vvEPA
United Slates
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park. NC 27711
EPA-454/R-92-015
October 1992
Air
SENSITIVITY ANALYSIS OF
A REVISED AREA SOURCE
ALGORITHM FOR THE
INDUSTRIAL SOURCE COMPLEX
SHORT TERM MODEL
and
-------
EPA-454/R-92-015
o
SENSITIVITY ANALYSIS OF
A REVISED AREA SOURCE
ALGORITHM FOR THE
INDUSTRIAL SOURCE COMPLEX
SHORT TERM MODEL
U.S. Environmental r ' ciion Agency
Region 5, Library (F:.-'.'-.;)
77 West Jackson BoLievard 12th
Chicago, IL 60604-3590
Office Of Air Quality Planning And Standards
Office Of Air And Radiation
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
October 1992
-------
report has been reviewed by the Office Of Air Quality Planning And Standards, *I. S.
'-Jvironmental Protection Agency, and has been approved for publication. Any mention 01 ir.de
names or commercial products is not intended to constitute endorsement or recommendation for use.
EPA-454/R-92-015
-------
PREFACE
The ability to accurately estimate pollutant concentration
due to atmospheric releases from r.rea sources is important to the
modeling community, and is of special concern for Superfund where
emissions are typically characterized as area sources.
Limitations of the Industrial Source Complex (ISC2) model (dated
92273) algorithms for modeling impacts from area sources,
especially for receptors located within and nearby the area, have
been documented in earlier studies. An improved algorithm for
modeling dispersion from area sources has been developed based on
a numerical integration of the point source concentration
function. Information on this algorithm is provided in three
interrelated reports.
In the first report (EPA-454/R-92-014) , an evaluation of the
algorithm is presented using wind tunnel data collected in the
Fluid Modeling Facility of the U.S. Environmental Protection
Agency. In the second report (EPA-454/R-92-015), a sensitivity
analysis is presented of the algorithm as implemented in the
short-term version of ISC2. In the third report
(EPA-454/R-92-016), a sensitivity analysis is presented of the
algorithm as implemented in the long-term version of ISC2.
The Environmental Protection Agency must conduct a formal
and public review before the Agency can recommend for routine use
this new algorithm in regulatory analyses. These reports are
being released to establish a basis for reviews of the
capabilities of this methodology and of the consequences
resulting from use of this methodology in routine dispersion
modeling of air pollutant impacts. These reports are one part of
a larger set of information on the ISC2 models that must be
considered before any formal changes can be adopted.
111
-------
ACKNOWLEDGEMENTS
This report was prepared by Pacific Environmental Ser-.ces,
Inc., under EPA Contract No. 68D00124. with Jawad S. Touma as the
Work Assignment Manager.
IV
-------
CONTENTS
PREFACE iii
ACKNOWLEDGEMENTS iv
1. PURPOSE 1
2. DESCRIPTION OF THE STUDY 1
3. RESULTS OF THE STUDY 4
3.1. Ground Level Sources With Downwind Receptors 4
3.2. Elevated Area Source 24
3.3. Ground-level Sources With Receptors Within and
Nearby the Area 29
4. LIMITED COMPARISON WITH FDM RESULTS 70
5. REFERENCES 73
v
-------
1. PURPOSE
The purpose of this study is to evaluate the sensitivity of
design concentrations across a range of source characteristics
for the new area source algorithm that has been incorporated into
the ISC2 Short Term (ISCST2) model (EPA, 1992) . Based on the
results of an evaluation of area source algorithms performed for
EPA by TRC Environmental Consultants (EPA, 1989), the finite line
segment algorithm used in the original ISCST model gives
physically unrealistic results for receptors located near the
edges and corners of the area. The new ISCST2 algorithm, which
implements an improved numerical integration approach to the
integrated line source algorithm used by the PAL model (Petersen
and Rumsey, 1987)., is compared to the finite line segment
algorithm used by the original ISCST model. Because the new
algorithm performs a numerical integration over the source area,
it is capable of explicitly handling receptor locations within
the area, whereas the finite line segment algorithm is limited to
determining impacts at receptors only located outside the area.
The integrated line source algorithm, as implemented in the
original PAL model, was also examined in the TRC report, and was
found to give physically reasonable results for all of the tests
performed. The conclusions of the TRC report apply as well to
the new area source algorithm implemented in the ISCST2 model
since it has been shown during development and testing to give
essentially the same results as the original PAL model (Erode,
1992) .
2. DESCRIPTION OF THE STUDY
To examine the sensitivity of the design concentrations
across a range of source characteristics, five ground-level area
sources were modeled, with sizes varying from 10 meters to 1,000
meters in width. An elevated source scenario consisting of a
100-meter wide area with a release height of 10 meters was also
modeled. An additional case involving a 1,000 meter wide ground
level area was also modeled with receptors located within and
nearby the area. The high and high-second-high (HSH) 1-hour,
3-hour and 24-hour averages and high annual averages were
determined for each of these source scenarios using a full year
of real time meteorological data. All of the sources were
modeled as square areas oriented N-S and E-W, since the original
ISC algorithm was limited to handling that source geometry. Each
scenario was run for one year of National Weather Service (NWS)
meteorological data from Pittsburgh, PA (1964) ; one year of NWS
data from Oklahoma City, OK (1988); and one year of NWS data from
Seattle, WA (1983) .
Each scenario was also run with both the rural and urban
mode dispersion options. The only difference between the rural
mode and the urban mode that effects the area sources modeled in
this st'jidy are the lateral and vertical dispersion coefficients,
-------
sigma-y and sigma-z. The dispersion coefficients are somewhat
larger for the urban mode to account for the increased dispersive
capacity of the atmosphere in the urban environment. The
regulatory default option was used for all scenarios. This
includes a procedure for calculating averages for rjriods that
include calm hours. A pollutant type of "OTHER" was specified,
so that no decay was used for either the rural or the urban mode.
For the sake of efficiency, all computer runs involving the
original algorithm were performed using the ISCST2 model, rather
than the original ISCST model. In this way, the same input
runstream file was used for both algorithms. It should also be
noted that the results presented in this report for the original
finite line segment algorithm reflect a correction to the finite
line segment equation as implemented in the original ISCST model.
This correction reduces all estimates for the finite line segment
algorithm by about 11.4 percent (a factor of 0.886 = SQRT(PI)/2.)
relative to the original uncorrected version.
A polar receptor network consisting of ground level
receptors at five distances and 36 directions (every 10 degrees)
was used to determine design concentrations. Since most area
sources are ground-level or low-level releases, the maximum
impacts can be expected to occur very near the source. However,
the finite line segment algorithm does not allow receptors within
the area itself, and is known to provide unreasonable
concentration estimates very close to the source. The guidance
in the ISC2 User's Guide states that if the source-receptor
distance is less than the width of the area, then the area should
be subdivided and modeled as multiple sources. Therefore, the
first distance ring in the polar network was placed at a downwind
distance (measured from the center of the area) of 1.5*XINIT
meters, where XINIT is the width of the area. This places the
nearest receptors at a distance of about one source width from
the edge of the area. Additional distance rings were placed at
approximately 2.0, 3.0, 5.0 and 10.0 times the initial distance,
for a total of 180 receptors. For the ground level sources, the
maximum ground level concentrations are expected to occur at or
near the downwind edge of the area, and to decrease beyond that
distance. Therefore the maximum concentrations for these
source-receptor geometries are expected to occur at the l.5*XINIT
distance. The concentrations at the larger receptor distances
were also examined for a few cases in order to compare the
algorithms downwind of the maximum concentration.
Additional receptor distances were used for the elevated
source to account for the fact that the maximum impact may occur
beyond the nearest distance ring. The SCREEN model was run for a
100 meter wide area source with a release height of 10 meters for
each stability class using both rural and urban dispersion
coefficients. Maximum impacts for the rural coefficients
occurred at downwind distances (measured from the downwind edge)
ranging fnm about 60 meters for A stability to 480 meters for F
stability, \;:.th a peak concentration at 116 meters for C
-------
stability. Maximum impacts for the urban coefficients occurred
at downwind distances ranging from 36 meters for A stability to
117 meters for E stability (SCREEN does not perform calculations
for F stability in the urban mode), with a peak concentration at
44 meters downwind of the edge for C stability. Additional
receptor rings were included at distances of 2.0*XINIT,
2.5*XINIT, and 4.0*XINIT for the elevated release height cases to
better represent the peak concentration from the refined model.
In order to assess the sensitivity of the design values for
receptors located close to and within an area source, an
additional scenario was modeled involving a 1,000 meter wide
(extra large) ground-level area source with receptors located
within the area and near the edge of the area. For the original
finite line segment algorithm, this source was subdivided into 4,
16, 64 and 100 separate areas of equal size. This was necessary
because the finite line segment algorithm cannot model impacts at
receptor locations within the area being modeled.
An emission rate equivalent to 1.0 g/s for the entire area
was used for all scenarios. The area source widths, heights of
release, emission rates in g/(sm2), and receptor distances are
shown in Table 1 for each scenario. Table 2 provides the source
inputs for the X-Large (XL), Close-in case for the 4-, 16-, 64-,
and 100-source treatment used with the finite line segment
algorithm. Figure 1 shows the location of the receptors used for
the X-Large source with receptors located within and nearby the
area.
Table 1. Area source Scenarios for Sensitivity Analysis
Source Type
X-Small, Ground- level
Small, Ground- level
Medium, Ground- level
Large, Ground- level
X-Large, Ground- level
Medium, Elevated
X-Large, Close-in,
Ground- I eve I
Width of
Area (m)
10.0
50.0
100.0
500.0
1000.0
100.0
1000.0
Height of
Release (m)
0.0
0.0
0.0
0.0
0.0
10.0
0.0
Emission
Rate
Cg/(sm2))
1.0E-2
4.0E-4
1.0E-4
4.0E-6
1.0E-6
1.0E-4
1.0E-6
Receptor Distances (m)
(measured from the center
of the area)
15, 30, 50, 75, 150
75, 150, 250, 400, 750
150, 300, 500, 750, 1500
750, 1500, 2500, 4000, 7500
1500, 3000, 5000, 7500, 15000
150, 200, 250, 300, 400, 500,
750, 1500
250, 500, 750, 1000, 1500
-------
Table 2. Area Source Inputs for X-Large, Close-in Scenario
(used for the original finite line segment algorithm only)
Scenario Description
XL, Close-in, 4- sources (2x2)
XL, Close-in, 16-sources (4x4)
XL, Close-in, 64 -sources (8x8)
XL, Close-in, 100-sources (10x10)
Wid'h of
Eacf ; -ib-
Area '-.0
500.0
250.0
125.0
100.0
Height of
Release
(m)
0.0
0.0
0.0
0.0
Emission
Rate
-------
1500.00
example Plot Showinc _ocation of Receptors
-1500.00 -100000 -500.00 000 500.00 100000 150000
1000.00
500.00
0 00
-50000
-1000.00
-1500.00
i 500.00
1000.00
500.00
0.00
-500.00
-1000 00
-1500.00 -1000.00
-500.00
0.00
500,00
-1500.00
1000.00
150000
Figure 1. Example Contour Plot Showing Location of Receptors
(Asterisks) Relative to the 1000 Meter Wide Ground
Level Source for the X-Large Close-in Case
-------
Table 3A
Comparison of Design Concentrations
Very Small Source (10m Width)
for the
- Rural
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24 -hr High
24-hr HSH
Annual
Numerical
Integration
(New)
204857.40000
169798.00000
124283.10000
118637.60000
45466.07000
31626.08000
4274.40000
238843.70000
210465.40000
125987.90000
94231.48000
40460.68000
31288.66000
7998.58900
205086.10000
200610.70000
101556.20000
83307.44000
29787.68000
26249.53000
6305.98900
Finite Line
Segment
(Old)
115807.46520
114605.51760
76954.48688
69151.30768
33218.78914
24526.59754
3336.20465
208284.60120
115809.68020
76546.80284
70575.28688
28647.34810
22365.44862
6122.28038
115440.48400
115108.32260
68040.57378
57482.91804
21684.13234
20813.33610
4814.46818
Ratio
(New/Old)
1.76895
1.48159
1.61502
1.71562
1.36869
1.28946
1.28122
1.14672
1.81734
1.64589
1.33519
1.41237
1.39897
1.30647
1.77655
1.74280
1.49258
1.44926
1.37371
1.26119
1.30980
-------
Table 3B
Comparison of Design Concentrations (jzg/m3) for the
Very Small Source (10m Width) - Urban
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Numerical
Integration
(New)
80215.95000
66346.86000
48728.32000
46834.78000
19892.41000
14349.51000
1997.23900
81917.46000
81917.46000
49956.39000
39824.69000
17721.98000
14468.18000
3756.77800
80292.38000
78857.11000
41170.05000
36746.55000
13349.30000
11331.30000
2855.42500
Finite Line
Segment
(Old)
45519.26004
45478.96476
30147.93972
27583.64308
13936.16512
10413.21116
1467.77152
45525.68354
45478.96476
31022.20022
29865.37660
12418.28232
10112.44960
2802.65190
45476.18272
45359.85978
28419.30942
23697.29268
9560.20580
8472.34222
2124.37549
Ratio
(New/Old)
1.76224
1.45885
1.61631
1.69792
1.42739
1.37801
1.36073
1.79937
1.80122
1.61034
1.33347
1.42709
1.43073
1.34044
1.76559
1.73848
1.44866
1.55066
1.39634
1.33745
1.34412
-------
Table 4A
Comparison of Design Concentrations (/ig/m- ) for the
Small Source (50m Width) - Rural
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Numerical
Integration
(New)
11092.09000
9191.13000
6724.00100
6413.65800
2420.26400
1652.91800
220.33050
47791.59000
11390.75000
16237.42000
5086.67400
2492.74500
1635.14000
412.03130
23028.41000
10855.31000
7676.13800
4574.56900
1536.58300
1325.26800
316.63400
Finite Line
Segment
(Old)
6560.99480
6168.35947
4147.70179
3721.89817
1771.20083
1277.19381
169.14086
41656.92024
6238.70521
14119.49092
3822.66295
2175.89373
1172.49785
314.56136
17203.88728
6195.51182
5734.62791
3153.53271
1123.35231
1047.97143
241.19002
Ratio
(New/Old)
1.69061
1.49004
1.62114
1.72322
1.36645
1.29418
1.30265
1.14727
1.82582
1.15000
1.33066
1.14562
1.39458
1.30986
1.33856
1.75212
1.33856
1.45062
1.36785
1.26460
1.31280
-------
Table 4B
Comparison of Design Concentrations (/ig/tn3) for the
Small Source (50m Width) - Urban
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Seattle 1983
l-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Numerical
Integration
(New)
3343.59000
2770.22400
2030.04800
1951.02400
827.10080
593.45340
82.16215
3413.99300
3413.99300
2080.32700
1658.01400
736.45720
598.99680
154.78950
3344.33000
3284.12400
1715.47200
1531.42300
548.15180
467.42150
116.43890
Finite Line
Segment
(Old)
1919.71392
1912.22279
1271.85743
1163.47837
585.66788
433.78206
60.43786
1920.01073
1918.13064
1308.67605
1260.04616
521.36608
422.03928
116.41287
1917.92331
1913.23637
1199.00785
999.57545
395.33116
352.87032
87.09868
Ratio
(New/Old)
1.74171
1.44869
1.59613
1.67689
1.41224
1.36809
1.35945
1.77811
1.77985
1.58964
1.31584
1.41255
1.41929
1.32966
1.74372
1.71653
1.43074
1.53207
1.38656
1.32463
1.33686
-------
Table 5 A
Comparison of Design Concentrations
Medium Source (100m Width) -
for the
Rural
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Numerical
Integration
(New)
4189.18800
2617.63200
1914.47100
1825.38300
684.94450
464.41880
61.57570
Finite Line
Segment
(Old)
3280.49740
1752.36890
1395.96654
1178.79022
502.12854
358.64412
47.03294
Ratio
(New/Old)
1.27700
1.49377
1.37143
1.54852
1.36408
1.29493
1.30920
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
23913.31000
3243.66700
8055.44500
1447.85100
1132.80900
459.60370
113.15580
20828.46012
1772.68931
7006.82734
1088.95336
990.78899
331.32874
87.97547
1.14811
1.82980
1.14966
1.32958
1.14334
1.38715
1.28622
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
11590.70000
3090.86700
3863.56600
1301.83300
540.42210
411.38410
87.45323
8601.94258
2298.72700
2867.31396
900.30447
411.73129
329.97918
66.63176
1.34745
1.34460
1.34745
1.44599
1.31256
1.24670
1.31249
10
-------
Table 5B
Comparison of Design Concentrations (j/g/m3) for the
Medium Source (100m Width) - Urban
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Numerical
ntegration
(New)
876.43690
727.42470
531.82200
511.08630
216.17090
154.20440
21.22755
894.73940
894.73940
544.71500
433.97790
192.38550
155.82930
40.05390
875.88820
860.01120
449.45410
401.30460
141.42290
121.11490
29.62306
Finite Line
Segment
(Old)
509.18588
507.27877
337.44054
308.65821
154.80414
113.56970
15.62494
509.26040
508.77886
347.26876
334.37011
137.62654
110.70676
30.38232
508.70665
507.51108
318.22462
265.16793
102.67934
92.36754
22.44348
Ratio
(New/Old)
1.72125
1.43397
1.57605
1.65583
1.39642
1.35780
1.35857
1.75694
1.75860
1.56857
1.29790
1.39788
1.40759
1.31833
1.72179
1.69457
1.41238
1.51340
1.37733
1.31123
1.31990
11
-------
Table 6A
Comparison of Design Concentrations
Large Source (500m Width) - Rural
for the
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Numerical
Integration
(New)
845.68710
339.78180
292.36470
113.26060
51.30783
25.66165
3.37200
4787.69200
184.33330
1600.98900
82.02331
206.67020
25.46798
6.31944
2352.58300
447.87980
784.19450
158.06740
107.69580
30.89734
4.67383
Finite Line
Segment
(Old)
656.09948
288.96367
235.58102
105.08270
39.73355
20.97978
2.61144
4165.69202
103.10914
1393.02679
63.83916
180.55466
18.97512
4.92970
1720.38873
440.01303
573.46279
146.67128
79.55786
24.21862
3.63307
Ratio
(New/Old)
1.28896
1.17586
1.24104
1.07782
1.29130
1.22316
1.29124
1.14931
1.78775
1.14929
1.28484
1.14464
1.34218
1.28191
1.36747
1.01788
1.36747
1.07770
1.35368
1.27577
1.28647
12
-------
Table 6B
Comparison of Design Concentrations
Large Source (500m Width) - Urban
for the
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
umerical
tegration
(New)
46.24926
38.69476
27.97519
26.82369
11.19910
7.76747
1.03987
47.18597
47.18597
28.60593
22.65347
9.96523
7.90206
1.97723
99.66319
59.07021
36.47066
23.86000
12.35849
6.40791
1.40481
Finite Line
Segment
(Old)
28
27
18
17
8
5
0
28
28
20
18
8
5
1
86
53
36
20
10
5
1
.04318
.93455
.58977
.02197
.42143
.90910
.79235
.04283
.00716
.49336
.46705
.46159
.80201
.55787
.12153
.74087
.53461
.50081
.86709
.65684
.09953
Ratio
(New/Old)
1.64922
1.38519
1.50487
1.57583
1.32983
1.31449
1.31239
1.68264
1.68478
1.39586
1.22670
1.17770
1.36195
1.26919
1.15724
1.09917
0.99825
1.16386
1.13724
1.13277
1.27764
13
-------
Table 7A
Comparison of Design Concentrations (/xg/m3) for the
Very Large Sc ^rce (1000m Width) - Rural
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Numerical
Integration
(New)
424.38890
170.89860
144.60010
56.96619
22.94715
9.44751
1.04146
2394.39400
59.10365
800.29010
27.27462
102.04460
7.88571
1.95303
1183.58100
224.21050
394.52700
77.40852
54.09274
11.31509
1.43274
Finite Line
Segment
(Old)
328.04974
144.48206
114.75047
50.98467
17.84420
8.56149
0.84477
2082.84601
51.25244
696.35383
24.09787
89.10697
6.07173
1.57807
860.19419
214.98834
286.73140
71.66283
39.77379
10.96250
1.16327
Ratio
(New/Old)
1.29367
1.18284
1.26013
1.11732
1.28597
1.10349
1.23284
1.14958
1.15319
1.14926
1.13183
1.14519
1.29876
1.23760
1.37595
1.04290
1.37595
1.08018
1.36001
1.03216
1.23164
14
-------
Table 7B
Comparison of Design Concentrations (/ig/m3) for the
Very Large Source (1000m Width) - Urban
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
umerical
tegration
(New)
14.40249
12.09639
8.68877
8.31132
3.43113
2.34182
0.31041
14.69298
14.69298
11.63643
8.39101
4.67000
2.39467
0.59116
50.83767
29.93746
18.61257
10.46676
5.38388
2.83220
0.41316
Finite Line
Segment
(Old)
8
8
5
5
2
1
0
11
10
10
7
4
1
0
43
26
18
8
4
2
0
.85869
.80733
.87223
.37184
.64040
.89132
.24362
.47803
.76066
.25001
.41392
.19282
.78293
.47608
.07947
.87695
.36330
.99064
.81261
.82427
.33395
Ratio
(New/Old)
1.62580
1.37345
1.47964
1.54720
1.29947
1.23819
1.27414
1.28010
1.36544
1.13526
1.13179
1.11381
1.34311
1.24172
1.18009
1.11387
1.01357
1.16418
1.11870
1.00281
1.23719
15
-------
Overall, the new integrated line source algorithm predicts
higher design concentrations than the original finite line
segment algorithm. The average ratio of the numerical
ir'-egration results over the finite line segment results
(a'<=raged over all three cities and for all averaging periods)
ranges from about 1.5 (i.e., 50 percent higher for the
integration method) for the 10 meter wide area to about 1.2 for
the 1000 meter wide area. This trend toward smaller ratios for
larger areas is illustrated in Figures 2 and 3, which show the
average ratios (averaged across the three meteorological data
locations) for the five ground-level sources for downwind
receptors only, for rural and urban dispersion, respectively.
Included in these figures are the average ratios for each of the
averaging periods. Note that only the high-second-high (HSH)
results are used for the short term averages presented in these
figures. The patterns are nearly identical for the 10-meter and
50-meter wide areas for both rural and urban dispersion, but the
pattern shifts as the size of the area increases. Figure 4 shows
the ratios by averaging period, averaged across all of the ground
level sources. As can be seen from these figures, the ratios
tend to be largest for the 1-hour averages, and then decrease
with longer averaging periods. The average ratios for the
24-hour HSH values and the high annual values are about the same.
The ratios are generally larger for the cases with urban
dispersion coefficients than for the cases with rural dispersion
coefficients.
The most notable feature about these results is that the
numerical integration method produces larger concentration
estimates than the original finite line segment algorithm. One
possible explanation for part of this difference is that the
finite line segment algorithm allows the vertical dispersion
coefficient, sigma-z, to grow from the upwind edge of the area.
This is done by adding a vertical virtual distance (XZ) equal to
the width of the area (XINIT) to the downwind distance when
calculating sigma-z. The downwind distance is measured from the
downwind edge of the area. In effect, for vertical dispersion,
the finite line segment is located at the upwind edge of the
area, whereas for lateral dispersion the finite line segment is
located at the downwind edge. Since the numerical integration
method integrates over the area, the vertical dispersion
coefficient for each element of the integration will be
representative of the actual distance from that element of the
area to the receptor location. Thus, for the portion of the area
that is closest to the receptor, and therefore having the
greatest impact on the receptor, the distance used for sigma-z
will essentially be the distance from the downwind edge of the
area to the receptor location. The result of this difference
will be a smaller overall "effective" vertical dispersion
coefficient for the numerical integration method than for the
finite line source algorithm. Since these are ground level
16
-------
releases and ground level receptors, a smaller effective vertical
dispersion coefficient would result in larger ground level
concentrations, other factors being equal.
To test this hypothesis, the 10 .neter wide ground level
source was modeled again for the Oklahoma City 1988 data with a
version of the finite line segment algorithm that used a vertical
virtual distance of one half the source width (XZ = XINIT/2). In
other words, the source-receptor distance for calculation of
sigma-z was measured from the center of the area. The ratios of
the numerical integration (new) algorithm to the finite line
segment (old) algorithm are presented below for the original
XZ=XINIT and the modified XZ=XINIT/2 versions for both rural and
urban dispersion coefficients. The ratios for the XZ=XINIT/2
case are much closer to 1.0 than the original XZ=XINIT case,
especially for the longer averaging periods. These results
provide an indication that a significant part of the
discrepancies between the two algorithms are related to the
treatment of the vertical dispersion coefficients, specifically
the use of a vertical virtual distance equal to the width of the
area for the finite line segment algorithm. In addition, since
the urban dispersion coefficients are larger than the rural
coefficients, this factor also explains in part why the ratios
are larger for the urban cases than for the rural cases.
Rural Dispersion Coefficients; 10-meter Ground Level Area,-
Oklahoma City, OK 1988 Data
Ratio New/Old Ratio New/Old
with XZ=XINIT with XZ=XINIT/2
1-Hr High 1.15 1.15
1-Hr HSH 1.81 1.41
3-Hr High 1.65 1.28
3-Hr HSH 1.34 1.04
24-Hr High 1.41 1.10
24-Hr HSH 1.40 1.09
Annual 1.31 1.01
17
-------
Urban Dispersion Coefficients,- 10-meter Ground Level Area;
Oklahoma City, OK 1988 Data
Ratio New/Old Ratio New/Old
with XZ=XINIT with XZ=XINIT/2
1-Hr High
1-Hr HSH
3-Hr High
3-Hr HSH
24 -Hr High
24-Hr HSH
Annual
1.80
1.80
1.61
1.33
1.43
1.43
1.34
1.33
1.33
1.19
0.99
1.06
1.06
0.99
Another notable feature of the results is that the ratios
show a larger variation from site to site and across averaging
periods for the cases with rural dispersion coefficients than for
the cases with urban dispersion coefficients. One of the major
factors in causing this variability for the rural cases is
thought to be the influence of limited mixing effects for very
low mixing heights. This is particularly noticeable for the
Oklahoma City cases, which show very large differences between
the high and HSH results for rural dispersion. The hourly
interpolation scheme used for urban mixing heights reduces the
likelihood of very low mixing heights for the urban cases.
In addition to examining the design values, which all
occurred at receptors located on the nearest distance ring for
the ground level sources, the results at distances located
further downwind were examined briefly to determine whether or
not the results converge with distance. Figures 5 and 6 present
the high concentration values versus distance downwind for the 10
meter wide ground level area source for the Oklahoma City data
for the case with rural dispersion coefficients. The HSH short
term values are presented in Figure 5, and the high annual
average values are presented in Figure 6. These figures show
that the two algorithms converge to nearly identical answers at a
distance of about 15 source widths for this example. The longer
period averages converge to within a few percent at a distance of
about 5 source widths downwind. This general pattern was also
apparent for other cases that were examined.
18
-------
(D
CD >*
o <;
(± (D
M H
n ju
(D IQ
W (D
H-
O
en
H-
ro o
h H
CO P.
So-
ro M
HI ro
hh JU
H-
n w
H- H-
(B N
3 (1)
rr
tn 'ii
c
o
h
o
c:
Area Source Sensitivity Analysis
Average Ratios by Area Size - Rural
5 1'8
6
CO
H
c5 1 p
DC '*
10
50 100 500
Width of Area (m)
1000
tr1
ro
<
n>
-- 1 -Hr HSH
3-Hr HSH
24-Hr HSH
ANNUAL
-------
to
o
M
CD
CO >
o <
C CD
H H
O Hi
(DU3
CO (D
(U
c! rt
H H-
cro
(U M
H-ro
to ^
c \
(P O
h M
(0 O.
H--"
Str
(D h
Mi (B
Hi Jl)
H-
O CO
H- H-
(D M
3 CD
c-r
cn rn
o
Q
^!
o
c
IT1
CD
CD
Area Source Sensitivity Analysis
Average Ratios by Area Size - Urban
10
50 100 500
Width of Area (m)
1000
1-HrHSH -*- 3-HrHSH
24-Hr HSH
ANNUAL
-------
p-
1
CD
4^
Q >
^ <
o ro
0 M
3 ju
q,uj
(D
(D !0
-------
CM
5: co
£ j=
* i
z o
HSH
Figure 5. High (HSH) Short Term Values Versus Distance for the
10 Meter Wide Ground Level Source for Rural Dispersion
ana Oklahoma City Data
22
-------
CO
CO to
eg
CD «
0 to
O >
O 1
CO D)
01
oooooooooo
ooooooooo
ooooooooo
CO
D
C
CO
D
C
Figure 6. High Annual Average Values Versus Distance for the 10
Meter Wide Ground Level Source for Rural Dispersion
and Oklahoma City Data
23
-------
3.2. Elevated Area Source
Tables 8A and 8B present comparisons of design va_-;er
obtained from the numerical integration algorithm and from the
finite line segment algorithm t^r the 100 meter wide elevated
source (10 meter release height). Part A of the table presents
the results using rural dispersion coefficients, and part B
presents the results using urban dispersion coefficients. The
ratios for the elevated source are smaller than the corresponding
ratios for the 100-meter ground level source (see Tables 5A and
5B). In fact, the ratios for the rural dispersion case for
longer averaging periods are actually less than 1.0, indicating
that the numerical integration algorithm estimates smaller
concentrations than the finite line segment algorithm. The
ratios follow a similar trend as the ground level sources with a
decrease for longer averaging periods. Urban ratios are larger
than rural ratios. This trend is shown in Figure 7, which
depicts the average ratios for each averaging period (averaged
across the three meteorological data locations).
One possible explanation for the lower ratios of design
values for the elevated source than for the ground level sources
is related to the differences in treatment of the vertical
dispersion parameter described in the previous section. The
ground level concentrations will tend to be smaller for the
numerical integration algorithm since it uses a smaller
"effective" vertical dispersion parameter than the finite line
segment algorithm, and since the receptors are located off the
plume centerline vertically. To test this hypothesis, the
modified finite line segment algorithm with a vertical virtual
distance of one half the source width (XZ=XINIT/2) was run on the
100 meter wide elevated source for the Oklahoma City 1988 data.
Once again, the ratios are much closer to 1.0, especially for
longer averaging periods, for the XZ=XINIT/2 case than for the
XZ=XINIT case. The ratios for the XZ=XINIT/2 cases are also very
similar to the corresponding ratios for the 10 meter ground level
source presented above. The results suggest that the use of
XZ=XINIT/2 for the finite line segment algorithm may better
represent an "effective" vertical dispersion coefficient that the
XZ=XINIT currently in use.
24
-------
Rural Dispersion Coefficients,-
100-meter Elevated Area (10m Release Height);
Oklahoma City, OK 1988 Data
Ratio New/Old Ratio New/Old
with XZ=XINIT with XZ=XINIT/2
1-Hr High
1-Hr HSH
3-Hr High
3-Hr HSH
24 -Hr High
24 -Hr HSH
Annual
1.02
1.12
1.01
0.99
0.92
0.93
0.84
1.06
1.22
1.06
1.03
1.13
1.04
1.02
Urban Dispersion Coefficients;
10-meter Elevated Area (10m Release Height);
Oklahoma City, OK 1988 Data
Ratio New/Old Ratio New/Old
with XZ=XINIT with XZ=XINIT/2
1-Hr High
1-Hr HSH
3-Hr High
3-Hr HSH
24-Hr High
24-Hr HSH
Annual
1.34
1.34
1.22
1.02
1.13
1.09
1.12
1.24
1.24
1.14
0.94
1.03
0.97
0.98
25
-------
Table 8A
Comparison of Design Concentrations (/zg/m3) for the
Medium Elevated Source (100m Width) - Rural
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Numerical
integration
(New)
462.32990
422.63290
234.75510
200.89060
91.53203
74.36944
11.16831
720.21940
507.60260
240.07310
200.39990
86.47166
67.74000
20.13812
502.80920
437.91930
265.34210
218.36560
101.61880
88.98233
21.29089
Finite Line
Segment
(Old)
393.69871
381.90658
214.93128
187.96933
107.91631
85.41618
14.37888
705.65284
451.47096
238.51368
202.80062
93.92442
72.81362
24.07448
416.25104
385.72169
253.70769
202.80062
93.92442
72.81362
26.61255
Ratio
(New/Old)
1.17432
1.10664
1.09223
1.06874
0.84818
0.87067
0.77672
1.02064
1.12433
1.00654
0.98816
0.92065
0.93032
0.83649
1.20795
1.13532
1.04586
1.07675
1.08192
1.22206
0.80003
26
-------
Table 8B
Comparison of Design Concentrations (/zg/m3) for the
"Medium Elevated Source (100m Width) - Urban
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Numerical
Integration
(New)
440.36300
387.56370
264.93450
253 .21430
116.70530
88.28399
13.34216
517.93870
517.93870
319.70040
257.91310
121.18880
96.71584
27.83596
502.52200
493.15130
262.68700
241.21780
102.85800
90.25307
23.19134
Finite Line
Segment
(Old)
333.21034
332.92531
221.74347
201.67743
103.46557
79.45618
11.61218
386.55693
386.55693
261.90585
253.84751
107.13822
88.77676
24.78361
386.02931
386.01780
239.52381
204.08399
87.11106
77.30210
19.54454
Ratio
(New/Old)
1.32158
1.16412
1.19478
1.25554
1.12796
1.11110
1.14898
1.33988
1.33988
1.22067
1.01602
1.13114
1.08943
1.12316
1.30177
1.27754
1.09671
1.18195
1.18077
1.16754
1.18659
27
-------
m
H-
2
l-l
fD
S. 3» >>
(D fD <
rt rt ro
fD (D H
H H ju
^ 51 fD
H-
fD (U
rt
W H-
M O
fD W
<
M fD
(0 H
(U [U
cn LQ
(D H-
n>
H- >tJ
U3 fD
^ fl
rr H-
O
O a
i-h
Mi
H O
O
O
Area Source Sensitivity Analysis
Average Ratios for Elevated Source
1 .0
1 A-
I .^f
"O 1 Q-
o 1'3
,L < 0 i
o i.^
CO
"5 1'1
(D -j
O no-
co u'y
n P-
.2
"rrt H 7-
VU II /
EC u''
0^?-
.O
n c_
>
i
-- ^
,
^ . . ....
7
- ,
t . ^
«
i
"^^^
\^
<
i
1 -Hr HSH 3-Hr HSH 24-Hr HSH
Averaging Period
ANNUAL
RURAL -*- URBAN
-------
3.3. Ground-level Sources With Receptors Within and Nearby the
Area
Tables 9A and 9B present comparisons of design values from
the numerical integration algorithm and from the finite line
segment algorithm for the 1000 meter wide ground level source
with receptors located within and nearby the area. Parts A and B
of the table present the results using rural and urban dispersion
coefficients, respectively. The results for the finite line
segment algorithm are presented for each of the subdivided
multiple-source scenarios examined using 4, 16, 64 and 100 areas
of equal size. The ratios for the cases with receptors within
and nearby the area are generally larger than the corresponding
ratios for the other ground level cases (see Tables 1 through 7).
In addition, the trend is for larger ratios for longer averaging
periods, which is the reverse of the trend seen for the other
ground level sources. This trend is shown in Figure 8, which
shows the average ratios (averaged over the three meteorological
data locations) for each averaging period. As with the other
sources examined, the ratios are larger for the case with urban
dispersion coefficients than for the case with rural dispersion
coefficients. The results in Tables 9A and 9B also show that, in
general, the design values for the old finite line segment
algorithm tend to increase as the number of subdivided areas
increases. Since the impact at any receptor located within the
area does not include any contribution from the subarea in which
the receptor is located, as the number of subareas increases and
the size of the subarea decreases, the amount of contribution not
accounted for will tend to decrease. In principal, as the number
of subareas approaches infinity and the individual subareas
approach point sources, the two algorithms should converge,
although no attempt has been made to verify this.
The reason for the ratios increasing with longer averaging
periods is also relatively simple. For the high 1-hour averages,
the amount of contribution from the subarea containing the
receptor location that is not accounted for will depend only on
the amount of the subarea upwind of the receptor for a single
wind direction. For the highest 1-hour average, it is likely
that the receptors are located near the upwind edge of the
subarea, where the lost contribution will be relatively small.
In fact, the high 1-hour values for the rural dispersion cases
(Table 9A) are quite similar for the two algorithms. For longer
period averages, the amount of contribution lost for the local
subarea will vary as the wind direction varies from hour to hour,
and the relative amount lost for the entire averaging period will
tend to be larger than for the high 1-hour values. This trend
should increase as the length of the averaging period increases
and the wind direction variation becomes larger. This trend is
clearly seen in Tables 9A and 9B.
29
-------
The receptor locations for the design values are also
included in Tables 9A and 9B for the numerical integration
algorithm and for the finite line segment case based on 1C 0
sources. The locations are given as direction (in degrees) and
distance .n meters). Thus, a location of ( 40,500) means a
receptor located along the 40 degree direction radial, measured
clockwise from North, at a distance of 500 meters from the center
of the area. The receptor locations show better agreement
between the algorithms for the longer averaging periods. A more
complete picture is provided in Figures 9 through 44, which
display contour plots of high concentrations across the receptor
grid for the numerical integration algorithm and for the finite
line segment algorithm based on 100 sources. Contour plots are
given for the HSH 1-hour, HSH 24-hour, and annual average
concentrations. The 3-hour average results were not included in
the contour plots since they are not expected to provide any
significantly different results. The rural results are presented
first, followed by the urban results, with the numerical
integration algorithm results and finite line segment
(100-source) results for the same location and averaging period
on facing pages to facilitate comparison. The four grid squares
at the center of the diagrams (between X = -500 to 500 and Y =
-500 to 500) define the location of the 1000 meter wide area
source. The source location and the distribution of receptor
points was shown in Figure 1 in Section 2.
Generally, the contour patterns between the two algorithms
compare better for the longer averaging periods than for the
1-hour averages. Some of the contour plots exhibit isolated
peaks and valleys, and some discontinuities (or "kinks") in the
contours. These may be due to the limited number of data points
(180) on which the plots are based, or may be an artifact of the
interpolation scheme used to generate a uniform grid of data by
the contouring program prior to determining the contours, or the
method used in contouring the data. Therefore, the fine-scale
details should not be given much credence in these plots,
although the overall patterns should be fairly reliable. The
numerical integration algorithm, which explicitly handles
receptors located within the area, generally shows reasonable
patterns across the area source itself, whereas the finite line
segment algorithm with the 100-source subdivided treatment of the
area shows some unusual patterns over the area. This is
particularly noticeable for 1-hour averages, such as in Figures
10, 16, 34 and 40. These unusual patterns for the finite line
segment algorithm are indicative of an inability of that
algorithm to adequately model the concentration distributions
within the area, even when the area is subdivided into 100 areas.
In a few cases the patterns are surprisingly similar, such as
Figures 21 and 22. But the overall conclusion evident from these
contour diagrams is that the numerical integration algorithm is
far superior to the finite line segment algorithm in handling
receptors within and nearby th^ area.
30
-------
Table 9A
Comparison of Design Concentrations (ug/m^) for the 1000m Wide Area
With Receptors Located Within and Nearby the Area - Rural
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Numerical
Integration
(New)
432.55590 ( 40,750)
207.18820 (280,500)
177.47750 ( 20,500)
134.43460 (360,500)
78.94704 (320,500)
74.63991 (290,500)
30.06087 ( 20,250)
2395.17100 (150,750)
210.16090 ( 60,500)
806.08260 (180,500)
186.00090 ( 50,250)
118.92130 (180,500)
57.16613 (290,250)
24.85228 (340,250)
1246.38900 (140,750)
553.65640 (210,250)
416.00020 (140,750)
201.34430 (190,500)
81.90249 (140,500)
71.92862 (150,250)
31.43343 (300,250)
Finite Line
Segment (Old)
4 Sources
491 .85377
204.15247
173.71970
75.71328
31.03172
18.91487
4.05184
2927.43792
83.93556
978.66585
41 .81989
126.18350
14.75862
4.37314
1286.65097
323.04225
431.85155
109.09121
60.53212
25.36595
4.62616
Finite Line
Segment (Old)
16 Sources
441 .24368
198.70738
159.83803
76.33872
31.18236
25.78640
8.33124
2474.81064
108.08660
828.29845
72.52082
110.04926
20.69101
7.19887
1303.16601
431.29435
434.38879
150.80349
60.87819
27.34034
8.59221
Finite Line
Segment (Old)
64 Sources
446.04837
194.18373
162.11036
75.61375
35.64745
32.86131
11.77271
2408.71504
118.18593
805.65300
94.45699
106.62629
26.84766
10.00716
1327.30242
474.10294
442.98414
165.82801
62.42380
34.48674
12.18340
Finite Line Ratio
Segment (Old) New/Old- 100
100 Sources
442.15387 ( 40,750)
191.51714 ( 50,750)
163.52689 ( 40,750)
76.89537 ( 50,750)
41.16997 (330,500)
38.21827 (320.500)
12.17878 ( 30,250)
2437.45954 ( 20,750)
134.81013 (300,500)
815.66666 ( 20,750)
92.22968 (100,250)
107.93943 (140,750)
27.59182 (310,500)
10.42655 (330,250)
1239.79929 (140,750)
516.40297 (210,250)
413.88205 (140,750)
182.86048 (210,250)
60.34736 (140,500)
33.90401 (210.250)
12.83955 (330,250)
0.97829
1.08183
1.08531
1.74828
1.91759
1.95299
2.46830
0.98265
1.55894
0.98825
2.01671
1.10174
2.07185
2.38356
1.00532
1.07214
1.00512
1.10108
1.35718
2.12154
2.44817
Note: Values in parentheses are receptor locations given as direction (degrees from North) and downwind distance (meters).
31
-------
Table 9B
Comparison of Design Concentrations (ug/m') for the 1000m Wide Area
With Receptors Located Within and Nearby the Area - Urban
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24 -hr HSH
Annual
Numerical
Integration
(New)
75.36378 ( 50,500)
74.17054 (280,500)
56.60395 (180,250)
47.63224 (320,500)
30.56433 (320,500)
29.19938 (290,500)
12.77044 ( 30,250)
75.31821 (300,500)
75.28640 (310,500)
69.43607 ( 20,500)
61.00494 (340,250)
23.90601 (330,500)
23.47807 (320,250)
10.75157 (340,250)
75.13278 ( 60,500)
75.09218 ( 60,500)
49.30389 ( 70,500)
48.29193 ( 70,500)
28.92939 (140,250)
27.20447 (290.250)
13.39305 (300,250)
Finite Line
Segment (Old)
4 Sources
22.87120
18.60340
13.61754
12.52957
6.51216
5.43449
1.23812
22.87172
22.68481
15.77947
12.86407
5.35411
4.44532
1 .39736
57.73794
32.12452
23.58462
16.09491
8.86869
5.13370
1 .42362
Finite Line
Segment (Old)
16 Sources
28.72459
24.24223
17.48827
15.82253
9.00127
7.70050
2.64040
28.47258
28.47258
19.90494
17.21320
6.27145
5.85526
2.33758
58.11319
31 .32098
24.22136
16.10262
10.03667
6.26770
2.73066
Finite Line
Segment (Old)
64 Sources
32.04270
30.78414
21.21428
18.05589
11.00509
10.10606
3.85130
32.14429
32.14429
26.64702
22.08193
8.12755
7.77340
3.34950
53.95553
31.46416
24.01670
19.32749
11.42896
8.39970
4.00072
Finite Line Ratio
Segment (Old) New/Old- 100
100 Sources
35.76426 (150,500)
32.90812 (270,500)
22.21074 (120,500)
21.24188 (300,500)
12.41048 (330,500)
11.65855 (320,500)
3.99965 ( 30,250)
36.81676 (300,500)
36.81676 (300,500)
27.92990 (360,500)
24.01311 ( 60,500)
8.86338 (310,500)
8.61436 (330,500)
3.52936 (330,250)
53.85180 (130.750)
36.41715 ( 60,500)
26.12551 (120,500)
20.61560 ( 90,500)
11.89885 (360,500)
8.94196 ( 60,250)
4.24711 (330,250)
2.10724
2.25387
2.54850
2.24237
2.46278
2.50455
3.19289
2.04576
2.04489
2.48608
2.54049
2.69716
2.72546
3.04632
1.39518
2.06200
1 .88719
2.34?49
2.43128
3.04234
3.i;345
Note: Values in parentheses are receptor locations given as direction (degrees from North) and downwind distance (meters).
32
-------
CD
fD O <
O O fD
fD O n
d $u
rt 3ua
O fD fD
H rt
tn fD 5*J
n fu
rt
S H-
H-0
QJ CO
fD
O
(D
.
o
tr1
n>
< cr
co?
O (D
c: i-«
H D)
n oj
fD H-
H-
rt ^
tf CD
K
n H-
H O
O d
to
fD Ml
I O
H-H
fD
Area Source Sensitivity Analysis
Average Ratios for Close-in Case
J.£-
t£> O Q
T3
-------
1500.00
HSH '-hour Averaaes, Pitts^urc", Rur^l, FAL
-15<~JOO -100000 -500 CO 000 50000 1000 00 150000
1000 00
500.00
000
-500 00
-1000 00
-150000
1500 00
1000.00
500.00
-150000 -100000 -50000
000
50000
1000 00
000
-500 00
-100000
-150000
1500.00
Figure 9. Contour Diagram of HSH 1-hour Average Rural
Concentrations (^ig/m3) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Pittsburgh 1964 Data.
34
-------
1500.00
HSH 1-hour Averages, Pittsburgh, Rural
-150000 -100000 -50000 000 50000 100000
' 500 00
1000.00
500.00
0.00
-500.00
-1000 00
-1500 00
1500.00
1000.00
500.00
0 00
-150000 -1000.00 -500.00
000
50000
-500.00
-1000.00
100000 1500.00
-150000
Figure 10. Contour Diagram of HSH 1-hour Average Rural
Concentrations (/^g/m^) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Pittsburgh 1964 Data.
35
-------
HSH 2--hour -^/eroces.
-100000 -50000
-1500 o:
1500 00
°:i:sb'jrcn, Rural, PAL
000 50000 100000
1000.00
50000
0 00
-500.00
-1000 00
-1500.00
150000
1500.00
1000.00
500.00
0.00
-500.00
-100000
-150000
-150000 -1000.00
500 00
3 00
50000
100000 1500.00
Figure 11. Contour Diagram of HSH 24-hour Average Rural
Concentrations (/ig/m3) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Pittsburgh 1964 Data.
36
-------
HSH 24-nour Averages, Pittsburgh. Rural
-150000 -100000 -50000 000 50000 1000 00 150000
500.00 i1II I I I I I I I IIII 1 11 I IIr/T-l'IIFT-1 1500.00
1000.00
500.00
1000.00
500.00
50000 100000 1500.00
-150000 -100000 -50000
Figure 12.
Contour Diagram of HSH 24-hour Average Rural
Concentrations (^g/m3) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Pittsburgh 1964 Data.
37
-------
-'5COOO -100000 -:DOCO 000
I 00
'jroi. FAL
50000 1 COO 00 1 50C 00
100000 r
50000
0 00
-50C 00
- 1000 00
-'SCO CC
-150000 -1000.00 -500.00
000
500.00
1500 00
1000.00
50000
0 00
-500.00
-1000 00
10COOO
1500.00
-150000
Figure 13.
Contour Diagram of Annual Average Rural
Concentrations (^ig/m3) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Pittsburgh 1964 Data.
38
-------
-nnual Averages, Pittsburgh, Rural
-isco.c:
1500.00
100000 -500.00
"IIIIT
000
1000.00
500.00
0 00
-500.00 -
-1000 00
1500.00 <
-1500
i i i I i i I i I I I I i i I i
1 500 00
Ti 1500 c:
1000.00
500.00
0.00
500.00
-10CO.CO
-1000.00 -500.00
0.00
500.00 1000.00 1500 00
-1500 00
Figure 14.
Contour Diagram of Annual Average Rural
Concentrations (^g/tn^) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Pittsburgh 1964 Data.
39
-------
HSH 1-hour Averages Oklahoma C;ty, Rural, PAL
ooo
-'500.00 -100000 -50000
'500.00
50000 100000 '50000
1500 DO
1000.00
500.00
0.00
-500.00
-100000
-1500 CO
i i i i i
1000.00
50000
0.00
*J -500.00
-1000.00
-1500.00 -1000.00 -50000
000
500.00 1000.00 150000
-150000
Figure 15.
Contour Diagram of HSH 1-hour Average Rural
Concentrations (^g/m^) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Oklahoma City 1988
Data.
40
-------
HSH 1-hour Averages, Oklahoma City, Rura!
-1500.00 -100000 -50000 000 50000 100000 1500.00
1500.00 i i i i 1 i ]^-r-4iL i ;r^ni i i _uai i i iiiir-i isoo.oo
1000.00
500.00
0.00
-500.00 -
-1000.00
-1500.00
1000.00
500.00
0.00
500.00
-1000.00
-1500.00 -1000.00 -500.00
0.00
500.I
-150000
1000.00 1500.00
Figure 16. Contour Diagram of HSH 1-hour Average Rural
Concentrations (/ig/m3) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Oklahoma City 1988
Data.
41
-------
HSH I-i-nr Avfoges. Oklahoma City, Rural, FAL
-150000 -1000.00 -500.00 0.00
50000 1000.00 1500.00
1500.00
1000.00
500.00
0.00
-50000
-100000
150000
1000 00
500.00
IIIlUIIII -1500 00
-150000 ' ! ' ' ' ' ' ' ' '
-150000 -1000.00 -500.00
0.00
500.00 1000.00 1500.00
Figure 17.
Contour Diagram of HSH 24-hour Average Rural
Concentrations (/ig/iti3) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Oklahoma City 1988
Data.
42
-------
1500.00
HSH 24-hour Averages, Oklahc-na City, Ruroi
-1500.00 -100000 -50000 0.00 50000 100000 1500.00
1000.00 -
500.00
0.00
-500.00
-100000
-1500.00
1500.00
- 1000.00
500.00
000
-500.00
-1000.00
-150000 -1000.00 -500.00 000 500.00 1000.00 150000
-150000
Figure 18.
Contour Diagram of HSH 24-hour Average Rural
Concentrations (^g/m3) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Oklahoma City 1988
Data.
43
-------
Annual Averages, Oklahoma City, Rural, PAL
-150000 -100000 -50000
1500.00
0 00
1000.00
500.00
0.00
-500.00
-1000.00
500 00 1000 00 15uO 00
1500 00
-1500.00
ill ill iiit
1000.00
500.00
000
-500.00
-1000 00
-1500.00 -1000.00 -500.00 0.00
500.00 1000.00 150000
-1500 00
Figure 19.
Contour Diagram of Annual Average Rural
Concentrations (^g/m^) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Oklahoma City 1988
Data.
44
-------
1500.00
Annual Averages, Oklahorr a City, Rural
-150000 -1000.00 -500.00 0.00 500.00 1000.00 1500.00
1000.00
500.00
0.00
-500 00
-100000
-1500 00
1500.00
1000.00
500.00
0.00
-500.00
-100000
-1500.00
-1500.00 -1000.00 -500.00 0.00 500.00 1000.00 1500.00
Figure 20. Contour Diagram of Annual Average Rural
Concentrations (jig/m3) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Oklahoma City 1988
Data.
45
-------
rSH 1-hour Averages, Seattle, Rural, PAL
-1500.0 -1000.00 -500.00 000
500.00 100000 150000
1500.00
1000.00
500.00
000
-500.00
-100000
-150000 L
1500.00
1000.00
500.00
000
-500.00
-1000.00
-150000
-150000 -1000.00 -500.00 0.00 500.00 1000.00 150000
Figure 21. Contour Diagram of HSH 1-hour Average Rural
Concentrations (/zg/m3) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Seattle 1983 Data.
46
-------
HSH 1-hour Averages, Seattle, Rural
1500.00
-150000 -1000.00 -50000 000
50000 100000 150000
1000.00
500.00
0.00
-500 00
-1000.00
-1500.00
150000
1000.00
500.00
000
-500.00
-100000
-1500.00 -1000.00 -500.00
0.00
50000
100000 150000
-1500.00
Figure 22. Contour Diagram of HSH 1-hour Average Rural
Concentrations (pig/m^) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Seattle 1983 Data.
47
-------
HSH 24-hour Ave :-jes, Seattle, Rural, PAL
-1500.00 -100000 -500.00
1500.00
0.00
500.00 100000 '50000
1500.00
1000.00
500.00
000
-50000
-1000.00
-150000
1000.00
500.00
000
-500 00
-1000.00
-1500.00 -1000.00 -500.00
0.00
50000 1000.00 "500.00
-150000
Figure 23. Contour Diagram of HSH 24-hour Average Rural
Concentrations (/xg/m3) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Seattle 1983 Data.
48
-------
HSH 24-hour Averages, Seattle, Rural
-150000 -100000 -50000 0.00
1500.00 I | | | | I | I| I I I I
50000 100000 1500.00
1500.00
1000.00
500.00
000
-500.00
-1000 00
-150000
1000.00
500.00
0.00
-500.00
-1000.00
-150000 -1000 CO -500.00 000
500.00 1000.00 150000
-150000
Figure 24. Contour Diagram of HSH 24-hour Average Rural
Concentrations (/xg/m3) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Seattle 1983 Data.
49
-------
1500.00
Annual Averages, Seattle, Rural.
150000 -100000 -50000 000 50000
1000.00
500.00
0.00
-500 00
-1000.00
1000.00 150000
150000
1000 00
500.00
0.00
-500.00
-100000
.150000
-1500.00 -1000.00
-1500CC
-500.00 000 50000 100000 150000
Figure 25.
Contour Diagram of Annual Average Rural
Concentrations (^g/m3) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Seattle 1983 Data.
50
-------
Annual Averages. Seattle. Rural
-150000 -100000 -50000 000 50000 1000.00 1500.00
'500.00 i i i ii:iii 7 \ \ i 1 I I I I I I k III1II I : I 1500.00
1000.00
500.00
0.00
-500.00
-1000 00
-1500 00
1000.00
500.00
0.00
-500.00
-1000.00
-150000 -1000.00 -500.00 0.00 500.00 100000 1500.00
-1500.00
Figure 26. Contour Diagram of Annual Average Rural
Concentrations (/zg/m3) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Seattle 1983 Data.
51
-------
HSH 1-hour Averages, Pittsourgh, Urban, PAL
-150000 -100000 -50000 000
1500.00
1000.00
500.00
000
-500.00
-100000
500.00 1000.00 1500.00
1500.00
-1500.00
-1500.00 -1000.00 -500.00
000
500.00
1000.00
500.00
0.00
-500 00
-1000.00
100000 1500.00
-1500.00
Figure 27.
Contour Diagram of HSH 1-hour Average Urban
Concentrations (^ig/m3) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Pittsburgh 1964 Data.
52
-------
HSH 1-hour Averages, Pittsburgh, Urban
-150000 -100000 -50000 000 50000 1000.00 1500.00
1500.00 iiiin=T=-ri:Ii < L i _ i . i . i _L-4-|||||| x. IIi 1500.00
1000.00
500.00
0.00
-500.00
-1000 00
-1500 00
-150000
1000.00
500.00
000
-500.00
-100000
500.00 1000.00 1500.00
-1500.00
Figure 28. Contour Diagram of HSH 1-hour Average Urban
Concentrations (^g/m3) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Pittsburgh 1964 Data.
53
-------
rSH 1- hour Ave,"^:es, Pittsouran. Urccn,
-1500 CD -100000 -5CC CO 000 50000
'500.00
1000.00
50000
COO
-500 00
-1000.00
-1 SCO CO
-500.00
-1COOCC
-1500 CO -1000.00 -500.00
0.00
500.00 1000.00 150000
-1500.CC
Figure 29.
Contour Diagram of HSH 24-hour Average Urban
Concentrations (/ig/m3) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Pittsburgh 1964 Data.
54
-------
HSH 24-hour Averages, DiUsburch, Urban
-1500 00 -1000.00 -50000 000 500.00 100000 1500.00
150000 iiiiiiiiiiiiiiii:l:IIIIIIII!!1I 1500.00
1000.00
500.00
000
-500 00
-100000
-1500 00
100000
500.00
0.00
-500.00
-100000
-1500.00 -1000.00 -50000 0.00 50000 100000 '50000
-1500 00
Figure 30. Contour Diagram of HSH 24-hour Average Urban
Concentrations (/xg/m3) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Pittsburgh 1964 Data.
55
-------
Annual Ave1 ~iqes, Pittsburgh, Urban, PAL
isoo.oo
-150000 -100000 -50000 000 500.00 ' COO 00 150000
I I i I I I | I I I I I| i I I I | I I I 1 I I I I I 150000
1000.00
500.00
0.00
-500.00
-1000.00
150000 '''''''L
-1500.00 -1000.00
1000.00
500.00
0.00
-500 00
-1000.00
-1500.00
-500.00 0.00 50000 '00000 1500.00
Figure 31.
Contour Diagram of Annual Average Urban
Concentrations (/ig/m3) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Pittsburgh 1964 Data.
56
-------
Annual Averages. D;ftsburgh, Urban
-150000
1500.00
-1000.00 -50000 000
-iiiiiiiiIiiir
1000.00
500.00
0.00
-500 00
-1000 00
500.00 100000 1500.00
Tirr~iir~r-\ 'soo.oo
-1500 00
I I I I III I 1 I I ' I I ! II
1000.00
500.00
0.00
-50000
-1000.00
-150000 -1000.00 -50000 000
500.00 1000.00 1500.00
-1500.00
Figure 32.
Contour Diagram of Annual Average Urban
Concentrations (/zg/m3) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Pittsburgh 1964 Data.
57
-------
'-hour Averages. Oklahoma City, Jrbar. DA'L
-150000 -1000.00 -500 CO : 00 50000 100000
50000
100000
500.00
0 00
-50000
-ooo.oo
-'50000
15CO 00
Ti 1500.00
100000
500.00
000
-500.00
-1000 00
-500.00 -1000.00 -50000
0 00
500.00 100000 '50000
-150000
Ficrure 33.
Contour Diagram of HSH 1-hour Average Urban
Concentrations (^tg/m^) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground level Source
with Close-in Receptors Using Oklahoma City 1988
Data.
58
-------
1 -hour Averages, Oklahoma City, Urban
-'50000 -100000 -50000
1500.00
000
500.00 100000
1000.00
500.00 -
0.00
-500.00 -
-100000 -
- 1500 00
500 00
1500.00
1000.00
500.00
-150000 -100000 -500.00
000
0.00
-500.00
1000.00
500.00 100000 150000
-150000
Figure 34.
Contour Diagram of HSH 1-hour Average Urban
Concentrations (^g/m3) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Oklahoma City 1988
Data.
59
-------
HSH 2--hour Averoc:s, Oklchoma City, Urban, -/
-150000 -1000.00 -5CDOO 000 £0000 1000 00
1500.00
1000.00
500.00
0.00
-50000
-100000
-150000
150000
150000
1000.00
500.00
000
-500 00
-100000
-1500 CO
-1500.00 -100000 -500.00
0.00
500.00 100000 150000
Figure 35. Contour Diagram of HSH 24-hour Average Urban
Concentrations (/zg/m^) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Oklahoma City 1988
Data.
60
-------
HSH 24-hour Averages, Oklahoma City, Urban
-1500.00 -1000.00 -50000 000 500.00 1000.00 150000
150000
1000.00
500.00
000
-500 00
-1000.00
-1500 00
1500.00
1000.00
500.00
0.00
-500.00
-1000.00
-150000 -100000 -500.00
0.00
500.I
1000.00 ' 500 00
-1500 00
Figure 36.
Contour Diagram of HSH 24-hour Average Urban
Concentrations (jxg/m3) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Oklahoma City 1988
Data.
61
-------
-nnual AverGces, Cklohomc City, Urb'jn. PAL
50000 100000 150000
-'=0000 -100000 -50000 000
- 15CO OC
-500.00
-1000 00
11' -1500 CO
-'50000 -1000.00 -50000
000
50000 100000 150000
Figure 27. Contour Diagram of Annual Average Urban
Concentrations (/ig/m3) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Oklahoma City 1988
Data.
-------
1500
Annual Averages, Oklahoma City, Urban
-150000 -1000.00 -50000 COO 500.00 1000 00 150000
1000.00
500.00
0.00
-500 00
-1000.00
150000
-1500.00 '!L
1000.00
500.00
000
-500.00
-100000
-1500.00 -1000.00 -500.00
50000 100000 150000
-1500.00
Figure 38. Contour Diagram of Annual Average Urban
Concentrations (/xg/m3) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Oklahoma City 1988
Data.
63
-------
HSH 1-hour Averages, Sectt'e, -ben, PAL
-150000 -100000 -50000 000 50000 100000 '50000
1500.00
100000
50000
000
-50000
-100000
-150000
1500 00
-150000 -100000 -500.00
000
1000.00
500.00
0.00
-500.00
-1000 00
-1500 00
500.00 1000.00 '50000
Figure 39.
Contour Diagram of HSH 1-hour Average Urban
Concentrations (pig/m3) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Seattle 1983 Data.
64
-------
1500.00
HSH 1-hour Averages, Seattle, Urban
-1500.00 -1000.00 -500.00 000 500.00 100000 150000
1000.00
500.00
0.00
-500.00
-1000.00
-150000
1500.00
1000.00
500.00
0.00
-500.00
-1000.00
-150000 -100000 -500.00
000
500.I
1000.00
1500
-1500.00
00
Figure 40. Contour Diagram of HSH 1-hour Average Urban
Concentrations (/zg/m3) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Seattle 1983 Data.
65
-------
HSH 24-hour Averages, S.ottle, Urban, PAL
-150000 -1000.00 -50000 000 500.00 1000.00 150000
150000
1000.00
500.00
0.00
-500.00
-1000.00
1500.00
1000.00
500.00
-1500.00
i I I I IIIIIIIII -150000
i i i I I I I IIII
-1500.00 -100000 -500.00 0.00 50000 100000 1500.00
Figure 41. Contour Diagram of HSH 24-hour Average Urban
Concentrations (/Kj/m3) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Seattle 1983 Data.
66
-------
HSH 24-hour Averages, Seattle, Urban
-150000 -1000.00 -500.00
1500.00
000
50000
1000.00 1500.00
T||i 1500.00
1000.00
500.00
0.00
-500.00
-100D.OD
-150000
i \i l i i \ i 1/1 ii /ill
1000.00
500.00
-1500.00 -1000.00 -500.00
0.00
0.00
-500.00
-1000 00
50000 1000.00 1500.00
-1500.00
Figure 42. Contour Diagram of HSH 24-hour Average Urban
Concentrations (jig/m3) from.the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Seattle 1983 Data.
67
-------
Annual Averages, Seattle, Urban, PAL
-150000 -1000.00 -500.00 0.00
1500.00
500.00 100000 1500.00
1500.00
1000.00
500.00
1000.00
500.00
i t i i i i i i i i i i i l l l ll l l l l
500.00 1000.00 1500.00
-1500.00 -1000.00 -50000 000
Figure 43.
Contour Diagram of Annual Average Urban
Concentrations (/zg/m^) from the Numerical Integration
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Seattle 1983 Data.
68
-------
1500.00
Annual Averages, Seattle, Urban
-1500.00 -1000.00 -500.00 0.00 500.00 1000.00
1000.00
500.00
000
-500.00
-1000.00
-1500.00
1500.00
1500.00
1000.00
500.00
0.00
-500.00
-1000.00
-1500.00 -1000.00 -500.00 0.00
500.00 1000.00 150000
-1500.00
Figure 44. Contour Diagram of Annual Average Urban
Concentrations (/zg/m^) from the Finite Line Segment
Algorithm for the 1000 Meter Wide Ground Level Source
with Close-in Receptors Using Seattle 1983 Data.
69
-------
4. LIMITED COMPARISON WITH FPM RESULTS
The Fugitive Dust Model (FDM) also includes an inte^med
line source algorithm for modeling impacts from area sources
(TRC, 1990). It was originally intended that the sensitivity
analysis presented in this report would include results of the
FDM model for the cases using rural dispersion coefficients (FDM
does not include the option for using urban dispersion
coefficients). However, comparison of the integrated line source
results based on the numerical integration method used in the new
ISCST2 model with initial FDM results generated by EPA Region X
showed unexpectedly large differences. The ISCST2 numerical
integration results were generally about 50 to 100 percent larger
than the FDM results, with larger differences in a few cases.
Upon further investigation, it was discovered that these
differences were attributable to three assumptions made by FDM in
its implementation of the integrated line source algorithm.
Specifically, FDM assumes:
l) a minimum mixing height of 100 meters;
2) a minimum release height of 0.5 meters,- and
3) that the rural dispersion coefficients are
representative of a 10-minute averaging period and a
surface roughness of 3 cm.
The FDM model uses the third assumption as the basis for
adjusting the lateral and vertical dispersion coefficients. Whc^n
the numerical integration algorithm in the new ISCST2 model was
modified to use the first two assumptions, and the FDM model was
modified to eliminate the third assumption (i.e., setting the
sigma adjustments factors = 1.0), the corresponding results of
the two models were very comparable, agreeing to within a few
percent in most cases. Table 10 presents these results for the
very small (10 meter wide) ground level area source case (with a
release height = 0.5 meters).
The largest difference in Table 10 is about 10 percent for
the Seattle 3-hour HSH. Upon further investigation it was
discovered that the FDM model includes an error in the code that
effects the calms processing routines for 3-hour averages. If
one hour during a three-hour period is calm, then the FDM model
sums the remaining two hours and divides by two for the average.
The correct procedure is to divide by three in this case, since
two hours is less than 75 percent of the 3-hour averaging period.
This error leads to larger 3-hour averages for cases including
calm hours from FDM than from ISCST2, and accounts for the larger
differences in Table 10 for 3-hour averages.
70
-------
Another difficulty in comparing ISCST2 results with FDM
results is related to the fact that the FDM model includes two
different modes of implementing the integrated line source
algorithm. One mode uses a 5-line integration to approximate the
area source, while the other mode "converges" to a more accurate
representation of the area source. The convergent mode begins by
comparing results for 5 lines versus 6 lines. If convergence is
not indicated, then the model continues by comparing results for
10 lines versus 11 lines, and then for 15 lines versus 16 lines,
and so on until convergence is reached, out to a limit of 901
lines. The 5-line and convergent algorithm were both executed
for the 10-meter wide area source, and gave comparable estimates
(to within a few percent difference). However, the convergent
mode could not successfully be executed on the X-Large, Close-in
case because of the extremely long execution time involved (it
was estimated that it would take at least 45 days to complete a
one year simulation with 180 receptors on a 33-MHz 486 computer,
compared to about 6 hours using the numerical integration method
implemented in the new ISCST2 model). Therefore, the FDM model
was only run for selected receptors in order to compare FDM
convergent results with results from the numerical integration
algorithm in the ISCST2 model. Based on these limited '
comparisons it is concluded that the numerical integration
algorithm gives results that are very comparable to the FDM
convergent results, to within about one percent difference. This
conclusion is based on comparisons for receptors located both
within the area for the close-in case and downwind of the area
for other ground level cases, and also includes the receptor
locations for the highest impacts as well as receptors with
relatively low impacts.
Comparisons were also made between the FDM convergent
results and the FDM 5-line results for selected receptors located
within the area. These comparisons show that the 5-line
integration is not reliable for receptors located within the
area. The 5-line results showed very large variations over
relatively short distances, especially near the center of the
area. Since FDM divides the area into 5 lines regardless of
where the receptor is located, the impacts for receptors located
within the area are estimated based only on the lines that are
located upwind of the receptor for a given hour.
The conclusion from all of these comparisons between the
ISCST2 numerical integration algorithm and the FDM integrated
line source algorithm is that ISCST2 provides a much more
efficient and reliable algorithm for modeling impacts at
receptors located within and nearby the area, and that it gives
comparable results to the FDM convergent algorithm when modeled
based on the same assumptions for release height, mixing height,
and dispersion parameters. Moreover, the current version of FDM
includes an error in the implementation of the calms processing
routines for 3-hour averages.
71
-------
Table 10
Comparison of ISCST2* Numerical Integration Results With
F^M** 5-Line Results
for the Very Smf.il Source (10m Width) - Rural
Pittsburgh 1964
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
ISCST2
Numerical
Integration
112172
95015
66169
62426
28378
21035
3208
FDM
5-Line
Integration
116993
94148
68600
64601
28745
21311
3205
Ratio
ISCST2/FDM
0.96
1.01
0.96
0.97
0.99
0.99
1.00
Okla. City 1988
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
113594
113594
68653
51986
25432
20537
5853
119314
119314
69308
51419
25442
20456
5891
0.95
0.95
0.99
1.01
1.00
1.00
0.99
Seattle 1983
1-hr High
1-hr HSH
3-hr High
3-hr HSH
24-hr High
24-hr HSH
Annual
107377
104723
55576
48897
21933
20231
5014
108872
105308
61648
52654
22547
20753
5002
0.99
0.99
0.90
0.93
0.97
0.97
1.00
**
ISCST2 results are based on a minimum mixing height of 100
meters and a release height of 0.5 meters.
FDM results are based on a minimum mixing height of 100
meters, a minimum release height of 0.5 meters, and no sigma
adjustment factors.
72
-------
5. REFERENCES
Erode, R.W., 1992: Summary of the Quality Assurance and
Equivalence Tests Performed on the Modified Area Source
Algorithm for the ISCST2 Model. Internal Project Report, WA
1-27, U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina.
Environmental Protection Agency, 1989: Review and Evaluation of
Area Source Dispersion Algorithms for Emission Sources at
Superfund Sites. EPA-450/4-89-020. U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina.
Environmental Protection Agency, 1992. User's Guide for the
Industrial Source Complex (ISC2) Dispersion Models.
EPA-450/4-92-008. U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina.
Petersen, W.B. and E.D. Rumsey, 1987. User's Guide for PAL 2.0 -
A Gaussian-Plume Algorithm for Point, Area, and Line
Sources. EPA/600/8-87/009. U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina.
TRC Environmental Consultants, 1990: User's Guide for the
Fugitive Dust Model (FDM), (Revised). EPA-910/9-88-202R.
U.S. Environmental Protection Agency - Region 10, Seattle,
Washington.
73
-------
TECHNICAL REPORT DATA
(Please read Instructions on reverse before completing)
1. REPORT NO.
EPA-454/R-92-015
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Sensitivity Analysis of tlie Revised Area Source
Algorithm for the ISC2 Short Term (ISCST2) Model
5. REPORT DATE
October 1992
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Pacific Environmental Services
5001 South Miami Boulevard
Post Office Box 12077
Research Triangle Park, NC 27709-2077
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO. WA No. 1-131
EPA Contract No. 68 D00124
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Technical Support Division
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final Report
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
EPA Work Assignment Manager; Jawad S. Touma
16. ABSTRACT
This report includes information on an improved algorithm for modeling dispersion
from area sources, which has been developed based on a numerical integration of the
point source concentration function. A sensitivity analysis is presented of the
algorithm as implemented in the short-term version of the Industrial Source Complex
(ISC2) model. To examine the sensitivity of the design concentrations across a range
of source characteristics, five ground-level area sources were modeled, ,;ith sizes
varying from 10 meters to 1,000 meters in width. An elevated source scenario
consisting of a 100-meter wide area with a release height of 10 meters was also
modeled. An additional case involving 1,000 meter wide ground level area was also
modeled with receptors located within and nearby the area. The high and high-second
(HSH) 1-hour, 3-hour and 24-hour averages and high annual averages were determined for
each of these source scenarios using a full year of real time meteorological data. All
of the sources were modeled as square areas oriented N-S and E-W, since the original
ISC algorithm was limited to handling that source geometry. Each scenario was run for
one year of National Weather Service meteorological data from Pittsburgh, PA (1964);
one year of NWS data from Oklahoma City, OK (1988); one year of NWS data from Seattle,
WA (1983). This report is being released to establish a basis for reviews of the
capabilities of this methodology and of the consequences resulting from use of this
methodology in routine dispersion modeling of air pollutant impacts.
17.
KEY WORDS AND DOCUMENT ANALYSIS
a.
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
c. COSAT I
Field/Group
Air Pollution
Toxic Air Pollutants
Air Quality Dispersion Models
Dispersion Modeling
Meteorology
Air Pollution Control
18. DISTRIBUTION STATEMENT
Release Unlimited
19. SECURITY CLASS (Report)
Unclassified
21. NO. OF PAGES
20. SECURITY CL^S (Page)
Unclassii: ,d
22. PRICE
EPA For* 2220-1 (Rev. 4-77)
PREVIOUS EDITION IS OBSOLETE
------- |