A A B
A
BC AAB AA A
AA DBA BDAA
A A AA
A AA A
A AA A
A. AAB
B A A
B
BAA A
A A
A
-AAAA'
A
A A
10 20
OBSERVED CONCENTRATION (ug/m3)
30
Figure 8. Fine sulfate residuals (D^ = 0-^ - P^) versus observed fine sulfate
concentrations for the twenty PEM evaluation days.
35
-------
25-
20-
b
z
o
Ed
Z
CO
Ed
Z
10-
5-
0-
P
0
0
0
P
0
P
0 P
0 OBSERVED
P CALCULATED
0
0
P
P
00
0 P P
pp P
000
P
P
0
315 365 50 100 150 200
JULIAN DAY
250
0
P
OP
300
350
NOTE; Both OBS (0) and CALC (P) of Day 357 coincide with the corresponding
values of Day 356.
Figure 9. Comparison of the observed and calculated mean daily centrations
of fine sulfate (averaged over all reporting RAMS stations) for
each of the twenty PEM evaluation days.
36
-------
calculated and observed daily mean fine sulfate concentrations (averaged
over all PEM evaluation days) in Figure 10. These results show that PEM is
capable of predicting, to within a factor of two, the average con-
centrations of fine sulfate resulting from direct emission and chemical
transformation of S02 over an urban area.
The model evaluation results for coarse sulfate (particle size > 3 urn)
concentrations are shown in Figures 11 to 14 and Table 10. These con-
centrations, resulting only from the direct emissions from sources, are
small (generally less than 3 yg/m^). The ratio of the means of calculated
and observed values of concentrations, TfO, is 0.52 and the ratio of the
corresponding standard deviations is 0.9. The correlation coefficient is
0.38 over the compared range of concentrations. The model slightly
underpredicts the concentrations with a bias TJ * 0.5 yg/m3 and average
absolute gross error of 0.66 ug/m^. The later is 59% of the mean of
observed concentrations (see Table 10). Thus, on the average, the calcu-
lated coarse sulfate concentrations are within about a factor of two of the
corresponding observed values. The model performed somewhat better in
winter than in summer (see Figure 13), though it generally underpredicted
the daily mean concentrations (averaged over all reporting RAMS stations)
for most days by about 50% or less. The model also tracks the station-to-
station variations of daily mean cbncentrations (averaged over all PEM eva-
luation days) fairly well (see Figure 14).
4.3 FINE AND COARSE TOTAL MASS
The model evaluation results for fine total mass are shown in Figures
15 to 18 and Table 11. The results clearly show that PEM overpredicts fine
37
-------
25-
0 OBSERVED
P CALCULATED
ro
e
60
20-
o
H-l
H
CJ
2
S 10-
_J
WJ
td
z
5-
P
0
P
P 0
0
P
0
P
0
0
P
0-
100
NOTE:
Figure 10,
105
110 115
RAMS STATION NUMBER
120
125
OBS (0) values coincide with the corresponding CALC (P) values
at Station Nos. 108, 115, and 118.
Comparison of the observed and calculated mean daily con-
centrations of fine sulfate (averaged over all PEM evaluation
days) at each of the reporting RAMS stations.
38
-------
TABLE 10
PEM Evaluation Statistics for Coarse Sulfates
Variable
01
Pi
Dl
|D±I
Mean
(Pg/m3)
1.12
0.58
0.54
0.66
Standard
deviation
(pg/m3)
0.58
0.51
0.61
0.49
RMSE =0.81 yg/m3
N - 261
Linear Regression
Slope 0.334
Intercept 0.200 pg/m3
Pearson's R 0.378
39
-------
a
w
1
3H
1-
0-
COARSE SULFATE CONCENTRATIONS (ug/m3)
LEGEND: A = 1 OBS, B = 2 OBS, ETC.
AA
A
A
AA
A A
A
AA
A
A
AAC
B
A A A
AAAA AA A A A AA A
A AACA A A A
AB AAABA A A A BA.
BAA BBABAB A^—iT"AA A
AA
A
A A
A A AAC AA ^Ar-arAAAA BAAAA A AA
BA AC AA
ABAAACAAAAAAA A
B AA B AD ACA AABAA A B A
AC BBD B AADAA A BA AA DA A
B AAADA CCBCBAABA A A B A
A A
AA
1
OBSERVED
Figure 11. Comparison of calculated and observed coarse sulfate concentrations
for the twenty PEM evaluation days. The solid line shows the linear
regression fit.
40
-------
n
W
OS
3-
2-
1-
0-
-1-
-2-
-3-
COARSE SULFATE
LEGEND: A - 1 OBS, B - 2 OBS, ETC.
A
AA
A A
AA A
A BA E AA
BB BCA BAADBA ABAAA
CBGBDCACACCAAA AABBAB B A
CECE EBBAB A A AABBBA A AAA
DBBDAAACABBBBACCA B B A A
-AAAA AA-ABBA-ABAC AA-AAB
A C C AB AA A A
AA A AA AA A A
A A AAA
AAA
B AAA
B
AA
A
A B
OBSERVED CONCENTRATION (yg/m3)
Figure 12. Coarse sulfate residuals (D^ =» 0^ - P^) versus observed coarse
sulfate concentrations for the twenty PEM evaluation days.
41
-------
3-
m
e
5P
vx
Z
0
H
**C O «
H
Z
w
o
8
w
H
CO
W 1
en 1
a!
-*!
o
o
0-
0 OBSERVED
P CALCULATED
0 0
0
000
0 0
0 0
PO 0 0
00 POO
P P . P
P P
P P
P PPPP
P P P
•
1 1 1 1
315 365 50 100 150 200 250
JULIAN DAY
NOTE; OBS (0) coincides with CALC (P) on Day 345.
00
P
P
P
300 350
Figure 13. Comparison of the observed and calculated daily mean concentra
of coarse sulfate (averaged over all reporting RAMS stations) for
each of the twenty PEM evaluation days.
42
-------
a
bo
Z
o
i
H
Z
W
Z
8
w
52
CO
w
CO
Bi
<:
o
0
0 OBSERVED
P CALCULATED
3-
2-
0
1-
0 0
P P
0
0
P
0-
100
Figure 14.
105
110 115
RAMS STATION NUMBER
120
125
Comparison of the observed and calculated mean daily concentrations
of coarse sulfate (averaged over all PEM evaluation days) at each of
the reporting RAMS stations.
43
-------
total mass concentrations. The observed concentrations are less than 80
pg/m^, but the corresponding calculated values range up to 300 vig/nH. The
larger calculated concentrations are generally associated with weak dif-
fusion conditions characterized by strong stabilities, low wind speeds, and
shallow mixing depths that were typical of several of the winter evaluation
days. The ratio of the means, "PYO~, is 3.1 and the ratio of the corresponding
standard deviations is 6.0. The model significantly overpredicts the con-
centration with a bias T) = -70.8 ug/m3 and average absolute gross error of
72.1 yg/m,3 which is 2.1 times the mean of observed concentrations (see
Table 11). The correlation coefficient of 0.45, however, is relatively
high indicating less randomness in the comparison of individual cases.
The day-to-day comparisons of observed and calculated daily mean fine
mass concentrations (averaged over all reporting RAMS stations) shown in
Figure 17 suggest that PEM performs relatively better in summer than in
winter. Figure 18 shows no significant trend in station-to-station
variations of the observed daily mean concentrations (averaged over all PEM
evaluation days); the model, however, overpredicts concentrations at sta-
tions within the city by a factor of 3 or less, while accurately modeling
the two outlying stations (122 and 124).
The model evaluation results for coarse total mass (particle size > 3
Mm) concentrations are shown in Figures 19 to 22 and Table 12. These
results are qualitatively similar to those obtained for fine total mass
evaluation discussed above.
The overprediction of both fine and coarse total mass concentrations by
the model is rather puzzling. One would expect prediction biases of oppo-
site signs for these two concentrations, since the emission rate of the
44
-------
TABLE 11
PEM Evaluation Statistics for Fine Total Mass
Standard
Mean deviation
Variable (pg/m3) (ug/m3)
34.4
105.2
-70.8
72.1
13.1
79.0
73.9
72.6
RMSE - 102.3 ug/m3
N = 281
Linear Regression
Slope 2.740
Intercept • 10.990 ug/m3
Pearson's R 0.445
45
-------
a
W
a
300-
250-
200-
150-
100-
50-
0-
FINE TOTAL MASS CONCENTRATIONS (jig/m3)
LEGEND: A - 1 OBS, B - 2 OBS, ETC.
A A
A AA
AA AA
AB A BBA
A AB
C CA B
A AAABABA
A DAA A/ A
AD A
AABBABB AAA
BB/ AA D
A AJJEBAC A
C
A BBGHCBA A
(DCGGAFBA
'BHAFHCD A A
AEDDHD A
BD BA
50
100
150
200
250
300
OBSERVED
Figure 15. Comparison of calculated and observed fine total mass concentrations
for the twenty PEM evaluation days. The solid line shows the linear
regression fit.
46
-------
FINE TOTAL MASS
LEGEND: A - 1 OBS, B - 2 OBS, ETC.
en
lo
3.
|
a
en
3
300-
250-
200-
150-
100-
50-
-
-50-
-100-
-150-
-200-
-250-
-300-
AA A A A
AA CEECABCJFEDBBBAAABA A
A ABAA BCABDABCC AAB C AAAB A A
A AAACB AAAA BAA A A A A
AAAAABBA AAAA
B CAAA BAAA BB AA BA A
A ABA AA BB B A
AAA ABAAAA AA A A
A A A "A
A A B AA A A
AA BA A B A A
A A A
1
50
100
OBSERVED CONCENTRATION (Mg/m3)
Figure 16. Fine total mass residuals (D^ = 0^ - P^) versus observed
fine total mass concentrations for the twenty PEM
evaluation days.
47
-------
en
_S
M
2
O
M
i
W
O
z
en
CO
Ed
2
125-J
100-
75-
50-
25-
0-
315
0
0
00
0
0 0
0 OBSERVED
P CALCULATED
P P
0
0 0
P P 0
OP 0
0 00
0
365
50 100 150 200
JULIAN DAY
250
P
0
300
350
NOTE: CALC (P) values of 184.8, 172.9, 223.3, 185.7, and 173.3 for Days
356, 34, 160, 180, and 232, respectively, are outside the range of
the plot and not shown.
OBS (0) of Day 345 coincides with OBS (0) of Day 343.
Figure 17. Comparison of the observed and calculated daily mean concentrations
of fine total mass (averaged over all reporting RAMS stations) for
each of the twenty PEM evaluation days.
48
-------
0 OBSERVED
P CALCULATED
150-
CO
125-
z
o
H
2
W
CO
100-
75-
s
H
W
2
50-
25-
0 0
0
0-
100
NOTE:
105
110
115
120
125
RAMS STATION NUMBER
Figure 18,
CALC (P) values of 164.3 and 155.1 at Station Nos. 106 and 112,
respectively, are outside the range of the plot and not shown.
OBS (0) values coincide with the corresponding CALC (P) values at
Station Nos. 122 and 124.
Comparison of the observed and calculated mean daily concentrations
of fine total mass (averaged over all PEM evaluation days) at each
of the reporting RAMS stations.
49
-------
TABLE 12
PEM Evaluation Statistics for Coarse Total Mass
Variable
Oi
Pi
Di
|D±|
Mean
(Ug/m3)
24.1
85.9
-61.7
65.3
Standard
deviation
(ug/m3)
14.2
63.0
60.5
56.6
RMSE =86.3 ug/m3
N - 264
Linear Regression
Slope 1.270
Intercept • 55.190 ug/m3
Pearson's R 0.285
50
------- An error occurred while trying to OCR this image.
------- An error occurred while trying to OCR this image.
-------
0 OBSERVED
P CALCULATED
150-
bp
~ 125-
2
O
M
H
2 100-
Cd
2
8
£ 75-
3
J
H
g 50-
w
3
o
0 25-
0-
3
P
P
P P
P P
P
P P
P
P
0 PP P
00 0
P . 00
00 00
000 00
0
0
15 365 50 100 ' 150 200 2
P
P
PO
0
50 300 350
JULIAN DAY
NOTE; CALC (P) values of 182.2 and 158.1 for Days 188 and 232, respec-
tively, are outside the range of the plot and not shown.
OBS (0) of Day 343 coincides with CALC (P) of day 345.
Figure 21. Comparison of the observed and calculated mean daily concentrations
of coarse total mass (averaged over all reporting RAMS stations) for
each of the twenty PEM evaluation days.
53
-------
150-
5P 125-
0 OBSERVED
P PREDICTED
w
o
CJ
to
to
100-
75-
50-
w
OT
25-
0-
100
Figure 22,
0
0
0
0
105
0
110 115
RAMS STATION NUMBER
120
0 0
P P
125
Comparison of the observed and calculated mean daily concentrations
of coarse total mass (averaged over all PEM evaluation days) at each
of the reporting RAMS stations.
54
-------
total mass is fixed for a given inventory. On the other hand, the model
performs better in predicting S02, fine and coarse sulfates. The fine and
coarse non-sulfate total mass emission data used in PEM Run II were checked
for possible program input errors and none were found. Since the urban
concentrations are strongly dependent on the emissions, we suspect that the
total mass emissions used in this evaluation must have been in error, i.e.,
overestimated significantly. Possible sources of errors in emissions and
other input parameters used in this evaluation are listed and discussed in
the next section.
55
-------
SECTION 5
CONCLUSIONS
This report described an evaluation of the Pollution Episodic Model
using twenty days of the St. Louis Regional Air Pollution Study data. This
evaluation was designed to test the model performance by comparing its con-
centration estimates for five pollutants to the measured air quality data,
using appropriate statistical measures of performance.
The emphasis in this evaluation was on S02 - fine sulfate runs with
chemical transformation and deposition, and comparison of the calculated
results with the data. For the twenty evaluation days, PEM predicted
average concentrations of S02, and fine and coarse sulfates to within a
factor of two, which is the best that may be expected considering the
natural variability in input meteorology and emission data (Hanna, 1981).
The model overpredicted the average concentrations of fine and coarse
total mass by a factor of three to four over the evaluation period. The
significant differences between the calculated and observed total mass con-
centrations may be attributed to a number of reasons:
1. Hourly point and area source emission inventories were available for
only one winter day and one summer day. These inventories were
further averaged over two 12-hour periods per day for use as input to
PEM. Analysis of the emission inventories indicated a core of steady
emission sources with various other area and point sources coming on or
56
-------
off line throughout the modeling period. Running PEM on an hour-by-
hour basis might account for this variability of emissions but the
modeling costs would be prohibitive. Despite this variability, both
fine and coarse sulfates are predicted to within a factor of two for
the total means as well as across the 12-hour averaging period.
However, the variability in emissions appears to be very important for
fine and coarse total mass, since these emission rates are signifi-
cantly larger, and dominated by ground-level sources, as shown below.
2. Table 13 shows the 12-hour average total emission rates of the five
pollutant species from area and point sources over the morning period
of the winter day.
TABLE 13
Average Total Emission Rates from Area and Point Sources
(January 22, 1976, 00 - 12 hours)
Pollutant Total Emission Rates (g/s)
Area sources Point sources
S02
Fine sulfate
Coarse Sulfate
Fine total mass
Coarse total mass
This table clearly shows that (a) point sources dominate the emissions
of S02, fine and coarse sulfates, while area sources dominate the
emissions of fine and coarse total mass, and (b) the sulfate com-
57
417.0
7.2
0.4
2353.0
6271.0
32556.0
67.8
162.6
200.2
913.0
-------
ponents of fine and coarse total mass emissions from area sources are
negligible compared to the non-sulfate components. The non-sulfate
total mass consists of fugitive dust, highway, residential, commercial,
industrial, and other particulate emissions of different sizes, which
are difficult to estimate accurately. No information is available on
the variability of these emissions. Any errors involved in the estima-
tion and location of these sources would significantly affect the
calculated concentrations due to the relatively large emissions from
area sources.
3. Because of the 12-hour averaging for periods 00-12 and 13-24 hours,
little can be said about the diurnal variation of model performance in
this evaluation. Table 14 shows the mean residuals (between the
observed and calculated 12-hour average concentrations) of the five
pollutants for the two averaging periods over the twenty evaluation
days. Obvious cases of large overprediction of fine and coarse total
mass may be attributed primarily to incorrect emission rates and
locations for area sources. There are also significant differences
between the first and second averaging periods in the mean residuals of
fine and coarse total mass. This may be associated with the diurnal
variability of area source emissions of these species and errors in
stability classification. The' first 12-hour averaging period is
generally characterized by stable conditions with weak diffusion con-
ditions. Hence the calculated concentrations and residuals are larger
for this period.
4. Constant deposition and settling velocities, and transformation rates
were used throughout the 12-hour averaging period. This ignores the
dependence of these variables on meteorological conditions such as wind,
58
-------
TABLE 14
Mean Concentration Residuals by 12-Hour Averaging Period
(Hours 00-12)
TF -17 -T
Pollutant _N (ug/m3)
S02 303 -10.80
fine sulfate 136 -1.47
coarse sulfate 127 0.54
fine mass 132 -91.77
coarse mass 125 -75.13
(Hours 13-24)
S02
fine sulfate
coarse sulfate
fine mass
coarse mass
309
144
134
149
139
-14.75
-0.56
0.54
-52.25
-49.70
59
-------
humidity, and thermal stratification. Also using one constant set of
values for deposition and settling velocities to describe the broad
particle size spectrum > 3 urn may not accurately represent the behavior
of particles of different sizes.
5. The wind speed and direction input to PEM were the RAMS network
resultant values. These are, obviously, approximations to real con-
ditions. Errors in wind direction may cause the model to affect par-
ticular receptors which may be completely ignored in reality. An
underestimation of the actual wind speed leads to overprediction of the
calculated concentrations.
Since one of the primary objectives of this study was to evaluate the
performance of the Pollution Episodic Model with emphasis on the 863 and
sulfate results, it is reasonable to conclude that PEM was able to simulate
the St. Louis RAPS data for the twenty evaluation days to within a factor
of two. Additional effort should be directed toward an examination of the
model response with respect to emission variability, stability classifica-
tion, and area source emissions and location. Experience has shown the
area sources to be the primary determinant in modeling urban ground level
concentrations of non-sulfate particulate matter.
60
-------
REFERENCES
Alkezweeny, A. J., and D. C. Powell, 1977: Estimation of transformation
rate of S02 to SO 4 from atmospheric concentration data. Atmos. Environ.
11, 179-182.
Alkezweeny, A. J., 1978: Measurement of aerosol particles and trace
gases in METROMEX. J. Applied Meteor. 17, 609-614.
Breeding, R. J., H. B. KLonis, J. P. Lodge, J. B. Pate, D. C. Sheesley,
T. R. Englert, and D. R. Sears, 1976: Measurements of atmospheric
pollutants in the St. Louis area. Atmos. Environ, 10, 181-194.
Brock, J. R., 1982: Personal communication to J. Shreffler, 8 pp.
Forrest, J., S. E. Schwartz, and L. Newman, 1979: Conversion of sulfur
dioxide to sulfate during the Da Vinci flights. Atmos. Environ. 13,
157-167.
Fox, D. G., 1981: Judging air quality model performance-A summary of the
AMS Workshop on dispersion model performance. Bulletin of the AMS 62,
599-609.
Hanna, S. R., 1981: Natural variability of observed hourly S02 and CO
concentrations in St. Louis. Atmos. Environ. 16, 1425-1440.
Hicks, B. B., 1983: Dry deposition of air pollutants in an urban environ-
ment. Air Specialty Conference on Air Quality Modeling of the
Nonhomogeneous, Nonstationary Urban Boundary Layer, Oct. 1983,
Baltimore, MD. ATDL Contribution File No. 83/15.
Rao, K. S., 1982: Plume concentration algorithms with deposition,
sedimentation, and chemical transformation. EPA-600/3-84-042, U. S.
Environmental Protection Agency, Research Triangle Park, NC; NOAA Tech.
Memo. ERL ARL-124, 87 pp. ATDL Contribution File No. 82/27.
Rao, K. S., and M. M. Stevens, 1982: Pollution Episodic Model User's Guide.
EPA-600/8-84-008, U.S. Environmental Protection Agency, Research
Triangle Park, NC; NOAA Tech. Memo., ERL ARL-125-. 186 pp. ATDL
Contribution File No. 82/28.
Ray, A. A., 1982: SAS User's Guide; Statistics, 1982 Ed. SAS
Institute, Inc., Cory, NC, 584 pp.
Schiermeier, F. A., 1978: Air Monitoring Milestones: RAPS' field
measurements are in. Environ. Sci. and Tech. 12, 644-651.
Tanner, R. L., W. H. Marlow, and L. Newman, 1979: Chemical composition
correlations of size-fractionated sulfate in New York City aerosol.
Environ. Sci. and Tech. 13, 75-78.
Texas Air Control Board, 1979: User's Guide: Texas Episodic Model
Permits Section, Austin, TX, 215 pp.
White, W. H., J. A. Anderson, D. L. Blumenthal, R. B. Hanson, N. V. Gillani,
J. D. Husar, and W. E. Wilson, 1976: Formation and transport of
secondary air pollutants: Ozone and aerosols in the St. Louis urban
plume. Science 194, 187-189.
61
-------
TECHNICAL REPORT DATA
{Please read Instructions on the reverse before completing)
1. REPORT NO.
2.
3. RECIPIENT'S ACCESSIOI*NO.
4. TITLE AND SUBTITLE
EVALUATION OF THE POLLUTION EPISODIC MODEL USING THE
RAPS DATA
5. REPORT DATE
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
William R. Pendergrass and K. Shankar Rao
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME ANO ADDRESS
Atmospheric Turbulence and Diffusion Division
National Oceanic and Atmospheric Administration
Oak Ridge, Tennessee 37830
10. PROGRAM ELEMENT NO.
CDTA1D/03-1606 (FY-84)
11. CONTRACT/GRANT NO.
IAG-AD-13-F-1-707-0
12. SPONSORING AGENCY NAME AND ADDRESS , ___ „-
Environmental Sciences Research Laboratory-RTP, NC
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final 10/82 - 2/84 .
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
This report describes an evaluation of the Pollution Episodic Model (PEM) using
the St. Louis Regional Air Pollution Study (RAPS) data. This evaluation is designed
to test the performance of the model by comparing its concentration estimates to the
measured air quality data, using appropriate statistical measures. Twenty days,
ten summer and ten winter, are selected from the RAPS data base for the PEM evalua-
tion. The model's performance is judged by comparing the calculated 12-hour average
concentrations with the corresponding observed values for five pollutant species,
namely, S07, fine and coarse sulfates, and fine and coarse total mass. A first-
order chemical transformation of S0? to fine sulfate is considered in the calcula-
tions in addition to the direct emission and dry deposition of all five pollutants.
The model domain, covering 125 x 125 km with a 50 x 50 receptor grid, includes 286
point sources and 36 area sources in the greater St. Louis urban area. Hourly
meteorological data and detailed emission inventories .for the five pollutants are
used as inputs to the model.
For the twenty PEM evaluation days, PEM predicted average concentrations of S02,
and fine and coarse sulfates 'to within a factor of two. The model overpredicted
the average concentrations of fine and coarse total mass by a factor of three to
four over the evaluation period. Th.is is attributed primarily to overestimation of
emission rates and incorrect location of area sources, which dominate the fine and
17.
coarse total mass eunssionsT
KEY WORDS ANO DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
18. DISTRIBUTION STATEMENT
19. SECURITY CLASS \JThis Report)
ECURITY CLASS \JTh
UNCLASSIFIED
21. NO. OF PAGES
RELEASE TO PUBLIC
20. SECURITY CLASS (Thispage)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
-------
------- |