-------
TIME (EST)
1151-1359
1709-1919
2308-0104
0704-0928
1309-1518
Figure 8. Aircraft sampling flight tracks and CL concentrations
(ppb) near the center of the mixed layer (dashed
lines) during the 3-4 August 1979 period, (from
Clarke et al., 1983)
19
-------
western New York and Pennsylvania, with the highest point values between 160
and 170 ppb. This is just to the east of the surface monitoring site at
Conneaut, Ohio where a maximum surface value of 159 ppb was recorded the same
day. The aircraft data add further evidence to a verification of the premise
that there existed an area of elevated 63 concentrations aloft in the boundary
layer to the south and east of the Lake Erie shoreline and possibly over the
lake itself on 3 August. The timing and location of this area of high 03
concentrations indicates that it was not generated on this day, but rather
advected here from another time and location, quite possibly from Detroit and
the heavy industrial areas to the west of Lake Erie. The wind flow over this
region the previous day had been from the west and southwest directions, with
a stronger westerly component immediately behind the weak cold front that
passed through on 2 August. Clarke et al. (1983) suggest that the higher
pollutant loading seen here may have originated as far west as Chicago, given
the results of a back trajectory analysis.
The first aircraft transect made on the morning of 4 August was a partic-
ularly interesting one because it sampled in an area of the NEROS region that
is sparse in surface monitoring locations. Flight E seen on Figure 7 was
conducted by BNL and proceeded from a point south of Syracuse, NY to easternmost
West Virginia. On this flight 03 values in the boundary layer were seen to
fluctuate in the range of 7U-8U ppb over New York and Pennsylvania and 60-70
ppb over Maryland and West Virginia. A similar pattern of concentrations was
seen on the return flight (G) that afternoon. The transect for flight G
proceeded from south to north and to the east of transect E. Again no pronoun-
ced increases in 03 concentration during the flight were observed here, and
the general boundary layer values were 70-80 ppb. The WSU aircraft conducted
a late afternoon flight on 4 August consisting of 2 legs. The first leg of
flight H proceeded from southeast Pennsylvania to West Virginia. Boundary
layer 03 concentrations along this leg were in the 70-80 ppb range. The
second leg of the flight went north from West Virginia to northeast Ohio.
Here the concentrations of 03 were more in the 50-70 ppb range within the
boundary layer. The RTI aircraft made 2 north-south transects along the same
path on 4 August. An afternoon flight from southern Ohio to Lake Huron
20
-------
showed 1)3 concentrations of 70-80 ppb generally in the boundary layer, with a
few exceptions. Concentrations of around 90 ppb were detected southeast of
Columbus and also west of Akron. The aircraft apparently encountered an
03 plume southeast of Detroit over Lake Erie extending into southwestern
Ontario. Concentrations of 90-lbO ppb 03 were detected in the plume.
The relative compactness of the plume and the distance downwind of Detroit
indicate that the plume probably represents photochemical activity from
that city on 4 August. The return flight occurring during late evening
along the same path showed a broad 03 plume now northeast of Detroit with
peak concentrations of 120 ppb extending over about a 35 km cross-section.
South of this plume 03 concentrations were generally in the 6b-7b ppb
range with an increase to about 90 ppb in southeast Ohio.
In general, the aircraft sampling during this 2-day episode over the
NEROS region indicated widespread boundary layer 03 levels of 70-80 ppb
with some areas slightly higher or lower than this range at times. The
aircraft sampling showed the presence of 2 pronounced 03 plumes during
the period in the areas covered by the transects. The first was a large
area of 03 from an aged plume over western New York on 3 August (see
Figure 8) and the second was an urban-scale plume, probably from Detroit,
over western Ontario on 4 August. Aircraft sampling downwind of other
major urban source areas in the NEROS region was either insufficient or
nonexistent in order to detect 03 plumes.
Section 3 presents detailed comparisons of the hourly averaged
surface 03 concentrations and the corresponding model predictions.
Measurements of 03 concentrations aloft from aircraft monitors are also
used in the analysis to help resolve particular features of the concentra-
tion pattern.
ROM1 Initial and Boundary Concentrations
Any mass-conservative grid model of atmospheric air pollution,
including ROM1, is sensitive to the initial field of mass within the
21
-------
modeling domain as well as the mass flux across inflow boundaries during the
simulation. The initial mass field specified as a model input can remain
within the model domain and exert considerable influence on simulation results
for hours or days, depending upon the spatial scale of the model, wind trans-
port speeds, and source emissions distributions and flux rates. In urban
scale modeling the initial field is typically advected out of the modeling
domain within a few hours after the start of simulation. Monitoring networks
on this scale are often sufficient to adequately specify the initial concen-
trations of the major pollutant species of interest. On the larger regional
scale however more problems are encountered in specifying a proper initial
concentration field.
One major concern is the density of monitoring stations over a 1000-km
regional scale model domain. Figure 1 demonstrates that the density of 03
monitors in the standard U.S. SAROAD network is not uniform. Major urban areas
are covered fairly well but large areas of the domain are unmonitored.
Interpolation across such large distances may not be reliable. Aircraft
sampling may help to some degree, but the transects available are relatively
few and those taken near model initialization time are fewer still. The
density of monitors for species other than 03, such as NOX and hydrocarbons, is
considerably less than that for 03, making the initialization problem worse
because these precursor species are important in the photochemical generation
of 03. The uncertainty in the initial concentration estimates is further
compounded by the transport speeds. Days may elapse before the initial mass
field is completely flushed out of a regional model domain, For instance, in
a IDOU-km model domain with a lateral transport wind speed of 5 m/sec it would
be 55 hours before the initial mass is completely transported out. In the
interim simulation period errors in the initial field specification would
propagate through the chemical interactions to the model species predictions.
The first attempt at simulating the 3-4 August 1979'period with the ROM1
included an estimate of the initial concentration field at 00 h, 1ST on 3
August over the model domain. Five "categories" of grid cells were determined,
based mainly on land use, where the same set of initial concentration values
22
-------
were used in all cells of a given category. The categories ranged from a
remote or wilderness class to an urban-industrial class. Ambient observations
of 03 near model initialization time were studied for each class of cells and
estimates of the initial 03 value within each class were made. Data observat-
ions of NOX were very sparse and those of hydrocarbons were essentially non-
existent, so estimates of these concentration values were made for each category
of grid cells based on chemical equilibrium considerations of all the reactive
species. It became clear in examining the simulation results for the 2-day
episode using this method of model initialization that it would not be possible
to distinguish between model 03 results based on precursor source emissions
input to the simulation or model results as artifacts of the specified initial
field. For example, the large area of 03 over western New York (see Figure 8)
appeared in the simulation results primarily because of heavy initial mass
loadings of 03, NOX, and hydrocarbons over the Ohio-Lake Erie vicinity. In
attempting to evaluate model results for 03 we are interested primarily in
analyzing the model's ability to generate 03 patterns from the given source
emissions distribution because the ROM1 will eventually be asked to judge the
effects on 03 of proposed changes in the source emissions input function.
The effects of initial conditions must therefore be minimized in any simulations
performed with a regional model.
We minimize the influence of initial conditions by specifying tropospheric
background "clean" conditions throughout the model domain. In conjunction
with this procedure it will be necessary to begin model simulation at a
relatively "clean" period of time throughout the domain also, such as after
the passage of a cold front with an attendant air mass exchange. In future
work with ROM2 we will use this initialization procedure at the start of a
long period (1-4 weeks) simulation. This longer period will also insure a
minimal dependence of model results on initial concentrations.
For the 2-day episode at hand, however, the simulation was performed
again with tropospheric clean initial conditions in order to judge the model 's
ability to produce 03 from the source emissions function. This is done with
the knowledge that the initial ambient field did not in fact, contain clean
23
-------
conditions. Surface and aircraft monitoring showed areas with higher 03
values, including the large 03 plume in the eastern Great Lakes area. Other
plumes though, formed during the 2-day period, should be included in the
simulation. Interpretation of model performance here must be done with this
understanding of the potential omissions in the initial" mass field. With
this newly acquired appreciation of the effects of initial conditions all
future simulations will begin at a clean period and proceed for one week or
more in time. It was, however, still thought useful to analyze the results
of this short test episode as an exercise in some of the model evaluation
techniques that will be used in longer term simulations.
Boundary concentrations, both at the lateral edges and the top of the
modeling region, are handled similarly to the initial concentrations for this
2-day episode. That is, tropospheric background values are used at all
boundaries for all times. The combination of the clean initial and boundary
concentrations here produced simulation results reflecting these clean concen-
trations, except where source emissions injections into the model domain
raised the level of pollutants above background. The background tropospheric
03 concentration specified in ROM1 was around 40 ppb, and model results show
this 03 value through large sections of the domain during the simulation.
Actual ambient boundary layer 03 values of 60-80 ppb 03 were measured by most
of the aircraft sampling performed during the 3-4 August period. Therefore
there is an inherent underpredictive bias of 30-40 ppb of 03 introduced into
this model simulation because of the inconsistency in background values.
Future applications of the regional model will provide for day/night temporal
variations in the lateral boundary concentrations.
24
-------
SECTION 3
RESULTS OF MODEL SIMULATION FOR 03
The ROM1 model was run for the 48-hour period of 3-4 August 1979 using
data from the NEROS and SAROAD data bases. Before the model itself was
started the series of required preprocessors was run to provide the necessary
model inputs to the ROM1 (Schere and Fabrick, 198b). The model was initialized
throughout its entire domain with a set of tropospheric background species
concentrations which were computationally adjusted for chemical equilibrium.
These concentrations are shown in Table 2. The same set of concentrations
was used for both lateral inflow and top boundary concentrations throughout
the model simulation period. These chemical species correspond to those in
the 36 reaction Demerjian chemical kinetic mechanism contained in ROM1.
The evaluation of the ROM1 results here for 03 occurs in several stages.
First we compare model predictions, interpolated to the surface SAROAD station
locations, with observed values over all hours and receptor sites. The model
predictions, which are available every 30 minutes, have been averaged into
1-hour intervals to correspond with the hourly averaged surface data. Next,
an analysis of the maximum concentrations at receptor locations is made, and
finally a close look at individual plumes from large source areas within the
domain is performed. Aircraft observations of 03 concentrations in urban plumes,
where available, are included in the analysis. We ignore the first 6 hours of
predictions from the model because of the strong influence of initial conditions
in this period. Therefore the first hour that comparisons of observed and
predicted data are made is 060U-070U h, LST on 3 August and the last hour of
the evaluation period is 2300-2400 h, LST on 4 August.
Hourly Data Comparisons
Figure 9 presents a histogram of the hourly 03 values observed by the
surface monitoring stations in the ROM1 domain during the model evaluation
period. There is a pronounced peak in the observations at very low values of
25
-------
TABLE 2. SET OF INITIAL AND BOUNDARY SPECIES
CONCENTRATIONS USED IN ROM1
Species
NO
N02
03
olefins
paraffins
aldehydes
aromatics
CO
HN02
HN03
PAN
RN03
Concentration
(ppb)
.00658
1.824
36.9
.3087
.3919
.04323
.1224
99.98
.001939
.1523
4.931 X 1C'5
1.803 X 10'8
Species
H202
0
N03
OH
H02
H02N02
RO
R02
R20
R102
R202.
Concentration
(ppb)
.002571
2.180 X 10'9
.01345
7.953 X 10-6
.003544
.002115
1.254 X 10-9
.001722
1.196 X 10'10
7.089 X 10"7
1.526 X lO'4
26
-------
OBSERVED OZONE
FREQUENCY
1700
1600-
1500
1400-
1300
1200
1 tOO
1000
900
800
700
600
500
400 H
300
200-
100-
0
12 24 38 48 60 72 84 96 108 120 132 144 156
OZONE. PPB
Figure 9. Histogram of observed 0., concentrations at surface
monitoring sites on 3-4 August 1979.
27
-------
03, between U and 18 ppb. The magnitude of this peak might be confusing.
These low values are generally occurring at night and are the result of near
surface depletion of 03 by ground deposition and chemical reaction. Instrument
sensitivity at such low concentration levels must also come into question.
Beyond these values, a broad peak in the distribution up to about 65 ppb is
seen, beyond which the frequency distribution drops off gradually to the
maximum values, near 160 ppb.
Figures 10a and lOb present the histograms of hourly jpredicted 03 values
interpolated to the surface monitoring site locations from the ROM1 layers 0
and I, respectively. These distributions appear more normally-shaped (Gaussian)
than the observed distribution and have sharp peaks in the 25-45 ppb vicinity.
This concentration range spans the specified background value for 03 of 37
ppb which is used by ROM1 for lateral and top boundary concentrations.
Apparently this background value dominates the ROM1 predictions at grid
locations of the surface monitoring sites over many of the simulation hours.
The peak in the layer 0 distribution appears in the 25-35 ppb range while the
peak in layer 1 predictions is 35-45 ppb. This slight shift toward lower
values in layer 0 reflects the effects of ground deposition and the sub-grid
scale processes that chemically deplete 03 in the surface "layer. In general,
the layer 0 ROM1 concentrations are a few ppb less than the corresponding
layer 1 concentrations over the grid.
The residual or bias in the concentration predictions, d, is defined as
the difference between the observed and predicted 03 concentrations. That
is, for layer 0, dg = c0b - CpQ, and for layer 1, di = c0b ~ Cpl- The
histograms of dg and di for the model simulation period are shown respectively
in Figures Ha and lib. The peak value in both figures appears slightly on the
negative, or overpredictive, side of the Gaussian shaped distributions. The
negative peaks in these figures are artifacts of the model.. These overpredicted
concentrations correspond to very low ambient 03 values observed mostly at
night. The majority of surface monitoring stations are located in or near
urban areas and are thus sensitive to rapid spatial or temporal changes in
28
-------
PREDICTED LAYER ZERO OZONE
FREQUENCY
2000-
1900-
1800-
1700-
1600
1500
1400-
1300-
1200-
1100
1000 H
900
800-
700
600 H
500
400-
300-
200-
100-
0
n
10 20 30 40 SO 60 70 80 90 100 110 120
OZONE. PPB
Figure Ida.
Histogram of CL
monitoring site
(ROM! layer 0).
concentrations predicted by ROM! at
locations for 3-4 August 1979
29
-------
PREDICTED LAYER ONE OZONE
FREQUENCY
1900
1800-
1700-
1600-
1500
1400-
1300-
1200
1100-
1000
900
800
700
600
500
400
JOO
200
too
0
10 20 30 40 50 CO 70 80 90 100 t 10 120
020NC, PPB
Figure lOb.
Histogram of CL
monitoring site
(ROM! layer 1).
concentrations predicted by ROM! at
locations for 3-4 August 1979
30
-------
OBSERVED MINUS PREDICTED OZONE, LEVEL ZERO
FREQUENCY
1400-
1300-
1200-
1100-
1000-
900-
aoo-
700-
600-
500-
400-
300-
200-
100-
n .
1 1
1
-60 -«4 -48 -32 -16 0 16 32 48 64 80 96 112
OZONE. PPB
Figure lla. Histogram of residual 0, concentrations at monitoring
site locations for 3-4 August 1979, ROM1 layer 0.
31
-------
OBSERVED MINUS PREDICTED OZONE, LEVEL ONE
FREQUENCY
1400-
1300-
1200-
1100
1000
900
800
700
600
500-1
400
300-
200
100
0
-80 -64 -48 -32 -16 0 16 32 48 64 80 96 112
OZONE. PPB
Figure lib. Histogram of residual 0, concentrations at monitoring
site locations for 3-4 ^August 1979, ROM1 layer 1.
32
-------
emissions patterns, especially the titration of 1)3 by NOX sources. On the
other hand the ROM1 predictions are volumetric averages over the 20 km X 20 km X
H size of the grid cell (where H is the depth of the grid cell). Therefore,
the sub-grid scale effects from urban area sources are typically only a small
perturbation on the volumetric average predicted cell concentration. This
will lead to overprediction under these circumstances and is reflected in the
negative peak in the histograms. Tne majority of daytime residuals corresponds
to underpredictions and we speculate at this point that they would be aligned
with higher 03 concentrations. A discussion of the trends in magnitude of
the bias with ambient concentration level is given later in the discussion of
concentration maxima.
The average bias in layer 0 03 predictions over all surface monitoring
locations is shown in Figure 12a as a function of time. The first data point,
at 6 hours after the start of model simulation, corresponds to the hour
period 06-U7 h, 1ST on 3 August 1979. The 95% confidence interval about each
average bias point is also shown. Figure 12b presents the same information,
except as the percent bias of the observed 03 concentration. Figure 12c is
another way of showing this information in a time series. It is a box plot
of the bias among the surface monitoring locations where the middle of each
box is the median, the top edge is the 75th percentile, the bottom edge is
the 25th percentile, and the minimum and maximum points are the bottom and
top, respectively, of the vertical projections from the box. All three time
series show a distinct diurnal trend in the average bias, with underpredictions
averaging 30-32% during the midday hours and overpredictions averaging as
much as 100+ % at night. These large overpredictions correspond to relatively
modest bias values, 10 to 25 ppb, as seen from Figure 12a. At those times
of high overpredictions, the bias and percent bias figures imply observed and
predicted 03 values on the order of 20 and 40 ppb, respectively, confirming
an earlier premise made from analyzing the histograms that the ROM1 overpred-
ictions correspond to quite low ambient concentrations.
From Figure 12c it is seen that the median underpredictions on the first
day of simulation peak at about 20 ppb and on the second day at nearly 30 ppb,
33
-------
BIAS AVERAGE ALL RECEPTORS, LAYER ZERO
10 20 30 40
HOURS FROM START OF SIMULATION
Figure I2a. Time-series over the 3-4 August 1979 simulation period
of average bias over all surface receptor locations.
34
-------
PERCENT BIAS AVERAGE ALL RECEPTORS, LAYER ZERO
1.00E+02-
P
E
R T £,C_ 1 K ,
" W. Ot Id
c
E
N
T
* -I.OE+02-
A
S
-2.0E+02-
j(< , swu.
* V i i.
,' 1,111 .,-
I.........|,....,..,|. ,.,... ..j... ......,,.,... ...j
0 10 20 30 40 SO
HOURS FROM START OF SIMULATION
Figure 12b. Time-series over the 3-4 August 1979 simulation period
of average percent bias over all surface receptor locations,
35
-------
BIAS ALL RECEPTORS, LAYER ZERO
0
B
2.00E+02-
1.00E4-02-
y
i
N
U
S
P
R
E
0
0 1.78E-15
0
N
E
-t.OE+02-
10 20 30 40
HOURS FROM START OF SIMULATION
50
Figure 12c. Time-series over the 3-4 August 1979 simulation period
of a box-plot distribution of bias over all surface
receptor locations.
36
-------
both during the daylight hours when 03 is expected to reach its daily maximum
values. Taken together with the percent underpredictions these bias values
imply that the 03 observations during the midday period on 3 August at the
monitoring site locations are in the 6U-65 ppb range, while the layer 0 pred-
ictions are in the 40-45 ppb range. On 4 August the corresponding observed
and predicted ranges are 65-70 ppb and 40-4b ppb, respectively. The layer 0
predictions are at a level generally near the specified background and initial
03 concentrations, indicating their influence over the grid locations corre-
sponding to the surface monitoring sites. It should be reiterated here that the
surface monitoring sites are not regularly spaced in the NEROS region, but are
concentrated near urban areas. The underprediction during this period may be
the result of the model's inability to correctly replicate all of the sub-grid
processes, transport inaccuracies, or other problems. A closer look at some
specific areas of the grid will be examined later. Figure 12a indicates that
there are only four hours during the simulation period when the average bias
is not significantly different from zero. These hours occur during transitional
time periods on both days of simulation. Although not shown here, the corre-
sponding time series plots from ROM1 layer 1 results are very similar to
those discussed for layer 0.
A tabulation of hourly averaged parameters for the simulation period is
given in Table 3. Parameter abbreviations are defined in the Glossary and
their significance is discussed in Schera and Fabrick (1985). For each hour
displayed there are 'n' monitoring station locations used in the averages.
The values in the table confirm the very low observed 03 concentrations at
night. The diurnal range in 03 concentrations is considerably greater in the
observations than in the model predictions, where a variation of only 10-15 ppb
is shown. The bias values confirm the data presented in Figure 12 with negative
values at night and positive values during the day. The absolute values of
bias (gross error) are all consistently rather high, indicating that the
hours with average bias near zero actually contain a wide variation of indivi-
dual values with opposite signs. The episodic average value of observed 03
for all hours at the surface monitoring sites was 38.9 ppb and the corres-
ponding model predictions for ROM1 layers 0 and 1 were 36.6 and 38.1 ppb,
37
-------
TABLE 3. SUMMARY OF ROM1 HOURLY RESULTS FOR 3-4 AUGUST 1979
R
DATE - HOUR n «cob «cp0 V
"TO"
(ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb)
79215 06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
79216 00
01
111
110
134
140
142
144
148
147
148
150
149
150
148
145
142
133
126
122
114
114
11.6
14.6
19.4
28.4
38.0
47.8
56.4
61.3
62.8
60.1
57.0
52.4
44.1
35.4
25.7
20.9
18.1
15.9
15.7
16.2
30.0
31.1
32.2
33.8
35.5
37.1
38.6
40.2
41.6
41.8
40.9
40.2
38.7
35.6
34.2
32.4
31.4
30.7
29.1
28.2
32.0
32.5
33.5
35.2
37.0
38.5
40.1
41.6
43.0
43.0
42.8
41.9
40.3
37.6
35.4
33.8
32.9
31.9
30.5
29.4
-18.4
-16.4
-12.8
- 5.5
2.6
10.7
17.6
21.1
21.2
18.3
16.1
12.1
5.3
- 0.2
- 8.5
-11.5
-13.3
-14.8
-13.4
-12.0
19.0
17.1
15.9
15.0
17.8
21.6
24.8
25.8
26.6
25.0
25.0
24.1
21.2
20.7
20.0
19.5
19.1
20.1
19.2
18.5
10.1
10.7
14.2
17.6
21.8
24.6
25.5
25.0
25.7
25.5
27.7
27.7
26.1
25.5
23.5
21.8
19.8
19.1
18.2
18.6
-20.3
-17.8
-14.2
- 6.8
1.0
9.3
16.3
19.7
19.8
17.1
14.1
10.5
3.7
- 2.1
- 9.8
-12.9
-14.8
-16.0
-14.8
-13.3
20.9
18.4
17.0
15.2
17.6
21.1
24.3
25.1
25.8
24.3
24.2
23.9
21.1
21.2
20.7
20.4
20.2
21.1
20.1
19.4
10.5
11.0
14.4
17.6
21.8
24.6
25.6
25.1
25.8
25.8
27.9
28.0
26.4
25.9
23.8
22.1
20.5
19.5
18.8
19.1
38
-------
TABLE 3. (continued)
DATE - HOUR n RcQb Rcp0
Rdn
>dO
'1
(ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb)
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
114
121
129
121
129
137
150
140
143
144
135
142
144
142
148
150
149
148
142
137
16.4
17.8
16.7
15.1
15.0
19.5
27.7
38.1
51.6
62.5
67.5
68.9
67.6
68.7
67.7
65.4
55.8
40.9
29.3
22.7
27.6
27.0
25.6
27.2
29.7
32.1
34.8
38.1
40.9
42.9
45.1
45.7
45.9
45.5
44.9
43.8
41.8
38.8
37.3
35.3
28.8
28.1
27.0
28.5
31.9
34.0
36.5
39.5
42.4
44.4
46.7
47.4
47.5
47.4
46.7
45.3
43.5
41.1
38.9
36.9
-11.2
- 9.2
- 8.9
-12.1
-14.7
-12.6
- 7.1
0.0
10.8
19.6
22.4
23.2
21.7
23.2
22.8
21.6
14.0
2.2
- 8.0
-12.7
17.6
16.8
16.2
16.2
17.5
17.0
17.0
18.8
22.5
28.6
32.4
33.6
33.5
34.4
33.4
32.9
30.0
25.4
21.9
22.4
17.6
17.6
16.9
14.9
14.9
16.2
19.1
23.0
26.0
28.3
30.3
31.7
32.4
32.6
32.2
32.7
33.2
31.9
27.9
25.1
-12.4
-10.3
-10.3
-13.4
-16.9
-14.5
- 8.8
- 1.4
9.2
18.1
20.8
21.5
20.1
21.3
21.1
20.0
12.3
- 0.1
- 9.5
-14.3
18.6
17.6
17.1
17.2
19.5
18.6
18.0
19.2
22.3
28.1
31.9
33.0
33.1
33.7
32.8
32.3
29.6
25.0
22.5
23.3
18.1
18.0
17.4
15.5
16.2
17.3
20.0
23.4
26.3
28.6
30.6
32.0
32.8
33.0
32.6
33.0
33.4
31.8
28.4
25.7
39
-------
TABLE 3. (continued)
DATE - HOUR n RcQb Rcp0 R^pl *SQ RTd^T RSd0 %
(ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb)
22
23
128
118
20.2
19.4
33.6
33.4
35.0
34.5
-13.5
-14.0
22.0
20.7
24.0
22.2
-14.8
-15.1
22.8
21.5
24.4
22.6
79215 ALL
79216 ALL
2489
3239
38.6
39.2
36.2
36.9
37.7
38.5
2.4
2.3
21.2
24.0
26.5
29.8
0.9
0.7
21.4
24.4
26.7
30.1
79215-
79216 ALL
5728
38.9
36.6
38.1
2.3
22.8
28.4
0.8
23.1
28.7
40
-------
respectively. These values represent a 5.9% underprediction for layer 0 and
a 2.1% underprediction for layer 1. Since these average 03 levels are near
tropospheric background values the significance of this analysis over all
hourly values in the simulation period is not particularly great.
The next step was to perform an hourly analysis with a subset of the
data values already used. The criterion for choosing the subset was based,
in part, on the tropospheric background value of near 40 ppb 03 used in the
ROMl's initial and boundary conditions and the fact that many of the surface
monitoring site locations showed 03 predictions near this value. Therefore
the data subset is based on only those receptor points where the observed
value jmd_ the level 0 and 1 predicted values of 03 are greater than 5U ppb.
This should limit the data to those sites most affected by source emissions
within the NEROS region. In fact the number of sites used in the analysis
dropped from near 150 during the midday hours to around 30, and from around
120 during the nighttime hours to 0, after this criterion was applied.
The choice of criterion for the data subset might not seem to be the most
obvious at first. Perhaps a more logical subset would be formed from choosing
all points where only the observed 03 concentration was greater than 50 ppb.
In practice, this was the first subset produced for the analysis. It was
found that this subset still contained far too many model predictions near
tropospheric background levels to be useful in the restricted subset analysis.
A number of factors could account for this including error in the location of
source plumes within the model domain through transport uncertainty or source
emission uncertainties. The overriding factor however is the assumption of
tropospheric 03 concentration values everywhere at the boundaries and initially
when, in fact, areas of higher 03 concentrations were observed. In any case the
more restrictive criterion that was ultimately chosen produced the desired
effect of reducing the data subset to those concentration values above the
level of tropospheric background 03, and presumably produced data points
more affected by local and transported photochemical effects.
41
-------
The histogram of observed 03 concentrations in this data subset is
presented in Figure 13. The concentration values along the axis of the
plot are midpoints of the range for each class. The distribution peaks
in the range 55-6b ppb and then falls off rapidly. The distribution
slowly diminishes in the range above 75 ppb. This distribution does not
contain the broad plateau of lower values of 03 shown in Figure 9 for the
full data set. The effect of eliminating the low data values is apparent
in comparing the two figures. The reason for the lower maximum value in
the data subset histogram is that we have screened out data points where
either the observed or predicted values are less than 50 ppb. Thus the
station that observed nearly 160 ppb showed a corresponding prediction of
less than 50 ppb and was not included in the data subset.
Figures 14a and 14b present the histograms of the ROM1 03 predictions
for layers 0 and 1, respectively, for the data subset. The predictions
are skewed toward the lower end of the concentration range in the data
subset. The number of model predictions above about 65 ppb is markedly
lower, although the higher values extend to the 120 - 130 ppb range,
nearly matching that on the observed concentration histogram. Comparing
these histograms with those for the full set shown in Figure 10, it can
be seen that the large number of predictions near the tropospheric back-
ground 03 value are now eliminated, although the effect of this background
value still persists to some extent in the peak for the range of 50-60
ppb of the data subset.
Histograms of d0 and d^ on the data subset are shown in Figures Iba
and Ibb. These distributions are Gaussian shaped with a peak at the zero
range residual. In general, both layers 0 and 1 residuals are symmetrically
distributed except for the larger tail on the underpredictive side.
These histograms show a similar distribution to those in Figure 11 for the
full data set, except they are not as asymmetrical as those in the latter
figure.
42
-------
OBSERVED OZONE
FREQUENCY
190-
180-
170
160
150-
140-
130-
120-
110
100
90
80 H
70
60
501
40
30-
20-
10-
0
60 70 80 90 100 110 120 130 t40
OZONE, PPB
Figure 13. Histogram of observed 03 concentrations
at surface monitoring sites (where obs.
and pred. 03 > 50 ppb) on 3-4 August 1979.
43
-------
PREDICTED LAYER ZERO OZONE
FREQUENCY
1JO-
120-
110-
100-
90-
80-
70-
60-
50-
40-
JO-
20
n
IIf ,_
50 60 70 80 90 100 110 120 130
OZONE. PPB
Figure 14a. Histogram of 03 concentrations predicted
by ROM1 at monitoring site locations (where
obs. and pred. 0, > 50 ppb) for 3-4 August
1979 (ROM1 layered).
44
-------
PREDICTED LAYER ONE OZONE
FREQUENCY
ISO
140
130-
120-
tto
100
90
80
70
60
30
40-
30-
20
56 64 72 80 88 96
OZONE. PPB
104 112 120
Figure 14b. Histogram of 0- concentrations predicted
by ROM1 at monitoring site locations (where
obs. and pred. 0- > 50 ppb) for 3-4 August
1979 (ROM1 layer13!).
45
-------
OBSERVED MINUS PREDICTED OZONE, LEVEL ZERO
FREQUENCY
180-
170-
180-
150-
140-
uo-
120-
1 10-
100-
90-
80-
70-
60
so-
40-
30-
20-
10-
o-
| 1
60
40
1 1! __
-20 0 20 40 80 80 100
OZONE. PPB
Figure 15a. Histogram of residual CL concentrations at
monitoring site locations (where obs. and pred.
03 > 50 ppb) for 3-4 August 1979 (ROM1 layer 0)
46
-------
OBSERVED MINUS PREDICTED OZONE, LEVEL ONE
FREQUENCY
140-
130-
120-
110-
100-
90-
80
70-
60-
50-
40-
30-
20-
10-
0-
-48
l~ lr i
-32 -16 0 16 32 48 64 80
OZONE. PPB
Figure 15b. Histogram of residual 03 concentrations at
monitoring site locations (where obs. and pred.
0, > 50 ppb) for 3-4 August 1979 (ROM1 layer 1)
47
-------
Plots of the time series of the concentration bias at the surface monit-
oring sites in the data subset are shown in Figure 16. The bias is depicted
in Figure 16a and the percent bias in Figure 16b, with the 95% confidence
interval about each point. It is immediately obvious that: at certain hours
there is little, if any, data in the subset being analyzed. This effect is a
result of the elimination of the low 03 values in this data subset, most of
which occur during the nighttime hours. The diurnal variation in the bias
for the full data set, seen in Figure 12, is also evident for the data
subset in Figure 16. There are some irregularities in the periodicity but
these are near the transition times when few data values have gone into the
average. The hours with underpredictions are again generally during the
daytime and the hours with overpredictions are during transitional periods or
at night. These latter hours however, contain very few stations with observed
or predicted concentrations above 50 ppb, so the significance of their bias
value is not great. The largest values of average bias shown here are about
19% underpredicted at 1300 h on 3 August, and 14% underpredicted at 1500 h on
4 August. The magnitudes of these average bias values are fairly similar for
both days of simulation. Although the plots are not shown here, the time
series of bias in ROM1 layer 1 for the data subset are very similar to those
shown for layer 0.
Table 4 presents a tabulation of hourly averaged parameters for values
in the data subset. The number of surface monitoring sites at any given hour
is considerably less here than in the full data set presented in Table 3.
The maximum at any hour is 38 stations, as compared to 150 in the full set.
The average observed concentration at the receptor locations varies from 51.0
ppb at 2000 h to 78.0 ppb at 1300 h on 3 August, and from 50.0 ppb at 0400 h
and 0800 h to 80.7 ppb at 1600 h on 4 August. The corresponding predictions
for layer 0 range from 56.0 ppb at 1900 h to 70.4 ppb at 1600 h on 3 August,
and from 54.4 ppb at 0400 h to 76.2 ppb at 2000 h on 4 August. The episodic
average value of observed 03 in the data subset for all hours was 72.2 ppb
and the corresponding ROM1 predictions for layers 0 and 1 were 68.9 and 71.2
ppb, respectively. This represents an average 03 underprediction of 4.6%
for ROM1 layer 0 and 1.4% for layer 1, a slightly smaller degree of underpre-
48
-------
BIAS AVERAGE ALL RECEPTORS, LAYER ZERO
0
B jo
S
20
10
I
N
U
S
P
R
E
0
2 -'0
0
N
E -20
P
p -JO-
B
to
20 30
HOURS FROM START OF SIMULATION
i '
40
r T
50
Figure 16a. Time-series over the 3-4 August 1979 simulation period
of average bias over all surface receptor locations
(where obs. and. pred. 03 > 50 ppb).
49
-------
PERCENT BIAS AVERAGE ALL RECEPTORS, LAYER ZERO
40-
P »
E
R 10-
C
E n
N °
T-,o-
a
1 -20
A
5 -30-
-40-
-50-
1
1
0
i
,
(
i
1
^
i
20 30
4
1
I
,
,
1
,
4
1
0
i
<
i
i
i
t
... i
SO
HOURS FROM START OF SIMULATION
Figure 16b. Time-series over the 3-4 August 1979 simulation period
of average percent bias over all surface receptor
locations (where obs. and pred. CL > 50 ppb).
50
-------
TABLE 4. SUMMARY OF ROM1 HOURLY RESULTS FOR 3-4 AUGUST 1979 WHERE
OBSERVED AND PREDICTED 03 > 50 PPB.
DATE - HOUR n RcQb Rcp0 Rcpl
5dO
'1
(ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb)
79215 06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
79216 00
01
0
0
0
0
5
6
13
21
29
24
16
13
8
2
1
0
0
0
0
0
_
_
_
62.2
56.2
67.5
78.0
74.4
73.0
69.4
66.7
59.4
61.0
51.0
_
_
-
_
_
60.0
68.1
61.1
63.3
62.2
65.6
70.4
63.3
66.5
56.0
63.3
_
_
_
-
61.5
69.4
63.4
65.4
64.5
67.4
73.1
65.9
68.4
58.9
65.8
_
_
«
-
_
_
2.2
-12.0
6.4
14.7
12.2
7.4
- 1.0
3.5
- 7.1
5.1
-12.3
_
_
_
_
_
_
9.5
19.2
16.5
21.6
19.0
17.8
13.7
12.6
16.1
5.1
12.3
_
_
_
_
_
_
_
14.3
18.5
22.0
24.1
20.6
20.9
18.1
18.7
19.3
2.6
_
_
_
^
«*
0.7
-13.2
4.1
12.6
9.9
5.6
- 3.7
0.7
- 9.0
2.1
-14.8
_
.
_
_
.,
9.9
20.4
16.1
21.6
18.2
17.3
14.1
12.8
16.2
2.1
14.8
>
_
_
»
_
^
»
15.0
18.6
22.1
24.8
20.8
21.2
18.2
19.1
18.8
0.6
^
_
51
-------
TABLE 4. (continued)
DATE - HOUR n
K - K
cob Cp0
pl
"dn
>dO
l
RWT
sdl
(ppb) (ppb) -(ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb)
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
0
0
1
0
0
0
1
8
19
29
32
38
37
32
31
27
18
12
6
8
_
50.0
^
50.0
53.7
64.1
66.9
72.9
75.1
78.0
80.4
80.7
74.6
75.1
69.6
67.5
61.0
*
54.4
.
mt
73.2
68.3
70.2
72.2
74.1
71.0
67.9
69.5
70.0
69.6
71.3
71.3
76.2
75.5
_
56.8
_
85.9
72.2
72.5
74.3
76.1
73.5
70.3
72.9
72.7
72.0
74.0
74.1
77.2
77.1
- 4.4
_
_
-23.2
-14.6
- 6.2
- 5.3
- 1.2
4.0
10.1
10.9
10.7
5.0
3.8
- 1.8
- 8.7
-14.6
4.4
»
_
23.2
14.6
13.6
16.0
21.7
23.8
26.0
27.4
23.8
18.8
19.6
13.6
17.4
16.8
_
^
_
_
10.9
17.1
18.1
26.9
31.7
31.8
33.2
31.2
27.5
24.0
18.5
20.5
16.1
-6.8
>
-35.9
-18.5
- 8.4
- 7.3
- 3.2
1.6
7.8
7.6
8.0
2.7
1.1
- 4.6
- 9.8
-16.1
6.8
w
^
^
35.9
18.5
15.0
16.5
22.4
23.8
25.7
26.6
22.7
18.5
20.0
13.4
17.6
17.6
_
_
_
_
fm
_
12.3
17.4
18.0
26.9
31.3
31.7
33.1
31.0
27.7
24.4
17.9
20.1
15.7
52
-------
TABLE 4. (continued)
DATE - HOUR n RcQb Rcp0
: % Rfd^ RSd0 Rd:
(ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb)
22
23
5
7
68.0
66.7
74.9
74.0
76.1
75.3
- 7.0
- 7.3
11.8
13.0
14.0
14.6
- 8.1
- 8.6
12.0
13.5
13.7
14.8
79215 ALL
79216 ALL
138
311
70.3
73.1
64.3
70.9
66.4
73.4
6.0
2.2
17.0
20.8
21.1
27.3
3.9
- 0.3
16.8
20.9
21.3
27.2
79215-
79216 ALL
449
72.2
68.9
71.2
3.4
19.6
25.6
1.0
19.6
25.6
53
-------
diction than the values computed for the full data set. The average values
in the data subset are however, significantly greater than the tropospheric
background 03 value.
Maximum Values Analysis
The most significant parameter of interest for the air quality policy
maker concerning ambient 03 is the daily hourly average maximum value
observed at a monitoring station. This is the value which is used, among
other measures, to determine the extent of source emissions controls needed
for the 03 precursor species. The model evaluation for maximum 03 concen-
tration takes place in two steps here. First, in the local maxima analysis
model predictions of maximum hourly average 63 at surface monitoring site
locations are compared with measured values at the sites. Second, in the
global maxima analysis the concentration magnitudes and position of
plumes of 03 within the NEROS region are compared in the ROM1 predictions
and in the observations at surface monitoring sites, to the extent that
these sites can define such plumes. The global maxima analysis is necessary
because small errors in the transport field can displace the predicted
area of maximum 03 concentrations just a short distance from the observed
field, and cause large apparent underpredictions in the local maxima
analysis. If the reason for the underprediction can be traced to a shift
in position of the predicted 03 plume from the observed plume, the problem
could be corrected, in theory, by a more accurate description of the wind
field. A model's ability to predict the correct maximum concentration,
not necessarily at the right location, is the most important aspect in
performance because the air quality standards apply only to the 03
concentration level.
The local daily maximum 03 concentration, cm^x, at a measuring site
can be compared with the corresponding model prediction in two different
ways. First, the predicted maximum at the site, cm^x for RUM1 layer
0 or cm|x for ROM1 layer 1, can be compared to cm^x. Second, a
more restrictive measure matches the prediction occuring at the same hour
54
-------
as the observed maximum, with the observed concentration (Cpg(hmbX) vs.
cjgx and cpl(h|Jgx) vs. cjjjx). Figure 17 presents a scatterplot
of the daily ROM1 layer 0 maximum 03 concentrations vs. observations at the
surface monitoring sites over the 2-day episode modeled. Figure 17a contains
the plot of cjjj)x vs. cjgx and Figure 17b shows the 'plot of c 0(hmgx)
vs. CQ^X. There is considerable scatter in both plots with little evidence
of a strong correlation or trend. In fact the computed correlation coefficient
between c[Jgx and cmgx is 0.08 and that between cp0(hmgx) and c^x
is 0.00.
Looking more closely at the plots in Figure 17 it is obvious that a
large cluster of the data points corresponds to an observed concentration
range of 6U-100 ppb and a predicted range of 3U-50 ppb. These points corres-
pond to receptor locations where the model has shown essentially only back-
ground 03 concentrations and the monitors have detected increased levels
above background. This effect may be due to a misplaced plume location in
the predictions or an underestimate of the effect from precursor emissions.
The plots also show a wide scatter in the data points at all concentration
levels.
The values shown for cmajx are greater than the corresponding value of
Cp0(hx) as expected, since CPQX is a less restrictive pairing of
maximum values. The upper end of the range of values for cmQX is about 30
ppb greater than that for cpQ(hx), attesting to a phase lag between the
time of the 63 maxima predicted by ROM1 and the maxima observed at the monitoring
stations. The results for ROM1 layer 1 are not significantly different than
those shown here for layer 0.
The corresponding plots of bias versus observed maximum concentration
m^
x=mx-mx
are shown in Figure 18. The measure of bias, dmx =
is plotted in 18a and d^x=cmgx-cpU(hmgx) is plotted in 18b. The
trends displayed in these scatterplots show a fairly strong correlation between
the bias and the observed maximum concentration with increased levels of
underprediction at higher ambient 03 concentrations. Much of this trend
5b
-------
LAYER ZERO MAX VS RECEPTOR MAX
0
3 120
P 110
ft
E 100-
D
I 90
C
T
E
0
80
70
M 60
A
X 50
f
9
8
40
30
20 H
a a
0 D
a
...Q-. __.--" " , ,
20 40 60 80 100 120 140
03 OBSERVED MAX. PP8
160
Figure 17a. Scatterplot of cQ vs. c for 03 concentrations at
all surface monitoring sites on 3-4 August 1979.
56
-------
LAYER ZERO VALUE AT TIME OF OBSERVED MAX
0
3
P
R
E
0
I
c
T
E
0
100
90
80
70
60-
50
40-
P 30
P
8 20
to-
DO
DO a
°aa a D
I**TTIT
20 40
TTrp. ,r, ..... J ... rTTTTT J»TI. ...... T .... ,~rrrjrr, . . ,TI
60 80 100 120 140 160
03 OBSERVED MAX, PPB
Figure 17b. Scatterplot of c Q(hx) vs.
for 03 concentrations
at all surface monitoring sites on 3-4 August 1979.
57
-------
LAYER ZERO BIAS VS RECEPTOR MAX
B
I
A
S
P
P -
B
100
90
80
70 H
60
40
30
20
10
OH
-20-
-30-
-40-
-50-
-60-
-70
20
40
60 80 100
03 OBSERVED MAX, PP8
120
140
160
Figure 18a. Scatterplot of dggX vs. cx for 03 concentrations at
all surface monitoring sites on 3-4 August 1979.
58
-------
LAYER ZERO BIAS AT TIME OF RECEPTOR MAX
120
no
100
90
80
70
B 60
I 50
A 40
5 30
' 20
p 10
P 0
B -10
-20
-JO
-40
-50
-60
-70
OQ
20 40 60 80 100
03 OBSERVED MAX. PP8
120
140
160
"x
Figure 18b. Scatterplot of
all surface monitoring sites on 3-4 August 1979.
vs. c for 0- concentrations at
59
-------
is understood to be caused by the narrow range of background concentrations
predicted at many of the monitoring site locations that showed a wide
range of observed maxima. Again, the problem relates to the initial and
boundary 03 tropospheric background concentration imposed on the model
and the subsequent pervasiveness of this value in the predicted model
results. The bias values are correspondingly larger for d^QX than
for dg[jx because of the more restrictive pairing implied in
A summary of statistics from the local maximum values analysis for the
model simulation is presented in Table 5. Results from the full data set as
well as the data subset of 03 concentrations where CQ^, CPQ, and cpj are all
greater than 50 ppb are given. The number of receptor stations dropped from
156 in the full data set to 47 in the data subset. Interestingly, the
average observed maximum value increased by only 2-4 ppb from the data
set to the subset because there were very few observed maxima less than
50 ppb at the monitoring site locations. Conversely, the average model
predictions increased by 20-30 ppb from the data set to the subset because
of the large number of values near the background 03 level. For the
looser pairing of observed and predicted maximum values the average bias
at all raceptor sites was 24.8 ppb on 3 August, representing a 34% under-
prediction and 25.3 ppb on 4 August, representing a 33% underprediction,
for model layer 0. For layer 1 the corresponding values are 23.2 ppb
(32% underprediction) on 3 August, and 23.3 ppb (30% underprediction) on
4 August. In the data subset where the lower 03 values are excluded the
results show a marked improvement. On 3 August, the average bias for
layer 0 was 11.3 ppb (15% underprediction) and on 4 August, it was 5.6
ppb (7% underprediction). For layer 1 the corresponding values are 9.2
ppb (12% underprediction) on 3 August, and 2.7 ppb (3% underprediction) on
4 August. A listing of all surface monitoring sites and their respective
observed and predicted maximum values on both days is given in Appendix B.
The purpose of the global maximum values analysis is to attempt to
isolate individual areas or plumes of high 03 concentrations in the model
predictions and the ambient observations for comparison. The perspective
60
-------
TABLE 5. SUMMARY OF ROM1 RESULTS FOR LOCAL MAXIMUM
VALUES FOR 3-4 AUGUST 1979
Full Data
79215
n 156
R^gx (ppb) 73.2
"cjJfP (ppb) 48.4
V"S6*> (ppb) 40.0
Rcpn?x (ppb) 50.0
%l(hobX) (PPb) 41'6
Raggx (ppb) 24.8
R^AQX (PPb) 33-2
Rd$fx (ppb) 23.2
"dflf* (ppb) 31.6
RS3fx (PPb) 26.9
RsdAO (PPb) 26-8
RSj|x (ppb) 26.9
RS5fx (ppb) 27.0
Set
79216
154
77.8
52.5
43.5
54.5
45.1
25.3
34.3
23.3
32.7
33.6
32.1
33.8
32.3
Data
(C(jb> Cp0»
79215
47
75.5
64.2
60.0
66.3
62.0
11.3
15.5
9.2
13.5
21.9
21.4
21.9
21.5
Subset
cpi > 50 ppb)
79216
48
81.8
76.2
68.9
79.1
71.4
5.6
12.9
2.7
10.4
32.6
30.9
32.3
30.7
61
-------
in this analysis is broader than that in the local maximum values analysis
in that model predictions at the same location as the observation are not
necessarily required. Instead, the eligible area from which the model pre-
diction is chosen is defined to be the coherent region of concentrations,
or plume, from which the observation comes. This less restrictive pairing
permits us to match observations and predictions based on supposedly simi-
lar phenomenological events in the physical processes producing the maximum
03 values. This perspective also permits us to include a larger area than the
single site location where the maximum value occurred.
A number of problems arise in the implementation of this analysis.
The number and spatial distribution of surface monitoring sites for 0^ is
not ideal for defining plumes over the NEROS region (see Figures 1,4, and 5).
The sites are primarily located in and very close to urban areas, limiting
the ability to define the downwind extent of an urban plume. We shall rely
on the interpolated wind field used by ROM1 to help define the projected
plume location from major source areas in the NEROS region. We desire to
include areas that show consistent plume transport as well as numerous
monitoring sites near the projected plume path.
Figure 19 shows the 12-hour projected forward trajectories starting
at 0600 h on 3 August from some of the major urban source areas in the
NEROS region. The stars indicate the starting location and the dots are
the monitoring site locations. The trajectories are based on the ROM1
layer 1 wind field. The analogous plot for 4 August is shown in Figure
20. On the first day the trajectories from cities in the western part of
the grid show fairly consistent directions while in the east the influence
of the offshore stationary front on the wind flows is evident by the
varying directions, especially in the New York and Philadelphia trajectories
which show recirculation patterns. On 4 August the eastern trajectories
all appear from the NNW, except for Boston which shows a"slower flow from
the SE. The western trajectories are from the west and SW.
62
-------
3
t
.
\
S!
s,
^
5;
p^
^'
^
^
^
*
c
h-
a
k,
h
'
»
s
t
3
^
t
f
»
*
5;
i»
*
y
B
9
\
ft
^
*
^
\
}
J
\
\
Jjj
%
rta
*
S
fl
.
(
>y
a
^
\
^
"1
t
,
M
1
«
'
,
^
Ok
^
**
*
^
,
S,
p
^
K
|k
s
k
y
^
^
i
i
^
«
1
"^
'
1
*
t
"
.
^
S
n
ii-
c
.1 ,
9
_1
J
V*
.
-
a
Fi
*
r^
t
l.s
ff
?
n
\
N
,
u^
10
^
Si
If?
«0
IS
tM
P
S
s
Jjj
g
s
"
«
j
s
"
s
*
9
«
J]
si
o
to
«D
*
o
«
«
<«
(!)
4->
(J
OJ
r-
O)
E
O
M-
«T)
i I
.
CD
£1
O
ai
^
0
o
^,
IB
i-
4J
T
2
"
^J
^
ci
i i
T3
ai
4J
O
O)
°
Q.
CT T3
C OJ
3 fO
O
10 O
r- O
CD
-H S
S -Q
O i-
o: 3
.
i i
t!
3
Ll_
63
-------
-o
O)
o
0)
4->
o
co o
T3 3
t- O
S- CO
CD
fl (O
21 -Q
O S-
o; 3
o
CM
OJ
3
CD
n
64
-------
It is immediately apparent that few of the trajectories have sufficient
monitoring sites in their vicinity to substantiate an 03 plume. Therefore,
in choosing which projected plumes to study further and determining their
spatial extent we examine the contoured field of 03 concentrations from ROM1
layer 1 on both days at 1400 h and 1600 h in Figures 21'and 22, as well as
the maximum observed values of 03 shown in Figures 4 and 5. In specific
instances the aircraft sampling measurements are useful in locating 03 plumes
also. On 3 August significant 03 plumes are projected downwind of Detroit
and Toronto and near the New York area. Observed concentrations are also high
along the south shore of Lake Erie so the Cleveland area will be included in
the analysis also. Some relatively high 03 values are observed also in the
Washington and Philadelphia areas, although the ROM1 predictions show only
minor perturbations in these areas. The variable cloud cover and occasional
rain showers may have allowed just enough solar radiation through to promote
some local 03 generation there, but the hour-averaged and spatially interpol-
ated data used by the model apparently did not allow for the same effect. The
ROM1 grid areas shown in Figure 19 with the darkened borders are the plume
areas chosen for further analysis on 3 August.
On 4 August significant 03 plumes are again projected downwind of Detroit
and Toronto. Since the greater portion of these plumes are over the Great
Lakes, comparison with monitoring data are not easily made. However, stations
near the periphery of the lakes do show higher values of 03 (Figure 5) so
these plumes will be considered for analysis. Higher 03 areas also exist
along coastal and southern New Jersey in both model predictions and observations.
These areas coincide with the projected downwind extent of the New York and
Philadelphia plumes, and they are included in the global maximum values
analysis. Relatively high 03 values (as high as 140 ppb) are observed in the
immediate Washington, D.C. area on 4 August, although the model's predictions
in this area are not much above background. The discrepancy here is unresolved
and may need more simulations to determine the adequacy of the emission data
and other inputs in this part of the grid. The areas with darkened borders in
Figure 20 are the plume areas chosen for further analysis on 4 August.
6b
-------
Figure 21a. Contours of hour-averaged predicted CL concentrations
in ROM1 layer 1 on 3 Aug. 1979 at 1400 h, LSI.
66
-------
Figure 21b. Contours of hour-averaged predicted 0_ concentrations
in ROM1 layer 1 on 3 Aug. 1979 at 1600 h, LSI.
67
-------
Figure 22a. Contours of hour-averaged predicted (L concentrations
in ROM1 layer 1 on 4 Aug. 1979 at 1400 h, LSI.
68
-------
Figure 22b. Contours of hour-averaged predicted CL concentrations
in ROM1 layer 1 on 4 Aug. 1979 at 1600 h, LSI.
69
-------
A more detailed depiction of the Toronto plume on 3 August is given
in Figure 23. The projected 03 plume is evident in the contoured model
predictions of daily maximum 03 downwind of the center location of Toronto,
indicated by the "+". The surface monitoring sites and the maximum con-
centration observed at each are also shown. Two of the downwind sites ob-
serving maximum 03 between 80 and 100 ppb show excellent agreement with model
predictions at those locations. Another site located in the area of the
projected plume appears to have an anomalously low maximum value of 03 of 55
ppb. This site may have been directly affected by a nearby source of NOX
emissions during the day. There were no surface monitoring stations in
the vicinity of the projected plume maximum of 144 ppb. Note that the
maximum values observed at the stations nearest to the central urban area
show considerable variation at close range to each other. This is not
unexpected given the heterogeneous emissions patterns of a large urban area.
The expected variance in concentrations within a grid cell is estimated with-
in the formulation for model layer 0 and will be evaluated with field data
using the second generation regional model, ROM2.
In estimating model performance in the global maximum perspective we
define three different comparisons of model predicted concentration with
the maximum observed concentration, cx, in the 03 plume. First, there
is the predicted maximum at the same location as the monitoring site of the
observed maximum. Second, there is the maximum predicted concentration
at any of the monitoring site locations in or near the projected plume, and
finally there is the maximum concentration found in the grid cell at the
center of the projected 03 plume. These three values are designated (for
ROM1 layer 1) as c^A, c^g, and c^c, respectively and proceed from
the most to the least strict pairing of concentrations in the global
maxima analysis. For the Toronto plume on 3 August, Cpi*A =
cpl-B = HO ppb at site ORO since the model -predicted maximum 03
occurred at the same site as the observed maximum, and Cpfx.r, = 144 ppb
at cell location (23,38). There are also corresponding values for ROM1
layer 0 that are not discussed here but will be tabulated later. The
measures of bias resulting from these pairings are
70
-------
40
39
38
37
36
35
34
1 T
70 * 41 TOR
62
I
I I I
TORONTO
3 AUG. 1979
80 ^-100
100
80
I I
Cm.3X = 98 (ORO)
ob
CpT-A =11° (ORO)
CpT-B =11° 38)
I I
J I
18 19 20 21 22 23 24
Figure 23. Predicted (ROM1 layer 1) and observed maximum CL concentrations
(ppb) downwind of Toronto on 3 Aug. 1979.
71
-------
and dmi*c and have values of -12, -12, and -46 ppb, respectively,
for this plume.
Figure 24 presents the time series of 03 concentrations for station
ORO on 3 August. It is evident that the predicted concentrations do not
fall to the low levels that the monitoring stations observe for the dark
hours. The predicted values shown here are for model layer 1, although
layer 0 values are not very different. Also, the peak value is predicted
earlier in the day than that observed at the station. A difference in time
between the predicted and observed peak 03 values is not uncommon in photo-
chemical dispersion models and is often the result of the chemical kinetics
in the model being more or less reactive than the existing atmospheric
case. Transport and dispersion of the precursor emissions also, of course,
can affect the timing of the peak at a given location. In this case the
model predicts the maximum 03 concentration within the plume to occur later
in the day at 17UO h downwind of ORO. That is closer to the time of the
observed maximum at ORO.
The case of the Detroit and Cleveland areas are considered together in
Figure 2b. The ROM1 does not develop an 03 plume downwind of Cleveland on 3
August as can be seen by the nearly flat gradient in concentrations. Surface
observations to the northeast however indicate two stations downwind, near
the Lake Erie shoreline, showing maxima of 159 and 95 ppb. As discussed in
Section 2, these higher concentrations are not the result of photochemical
generation from Cleveland, but rather reflect the previous day's 03 in an aged
plume from Detroit or elsewhere that had been advected over Lake Erie and
fumigated to the surface during the morning hours. Both the time-series
records of surface concentrations at these stations and aircraft observations
of 03 aloft made by the WSU aircraft (flight 8 - see Figure 6) later in the
day on 3 August over the eastern end of Lake Erie and south of Buffalo help
substantiate this. Also, the maximum concentration in the model-predicted
plume downwind of Detroit was 158 ppb on 3 August, very close to the value
observed at the CNE station on the south shore of Lake Erie. Presumably if
the maximum 03 generated in the Detroit plume from the previous day was of a
-------
I I I
I I I I
O
o:
O
CO
LO
CN
Q
o
in
, in
I1II CM
o-
o -
o
CM
I I I
J I
o
o
o
m
qdd
O-
I I II
CO
Ld
to
O
c
s_
01
1/5
(O
S-
O)
(O
CD
i I 3
s: <:
O
o: co
' c
T3 O
OJ
P O
O C£.
r- O
T3
O) 0)
S- +->
Q.T-
CD CD
OJ S-
i- O
S- +->
cu -
00 £
ii
CNJ
-------
31
30
29
28
27
26
25
24
23
22
21
20
CLEVELAND
3 AUG. 1979
Cmax =159 (CNE)
ob
CpT*A = 48 (CNE)
DETROIT
3 AUG. 1979
cmax
ob
(SR2)
= 90 (SR2)
= 122 (LND)
= 158(10.28)
0
= 53 (CL2)
= 58 (10,22)
i , I
6
10 11 12 13 14 15 16
Figure 25. Predicted (ROM1 layer 1) and observed maximum CU concentrations
(ppb) downwind of Detroit and Cleveland on 3 Aug. 1979.
74
-------
similiar magnitude, the more westerly winds on 2 August would have carried
the plume over Lake Erie and to the vicinity of these south shore stations.
The test simulation discussed here was initialized with clean tropospheric
concentrations of all species and thus does not address the issue of
carry-over of concentrations from 2 August. The Cleveland area is therefore
dropped from further consideration because the model predictions and the
surface monitors do not support plume generation here.
The ROM! 63 predictions clearly show a plume emanating from the Detroit-
Windsor area northeast into Ontario on 3 August. The maximum predicted
concentration within this plume, cma^£, is 158 ppb at grid cell (10,28)
at 1700 h, 1ST. The one surface station closest to the area of maximum
predicted concentrations shows a measured maximum of only 64 ppb, while
the model-predicted 03 value for the site is nearly twice as much. Stations
along the northern edge of the plume show somewhat closer agreement. For
instance, the maximum value observed in the projected plume area, CQ^X, was
at Sarnia, Ontario (SR2) at 100 ppb and the maximum predicted value at that
site was 90 ppb, a 10% underprediction. The time-series of observed and
predicted 03 concentrations for SR2 is shown in Figure 26. The observed
peak is attained rather early in the day, at 1200 h, indicating the like-
lihood of even higher concentrations farther downwind. The model predict-
ions again stabilize at higher levels during the nighttime hours than do
the observations.
The final plume analyzed for this day is that for New York. The
projected 12-hour forward trajectory for New York shown in Figure 19
indicates some looping and recirculation in the air parcels in the vicinity
of the stationary front lying very close to New York. This introduces a
larger degree of uncertainty in establishing an accurate trajectory path,
but we do attempt to locate an 03 plume here because observed concentrations
were relatively high at several surface monitoring stations to the northeast
of New York. Figure 27 shows the contoured predictions of maximum 03
concentrations along the projected trajectory path from New York which
has been expanded and lengthened somewhat for this analysis to allow for
75
-------
I I I I I I I I I I- I I I
I I I I I I I I I
co
c
o
(O
(II
u
c
o
o
CO
o
T3
s-
Ol
CO
a
03
s-
d)
01
(O
01
I 3
-a o
OJ
4-> CNJ
u Q£
r- OO
-o
(U O)
S- -4->
CL'i-
co
<4-
o cn
c
CO <
CD S-
^ O
i. -M
0) !-
CO C
I O
i- 4J
I <0
a>
s-
qdd
76
-------
15
NEW YORK
3 AUG. 1979
cmax
OD
= 97 (GRW)
= 97 (GRW)
= 117 (42,19)
40 41 42 43 44 45
Figure 27. Predicted (ROM1 layer 1) and observed maximum CL concentrations
(ppb) downwind of New York on 3 Aug. 1979.
77
-------
the greater uncertainty in the looping path. An 03 plume is observed in the
predictions to the northeast of New York with the maximum, C^^Q, of 117 ppb
at grid cell (42,19). Fortuitously there are several monitoring sites in the
vicinity of the predicted plume, and it appears that its predicted location is
quite good. The maximum observed concentration, cm^x, occurred at GRW in
Connecticut, located in the same grid cell as cm^o Tne magnitude of the
observed maximum was 130 ppb, implying a 10% underprediction when cx is
paired with cmf*c. The model prediction at GRW, cmf*A (and cmfxB), is 97
ppb, giving a 25% underprediction with the more strict pairing. A station at
the northern edge of the projected 03 plume shows a maximum concentration of
120 ppb while the prediction is around 60 ppb. Apparently, while the center
portion of the plume has been well-located the breadth of the high concentration
area has been underestimated. Figure 28 presents the concentration time-series
at station GRW on 3 August 1979. The timing of the peak concentration coincides
within one hour between observed and predicted 03 maxima in accordance with
the well-located predicted plume position. The temporal concentration gradients
are smaller in the ROM1 predictions at this location than the observations
and, in fact, the predicted 03 concentration appears to level off into the
evening hours while the surface monitor indicates a rapid depletion of 03
toward evening.
Another interesting aspect of the concentration field near New York
occurs to the south and southwest of the city where relatively high 03 values
(up to 123 ppb) are observed at monitoring stations there. The initial path of
the trajectory from New York is in a southerly direction before looping to
the northeast so it is not entirely surprising to find higher concentrations
there. The model predictions, however, do not exhibit this secondary peak to
the southwest, possibly because the model has advected the precursor emissions
farther along the trajectory before photochemical generation could get well
underway.
Over the evening and nighttime periods from 3 August into 4 August, the
ROM1 advects the predicted 03 plume from Detroit east and northeast over Lake
Erie and eventually to the south of Toronto over Lake Ontario. The more stable
78
-------
i i i
o
II
z:
o
5
t/)
LO
CN
O)
r^
II
£
Q
o -
o -
o
o
i i i
i i i
in
CN
LJ
o
m
O
O
o
in
c
o
(O
O)
o
c
o
o
CO
o
o
OJ
>
s_
(U
s- r^
OJ CTl
qdd
ro
en
-H 3
s: <
o
Of. CO
0)
+-> 3
o a:
<-
d) -r-
(^ C
0) 1
CO
CM
a>
en
u_
79
-------
atmosphere over the lake waters is modeled by ROM1 and manifests in allowing
the concentrations within the now-aging Detroit plume to remain rather high.
The Toronto forward trajectory for 4 August shows a movement, to the east of
the urban area. Contours of the maximum predicted hourly 63 for model layer
1 downwind of Toronto are shown in Figure 29. The highest predicted values,
in excess of 220 ppb, are east of Toronto over Lake Ontario. The Toronto
plume on 4 August is shown by the simulation to have been transported into
the area of the day-old Detroit plume over the lake. The synergism between the
new and older plumes has produced rather high predicted levels. None of the
surface monitoring stations is located in a position to validate the highest
03 levels. Site ORO, northeast of Toronto, had an observed maximum 03 value
of 101 ppb on 4 August. The ROM1 predictions show this site location to be just
on the periphery of the 03 plume, and the time-series of concentrations here,
shown in Figure 30, clearly shows not only an underprediction, but also little
correspondence in the temporal pattern. The observations appear to show that
ORO was more directly affected by the Toronto plume than the predictions
indicate. The maximum predicted concentration at ORO was 66 ppb.
Examining the monitoring sites upwind of Toronto might indicate the
existence of the transported Detroit plume on 4 August. The results here are
mixed. Two sites near the lakes, St. Catherines in Ontario and Niagara Falls
in New York, indicate maximum 03 values less than 60 ppb. The other sites
show a range of 76-92 ppb, certainly more than background but not at the 100
and over ppb level of the predictions. It appears that the effect of the
older transported plume has been overemphasized by the model. The highest 03
prediction occurring at monitoring site locations, Cpf^, was 107 ppb at
STC and NG2, the same two sites that showed the lower maximum observed values.
WSU conducted several aircraft transects on 3-4 August that approached
the area of interest of the Toronto-(old) Detroit predicted plume. The shaded
grid cells in Figure 29 are those where the WSU aircraft-03 monitor observed
values between 100 and 120 ppb at about 1000 m above ground between the hours
of 2000 on 3 August, and 0000 on 4 August. It thus appears that an older
plume was advected into this area aloft, although judging from the sampling
80
-------
= 101 (ORO)
= 66 (ORO)
= 107 (STC.NG2)
= 227 (24,35)
TORONTO
4 AUG. 1979
31
18 19 20 21 22 23 24 25 26
Figure 29. Predicted (ROM1 layer 1) and observed maximum 0. concentrations
(ppb) downwind of Toronto on 4 Aug. 1979.
81
-------
O
o
II
z
g
CO
(N
CD
r--
II
I I
I I
o
in
o
o
o
o
I I I
o
o
I II
J I
I I
I I l I
O
CM
LJ
o
o
qdd
O)
u
O
O
CO
O
(U
>
(U
t/1
JD
O
s_
O)
01
O)
O
o:
o o
O)
-(-> O
o a:
f O
T3
0) (U
S- (->
Q.-I-
V)
4-
O 01
C
(/> 1-
o> s-
r- O
S- 4->
0) -t-
«n c:
i o
fO
O
CO
1-
3
r-
U.
82
-------
time and location this may have been the remnants of a plume from the Detroit
area on 2 August. The surface monitoring site at Monroe, New York, near
Rochester, showed maximum hourly 03 concentrations of 61 ppb on 3 August and
69 ppb on 4 August, while the WSU aircraft sampled 03 as high as 127 ppb aloft
in the same vicinity during the night hours spanning 3-4 August. It appears
that the effects of the plume aloft were not entirely felt at the surface at
this particular site. If the higher observed concentrations aloft are, in
fact, from the Detroit plume of 3 August, the ROM1 has predicted its location
somewhat to the north of its true position.
The forward trajectory from Detroit on 4 August travels to the east of
the city and later takes a northeast turn. The axis of the maximum predicted
03 concentration area from Detroit, as seen in Figure 31, is oriented southeast
from the city over Ontario and western Lake Erie. The maximum concentration,
cpl-C» 1n t'1e Plume 1S J-62 ppb at grid cell (8,26). The surface monitoring
stations showing the largest hourly level of 03 are SR2 and PTR, northeast of
Detroit, where 99 ppb was observed. There appears to be a bifurcation in
the predicted plume with a secondary maximum just west of station PTR. The
maximum 03 concentration predicted at the PTR site, cm^B , is 88 ppb, an 11%
underprediction. The time-series of observed and predicted 03 concentrations
at PTR is shown in Figure 32. The hourly observed values show a bimodal
peak with the greater value occuring at 1800 h, LST. The predicted pattern
of values is smoother with the peak occurring at 1600 h, LST. There are
only 2 monitoring stations south of Detroit, near the periphery of the
predicted plume. At WNS and DT4 the predicted 03 peak values were both 94
ppb, while the observed peaks were 39 and 74 ppb, respectively. The wide
variation in values at these near-by stations attests to their central urban
locations. No surface monitoring sites lie close to the central area of the
predicted plume.
The RTI aircraft, however, flew a sampling transect' in a north-south
line during the afternoon of 4 August, and measured elevated 03 concentrations
in the exact vicinity of the predicted Detroit plume. The specific flight is
labeled F in Figure 7, and corresponds to a north-south run along column 8 of
83
-------
31
30
29
28
27
26
25
24
I I
DETROIT
4 AUG. 1979
80
60
68
max
ob
max
P1 -A
max
P1 -B
max
PI - C
= 99 (SR2.PTR)
= 88 (PTR)
= 94 (WNS,DT4)
= 162 (8,26)
120
10
Figure 31. Predicted (ROM1 layer 1) and observed maximum CL concentrations
(ppb) downwind of Detroit on 4 Aug. 1979.
84
-------
I I I i I I I I i I I I I
_, . If)
T CM
CL
I!
~ZL
a
s
CO
CD
CN
CD
r^
II
5
Q
I I
I I I I I
I I I I I I I
o -
1111
O
CM
CO
o
o
O
(O
a;
u
o
u
oo
O
cu
>
a)
en
JO
o
-o
rO
1. r*^
a)
en
3
<=£
c
-o o
a>
-p o:
O h-
r- CL.
o
a) ai
s- (->
Q.-I-
00
r- O
S- +J
a; -
w c
i o
0) E
"3
CM
CO
3
cn
qdd '
85
-------
cells in the RUM1 model domain. The shaded cells in Figure 31 correspond to
the locations where the aircraft sampled 03 above the 60 ppb background level
that existed on either side of the plume. Within the shaded cells the integ-
rated observed values along the aircraft trajectory were about 90 ppb in cell
(8,25), 105 ppb in cell (8,26), and 110 ppb in cell (8,27), although individual
03 peak values within this sampling area reached 150 ppb. The model predictions
and the aircraft observations are in rough agreement here in terms of plume
location, width, and approximate 03 concentrations. The maximum level predicted
in the model (162 ppb in grid cell 8,26) occurred at 1600 h, 1ST while the
aircraft sampling in the vicinity occurred at 1400 h, 1ST.
The 03 plumes from the New York and Philadelphia areas on 4 August are
shown in Figure 33. The forward trajectories for that day predict the plumes
to be located SE of Philadelphia and SSE of New York. Maximum predicted 03
concentrations in those areas are 135 ppb downwind of Philadelphia and 114
ppb downwind of New York. There is also a maximum concentration area to the
east of New York that is not in a downwind direction of the city. Recirculation
effects from the looped trajectory out of New York on the previous day have
most likely had an additive effect with fresh emissions to the north of New
York on the current day to produce this maximum concentration area off Long
Island. The only surface station within this predicted plume is BAB and the
maximum level recorded there was 58 ppb. There is no evidence that there
was, in fact, a plume in this vicinity. The meteorological conditions of
clouds and showers here on 4 August also do not support the existence of a
plume. The interpolated field of cloud cover however did show some anomalous
areas in the data-sparse regions over the ocean and these areas may have
underestimated the cloud cover to the east of New York going into the ROM1.
This would lead to an overestimate of photochemical reactivity.
The surface stations showing the highest maximum hourly 03 concentrations
near New York lie southwest in New Jersey. A maximum value of 144 ppb was
recorded at site MER, while the maximum layer 1 prediction here was 59 ppb.
Figure 34 presents the time-series of predicted and observed concentrations
at this site. Either the predicted plume is badly positioned or recirculation
86
-------
18
17
16
15
14
13
12
11
10
9
8
7
6
5
NEW YORK
4 AUG. 1979
1201
23*
= 144 (MER)b52
NV+
= 59 (MER)
= 88 (BAB)
71«
50
111
123*
125*
130
90
I PHL,
80| 100
[1008060 [ I
PHILADELPHIA
4 AUG. 1979
max
ob
max
P1 -A
max
P1 - B
max
P1 -C
l
= 132
= 101
= 130
= 135
I
{VIN)
(VIN)
(AC1)
(38,9)
J
35 36 37 38 39 40 41 42 43 44 45
Figure 33. Predicted (ROM1 layer 1) and observed maximum CL concentrations
(ppb) downwind of New York and Philadelphia on 4 Aug. 1979.
87
-------
I I I I
o:
LU
II
z
g
5
CO
CD
CN
O>
r^
II
^
Q
I r
m
1I1II CM
O
O
O
o
O
I I I
I I I
CO
o
O
O
m
(/I
c
o
as
0)
o
CO
o
-o
Ol
>
cu
t/5
-Q
o
qdd
o
o; <*
c
T3 O
a;
(J LU
- S
T3
O) (1)
i- -U
Q.-I-
co
M-
o cr>
c
in <
O) S_
i- O
i- +->
O) -r-
10 C
I O
-------
of previous day's pollutants from Philadelphia have had an effect in this area
also. It is difficult to specify because of the complex wind situation ex-
isting here during the period. Evidence of recirculation also exists upwind
of Philadelphia where surface stations recorded 03 maxima up to 100 ppb on 4
August. Downwind of Philadelphia surface monitors do indicate an 63 plume in
the vicinity of the predicted maximum. Site VIN in south-central New Jersey
recorded a maximum concentration of 132 ppb on 4 August. The relatively
early hour (1300 h) at which- this concentration occurred may indicate that
still higher concentrations existed later in the day downwind of the site.
The highest concentration predicted at VIN, cm|*A> was 1U1 ppb, while the
model predicted a maximum, cmf*c, of 135 ppb at cell (38,9) less than 40 km
ESE of VIN. Site AC1 is located very close to the position of the ROM1 pred-
icted maximum and shows the highest predicted 03 maximum at a surface monitoring
location, cm|^g, at 130 ppb. However, the observed maximum here on 4 August
was 73 ppb. Therefore, while the position of the predicted plume appears
fairly good, the area of the greatest concentrations is somewhat misaligned.
The time-series of predicted and observed 03 concentrations at VIN is shown
in Figure 35.
Table 6 presents a summary of statistics from the global analysis of 03
plume maximum values discussed here. Seven plumes were considered, not
including the Cleveland area on 3 August. The average value of CQ^X recorded
at surface monitoring sites in these plumes was 114.9 ppb. Three methods of
pairing this observed value with ROM1 predictions were made. From the most
restrictive to the least restrictive pairing the corresponding layer 1 model
predictions averaged over the 7 cases are cma^ = 87.3 ppb, cm^B = 106.9 ppb,
and cma^Q = 151.0 ppb. The last predicted value is not restricted to measure-
ment site locations, but can be at any grid cell within the predicted plume
area. This measure has the most potential for overprediction, and in fact we
would expect this to be the case because the density of measurement networks is
rarely sufficient to capture the true 03 peak value. The average bias for
this pairing is -36.1 ppb, implying an average overprediction of about 31%.
Note that the table shows the average of the normalized bias values over the
individual days to be about 38%. The difference represents somewhat different
89
-------
i r i r
i r i i
IIIII CM
II
Z
o
I
CO
(£>
CM
CT>
r^
II
£
O
L I I I
o -
1 I 1
o
CM
CO
LU
o
m
o
o
O
m
qdd
IS)
c
o
to
O)
o
c
o
o
CO
o
o
0)
>
cu
,r-H
to
*"" O)
i I 3
S <
O
o: =!
T3 O
0)
4-> 2:
CJ i i
r- 5>
T3
O) d)
S- 4->
Q.T-
O CD
c
CO -r-
QJ i-
- O
S_ 4->
CU -r-
CO C
I O
|E
r- 4->
CO
3
cn
90
-------
TABLE 6. SUMMARY OF ROM1 RESULTS FOR GLOBAL MAXIMUM VALUES
FOR 3-4 AUGUST 1979.*
DET
3 Aug
TOR
3 Aug
NY
3 Aug
DET
4 Aug
TOR
4 Aug
NY
4 Aug
PHL
4 Aug AVG S.D.
..max
cob
(site-hr)
cmax
cpl-A
(site-hr)
dpT-A
Idpl-Al
(^!-A/cSbx)xlao
CpT-B
(site-hr)
dp?-B
ldpl-8l
(d^vX)*100
cpf-C
(grid cell)
(hr)
G i P
p X *L
Idpf-Cl
(dpi-c/cSbx>*100
100
(SR2-12)
90
(SR2-16)
10
10
10%
122
(LNO-21)
-22
22
-22%
158
(10,28)
(17)
-58
58
-58%
98
(ORO-16)
110
(ORO-14)
-12
12
-12.2%
110
(ORO-14)
-12
12
-12.2%
144
(23,38)
(17)
-46
46
-46.9%
130
(GRW-14)
97
(GRW-15)
33
33
25.4%
97
(GRW-15)
33
33
26.4%
117
(42,19)
(16)
13
13
10%
99
(PTR-18)
88
(PTR-16)
11
11
11.1%
94
(WNS-10)
(DT4-10)
5
5
5.1%
162
(8,26)
(16)
-63
63
-63.6%
101
(ORO-15)
66
(ORO-10)
35
35
34.7%
107
(STC-18)
(NG2-19)
-6
6
-5.9%
227
(24,35)
(17)
-126
126
-125%
144
(MER-16)
59
(MER-16)
85
85
59%
88
(BAB-11)
56
56
38.9%
114
(41,15)
(17)
30
30
20.8%
132
(VIN-13)
101
(VIN-19)
31
31
23.5%
130
(AC1-15)
2
2
1.5%
135
(38,9)
(15)
-3
3
-2.3%
114.
87.
27.
31.
21.
106.
8.
19.
4.
151.
-36.
48.
-37.
9
3
6
0
6%
9
0
4
4%
0
1
4
9
19.
18.
30.
26.
22.
15.
27.
19.
21.
38.
53.
40.
51.
7
5
4
3
3%
2
3
5
3%
2
6
8
2
*A11 concentrations are 0 (ppb); all times are hours, LST.
91
-------
ways to calculate the average degree of variation of the predicted from the
observed maximum value. Note that if we drop the Toronto plume on 4 August
from consideration the average overprediction changes to about 23%. The very
high predicted value over Lake Ontario, out of the range of the surface
monitors, lends some support for this action.
The average bias in the most restrictive pairing where the model prediction
is interpolated to the same location as the monitoring station is 27.6 ppb
here, and the corresponding degree of underprediction is 21.6%. For the
middle case, where the maximum model prediction is restricted to monitoring
site locations in or near the 63 plume, but not necessarily at the site of
the observed maximum, the average bias is 8.U ppb with a corresponding degree
of underprediction of 4.4%.
The variance in the values of over- or underprediction among the 7 plume
cases considered is indicated by the standard deviation (s.d.) about the
average. For the 2 most restrictive pairings the s.d. is about the same in
each case, 21-22%. For the least restrictive pairing the s.d. is about 54%,
although much of that variance is attributable to the 4 August Toronto plume
case. Although layer 0 predicted values were not considered in the global
maximum values analysis the results would not have been very different from
those shown for layer 1. The layer 0 predicted maximum values were typically
only a few ppb less than the corresponding layer 1 values.
-------
SECTION 4
SUMMARY AND CONCLUSIONS
The first generation Environmental Protection Agency Regional Oxidant
Model (ROM1) has been evaluated for 03 concentrations using the 2-day test
period, 3-4 August 1979, from the NEROS data base. Relatively high 03 concen-
trations were measured at the surface and aloft during this period near the
Great Lakes area of the United States and southern Ontario. Cloudy and
showery conditions along the immediate East coast held 03 production to low
levels except for isolated times and locations in the period. The highest
measured surface 03 values were about 160 ppb south of Lake Erie on 3 August.
Aircraft monitoring also confirmed about the same concentration levels aloft
late on that day. An earlier study (Clarke and Ching, 1983) indicated that
this large area of 03 over the Great Lakes probably had its genesis earlier
than 3 August west of the NEROS region.
The ROM1 was initialized with a chemically equilibrated mix of tropos-
pheric background species concentrations, including about 40 ppb of 03. The
rationale behind the clean initialization was that the model would start
simulation sufficiently ahead of the time of prime interest so that source
emissions within the model domain would have time to accumulate and participate
in photochemical smog reactions producing 03. A relatively long lead time is
required with this methodology. We have used 6 hours in this 2-day test
episode, but 24-48 hours of lead time would be preferable in a longer period
of simulation. This lead time is a buffer during which the initial conditions
are advected out of the system and model results more properly reflect the
outcome of physical and chemical processes occurring within the model domain.
The purpose of this is to minimize the effect of initial conditions on model
results and allow us to initialize with a spatially homogeneous field of
background concentrations. If this were not the case, an initial concentration
field accurately reflecting the existing ambient concentrations would need to
be established. This is a particularly difficult task to accomplish on the
regional scale with relatively sparse data observations.
93
-------
In retrospect, the 2-day episode modeled here was not long enough to
justify initialization with clean conditions. Apparently the aged 03 plume
existing in the western model domain at the start of model simulation, but
not represented in the initial concentration field, caused the ROM1 to overlook
a major contribution to the high concentrations observed in the southern
Great Lakes area during the first day of simulation. This preliminary model
evaluation exercise is therefore more a test of the procedures that will be
used in later evaluations of the second generation model, ROM2, than it is of
the ability of the current model to accurately simulate plumes of 03 on the
regional scale. Future, evaluations will simulate over an extended period of
time (up to 4 weeks), circumventing the initial condition problem.
The evaluation of ROM1 on the test episode proceeded in two stages.
First, an overall analysis of 03 observations and predictions at receptor
monitoring locations in the model domain was made for all hours of the
simulation except the first six. Next an analysis of the model to predict
daily maximum values of 63 was performed. In the first stage where all data
were involved, results showed that the ROM1 predictions averaged under 6%
less than measurements at monitoring sites. There is, however not much
significance to this result since averaging data over all monitoring stations
for all hours weights the result near the tropospheric background level. We
are more concerned with the model's ability to predict higher values of 03.
Therefore in a second part to the first stage of analysis the same manner of
evaluation was applied to a subset of the data where only simultaneous 03
values above 50 ppb were included for observed and predicted concentrations
at monitoring sites. In this case the average observed 03 concentration over
all receptors and hours was 72.2 ppb, above the level of general background
03. The average ROM1 performance for the data subset showed a 4.6% under-
prediction for layer 0 and a 1.4% underprediction for layer 1. Beyond
these averaged statistics, the graphical analyses used in interpreting model
evaluation results were extremely useful. These analyses included histograms
of observed, predicted, and residual data as well as time series of the model
bias. The inherent structure of the data "as well as trends in model bias can
be seen from these type plots.
94
-------
The second stage of model evaluation, using daily 03 maxima, also
proceeded in 2 steps. In the first step the daily maximum value at each
receptor location was compared to the predicted maximum at that location as
well as the prediction at the same time as the observed maximum. For layer
1, the ROM1 performance showed an average 31% underprediction for the first
comparison and a 43% underprediction for the second comparison. Results for
layer 0 were similar. When data were restricted to those observed and pre-
dicted 03 pairs above 50 ppb the model performance for the local maximum
analysis improved by showing average underpredictions of 8% and 15%, respect-
ively for the above layer 1 comparisons. The improvement in model predictions
in this case is principally due to restricting the analysis to comparisons at
monitoring station locations where source emissions had a larger effect on 63
levels than material that was present initially and overlooked by the clean
initialization procedure.
The second step of evaluating the ROMl's performance for 03 maxima
predictions involves an analysis of individual plumes of 03 from major source
areas within the NEROS region. This is perhaps the most challenging and
potentially, the most rewarding aspect of the model evaluation for it ad-
dresses the major aspect of the regional model's function, resolving plumes
of pollution from urban scale and smaller sources on a regional scale grid.
A difficult aspect of this step in the evaluation is determining the location
and magnitude of the source plumes from a monitoring network that was not
designed specifically for regional scale purposes. Monitors are most often
located close to urban areas and downwind spatial coverage is usually very
limited. Nevertheless, 7 cases of such plumes were identified and analyzed
for the 3-4 August 1979 episode. The evaluation used three types of compa-
risons of 03 maxima; only layer 1 model predictions were considered in this
part of the analysis. The first comparison matched the observed daily maximum
03 level at any monitoring site within a plume with the daily maximum 03
prediction at that site. The second comparison matched the observed 03
maximum with the predicted maximum at any site within or near the plume, and
the third comparison matched the observed maximum with the predicted maximum
at any grid cell within or near the plume. Results averaged over the 7
95
-------
plumes in the test case showed a 22% underprediction for the first comparison,
a 4% underprediction for the second comparison, and a 38% overprediction for
the third comparison. In -interpreting these results one must bear in mind
that the monitoring network is probably not sufficently dense to capture the
actual ambient maximum 03 value within a plume, and also the predicted plume
location and magnitudes of 03 may not be quite accurate.
There are numerous reasons why the ROM1 may not have performed well for
a given plume. We have already discussed the fact that the model was
initialized with precursor species concentrations considerably below ambient
conditions for some areas of the grid. Another condition likely to cause
problems in model performance is a wind field with poorly defined organization.
This can lead to large uncertainties in the transport and dispersion of a
plume. A third condition under which the ROM1 may perform poorly is that of
"patchy" cloudiness, or large spatial variability in cloud cover. This
condition has the potential for calculated errors in the amount of sunlight
available to the photochemical precursor species for 63 formation. Examining
the ROM1 performance for 63 in light of these 3 conditions may provide an
interesting perspective.
Both the surface data and aircraft sampling have shown that the location
and magnitude of the predicted 03 plume from Detroit on 4 Aug. is fairly
accurate. The wind field in this part of the domain is well-organized with a
predominant flow from the WNW during the early part of the day, shifting to
more westerly and WSW directions later. Likewise, the skies were mostly
clear throughout this part of the simulation period here. An examination of
the ambient concentration levels of 03 and precursor species on 2 Aug. near
and upwind of Detroit does not show the presence of high concentration levels,
indicating that the assumption of clean tropospheric initial concentrations
was probably not in gross error. We might speculate then that the good
performance of ROM1 in this case is not due to chance, but rather because
certain identifiable physical conditions that are associated with the like-
lihood of consistent model behavior have been met. The concurrent aircraft
sampling of 03 in this plume with high levels of photochemical activity
96
-------
provides a definitive measure of the model performance. Similar conditions
of steady flow and mostly clear skies also existed with the Detroit plume of
3 Aug. and the Toronto plumes on both days. Model performance for 63 in
these plumes was also quite good at the monitoring station locations.
The flow field along the East coast was much less organized during the
period than it was in the Great Lakes area because of the presence of a
stationary front. Highly variable cloud cover was also seen to exist there.
The rather poor performance of the model in the vicinity of the New York
plume on 4 Aug. is not surprising in view of this. The predicted plume
location and magnitude is not supported by any of the monitoring data. Model
performance for the New York plume on 3 Aug. is mixed, with apparently
reasonable agreement to the northeast in Connecticut and poor performance to
the south and southwest. The looping trajectory of this plume, shown in
Figure 19, indicates some of the disorganization of the flow. The fairly
good agreement shown between 03 observations and predictions in the Philadelphia
plume on 4 Aug. is one case that may be ascribed to chance as several of the
conditions described above associated with potentially poor model performance
were in effect. Also, model performance for 03 was seen to be poor for the
first day of simulation where heavier 03 or precursor concentrations were
relatively high on 2 Aug., such as to the south of Lake Erie, between Cleveland,
OH and Erie, PA and also on the East coast between New York and Philadelphia.
Overall, the performance of ROM1 for 03 on this test episode is quite
good. Judging from the performance of the model in the vicinity of the
analyzed 03 plumes for maximum 03 concentrations the model performance is
within the range of 4-22% underprediction, corresponding to the 2 stricter
methods of paired comparison between model prediction and observation. The
least strict pairing is more appropriate for a denser monitoring network, but
it is useful for examining the full predictive photochemical potential from a
source emissions region. The model results for prediction of maximum 03 at
receptor sites where both the observed and predicted values were over 50 ppb
were also quite good, in the range of 3-15% underprediction. These are the
sites most likely to have been affected by plumes generated from emissions
sources within the model domain.
97
-------
In summary, the ROM1 performed well and in a consistent pattern in areas
of the domain least affected by high boundary or initial species concentrat-
ions, by large spatial gradients in meteorological factors,"and in areas most
affected by large emissions source areas within the domain. Applications
with the second generation regional model, ROM2, will address some of the
deficiencies found with this test application of the ROM1. The ROM2 applic-
ations will begin at a time of relatively clean conditions throughout the
model domain so the assumption of clean tropospheric conditions will be fair.
Also the simulation period will be one of several weeks duration to minimize
the impact of the initial conditions. The probabilistic capability of the
model's wind field processor will be explored in the ROM2 evaluation. This
capability may address some of the concerns about model performance in regions
of less organized flow.
The ROM2 will also incorporate a newer and more complete chemical kinetic
mechanism than ROM1 allowing it to include biogenic organic species in the
emissions inventory, a possibly important component in the photochemical smog
equation on the regional scale. The vertical depth of the grid cells in ROM2
will be variable in space and time and will respond to the physical phenomena
influencing the depth and stability of the boundary layer. Terrain features
and cell-specific cloud fluxes will be more accurately treated. The ability
of ROM2 to estimate sub-grid scale concentration fluctuations near the surface
will be explored. Current plans include modeling approximately 4 weeks from
the summer 1980 period of the NEROS and SAROAD data bases.
98
-------
REFERENCES
Clarke, J.F. and J.K.S. Ching, 1983: -Aircraft observations of regional
transport of ozone in the northeastern United States. Atmospheric
Environment, 17(9): 1703-1712.
Freas, W.P., 1983: Northeast corridor modeling project-data base description.
EPA-450/4-83-015a, U.S. Environmental Protection Agency, Research Triangle
Park, NC.
Lamb, R.6., 1983: A regional scale (1000 km) model of photochemical air
pollution. Part I - Theoretical formulation. EPA-600/3-83-035,
U.S. Environmental Protection Agency, Research Triangle Park, NC.
Lamb, R.G., 1984: A regional scale (1000 km) model of photochemical air
pollution. Part II - Input processor network design. EPA-600/3-84-085,
U.S. Environmental Protection Agency, Research Triangle Park, NC.
Lamb, R.G. and G.F. Laniak, 1985: A regional scale (1000 km) model of photo-
chemical air pollution. Part III - Tests of the numerical algorithms.
EPA/600/3-85/037, U.S. Environmental Protection Agency, Research Triangle
Park, NC.
Possiel, N.C. and W.P. Freas, 1982: Northeast corridor regional modeling
project-description of the 1980 urban field studies. EPA-450/4-82-018,
U.S. Environmental Protection Agency, Research Triangle Park, NC.
Schere, K.L. and A.J. Fabrick, 1985: EPA Regional Oxidant Model: description
and evaluation plan. EPA/600/3-85/068, U.S. Environmental Protection
Agency, Research Triangle Park, NC.
Vaughan, W.M., 1985: Transport of pollutants in plumes and PEPES.
EPA/600/3-85/033, U.S. Environmental Protection Agency, Research Triangle
Park, NC.
99
-------
APPENDIX A
LISTING OF SURFACE MONITORING STATIONS
This appendix contains a list of all surface monitoring sites for 03 in
the SAROAD and NEROS programs that were in operation during any part of 1979
and 1980. Ontario (Canada) sites are also included. The listing is limited
to those sites within and just outside of the ROM modeling domain for the
NEROS application. Site locations are listed both by latitude-longitude and
grid coordinates.
100
-------
SITE
AA1
AA2
AA3
AA4
AC1
AC 2
AG1
AG2
AG3
AG4
AGA
AK1
AK2
ALC
ALN
ALP
ALT
ALX
AMH
AMR
ARL
AS1
AS2
AS3
ASB
ATL
BAB
BAG
BAD
BAR
BAY
BC1
BC2
BCK
BER
BET
BF1
BF2
BF3
BGR
BL1
BL2
BL3
BL4
BL5
BL6
BL7
BLG
BN1
BN2
LOCATION
ANNE ARUNDEL CO
ANNE ARUNDEL CO
ANNE ARUNDEL CO
ANNE ARUNDEL CO
ATLANTIC CO
ATLANTIC CITY
ALLEGHENY CO
ALLEGHENY CO
ALLEGHENY CO
ALLEGHENY CO
AGAWAM
AKRON
AKRON
ALLEN CO
ALLENTOWN
ALLEN PARK
ALTOONA
ALEXANDRIA
AMHERST
AMHERST
ARLINGTON CO
ASHLAND
ASHLAND
ASHLAND
ASBURY PARK
ATTLEBORO
BABYLON
BAY COUNTY
BADEN
BARBERTON
BAYONNE
BOONE CO
BOONE CO
BRACKENRIDGE
BEREA
BETHLEHEM
BUFFALO
BUFFALO
BUFFALO
BANGOR
BALTIMORE
BALTIMORE
BALTIMORE
BALTIMORE
BALTIMORE
BALTIMORE CO
BALTIMORE CO
BURLINGTON
BERLIN
BERLIN
STATE OR
PROVINCE
MD
MD
MD
MD
NJ
NJ
PA
PA
PA
PA
MA
OH
OH
OH
PA
MI
PA
VA
NY
MA
VA
KY
KY
KY
NJ
MA
NY
MI
PA
OH
NJ
KY
KY
PA
OH
PA
NY
NY
NY
ME
MD
MD
MD
MD
MD
MD
MD
ON
NH
NH
LATITUDE
(deg, N)
39.21
39.17
38.90
39.10
39.52
39.36
40.38
40.44
40.44
40.47
42.04
41.08
41.11
40.76
40.62
42.23
40.51
38.81
42.99
42.39
38.86
38.44
38.48
38.46
40.22
41.93
40.74
43.64
40.64
41.00
40.68
38.92
39.05
40.61
41.37
40.62
42.88
42.93
42.89
44.80
39.30
39.24
39.31
39.29
39.29
39.41
39.51
43.32
44.48
44.47
LONGITUDE
(deg, W)
76.65
76.63
76.65
76.73
74.46
74.44
80.19
79.95
79.99
79.82
72.64
81.52
81.50
84.07
75.45
83.21
78.39
77.04
78.78
72.52
77.06
82.61
82.52
82.64
74.01
71.26
73.41
83.85
80.23
81.63
74.12
84.85
84.88
79.75
81.85
75.36
78.81
78.87
78.89
68.77
76.61
76.58
76.61
76.63
76.61
76.77
76.43
79.80
71.18
71.18
X-GRID Y-GRID
29.4 7.3
29.5 7.0
29.4 5.4
29.1 6.6
38.2 9.1
38.2 8.1
15.3 14.3
16.2 14.6
16.0 14.7
16.7 14.8
45.5 24.2
9.9 18.5
10.0 18.6
OUTSIDE ROM GRID
34.2 15.7
3.2 25.4
22.4 15.1
27.8 4.9
20.9 30.0
45.9 26.3
27.8 5.1
5.5 2.7
5.9 2.9
5.4 2.8
39.9 13.3
50.9 23.6
42.4 16.4
0.6 33.9
15.1 15.8
9.5 18.0
39.5 16.1
OUTSIDE ROM GRID
OUTSIDE ROM GRID
17.0 15.7
8.6 20.2
34.5 15.7
20.8 29.3
20.5 29.6
20.4 29.3
OUTSIDE ROM GRID
29.5 7.8
29.7 7.4
29.5 7.9
29.5 7.7
29.6 7.7
28.9 8.4
30.3 9.1
16.8 31.9
51.3 38.9
51.3 38.8
101
-------
SITE
BNG
BNW
BR1
BR2
BRI
BRL
BS1
BS2
BS3
BS4
BTH
BUI
BU2
BUL
BUT
BVF
CAN
CAR
CB1
CB2
CBS
CB4
CBS
CBN
CCL
CHL
CHP
CHR
CHS
CHT
CHV
CKY
CL1
CL2
CLK
CLN
CLO
CLR
CM1
CM2
CMS
CMB
CN1
CN2
CN3
CNE
CNS
CPE
CPM
LOCATION
BINGHAMTON
BINBROOK WEST
BRIDGEPORT
BRIDGEPORT
BRISTOL (BOROUGH)
BURLINGTON
BOSTON
BOSTON
BOSTON
BOSTON
BETHESDA
BURLINGTON
BURLINGTON
BURLINGTON CO.
BUTLER CO
BEAVER FALLS
CANTON
CARIBOU
COLUMBUS
COLUMBUS
COLUMBUS
COLUMBUS
COLUMBUS
CARBONDALE
CECIL CO
CHARLEROI
CHESAPEAKE
CHARLTON
CHARLESTON
CHESTER
CHEVERLY
COCKEYSVILLE
CLEVELAND
CLEVELAND
CLARK CO
CLINTON CO
GALLOWAY CO
CLERMONT CO
CAMDEN
CAMDEN
CAMDEN CO
CUMBERLAND
CINCINNATI
CINCINNATI
CINCINNATI
CONNEAUT
CONSHOHOCKEN
CAPE ELIZABETH
CAPE MAY CO
STATE OR
PROVINCE
NY
ON
CT
CT
PA
NJ
MA
MA
MA
MA
MD
VT
VT
NJ
PA
PA
OH
ME
OH
OH
OH
OH
OH
PA
MD
PA
VA
MA
WV
PA
MD
MD
OH
OH
OH
OH
KY
OH
NJ
NJ
NJ
MD
OH
OH
OH
OH
PA
ME
NJ
LATITUDE
(deg, N)
42.10
43.12
41.18
41.19
40.25
40.08
42. 3b
42.37
42.38
42.35
39.00
44.48
44.46
39.96
41.01
40.75
40.80
46.87
40.00
40.06
39.96
39.96
40.09
41.57
39.72
40.15
36.67
42.31
38.34
39.84
38.92
39.46
41.50
41.55
40.00
39.43
36.73
39.08
39.92
39.95
39.67
39.65
39.14
39.21
39.14
41.96
40.07
43.57
39.10
LONGITUDE
(deg, W)
75.89
79.88
73.19
73.18
75.00
74.86
71.10
71.04
71.03
71.06
77.11
73.21
73.19
74.79
79.72
80.32
81.38
68.01
83.04
82.98
82.99
82.99
82.96
75.51
76.11
79.90
76.33
71.97
81.62
75.37
76.90
76.63
81.62
81.57
83.80
83.79
88.12
84.18
75.10
75.12
74.86
78.76
84.55
84.48
84.51
80.57
75.30
70.20
74.82
X-GRID Y-GRID
32.4 24.6
16.5 30.7
43.2 19.1
43.3 19.1
36.0 13.5
36.6 12.5
51.6 26.1
51.8 26.2
51.9 26.3
51.7 26.1
27.6 6.0
43.2 38.9
43.2 38.7
36.8 11.8
17.1 18.0
14.7 16.5
10.5 16.8
OUTSIDE ROM GRID
3.8 12.0
4.1 12.4
4.0 11.8
4.0 11.8
4.2 12.5
34.0 21.4
31.6 10.3
16.4 12.9
OUTSIDE ROM GRID
48.1 25.9
9.5 2.1
34.5 11.0
28.4 5.5
29.5 8.8
9.5 21.0
9.7 21.3
0.8 12.0
0.9 8.6
OUTSIDE ROM GRID
OUTSIDE ROM GRID
35.6 11.5
35.5 11.7
36.6 10.0
20.9 9.9
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
13.7 23.8
34.8 12.4
55.2 33.4
36.7 6.6
102
-------
SITE
CR1
CR3
CTF
CUM
CV1
CV2
DAI
DA2
DA3
DA4
DA 5
DAN
DAY
DC1
DC2
DCS
DC4
DC 5
DC6
DMS
DNR
DNV
DOV
DOW
DRB
DTI
DT2
DT3
DT4
DT5
DUM
DVR
EAS
EB1
EB2
EDG
ELM
ENF
ERE
ESO
ESS
EST
ESX
EXC
EY1
EY2
EZ1
EZ2
FAY
LOCATION
CORUNNA
CORUNNA
CHESTERFIELD CO
CUMBERLAND CO
COVINGTON
COVINGTON
DAYTON
DAYTON
DAYTON
DAYTON
DAYTON
DANBURY
DAYTON
WASHINGTON
WASHINGTON
WASHINGTON
WASHINGTON
WASHINGTON
WASHINGTON
DAMASCUS
DONURA
DANVILLE
DOVER
DOWNINGTON
DERBY
DETROIT
DETROIT
DETROIT
DETROIT
DETROIT
DUMONT
DANVERS
EASTON
ETOBICOKE
ETOBICOKE
EDGEWOOD
ELMIRA
ENF I ELD
ERIE
EAST ORANGE
ESSEXVILLE
EASTON
ESSEX
ESSEX CO
ELYRIA
ELYRIA
ELIZABETH
ELIZABETH
FAYETTE CO
STATE OR
PROVINCE
ON
ON
VA
ME
KY
KY
OH
OH
OH
OH
OH
CT
KY
DC
DC
DC
DC
DC
DC
MD
PA
KY
DE
PA
CT
MI
MI
MI
MI
MI
NJ
MA
PA
ON
ON
MD
NY
CT
PA
NJ
MI
MA
MD
NY
OH
OH
NJ
NJ
KY
LATITUDE
(deg, N)
42.88
42.91
37.36
43.91
39.07
39.09
39.77
39.76
39.81
39.76
39.78
41.40
39.11
38.90
38.98
38.89
38.88
38.93
38.94
39.27
40.17
37.67
39.15
40.02
41.32
42.40
42.36
42.43
42.33
42.33
40.94
42.59
40.68
43.65
43.62
39.41
42.11
42.00
42.14
40.76
43.62
42.07
39.35
44.37
41.37
41.37
40.66
40.64
38.13
LONGITUDE
(deg, W)
82.45
82.45
77.59
70.23
84.53
84.51
84.21
84.19
84.19
84.20
84.20
73.44
84.48
77.05
77.02
77.01
76.97
77.06
77.03
77.22
79.86
84.77
75.50
75.71
73.09
83.24
83.10
83.00
83.03
83.05
74.00
70.98
75.22
79.58
79.52
76.30
76.80
72.57
80.05
74.20
83.85
71.09
76.48
73.90
82.11
82.11
74.21
74.21
84.47
X-GRID Y-GRID
6.2 29.3
6.2 29.5
OUTSIDE ROM GRID
55.1 35.5
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
42.2 20.4
OUTSIDE ROM GRID
27.8 5.4
27.9 5.9
27.9 5.3
28.1 5.3
27.8 5.6
27.9 5.7
27.1 7.6
16.5 13.0
OUTSIDE ROM GRID
34.0 6.9
33.2 12.1
43.6 19.9
3.1 26.4
3.6 26.1
4.0 26.6
3.9 26.0
3.8 26.0
40.0 17.6
52.1 27.6
35.1 16,1
17.7 33.9
17.9 33.7
30.8 8.5
28.8 24.7
45.7 24.0
15.8 24,8
39.2 16.5
0.6 33.7
51.6 24.4
30.1 8.1
40.4 38.2
7.6 20.2
7.6 20.2
39.1 16.0
39.2 15.8
OUTSIDE ROM GRID
103
-------
SITE
FDR
FLM
FLN
FLO
FOL
FRD
FRE
FX1
FX2
FX3
GAR
GLN
GLO
GMT
GNB
GNS
GRD
GRE
GRG
GRN
GRR
GRW
HAK
HAM
HAR
HEN
HF1
HF2
HGR
HM1
HM2
HML
HMP
HN1
HN2
HNC
HNR
HNV
HOP
HOW
HRD
HRF
HRO
HRP
HT1
HT2
HT3
JF1
JF2
LOCATION STATE OR
PROVINCE
FREDERICKSBURG
FLEMINGTON
FLINT
FLORENCE
FOLCROFT
FREDERICK CO
FREEHOLD
FAIRFAX
FAIRFAX CO
FAIRFAX CO
GARDNER
GLENS FALLS
GLOUCESTER
GREEN MOUNTAIN NAT.
GREEN8RIER CO
GENESEE CO
GARDINER
GREENFIELD
GEORGETOWN
GREENBELT
GRAND RAPIDS
GREENWICH
HACKENSACK
HAMILTON
HARRISBURG
HENRICO CO
HARTFORD
HARTFORD
HAGERSTOWN
HEMPSTEAD
HEMPSTEAD
HAMILTON
HAMPTON
HENDERSON
HENDERSON
HANCOCK CO
HENRICO CO
HANOVER CO
HOPKINS CO
HOWARD CO
HARDIN CO
HARFORD CO
HARRODSBURG
HURON PARK
HAMILTON CO
HAMILTON CO
HAMILTON CO
JEFFERSON CO
JEFFERSON CO
VA
NJ
MI
KY
PA
MD
NJ
VA
VA
VA
MA
NY
MA
VT
WV
MI
ME
MA
MA
MD
MI
CT
NJ
MA
PA
VA
CT
CT
MD
NY
NY
ON
VA
KY
KY
KY
VA
VA
KY
MD
KY
MD
KY
ON
OH
OH
OH
KY
KY
LATITUDE
(deg, N)
38.31
40.52
43.02
39.00
39.89
39.48
40.26
38.84
38.73
38.74
42.57
43.32
42.62
43.93
37.77
43.17
44.22
42.57
42.72
39.02
42.97
41.08
40.88
42.63
4U.24
37.59
41.77
41.76
39.65
40.75
40.89
43.26
37.00
37.86
37.88
37.94
37.59
37.64
37.53
39.17
37.76
39.63
37.76
43.29
39.28
39.28
39.16
38.14
38.31
LONGITUDE X-GRID Y-GRID
(deg, W)
77.45
74.81
83.69
84.63
75.28
77.23
74.27
77.31
77.10
77.08
71.99
73.62
70.71
73.03
80.33
83.46
69.79
72.60
7U.99
76.83
85.68
73.71
74.04
70.82
76.85
77.42
72.68
72.68
77.72
73.59
73.60
79.84
76.40
87.58
87.57
86.90
77.50
77.32
87.35
76.90
85.72
76.33
84.83
81.49
84.37
84.37
84.80
85.58
85.58
26.2 1.9
36.8 15.1
1.3 30.1
OUTSIDE ROM GRID
34.9 11.3
27.1 8.9
38.9 13.6
26.8 5.1
27.6 4.4
27.7 4.5
48.0 27.4
41.5 31.9
53.2 27.7
43.9 35.6
OUTSIDE ROM GRID
2.2 31.0
56.8 37.3
45.6 27.4
52.1 28.3
28.7 6.1
OUTSIDE ROM GRID
41.2 18.5
39.8 17.3
52.7 27.8
28.6 13.5
OUTSIDE ROM GRID
45.3 22.6
45.3 22.6
25.1 9.9
41.7 16.5
41.6 17.4
16.6 31.6
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
28.4 7.0
OUTSIDE ROM GRID
30.7 9,8
OUTSIDE ROM GRID
10.0 31,7
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
104
-------
SITE
JF3
JF4
JNS
JRS
KCH
KEE
KEN
KNT
KUT
LAN
LAW
LCH
LEB
LEW
LEX
LF1
LF2
LF3
LIC
LIN
LIV
LND
LOG
LOW
LPT
LSI
LS2
LV1
LV2
LV3
LV4
LV5
MAC
MAN
MAM
MAR
MCI
MC2
MCG
MCL
MDU
MDL
MDN
MED
MER
MFD
MG1
MG2
MID
LOCATION
JEFFERSON CO
JEFFERSON CO
JOHNSTOWN
JERSEY CITY
KITCHENER
KEENE
KENT CO
KENT CO
KUTZTOWN
LANCASTER CITY
LAWRENCE
LITCHFIELD CO
LEBANON
LEWISTON
LEXINGTON
LEXINGTON-FAYETTE
LEXINGTON-FAYETTE
LEXINGTON-FAYETTE
LINCOLN
LINDEN
LIVONIA
LONDON
LOGAN
LOWELL
LONG POINT PARK
LANSING
LANSING
LOUISVILLE
LOUISVILLE
LOUISVILLE
LOUISVILLE
LOUISVILLE
MACOMB CO
MAHONING CO
MAMARONECK
MARION
MC CRACKEN CO
MC CRACKEN CO
MC GUIRE AFB
MC LEAN
MIDLAND
MIDDLETOWN
MEDINA CO
MEDFIELD
MERCER CO.
MEDFORD
MONTGOMERY CO
MONTGOMERY CO
MIDDLETOWN
STATE OR
PROVINCE
KY
KY
PA
NJ
ON
NH
MI
RI
PA
PA
MA
CT
OH
ME
MA
KY
KY
KY
MA
NJ
MI
ON
OH
MA
ON
MI
MI
KY
KY
KY
KY
KY
MI
OH
NY
VA
KY
KY
NJ
VA
PA
CT
OH
MA
NJ
MA
OH
OH
OH
LATITUDE
(deg, N)
38.11
38.17
40.31
40.73
43.46
42.92
43.04
41.62
40.51
40.04
42.71
41.67
39.43
44.10
42.45
38.05
38.06
38.06
42.43
40.60
42.38
42.99
39.53
42.65
42.58
42.74
42.74
38.25
38.23
38.26
38.26
38.23
42.72
41.11
40.93
36.85
37.16
37.14
40.05
38.93
40.63
41.55
41.15
42.21
40.22
42.40
39.80
39.80
39.53
LONGITUDE
(deg, W)
85.87
85.88
78,92
74.07
80.47
72.32
85.41
71.72
75.79
76.29
71.17
73.14
84.20
70.22
71.22
84.50
84.50
84.46
71.30
74.25
83.40
81.22
82.38
71.31
80.39
84.55
84.54
85.81
85.70
85.71
85.76
85.76
82.79
80.98
73.76
81.51
88.79
88.80
74.58
77.20
80.44
72.63
81.90
71.34
74.59
71.08
84,13
84.12
84.39
X-GRID Y-GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
20.3 13.9
39.7 16.4
14.1 32.7
46.7 29.5
OUTSIDE ROM GRID
49.1 21.7
32.9 15.1
30.9 12.3
51.3 28.2
43.4 22.0
OUTSIDE ROM GRID
55.1 36.6
51.1 26.7
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
50.8 26.6
39.0 15.6
2.4 26.3
11.1 29.9
6.5 9.2
50.8 27.9
14.4 27.5
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
4.8 28.3
12.1 18.7
40.9 17.6
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
37.7 12.3
27.2 5.6
14.2 15.8
45.5 21.3
8,4 18.9
50.7 25.3
37.6 13.3
51.7 26.4
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
105
-------
SITE
MK1
MK2
MK3
MN1
MN2
MNR
MNT
MON
MRC
MRL
MRQ
MRT
MSG
MU1
MU2
MU3
N10
Nil
N12
NBR
NCI
NC2
NC3
NC4
NCL
NG1
NG2
NH1
NH2
NHP
NKS
NOR
NPN
NRF
NRS
NRT
NWB
NWR
NY1
NY 2
NY3
NY4
NY5
NY6
NY7
NY8
NY9
NYK
OCN
LUCATIUN
MUSKEGON
MUSKEGON .
MUSKEGON CO
MANCHESTER
MANCHESTER
MONROE CO
MONTAGUE
MONMOUTH CO
MORRIS CO
MERLIN
MARQUETTE CO
MORRISTOWN
MISSISSAUGA
MUHLENBERG CO
MUHLENBERG CO
MUHLENBERG CO
NEW YORK CITY
NEW YORK CITY
NEW YORK CITY
NEW BRITAIN
NEW CASTLE CO
NEW CASTLE CO
NEW CASTLE CO
NEW CASTLE CO
NEW CASTLE
NIAGARA CO
NIAGARA FALLS
NEW HAVEN
NEW HAVEN
NORTHAMPTON CO
NEW KENSINGTON
NORWALK
NEWPORT NEWS
NORFOLK
NORRISTOWN
NORTHAMPTON
NEW BRUNSWICK
NEWARK
NEW YORK CITY
NEW YORK CITY
NEW YORK CITY
NEW YORK CITY
NEW YORK CITY
NEW YORK CITY
NEW YORK CITY
NEW YORK CITY
NEW YORK CITY
NORTH YORK
OCEAN CO.
STATE OR
PROVINCE
MI
' MI
MI
NH
NH
NY
MA
NJ
NJ
ON
MI
NJ
ON
KY
KY
KY
NY
NY
NY
CT
DE
DE
DE
DE
PA
NY
NY
CT
CT
VA
PA
CT
VA
VA
PA
PA
NJ
NJ
NY
NY
NY
NY
NY
NY
NY
NY
NY
ON
NJ
LATITUDE
(deg, N)
43.24
43.24
43,28
42.99
42.99
43.07
42.57
40.33
40.79
42.25
46.81
40.80
43.57
37.21
37.22
37.31
40.60
40.78
40.67
41.67
39.81
39.78
39.70
39.53
41,00
43.22
43.09
41.31
41.31
37.29
40.55
41.11
37.07
36.85
40.11
40.69
40.48
40.73
40.83
40.74
40.77
40.59
40.74
40.73
40.67
40.75
40.72
43.72
40.11
LONGITUDE
(deg, W)
86.20
86.25
86.22
71.46
71.46
77.71
72.53
74.27
74.68
82.22
87.73
74.48
79.61
86.94
87.05
87.00
74.13
73.91
73.98
72.78
75.45
75.52
75.68
75.69
80.35
78.48
79.00
72.92
72.92
75.97
79.76
73.41
76.49
76.28
75.31
75.49
74.44
74.18
73.90
73.82
73.97
73.94
73.99
73.95
73.97
73.98
74.00
79.34
74.31
X-GRID Y-GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
50.1 30.0
50.1 30.0
25.2 30.4
45.9 27.4
38.9 14.0
37.3 16.7
7.1 25.5
OUTSIDE ROM GRID
38.1 16.8
17.6 33.4
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
39.5 15.6
40.4 16.7
40.1 16.0
44.9 22.0
34.2 10,8
33.9 10.7
33.3 10.2
33.3 9.2
14.6 18.0
22.1 31.3
20.0 30.5
44.3 19.8
44.3 19.8
OUTSIDE ROM GRID
17.0 15.3
42.4 18.7
OUTSIDE ROM GRID
OUTSIDE ROM GRID
34.8 12.7
34.0 16.1
38.2 14.9
39.3 16.4
40.4 17.0
40.7 16.4
40.1 16.6
40.3 15.6
40.1 16.4
40.2 16.4
40.1 16.0
40.1 16.5
40.0 16.3
18.6 34.3
38.8 12.6
106
-------
SITE
OH1
OH2
OH3
OKI
OK2
ONI
ORN
ORO
OSH
OWN
P10
Pll
PA I
PAT
PAU
PB1
PB2
PBS
PD1
PD2
PD3
PEN
PHI
PH2
PH3
PH4
PH5
PH6
PH7
PH8
PH9
PLN
PLS
PLY
P01
P02
POR
POU
PRB
PRH
PRY
PSB
PSQ
PT1
PT2
PT3
PTG
PTH
PTR
LOCATION
OHIO CO
OHIO CO
OHIO CO
OAKVILLE
OAKVILLE
ONEIDA CO
ORONO, MAINE
ORONO
OSHAWA
OWENSBORO
PHILADELPHIA
PHILADELPHIA
PAINESVILLE
PATERSON
PAULSBORO
PENOBSCOT CO
PENOBSCOT CO
PENOBSCOT CO
PADUCAH
PADUCAH
PADUCAH
PENNS GROVE
PHILADELPHIA
PHILADELPHIA
PHILADELPHIA
PHILADELPHIA
PHILADELPHIA
PHILADELPHIA
PHILADELPHIA
PHILADELPHIA
PHILADELPHIA
PLAINFIELD
PULASKI CO
PLYMOUTH
PORTSMOUTH
PORTSMOUTH
PORTLAND
POUGHKEEPSIE
PREBLE CO
PORT HURON
PERRY CO
PHILLIPSBURG
PRESQUE ISLE
PITTSBURGH
PITTSBURGH
PITTSBURGH
PORTAGE CO
PERTH AMBOY
PETROLIA
STATE OR
PROVINCE
KY
KY
KY
ON
ON
NY
ME
ON
ON
KY
PA
PA
OH
NJ
NJ
ME
ME
ME
KY
KY
KY
NJ
PA
PA
PA
PA
PA
PA
PA
PA
PA
NJ
KY
NH
NH
NH
ME
NY
OH
MI
PA
NJ
ME
PA
PA
PA
OH
NJ
ON
LATITUDE
(cleg, N)
37.32
37.30
37.56
43.42
43.42
43.30
44.90
43.97
43.90
37.78
39.96
39.95
41.72
40.93
39.83
45.01
44.73
45.25
37.08
37.09
37.06
39.73
40.01
40.05
39.98
39.88
39.92
39.91
40.08
39.95
40.01
40.60
37.16
43.78
43.08
43.08
43.66
41.70
39.84
42.95
40.46
40.68
46.69
40.44
4U.47
40.44
41.18
40.51
42.96
LONGITUDE
(deg, W)
86.93
86.95
86.77
79.70
79.70
75.72
68.67
78.62
78.85
87.07
75.17
75.16
81.24
74.16
75.24
68.63
68.98
68.57
88.60
88.60
88.57
75.47
75.10
75.24
75.10
75.23
75.19
75.15
75.01
75.16
75.15
74.44
84.48
71.75
70.76
70.76
70.26
73.94
84.72
82.44
77.17
75.19
68.00
80.00
79.96
79.95
81.33
74.27
82.19
X-GRID Y-GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
17.2 32.5
17.2 32.5
33.1 31.8
OUTSIDE ROM GRID
21.5 35.8
20.6 35.4
OUTSIDE ROM GRID
35.3 11.7
35.3 11.7
11.0 22.3
39.4 17.6
35.0 11.0
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
OUTSIDE ROM GRID
34.1 10.4
35.6 12.1
35.0 12.3
35.6 11.9
35.1 11.3
3b.2 11.5
35.4 11.4
36.0 12.5
35.3 11.7
35.4 12.1
38.2 15.6
OUTSIDE ROM GRID
49.0 34.7
52.9 3D, 5
53.0 30.5
55.0 34.0
40.3 22.2
OUTSIDE ROM GRID
6.2 29.7
27.3 14.7
35.3 16.1
OUTSIDE ROM GRID
16.0 14.6
16.2 14.8
16.2 14.6
10.7 19.1
38.9 15.1
7.2 29.7
107
-------
SITE
PTS
PV1
PV2
PV3
PV4
PWM
QIN
RC1
RC2
RCK
RDG
REN
RIC
RIM
RVB
SAG
SAL
SCI
SC2
SCH
SCR
SDB
SHP
SLM
SMC
SMT
SMV
SOM
SP1
SP2
SR2
STC
STF
STH
STK
STM
SIR
STU
STV
SUS
SUT
SVC
SY1
SY2
SY3
SY4
SYB
TEW
TL1
LOCATION STATE OR
PROVINCE
PITTSFIELD
PROVIDENCE
PROVIDENCE
PROVIDENCE
PROVIDENCE
PRINCE WILLIAM CO
QUINCY
ROCHESTER
ROCHESTER
ROCKVILLE
READING
RENSSELAER
RICHMOND
RICHMOND
RIVIERA BEACH
SAGINAW
SALINE
SCRANTON
SCRANTON
SCHENECTADY
SCARBOROUGH
SUDBURY
SHEPHERDSVILLE
SALEM
SIMCOE
ST MATTHEWS
SOMERVILLE
SOMERVILLE
SPRINGFIELD
SPRINGFIELD
SARNIA
ST. CATHERINES
STAFFORD
SOUTHFIELD
STARK CO
STAMFORD
STRATFORD
STEUBENVILLE
STOUFFVILLE
SUSSEX CO
SUITLAND-SILVER HILL
SEVEN CORNERS
SYRACUSE
SYRACUSE
SYRACUSE
SYRACUSE
STONY BROOK
TEWKSBURY
TOLEDO
MA
RI
RI
RI
RI
VA
MA
NY
NY
MD
PA
NY
VA
VA
MD
MI
MI
PA
PA
NY
ON
MA
KY
VA
ON
KY
NJ
MA
MA
MA
ON
ON
CT
MI
OH
CT
CT
OH
ON
DE
MD
VA
NY
NY
NY
NY
NY
MA
OH
LATITUDE
(deg, N)
42.45
41.82
41.82
41.83
41.83
38.64
42.24
43.16
43.17
39.11
40.32
42.63
37.54
37.49
39.16
43.45
42.16
41.44
41.59
42.80
43.75
42.38
38.00
37.29
42.86
38.25
40.57
42.40
42.09
42.14
42.98
43.17
41.98
42.45
40.87
41.06
41.15
40.36
43.97
38.58
38.85
38.87
43.05
43.06
43.06
43.06
40.91
42.61
41.65
LONGITUDE X-GRID Y-GRID
(deg, W)
73.20
71.41
71.41
71.41
71.41
77.53
70.97
77.60
77.52
77.11
75.93
73.75
77.43
77.47
76.51
83.91
83.78
75.62
76.07
73.94
79.27
71.39
85.70
80.05
80.27
85.81
74.61
71.11
72.59
72.49
82.41
79.24
72.39
83.22
81.33
73.54
73.15
80.62
79.27
75.27
76.93
77.14
76.15
76.18
76.13
76.18
73.13
71.22
83.53
43.2 26.7
50.4 22.9
50.4 22.9
50.3 23.0
50.4 23.0
25.9 3.9
52.1 25.5
25.6 31.0
25.9 31.0
27.6 6.7
32.3 13.9
41.0 27.8
OUTSIDE ROM GRID
OUTSIDE ROM GRID
30.0 7.0
0.4 32.7
0.9 24.9
33.5 20.7
31.7 21.6
40.2 28.8
18.9 34.5
50.4 26.3
OUTSIDE ROM GRID
OUTSIDE ROM GRID
14.9 29.1
OUTSIDE ROM GRID
37.5 15.4
51.6 26.4
45.6 24.5
46.0 24.8
6.4 29.9
19.0 31.0
46.5 23.9
3.1 26.7
10.7 17.2
41.8 18.3
43.4 18.9
13.5 14.2
18.9 35.8
34.9 3.5
28.3 5.1
27.4 5.2
31.4 30.3
31.3 30.4
31.5 30.4
31.3 30.4
43.5 17.5
51.1 27.6
1.9 21.9
108
-------
SITE
TL2
TL3
TL4
TOM
TR1
TR2
TRE
TRM
TRN
TV2
UTC
VAB
VIN
WAT
WAY
WCH
WH1
WH2
WIL
WL1
WL2
WLK
WLO
WNS
W01
WO 2
WPL
WRN
WSL
YG1
YG2
YOR
LOCATION
TOLEDO
TOLEDO
TOLEDO
TOMS RIVER
TORONTO
TORONTO
TRENTON
TRIMBLE CO
TRENTON
TIVERTON
UTICA
VIRGINIA BEACH
VINELAND
WATERTOWN
WAYNE CO
WINCHESTER
WHEELING
WHEELING
WILMINGTON
WILLIAMSPORT
WILLIAMSPORT
WILKES-BARRE
WILLOUGHBY
WINDSOR
WORCESTER
WORCESTER
WHITE PLAINS
WARREN
WESTLAKE
YOUNGSTOWN
YOUNGSTOWN
YORK
STATE OR
PROVINCE
OH
OH
OH
NJ
ON
ON
MI
KY
NJ
ON
NY
VA
NJ
MA
NY
KY
WV
WV
DE
PA
PA
PA
OH
ON
MA
MA
NY
MI
OH
OH
OH
PA
LATITUDE
(deg, N)
41.66
41.66
41.72
39.95
43.66
43.65
42.14
38.71
40.22
44.30
43.10
36.89
39.49
42.37
43.23
38.00
40.07
40.12
39.73
41.25
41.25
41.24
41.64
42.32
42.26
42.30
41.05
42.51
41.48
41.10
41.10
39.97
LONGITUDE X-GRID Y-GRID
(deg, W)
83.53
83.57
83.48
74.20
79.39
79.37
83.19
85.42
74,76
81.58
75.20
76.18
75.02
71.18
77.17
84.17
80.72
80.70
75.55
76.99
76.99
75.90
81.41
83.04
71.80
71.75
73.76
83.01
81.88
80.65
80.65
76.70
1.9
1.7
2.1
39.2
18.4
18.5
3.3
OUTSIDE
36.9
9.7
35.2
OUTSIDE
35.9
51.3
27.3
OUTSIDE
13.1
13.2
33.8
28.0
28.0
32.4
1U.4
3.8
48.8
49.0
40.9
4.0
8.5
13.4
13.4
29.2
21.9
22.0
22.3
11.7
34.0
33.9
24.8
ROM GRID
13.3
37.8
30.6
ROM GRID
9.0
26.2
31.4
ROM GRID
12,4
12.7
10.4
19.5
19.5
19.4
21.8
25.9
25.6
25.8
18.3
27.1
20.9
18.6
18.6
11.8
109
-------
APPENDIX B
MAXIMUM 03 VALUES AT SURFACE MONITORING STATIONS
This appendix contains a list of the surface monitoring sites operating
during the 3-4 August 1979 period along with the observed and predicted
maximum 03 values at these sites for each of the days and the hour that the
maximum value occurred. All concentrations are in units of ppb and times are
in hours, 1ST.
110
-------
DATE
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
SITE
AA1
AA2
AC1
AG4
AGA
AK1
ALN
ALX
AMH
AMR
ARL
ATL
BAB
BAD
BAY
BER
BET
BL1
BL3
BL6
BLG
BN1
BNG
BMW
BR2
BRI
BTH
BUL
BVF
CAN
CB1
CB4
CHL
CHR
CHS
CHT
CHV
CKY
CL2
CLK
CM1
CM3
CNE
CPE
DAN
DC1
DC4
DC5
DC 6
DNR
max
cob
69.0
71.0
76.0
57.0
100.0
58.0
59.0
75.0
69.0
99.0
100.0
87.0
83.0
31.0
91.0
53.0
55.0
37.0
38.0
70.0
55.0
41.0
79.0
66.0
117.0
100.0
61.0
87.0
29.0
55.0
85.0
66.0
79.0
100.0
60.2
83.0
. 64.0
66.0
63.0
105.0
89.0
84.0
159.0
105.0
120.0
15.0
10.0
125.0
100.0
95.0
hob
14
14
12
15
16
18
17
14
18
14
14
11
11
14
12
9
17
14
14
14
13
16
14
16
14
14
14
12
18
16
19
19
16
17
12
13
14
15
12
15
13
14
16
18
16
10
15
16
17
16
Cp0(hob ^
36.4
36.4
31.6
33.9
16.0
37.6
20.8
69.5
39.8
33.6
69.5
42.1
10.2
23.1
47.1
32.3
18.4
26.0
26.0
31.8
40.3
40.7
37.8
45.6
42.6
46.4
52.5
36.9
26.6
38.2
32.2
32.2
24.8
51.3
29.2
50.1
52.5
33.5
39.1
33.2
41.3
32.0
41.5
38.1
38.7
39.9
70.0
41.4
40.2
26.7
Cp](hob:
38.5
38.5
32.1
33.9
16.2
38.0
22.3
72.0
41.4
33.9
72.0
43.7
10.6
23.5
47.8
34.5
19.5
26.3
26.3
33.0
40.5
41.6
38.0
48.3
43.0
49.1
55.0
39.5
27.2
38.4
33.2
33.2
25.0
52.0
33.8
54.3
55.0
36.4
39.9
36.0
45.1
33.7
41.4
38.5
38.8
41.1
72.3
43.9
42.0
27.3
>u max
> Cp0
42.8
42.8
37.0
35.1
37.7
40.0
33.1
70.0
40.4
37.5
70.0
42.1
24.6
26.4
59.4
47.5
30.2
28.2
31.0
36.3
48.0
41.3
38.5
55.6
92.2
62.8
56.2
51.2
26.6
38.2
34.2
34.2
31.1
51.3
31.9
51.9
56.2
33.5
51.7
34.2
59.2
41.2
47.7
42.2
49.5
39.9
70.0
56.2
56.2
33.1
.max
hpO
17
17
9
13
13
16
9
15
17
17
15
11
20
17
16
17
9
10
8
17
22
18
15
20
18
17
13
15
18
16
14
14
8
17
13
14
13
15
18
20
15
12
19
23
17
10
15
13
13
7
.max
Cpl
46.0
45.9
37.5
35.1
38.3
40.4
35.1
72.3
41.5
37.9
72.3
43.7
25.1
26.9
59.8
49.8
31.4
28.3
31.2
38.8
48.8
41.9
38.4
57.3
93.0
67.8
58.9
53.5
27.2
38.4
34.8
34.8
32.0
52.0
35.3
55.9
58.9
36.4
52.8
36.2
64.5
43.4
47.7
42.8
49.7
41.1
72.3
58.9
58.9
34.5
.max
hpl
18
16
9
13
13
16
8
15
17
17
15
11
20
17
16
17
9
10
8
18
22
17
15
20
18
17
13
15
18
16
14
14
8
17
9
14
13
15
18
9
16
12
19
23
17
10
15
13
13
7
111
-------
DATE
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
SITE
DOW
ORB
DT3
DT4
EB1
EB2
ERE
ESS
ESX
EXC
EY1
FDR
FLN
FX1
FX2
GAR
GLN
GLO
GMT
GRE
GRG
GRW
HAR
HF2
HGR
HM1
HML
HRP
JNS
KCH
KNT
LAN
LEX
LIC
LIV
LND
MAM
MCL
MED
MER
MFD
MN1
MNR
MRC
MSG
N10
NCI
NC4
NCL
max
cob
96.0
104.0
43.9
25.0
75.0
62.0
95.0
63.3
80.0
67.0'
75.0
100.0
57.7
92.3
115.0
78.0
65.0
102.0
45.0
38.0
69.0
130.0
69.0
100.0
46.0
54.0
68.0
63.0
65.0
76.0
115.0
56.0
70.0
76.0
39.8
64.0
54.0
50.0
50.0
102.0
66.0
75.0
61.0
69.0
70.0
123.0
92.0
31.0
kmax
hob
14
15
15
15
15
15
14
17
15
23
17
13
12
14
14
14
15
14
10
18
12
14
17
14
13
11
17
12
16
17
13
16
14
16
12
13
15
13
9
12
14
17
17
14
15
12
12
18
CpO(hobX) '
31.9
33.7
44.5
30.1
46.0
46.0
39.6
36.5
25.3
36.1
40.2
35.3
35.7
52.4
68.2
41.7
43.8
38.9
35.0
22.5
39.1
88.5
41.8
27.2
31.0
10.2
47.2
34.6
26.3
31.4
35.9
33.5
47.4
47.5
38.0
66.6
95.0
38.3
39.8
31.7
53.9
33.2
33.2
33.7
48.9
44.7
MISSING
35.1
35.6
=pl Cp0
33.2
85.0
72.7
70.2
59.0
59.0
40.0
37.2
31.0
38.8
42.8
38.5
35.7
54.3
81.7
42.6
43.8
50.0
39.2
36.4
52.6
95.0
44.5
38.0
32.6
24.6
48.0
34.6
30.1
64.9
40.7
43.1
54.6
54.6
39.3
119.0
95.0
38.3
39.8
40.1
53.9
37.3
34.4
33.8
56.8
65.7
45.9
37.0
umax
hPo
15
19
10
9
10
10
15
7
6
18
16
15
11
13
17
17
15
17
18
6
16
15
18
11
9
20
22
12
9
21
17
20
15
15
16
21
15
13
9
15
14
15
12
- 13
10
15
20
17
max
Cpl
37.5
86.0
73.4
71.1
61.0
61.0
40.1
37.2
31.6
39.3
45.4
40.9
37.4
54.8
87.1
43.3
45.3
50.9
40.5
36.8
53.4
96.7
44.6
38.6
36.2
25.1
48.8
37.6
33.6
68.2
41.0
45.9
55.3
55.3
39.9
122.0
96.7
41.3
40.9
44.0
56.0
37.7
36.3
37.7
56.4
67.3
48.6
39.2
.max
hpl
15.
19
10
9
10
10
16
8
7
17
15
16
11
13
16
17
15
17
17
6
16
15
18
11
16
20
22
12
9
21
17
20
15
15
16
21
15
12
9
16
14
15
11
13
10
15
20
17
112
-------
DATE
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
SITE
NG2
NH2
NRS
NY2
NY4
NY5
NYK
OCN
OKI
ORO
OSH
P10
PA I
PHI
PH2
PH4
PH5
PH6
PH7
PH8
PLY
POU
PRH
PT2
PTG
PTR
PTS
PV3
RDG
REN
RVB
SCI
SCH
SCR
SMC
SMV
SR2
STC
STH
STU
STV
SVC
SY1
SY2
SY4
TL1
TL4
TR1
TR2
.max
cob
72.0
30.0
110.0
87.0
101.0
83.0
48.0
95.0
64.0
98.0
55.0
80.0
54.0
80.0
90.0
60.0
60.0
40.0
110.0
50.0
41.0
65.3
76.0
77.0
87.0
62.0
75.0
36.0
78.0
73.0
113.0
71.0
66.0
84.0
51.0
100.0
72.0
26.5
100.0
81.0
97.4
72.0
78.0
78.0
70.0
65.0
64.0
41.0
.max
hob
17
8
14
13
12
13
13
11
13
16
16
16
13
12
12
15
12
12
12
13
17
13
15
14
16
14
13
14
15
13
13
14
15
14
13
12
17
12
17
15
14
15
15
15
19
14
17
11
GpO(hobX)
40.0
29.9
37.4
31.3
47.1
53.2
53.8
38.0
41.1
92.5
58.0
48.7
41.2
35.3
35.1
46.0
35.1
46.0
35.3
MISSING
39.3
3.8
66.6
29.4
32.2
116.0
10.3
36.7
33.5
10.9
31.7
37.2
38.4
51.0
39.7
41.2
57.1
39.4
42.8
22.4
58.9
52.4
23.6
23.6
23.6
34.9
37.3
40.6
72.2
CpXb
40.1
30.7
41.8
31.9
47.8
53.9
53.6
42.7
42.7
99.3
57.6
55.3
41.3
38.0
35.9
48.3
35.9
49.1
38.0
DATA
39.9
4.0
67.2
-29.4
33.3
120.0
10.4
38.6
36.7
11.4
32.7
39.5
38.8
50.9
41.4
43.2
57.4
40.4
43.9
23.1
66.0
52.9
24.1
24.1
24.1
35.9
38.1
41.7
72.4
Xx max
} Cp0
40.0
31.4
46.6
39.5
59.4
59.4
72.2
42.5
42.0
104.0
93.7
54.4
47.7
59.2
54.4
51.9
54.4
51.9
59.2
39.3
32.6
89.1
29.5
38.8
116.0
23.1
40.7
36.7
26.1
34.0
41.3
38.4
72.2
46.4
46.8
89.1
40.9
45.2
29.4
68.6
54.3
36.8
36.8
36.8
42.5
40.3
59.0
72.2
.max
hpO
17
7
18
18
16
16
11
15
16
14
12
15
19
15
15
14
15
14
15
13
9
16
16
17
16
7
19
17
7
12
16
14
11
23
23
16
23
10
13
13
13
10
- 10
10
11
12
10
11
max
Cpl
40.3
32.1
52.9
40.0
59.8
59.8
72.4
45.3
43.8
110.0
93.6
55.5
48.2
64.5
55.5
55.9
55.5
55.9
64.5
39.9
33.6
89.9
29.6
40.2
120.0
23.5
41.9
40.2
26.4
35.6
42.4
38.9
72.4
48.0
49.1
89.9
41.7
46.5
30.4
77.0
54.8
38.0
38.0
38.0
43.5
41.1
61.0
72.4
.max
hpl
18
7
18
18
16
16
11
15
16
14
12
15
19
16
15
14
15
14
16
13
9
16
16
17
16
7
19
17
7
12
16
13
11
23
23
16
23
10
13
13
13
10
10
10
11
12
10
11
113
-------
DATE
79215
79215
79215
79215
79215
79215
79215
79215
79215
79215
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
SITE
TRN
UTC
VIN
WH2
WLK
WNS
W01
W02
WRN
YG2
AA1
AA2
AC1
AG4
AGA
AK1
ALN
ALX
AMH
AMR
ARL
BAB
BAD
BAY
BER
BET
BL1
BL3
BL6
BLG
BN1
BNG
BNW
BR2
BRI
BTH
BUL
BVF
CAN
CB1
CB4
CHL
CHR
CHS
CHT
CHV
CKY
CL2
CLK
_max
cob
81.0
83.0
77.0
69.9
72.0
92.0
52.0
65.0
62.2
70.0
94.0
88.0
73.0
78.0
90.0
80.0
75.0
75.0
77.0
68.0
100.0
58.0
43.0
71.0
84.0
75.0
71.0
67.0
96.0
57.0
76.0
88.0
50.0
68.0
123.0
80.0
105.0
28.0
50.0
75.0
64.0
85.0
75.0
77.0
78.0
78.0
97.0
108.0
90.0
.max
hob
13
17
13
18
13
10
15
18
13
16
13
12
11
13
13
13
17
11
22
16
12
12
14
12
11
14
15
15
12
11
9
11
15
12
19
13
17
14
15
15
16
16
15
16
15
13
14
13
17
Cp0(hob '
36.9
37.5
29.4
23.2
29.8
65.9
53.7
55.1
53.4
36.1
35.6
36.7
85.8
36.8
36.6
32.8
28.0
41.9
80.5
15.1
45.3
85.4
45.7
9.1
40.5
23.2
39.8
39.8
35.8
84.6
40.4
36.6
92.9
16.6
14.3
41.1
75.9
45.7
38.9
32.2
32.6
17.0
19.9
15.3
47.8
41.1
35.0
50.6
33.7
' cPi
-------
DATE
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
SITE
CM1
CMS
CNE
CPE
DAN
DC4
DC 5
DC6
DNR
DOW
ORB
DT3
DT4
EB1
EB2
ELM
ERE
ESS
ESX
EXC
EY1
FDR
FLN
FX1
FX2
GAR
GLN
GLO
GMT
GRE
GRG
GRW
HAR
HF2
HGR
HM1
HML
JNS
KCH
KNT
LAN
LEX
LIC
LIV
LND
MAM
MCL
MER
MFD
max
cob
90.0
102.0
136.0
55.0
79.0
25.0
140.0
95.0
91.0
50.0
66.0
68.9
38.8
82.0
78.0
2.0
88.0
62.2
98.0
95.0
125.0
100.0
64.3
128.0
117.0
78.0
91.0
33.0
75.0
40.0
37.0
65.0
85.0
85.0
51.0
30.0
60.0
76.0
75.0
50.0
66.0
60.0
69.0
66.8
73.0
23.0
67.9
144.0
29.0
.max
hob
12
17
15
10
15
13
13
15
10
13
12
13
13
12
11
23
16
13
16
22
11
13
12
12
12
14
13
16
13
1
9
11
14
14
15
9
21
16
23
2
18
17
17
12
18
16
12
16
15
/. max^
Cp0(hob '
47.0
95.2
58.7
45.4
0.4
48.2
41.1
38.1
21.1
36.5
1.6
78.8
60.1
67.8
68.6
62.7
66.1
37.2
42.0
40.3
48.0
31.1
35.7
40.4
49.5
13.6
42.5
56.2
42.3
22.4
25.3
15.7
41.9
40.5
24.8
70.0
68.5
23.1
44.1
24.7
46.2
46.9
46.9
37.8
45.8
2.1
36.4
53.0
34.8
cp](hob
51.7
99.7
58.6
47.1
0.4
49.7 ;
42.9
39.7
21.7
41.7'
1.6
79.4
60.9
69.7
70.8
65.3
66.3
37.2
42.3
44.0
50.0
32.6
37.5
40.8
52.7
14.1
44.5
57.5
43.8
23.7
26.1
15.9
41.7
41.7
26.7
72.5
69.2
24.5
46.5
24.9
49.5
48.2
48.2
38.3
59.0
2.1
38.3
58.9
36.8
N max
> CpO
55.9
99.7
68.8
57.4
19.9
50.4
41.1
41.1
21.1
39.6
25.4
78.8
92.5
68.6
68.6
62.7
68.8
37.3
42.6
44.1
62.3
33.9
36.3
40.7
49.5
23.9
43.7
57.2
43.0
25.6
52.3
58.5
43.8
41.6
28.4
70.0
109.0
23.1
73.2
100.0
49.4
49.4
49.4
46.1
99.7
35.5
37.2
53.0
39.2
.max
hpO
16
16
23
13
1
14
13
13
10
14
4
13
10
11
11
23
15
9
17
17
18
10
13
13
12
0
16
17
15
0
17
0
12
13
2
9
15
16
8
14
16
16
16
' 18
0
3
13
16
12
max
Cpl
60.3
105.0
69.1
58.2
20.6
52.1
42.9
42.9
21.7
44.6
26.7
79.4
93.7
70.8
70.8
65.3
68.9
37.3
42.9
44.6
65.1
35.2
38.4
41.4
52.7
24.2
47.6
59.0
44.8
26.1
55.4
62.0
43.7
42.5
31.4
72.5
110.0
24.6
86.4
102.0
52.5
50.7
50.7
47.0
107.0
36.5
39.4
58.9
40.1
.max
16
16
23
13
1
14
13
13
10
15
4
13
10
11
11
23
15
9
17
17
18
10
9
13
12
0
18
17
16
0
15
0
12
13
2
9
15
15
7
13
16
16
16
18
0
3
13
16
12
115
-------
DATE
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
SITE
MN1
MNR
MRC
MSG
N10
NCI
NC4
NCL
NG2
NRS
NY2
NY4
NY5
NYK
OCN
OKI
ORO
OSH
P10
PAI
PHI
PH2
PH4
PH5
PH6
PH7
PLY
POU
PRH
PT2
PTG
PTR
PTS
PV3
RDG
REN
RVB
SCI
SCH
SCR
SMC
SMV
SR2
STC
STH
STU
STV
SVC
SY1
max
cob
96.0
69.0
53.0
76.0
89.0
100.0
94.0
38.0
83.0
76.0
67.0
45.0
50.0
55.0
85.0
54.0
101.0
79.0
90.0
62.0
90.0
90.0
70.0
60.0
40.0
130.0
65.0
42.0
63.8
76.0
87.0
99.0
71.0
32.7
54.0
105.0
96.0
99.0
85.0
83.0
85.0
112.0
99.0
59.0
68.4
96.0
89.0
110.0
84.0
,max
hob
11
14
17
10
13
16
14
18
23
13
11
10
11
12
13
12
15
12
18
15
18
16
13
18
18
18
11
18
18
14
11
18
14
2
16
15
14
14
12
13
18
17
18
12
13
12
13
12
16
Cp0^hob '
22.8
45.2
47.5
95.2
25.6
41.0
52.9
44.2
98.7
32.7
32.5
35.1
27.1
84.7
48.5
79.1
31.3
98.4
27.7
49.6
31.8
32.8
35.6
27.7
33.5
31.8
39.9
11.0
69.4
37.6
31.2
82.9
13.9
27.0
33.5
23.6
46.3
37.2
50.6
82.0
84.5
23.8
69.4
73.7
39.2
33.9
36.4
40.4
38.9
1 cPi(hSbx
23.3
48.7
55.7
94.6
26.5
43.5
54.7
47.1
99.9
36.2
33.2
35.6
27.4
84.9
54.2
83.7
33.4
98.4
28.3
49.8
35.2
33.3
38.2
28.3
36.2
35.2
40.5
12.0
70.0
37.5
32.5
85.3
14.3
28.0
35.3
24.1
47.4
37.9
51.7
82.4
90.2
25.5
70.0
75.9
40.6
35.1
41.8
40.8
39.5
\ max
>CP0
34.7
62.1
48.2
107.0
45.1
41.0
52.9
50.1
107.0
32.7
32.5
35.1
35.1
85.1
52.7
90.4
62.3
99.7
42.9
68.7
55.9
42.9
47.8
42.9
47.8
55.9
42.0
31.2
75.3
42.1
46.3
85.0
25.4
36.5
37.1
32.4
46.3
38.3
55.5
85.1
95.6
48.1
75.3
104.0
64.5
38.5
36.7
40.7
41.7
max
hpO
19
18
16
17
10
16
14
16
19
13
11
10
10
11
15
15
10
13
14
19
16
14
15
14
1.5
16
14
8
16
17
23
17
12
6
10
19
13
15
16
11
15
3
16
- 18
18
17
12
13
20
max
CP1
36.0
66.1
55.7
107.0
46.2
43.5
55.3
53.7
107.0
36.2
33.2
35.6
35.6
85.3
59.7
94.7
65.7
99.6
43.6
71.6
60.3
43.6
51.4
43.6
51.4
60.3
43.0
32.6
77.6
42.2
47.6
87.9
26.0
39.7
39.3
33.2
47.4
39.1
56.4
85.3
101.0
49.9
77.6
107.0
66.6
39.8
41.8
41.4
42.8
.max
hpl
18
18
17
17
10
16
15
16
19
13
11
10
10
11
15
15
10
13
14
20
16
14
15
14
15
16
15
8
16
17
23
16
12
6
10
19
14
15
16
11
15
2
16
18
18
16
13
13
20
116
-------
DATE
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
79216
SITE
SY2
SY4
TL1
TL4
TR1
TR2
TRN
UTC
VIN
WH2
WLK
WNS
W01
W02
URN
WSL
Y62
max
cob
91.0
91.0
95.0
100.0
92.0
65.0
125.0
104.0
132.0
63.8
73.0
74.0
55.0
68.0
100.0
120.0
81.0
hob
16
16
14
12
12
13
17
16
13
14
13
11
16
17
15
10
15
/ i insx i
Cp0(hob '
38.9
38.9
40.0
41.4
67.8
82.0
42.1
41.4
50.7
36.5
28.0
89.0
28.2
29.1
58.3
40.1
36.2
1 cpl(hobX:
39.5
39.5
41.0
42.6
69.7
82.4
42.5
42.2
56.8
39.3
29.2
89.9
28.9
29.7
58.9
40.1
37.2
\ Jnax
' Cp0
41.7
41.7
41.7
41.7
68.6
85.1
56.3
41.4
86.1
40.0
32.4
92.5
29.1
29.1
78.8
75.9
39.8
.max
hpo
20
20
13
13
11
11
16
16
20
16
9
10
17
17
13
15
19
max
Cpl
42.8
42.8
42.9
42.9
70.8
85.3
57.0
42.4
101.0
43.0
33.8
93.7
29.7
29.7
79.4
76.0
41.2
.max
hpl
20
20
13
13
11
11
16
19
19
16
9
10
17
17
13
15
19
117
-------
GLOSSARY
Symbol
Cob
cob
.max
cob
Rpnax
cob
-po
7
Cp0
.max
Cp0
Rpnax
Cp0
pl
max
cpl
Rpnax
cpl
R / hma
cpl(hob
Definition
Observed hour-average 03 concentration at a given
receptor at a given hour
Average value of c^ over all receptors at a given
hour
Maximum value of
receptor
Average value of
hour
during a daily period at a given
over all receptors at a given
Predicted ROM1 layer 0 hour-average 03 concentration
interpolated to a given receptor location at a given hour
Average value of CPQ over all receptors at a given hour
Maximum value of cpQ during a daily period at a given
receptor location
Average value of Cp§x over all receptors at a given
hour
Value of cpQ at the hour of occurence of
Average value of cpQ(hgj^x) over all receptors at
a given hour
Predicted ROM1 layer 1 hour-average 03 concentration
interpolated to a given receptor location at a given
hour
Average value of cpi over all receptors at a given
hour
Maximum value of cp^ during a daily period at a given
receptor location
Average value of cpfx over all receptors at a given
hour
Value of
Average value of
a given hour
at the hour of occurrence of
over-all receptors at
118
-------
Symbol Definition
c1!13^ Maximum ROM1 layer 1 predicted 03 concentration during
a daily period at the same receptor location where
Cgj*x occurred within a given 03 plume
Maximum ROM1 layer 1 predicted 03 concentration during
a daily period at any receptor location within or near
a given 03 plume
Cp^X£ Maximum ROM1 layer 1 predicted 03 concentration during
a daily period at any grid cell within or near a
given 03 plume
dg = c0t,-CpQ. Residual difference (bias) between observed
and predicted 03 concentrations at a given receptor
location at a given hour
^dg Average value of dg over all receptors at a given hour
Average value of absolute dg values (gross error) over
all receptors at a given hour
...«x _ rmax r fumax\
AO ob puv ob '
Average value of d^g^ over all receptors at a given
hour
_ rmax rmax
" cob " Cp0
Average value of dg^x over all receptors at a given
hour
= c0b - Cpi. Residual difference (bias) betwen observed
and predicted 03 concentrations at a given receptor
location at a given hour
Average value of d^ over all receptors at a given hour
Average value of absolute d^ values (gross error) over
all receptors at a given hour
Average value of d^x over all receptors at a given
hour
Hmax _rmax rmax
dBl "cob cpl
Average value of d^|x over all receptors at a given
hour
119
-------
Symbol Definition
_ rmax rmax
' cob ' cpl-A
- rmax _ rmax
cob cpl-B
_ rmax rmax
' cob cpl-C
Percent bias between c^x and
Percent bias between c^x and
100 Percent bias between cJJ* and
Hour of occurrence of
Hour of occurrence of
Hour of occurrence of
P j. pj.
n Number of receptor monitoring sites for 03 reporting
values at a given hour
^S^Q Standard deviation of the residual differences
(noise) over all receptors at a given hour for ROM1
layer 0
S^l Standard deviation of the residual differences
(noise) over all receptors at a given hour
for ROM1 layer 1
^S^*) Standard deviation of d^x over all receptor
locations
Sjjjn Standard deviation of d^x over all receptor
locations
RS^|X) Standard deviation of dg^x over all receptor
locations
RS^|^ Standard deviation of d"§^x over all receptor
locations
120
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT MO.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
EPA REGIONAL OXIDANT MODEL: ROM1 EVALUATION FOR
3-4 AUGUST 1979
5. REPORT DATE
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
Kenneth L. Schere
9. PERFORMING ORGANIZATION NAME AND ADDRESS
(same as 12.)
10. PROGRAM ELEMENT NO.
CDWA1A/02-4021 (FY-86)
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
ATMOSPHERIC 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
In-house
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The first generation U.S. Environmental Protection Agency Regional Oxidant Model
(ROM1) has been tested and evaluated for 03 predictions on a two-day test case episode
in the northeast U.S. during 3-4 August 1979. The period was characterized by relat-
ively high 0, concentrations in the southern Great Lakes area where clear skies
persisted. The highest observed hour-average 03 level monitored at a surface site
during the period was 159 ppb. The ROM1 incorporates some simplifying assumptions and
algorithms as compared to the second generation version, ROM2, which is in preliminary
testing stages now and will eventually become the production version of the model.
Evaluation results for this test episode showed that the ROM1 had approximately a 6%
average underprediction of 0, when all hours and surface monitoring sites were consid-
ered. When the data were restricted to only those observed and predicted pairs of 0,
values greater than 50 ppb the average performance improved to a 1% underprediction.
The evaluation aspect concerned with estimating maximum daily 0- values showed a 8%
average underprediction of the maximum value for the restricted data subset. An
analysis of individual 0- plumes during the episode showed average model performance
for predicting the plume maximum concentration level to lie between 22% underpredict-
ion and 38% overprediction.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (Tins Report I
UNCLASSIFIED
21. NO. OF PAGES
20. SECURITY C
22. PRICE
EPA Farm 2220-1 (R«v. 4-77) PREVIOUS EDITION is OBSOLETE
-------