-------
5.2 ASSUMPTIONS
Below is a list of key factors requiring assumption in the BOM
modelling, the problem with each factor, the assumption made by previous BOM
modellers (Dodge and Demerjian) and the choices made in this study.
Assumptions necessary to model the Bureau of Mines smog chamber data-
base:
1) Light Intensity
- only pseudo-first order N09 photolysis rate
1
- Dodge assumed 0.35 min photolysis rate for N09
-1
- Demerjian assumed 0.43 min photolysis rate for NOp
- spectral distribution not measured assumed to be same as sunlight,
but two different sets of quantum yields have been used (see
Section 4.3, Table 10)
- this study tested both assumptions for Demerjian (see Section 5.3),
used "old" (OZIPP) values for Dodge, Carbon Bond, and CIT supplied
values for CIT.
2) Temperature
- wall temperature measured, air temperature gradients not characterized
- assumed to be constant at 85 F
- this study assumed a constant temperature of 85 F in all mechanisms
3) HC Composition
- known to vary from test to test
- detailed data not available during run
- was assumed by Demerjian to be constant over range of experiments (see
Section 4.4, Table 13)
- Dodge "tuned" model composition fractions of n-butane and propylene
to obtain good fit
- this study accepted Demerjian's values and applied them in Carbon Bond
and CIT (see Section 4.4)
4) 03 Concentrations
- known to be subject to interference
- Dodge assumed truefO,] = 0.9 [03] measured
- Demerjian assumed true[03J = [03] measured
- this study accepted Dodge assumption
75
-------
5) Aldehyde Initial Values
- large variation in initial measured aldehyde to measured HC ratio
- Dodge assumed true aldehyde - measured[aldehyde]
- Demerjian assumed true aldehyde = 0.05 initial[NHMC]
- this study used Dodge assumption for Dodge and Demerjian's assumption
for Demerjian, Carbon Bond and CIT simulations
6) Wall Effects and Background Reactivity
- Dodge assumed wall was a source of propylene
- Demerjian did not assume wall effects
- this study used Dodge's assumption for Dodge and Demerjian's
assumption for Demerjian, Carbon Bond, and CIT simulations
76
-------
5.3 SIMULATION RESULTS
Appendix C shows the NO, NO-, PAN, and 03 data and model predictions
for each BOM run for each mechanism. In these plots the reported BOM 0^
values are used (i.e. the data were not adjusted) and the original Demerjian
mechanism and light intensity assumptions were used. The simulated ozone
maximum and the time of occurrence for each mechanism for each experiment
are given in Table 14. Figure 19 shows the scatter diagrams for 0^ max for
all four mechanisms. (In Figure 19, the second set of assumptions was used
for the Demerjian mechanism.) These figures assume that the BOM Oo was 90%
of what was reported. Figure 20 shows the scatter diagrams for time to 0-
max.
These figures show that none of the four mechanisms is obviously
superior. The CIT mechanism, however, had a large number of low predictions,
late times, and more scatter than the others.
Runs 25, 27, and 28 for Carbon Bond and Demerjian were high compared to
the Dodge mechanism. This was a direct consequence of the initial aldehyde
assumptions for the simulations. Dodge assumed that the measured initial
aldehyde was the actual aldehyde and Demerjian assumed that the aldehydes
were proportional to the initial NMHC. UNC used Demerjian's assumption for
Carbon Bond. For runs 25, 27, and 28 these assumptions resulted in a 5-fold
difference in initial aldehydes (Dodge being less). Run 25 for Carbon Bond
was resimulated with the same aldehyde as Dodge and the predicted 0, was 2.5
times lower and similar to Dodge's prediction. These runs are at high NO
A
concentrations and are very sensitive to radical sources and thus the results
are more a reflection of the assumptions than of the mechanism behavior.
At the request of Demerjian, two sets of simulations were performed
with the Demerjian mechanism. The first set used the rate constants from
77
-------
the original mechanism (including some now known to be wrong) and an assumed
light intensity equivalent to an NOo photolysis rate of 0.43 min~. The
second set used lower rate constants for reactions 11 and 12 in Table 7 (the
peroxynitric acid formation and decomposition reactions) resulting in less
NCL loss and thus requiring a lower NCL photolysis rate to avoid overpre-
dicting the 03 formation. UNC assumed the same value for light intensity as
Dodge had assumed (0.35 min~ ) for this second set of Demerjian simulations.
The scatter diagrams for these two sets are shown in Figure 21. The
older rates and assumptions produced consistent overpredictions. The newer
rates and assumptions still result in a slight overprediction. The BOM
observed 0~ in these plots, however, ib only 90% of the reported Oo value
(using the DODGE assumption). Therefore, if the reported 03 is the true 03
than the Demerjian predictions would be more centered on the data set. The
newer rates and lower light intensity assumption were used for all
subsequent work.
78
-------
0.80
71 O.BO
- 0.70
- 0.60
O.SO
I ' I ' I ' I ' '
CflRIION BUND MCC
DEMERJJflN MCCHfiNJSM
LIGHT INTENSITY - 0.3S
- 0.10
0.00 0.10 0.20 0.30 0.10 0.50 0.60 0.70 0.8
BOM CHflMBER OBSERVED Q, MflX
0.00 0.10 0.21) 0.30 0.10 0.50 0.60 0.70 0 to
BOM CHflMBER OBSERVED 0, MflX. PPM
0.00
0.80
0.70 -
DODGE MECHflNISn
BOH CHflMBER DRTH
0.00
- 0.30
- 0.20
- 0.10
•—' 0.00
0.00 0.10 0.20 0 30 0.10 0.00 0.60 0.70 0 80
BUM nllllllll.H UBiUiVLI) 0, I1IIX
0.30
- 0.70
0.00
0.00 0.10 0.20 0.30 C.40 0 50 0.60 0 70 O.bB
BOM CiiniillLR OBSERVED G, mX
0.00
Figure 19. Scatter diagrams for mechanism simulations and BOM data,
ozone maxima. Run numbers are identified in Table 14. Replicates
are connected by broken line. Dashed lines are +25%. It is
assumed that the observed 0^ maximum is 90% of the oxidant value
reported.
79
-------
100
350. -
CmiBON BOM) MtCllilMSM
BOM CHMMBLR L'fHfl
100.
350. -
DEMERJIfW MECHflNISM
LIGHT INTENSITY - 0.35
BOM CHHM8ER DflTfl
0. 50. 100. 150. 200. 250. 300. 350. 100.
BOM CHfllffitR UOSERVED TIME 0, MflX
0. 50. 100. 150. 200. 250. 300. 350. 100.
BOM CHRMBER OBSERVED TIME 0, MflX. MIN
DODGE MECHflNISM
BUM CHflKBER DflTfl
100.
350.
300.
250.
ISO.
ICO.
SO.
100. I—
0. SO. 101). 11)0. 200. 250. 300. 350. 100.
BUM CHflMIIER UliSERVtn TlrE 0, mx
1 ' ' I ' ' • • ' ' /
_ CPU TECH MECHflNISM ti l
0. t
0, 50. 100. 150. 200. 250. 300. 350. 100.
6011 CHflimiH CRStRVED TIKE 0> I'fiX. MIN
Figure 20. Scatter diagrams for mechanism simulations and BOM data,
time to ozone maxima. Run numbers are identified in Table 14.
Replicates are connected by broken lines. Dashed lines are +25%.
80
-------
£
c5
P \-
"- Jo
NIW 'XbW "0 3WI1 Q31DI03ad 1300W
isaou
> 13
E
O •'-
KdJ 'XHW '0 a310IC3a=) 13QOW
1300UI
81
-------
5.4 BOM ISOPLETH AND CROSS SECTION DIAGRAMS
In addition to simulating the individual BOM runs, the mechanisms were
used to generate isopleth diagrams for the BOM conditions. These are shown
in Figure 23. The BOM datapoints have been superimposed on the isopleths.
The differences among the mechanisms are more evident in this figure.
The HC-NOX-ratio for maximum 03 formation is much higher for the CIT mechan-
ism than for the others which are reasonably similar. This accounts for the
substantial number of underpredictions by the CIT mechanism. The top
isopleth for Carbon Bond and Demerjian simulations is 0.65 ppm compared to
0.60 ppm for Dodge and CIT.
To obtain more insight into the differences among the mechanisms, cross
sections of the 0.,-precursor surface were generated at constant NO
•3 X
levels and at constant HC levels. These are shown in Figures 24 and 25.
These figures suggest that Carbon Bond and Demerjian mechanisms are more
sensitive to both HC and NOV change than is Dodge. The CIT mechanism is
X
less sensitive to HC change and more sensitive to NO change than the Dodge
""""- •--™" • ' X
mechanism.
In Figure 26 all four mechanisms are compared in two sets of cross sec-
tional diagrams, one at the maximum condition and one at lower, more urban-
like conditions. Also shown at the lower conditions are BOM experimental
datapoints and replicate experiments.
At NO above about 0.2 ppm the CIT mechanism is inconsistent with the
X
other mechanisms and with the experimental data: it appears to be very
inhibited by NO . An examination of the mechanism's reactions (Table 9)
/\
shows that one olefin molecule only gives one aldehyde molecule (the usual
practice is two aldehydes) and that the aromatic molecules only produce a
higher aldehyde which has a lower photolysis rate than formaldehyde. These
82
-------
OOt 0 009 0 DOS 0 OOti 0 CQE 0 OQZ 0 001 0 0
OOi 0 009 0 DOS 0 OOh 0 DOC 0 002 0 001 0 '0
0 700 0 600 0 500 0 tOO 0 300 0 20J 0 100 0
NOX.PPM
o
+->
re
OOi 0 009 0 OOS 0 OOh 0 OOC 0 002 0 001 0
OOi. 0 009 0 DOS 0 OOh 0 OOE 0 002 0 001 0
0 700 0 600 0 500 0 100 0 300 0 200 0 100 0
NOX PPM
0 700 0 tiOO 0 jOO 0 400 0 300 0 200 0 IC? 0
O
CO
o
M-
-i->
O)
Q_
o
01
c:
o
M
CZ3
OO
CSJ
OJ
i.
cn
83
-------
001 0 D0£ 0 009'0 OOS'O OOt 0 OOC 0 002 0 00! 0 0 '
CO
c
o
4J
0}
3
E
O
ca
o
o
+J
10
ts>
c
o
Ol
co
CO
CO
O
O
s_
(O
o
o
-o
i
O)
c
o
N
O
CO
0)
3
o>
-------
008*0 004. 0 009 0 DOS 0 OOi 0 DOC 0 OOZ 0 OOl'O
008 0 OOL 0 009 0 DOS 0 OOfc 0 OOC 0 00"; 0 001 0 0
0 800 0 700 0.600 0 500 0.100 0 300 0 200 0 100 0
03.PPM
0 600 0 700 0 600 0 SOO 0
ra
3
O
CO
c
o
o
D
008 0 00' 0 009 0 OOS 0 OOi 0 OOC 0 00£ 0 001 0 0 .
0 800 0.700 0 600 0 SOO 0 '
; 03.PPM
0 800 0 700 0 600 0 SOO 0 100 0 300 0 200 0 100 0.
| 03.PPM
O
u
to
c
o
o
Ol
to
O
O
X
o
z:
O)
c:
o
IM
O
If)
CVJ
cu
s-
3
en
85
-------
COS 0 OOt D (IDt 0 005 0 00k 0 DOC 0 OOE 0 001 • 0
1 t 001 0 004 0 DOS t DOl'O OOC'O 002 0 001 0 0
CO
c:
o
(0
ll
to
g -!->
O C
CO -i-
t- i.
O
•+- r—
10
I/) 4J
C C
O Ol
<-» i.
O) O)
C/> Q.
X
CO Q>
CO
O 0)
i- S.
O ns
o c
N -r-
O r-
XC
O O)
X
o
Q.
Q.
00
CM
X
o
•a
o
-Q T3
S- O)
ns -*J
CJ o
O Q)
s- c:
-o c:
0 200 0 100 0
200 0 100 0
o
O
«*- CO
O -
o -o
•>- to re
s- o
re + -r-
Q. r—
E cu o.
O .C OJ
QJ
t-
-------
differences suggest that the initial reactivity in the mechanism would
decrease very rapidly in the simulation. On the other hand, there would not
be enough NO loss at high HC-to-NO ratios, so that although the 0, may be
XX o
reasonably predicted, the other secondary products would not be. Careful
adjustment of the mechanism's rates and compositional inputs, however, could
lead to good fits over a narrow range of conditions such as those shown in
Figure 15.
In Figure 26 it can be seen that Carbon Bond and Dodge converge at low
HC and at high NO for medium HC. Carbon Bond and Demerjian converge at
y\
high HC and at high NOV for high HC. Compared to the experimental points,
X
however, it would seem to be difficult to say which mechanism is a better
description of BOM results.
87
-------
5.5 SUMMARY
Four chemical mechanisms, potentially useful in photochemical simple
trajectory models to calculate hydrocarbon and NOV control requirements,
A
have been compared with the Bureau of Mines auto exhaust database. The
mechanisms were: the Dodge propylene-butane mechanism used in the OZIPP
program; the Demerjian mechanism used in his photochemical box model;
the Carbon Bond II mechanism (CB2) used in the SAI airshed grid model;
the California Institute of Technology mechanism (CIT) designed for use
in their airshed grid model.
The BOM database is the only complete and well documented set of
auto exhaust smog chamber experiments. It does, however, contain enough
uncertainties in experimental conditions, analytical methods, and replicate
experiments that comparisons of models and data are not unambiguous.
That is, the "noise" in the experimental results is greater than the
differences shown among the Dodge, Demerjian, and Carbon Bond II mechanism
simulations which have large differences in construction assumptions. At
lower NOV concentrations, this is also true for the CIT mechanism. At
X
higher NO concentrations, the CIT mechanism 0-, reactivity fall s signifi-
X o
cantly below that of the other three mechanisms and the BOM chamber. Thus,
the suitability of the CIT mechanism for control strategy calculations at
medium to high initial NO conditions may be questionable. Given the range
X
of uncertainty in the BOM data (e.g. 03 maxima and initial aldehydes) and
the sensitivity of mechanisms to necessary assumptions, it is concluded
that the Demerjian, Dodge, and CB2 mechanisms were all capable of providing
adequate descriptions of the 03 formation in the BOM chamber; the CB2
mechanism is probably slightly better at describing the other measured
secondary products.
88
-------
6.0 REGIONAL AIR POLLUTION STUDIES DATABASE:
EPA SUPPLIED INFORMATION
6.1 GENERAL DESCRIPTION
The Regional Air Pollution Study (RAPS) was conceived early in 1970 to
provide a rational, scientific basis for the management of air quality, as
mandated by the Clean Air Act (as amended). The basic premise of the Act
is that desired air quality standards can be obtained by setting appropriate
emission standards. The development of Implementation Plans, called for by
the Act, assumes that existing knowledge was at least minimally adequate
for planning.
A basic tool for the development of air quality management is the
simulation model, a mathematical description of the complex relationship
between emissions, atmospheric dispersion and transformation, and ambient
concentration. The development of any model presupposes: 1) a detailed
understanding of the physical, chemical and meteorological process involved;
and 2) availability of adequate emission data, meteorological information
and measurements of ambient concentrations of the pollutants under investi-
gation.
At the beginning of the RAPS study, a number of simulation models had
been developed, but few -if any- had been verified in the field. The pri-
mary reason for this was the absence of an adequate database, which would
contain accurate, high resolution data covering a sufficiently large area.
Ambient data were available with adequate time resolution -one hour or less-
but the stations providing such data were typically too few and improperly
spaced to cover a given area. Meteorological data were usually available
89
-------
only at very few points in a given area, such as an airport, where they
were gathered for other purposes. Micrometeorological data related directly
to ambient measurements were generally unavailable.
Emission inventories have been in existence for some years, and owing
to the efforts of the National Air Data Branch of OAQPS, were being collected
in a uniform, machine readable format known as National Emission Data System
(NEDS). However, the NEDS inventory contained essentially only annual data,
which cannot readily and reliably be converted to hourly values over the
two year RAPS program period.
Clearly, what was needed as a first step in the development of a rational
approach to the management of air pollution was an extensive, detailed data-
base containing all these elements: emission, meteorological, and ambient
data, with a resolution in time and space and an accuracy adequate to pro-
vide an input to simulation models.
The St. Louis Interstate Air Quality Control Region (AQCR 70) was chosen
as the site for RAPS (Figure 27). The selection was based on the need to
find a large city within the continental United States, which was away from
oceans and mountains and which typified the coal-burning industrial nature
of many urban areas, yet which lay in an extended region of rural country.
Of the 33 Standard Metropolitan Statistical Areas larger than 400,000 popu-
lation, St. Louis emerged as the clear choice based on the following criteria:
- Surrounding Area
- Heterogeneous Emission
- Area Size
- Pollution Control Program
- Historical Information
- Climate
90
-------
ircles denote radius
in km from Jefferson
Arch Memorial in down-
town St. Louis.
50
FIGURE 27. THE REGIONAL AIR MONITORING STATIONS NETWORK
91
-------
CD
CD
CD
IO
UJ
O
s:
TOTU
4330
4320
4310
4300
4290
4280
4270
4260
4250
4240
*o^n
RftMS STRTION LOCflTJONS -
~~ +22 ~~
— —
»44 ~~
+14 +15
• O
— +13 *Q +16 —
+20 +2 23 + _
+ 25 +7 +3
A"\ 0* 1 *-7 *"""""
6^* *r ^4
+ 11 +10 +17
- +19 —
»42
+ 18
« 43
+ - RflMS STflTIQN
— o - URN STRTIQN —
+24
— «••
,1,1,1.1,1,1.1,1.
I OTU
4330
4320
4310
4300
4290
4280
4270
4260
4250
4240
493 0
690 700 710 720 730 740 750 760 770 780
UTM ZONE 15 1000M
Figure 28. The Regional Air Monitoring Station UTM coordinates.
92
-------
In the RAPS program 25 regional air monitoring stations (RAMS) were
placed concentrically throughout the study region (see Figure 27). The
stations, which were numbered 101 to 125, were thought to be located where
they would riot be unduly influenced by any one source or group of sources.
The network was operated for two years, 1975 and 1976.
The RAPS database is maintained by EPA in several forms. Most readily
available for each station are hourly average concentrations for: ozone,
nitrogen dioxide, nitric oxide, total oxides ot nitrogen, nonmethane hydro-
carbons, sulfur dioxide, and carbon monoxide. Each station also measured
wind speed, wind direction, and temperature; these too are available as
hourly averaged values. The RAPS program also included an Upper Air Sounding
Network (UASN). In this network, radiosondes were released three times per
day, five times per week, at at least two stations. From these soundings a
vertical temperature profile could be derived and the extent of atmospheric
mixiny could be estimated.
Figure 28 gives the locations of the RAMS stations and the four UASN
stations on a UTM grid coordinate system (the RAMS stations have been
numbered 1 to 25, the UASN stations are numbered 41 to 44). Station 1 was at
the Arch. This plot will be the basis for showing the trajectories and
emission areas in Section 7.1.2. The RAPS program also included a new and
hourly resolved area and point source emissions inventory for a number of
pollutants (Littman, 1979). Figure 29 identifies the major point sources.
The area source inventory was spatially resolved using a variable-sized grid
2 2
system in which the smallest grids were 1 km and the largest were 100 km .
The grid system is shown in Figure 30. A detailed section of the grid
system near the center of the city will be shown later as Figure 65.
Estimates of emission rates from both area and point source's are available
by hour and by grid for any day in 1975 and 1976.
93
-------
6.2 DAYS SELECTED
The days used in this study were the same days selected by OAQPS/EPA in
their evaluation of EKMA (Gipson, 1980). These days were selected to include
the days with the highest ozone levels and to include a range of atmospheric
conditions (season, wind direction, mixing heights, ozone aloft, etc.)- In
addition, a few days with lower ozone concentrations were included to test
for performance near the 03 standard. Table 15 identifies the days selected.
Table 15. Model Test Days
Time of Peak 0,
Peak
Concentration,
Date
10/1/76
7/13/76
6/8/76
6/7/76
6/8/76
8/25/76
10/2/76
9/17/76
7/19/76
8/8/76
Julian Day
275
195
160
159
160
238
276
261
201
221
RAMS Site
102
114
115
122
103
115
115
118
122
125
Local Daylight Time
1500-1600
1600-1700
1700-1800
1600-1700
1400-1500
1400-1500
1700-1800
1300-1400
1300-1400
1800-1900
ppm
.24
.22
.22
.20
.19
.19
.19
.15
.15
.12
94
-------
ii.iu.ii! ,!...!„!!,
5 H . ? -- I fM^f'ill-S.si"'* !
"*"i?1«35i;-s's-!?*M3aifii;
5fSgii°E~ls
-------
96
-------
6.3 INFORMATION SUPPLIED BY EPA
For each of the ten days during the 1976 RAPS study, EPA supplied to UNC
the data used to run the DODGE model (OZIPP) and additional information that
was needed for the different mechanisms. This information included:
a) trajectories for each day as both UTM coordinates and maps.
b) the emissions along the trajectory as used in the DODGE/OZIPP
simulations and also in as much detail as is present in the emissions
inventory.
c) the boundary conditions, both transport and aloft,used in all the
DODGE simulations and the data used to generate these conditions.
d) the ambient concentrations estimated along each trajectory and the
individual data points used to estimate these values.
e) the mixing height data used in the DODGE simulation and the sounding
data used to generate this data.
f) example results of the DODGE/EKMA simulations for selected days and
copies of each isopleth figure generated.
g) other data needed to accomplish the tasks.
97
-------
7.0 INPUT DATA FOR SIMPLE TRAJECTORY MODELS FOR RAPS DAYS
7.1 TRAJECTORIES
7.1.1 Methods of Determining Trajectories
The determination of a trajectory is an implicit input to a simple tra-
jectory model. An accurate trajectory is important because ambient data and
emissions into the parcel are determined along its path. The first problem
in determining the trajectory is to select the height at which it is to be
determined. In most cases the 10-m wind is chosen; this is the usual obser-
vational height. Being in the surface layer, however, this height may not be
the best one.
The surface layer is generally characterized by strong gradients of wind,
temperature, and humidity. Thus a level in the higher mixed layer where
the gradients are more uniform may describe the true winds more accurately.
Measurements of these winds can be obtained using bistatic acoustic sounder,
lidar, towers, balloons and aircraft.
In most cases the 10-m wind and possibly one or two rawinsonde obser-
vations may be available. From this information, models and smoothing
techniques can give a mass consistent wind field from which a trajectory can
be determined (Dickerson, 1978, Goodin, McRae and Seinfeld, 1980). The
concept of a mass consistent wind field is important since in a convective
boundary layer strong convergence can occur (Shreffler, 1978). The
neglecting of the vertical wind component in a trajectory model can produce
significant errors (Liu, Seinfeld, 1975).
98
-------
An even more simple technique of calculating trajectories is to use a
time averaged wind vector from the observing stations, weighing the closest
2
stations to the point of interest by 1/r.
If sufficient wind data are not available but there is sufficient atmo-
sphere pressure data available for a region, then a wind flow may be estimated
assuming a geostrophic balance. In an urban area, however, the difference
in pressure measurement may not be resolvable within the accuracies of the
instrumentation.
7.1.2 Trajectories for RAPS Days
The trajectories for the RAPS days were determined by OAOPS/EPA personnel
and were supplied to UNC on request. The general procedure used by EPA was
to perform backward trajectory calculations from the vicinity of the RAPS
station with the high CL, starting at the time of the observed high CL. Two
different computer programs were used depending upon station density (Gipson,
2
1980). In general, these programs performed a 1/r weighting of the three
closest wind stations to generate a wind vector, then moved backwards along
this vector and repeated the process.
Five trajectories were run for each site-day, one from the station
itself and four displaced 5 Km North, South, East, and West around the
station. As will be shown below, these trajectories sometimes converged,
sometimes crossed, and sometimes diverged, leading to a large uncertainty
in the initial location and path history of the air parcel.
Given the uncertainties in the trajectories, OAQPS/EPA personnel gene-
rated a "box" for each hour by connecting the extreme positions of the
trajectories for that hour. Thus, the probable trajectory was assumed to
be contained somewhere in the box. The "box" was used to select and average
the emissions sources for that hour. This process will be described in the
99
-------
next section. On some days, a bigger "box" was drawn around multiple hours
of trajectories when little motion had taken place.
In the following discussion, the maps with boxes were supplied by
OAQPS/EPA. They also supplied tables of trajectory and box coordinates.
These have been plotted to more clearly demonstrate the basis of the
calculations.
100
-------
4340
4330
4320
4310
4300
4290
4280
4270
4260
4250
4240
700 710 720 730 740 750 760 770 780
Figure 31- Trajectory for Day 159, Juno 7/Site 122
101
-------
4330
1330
730
Figure 32.
750
UTM ZONE 15 1000M
Individual trajectories, Day 159.
-I ^320
I
DRY 159 JUNE 7 1976
M
D
rn
I-*
in
1330
4 r) q
40
/SO
^QNC il) lOOQM
no
Figure 33. Boxes generated from trajectories, Day 159.
102
-------
4340
4330
4320
4310
4300
4290
4280 />-'f
4270
4260
4250
4240
700 710 720 730 740 750
Figure 34. Trajectory for Day 160, June 8/Site 115
760 770 780
103
-------
1330
1320
1310
S 1300
M
1290
I 'i^T1I ' I ^
DRY 160 JUNE 8 1316 SITE 115
44
13
16
* 20
12
f7 3* i
I I I
^330
1310
n
f2SO
700 710 720 730 740 750 760 710
UTM ZONE 15 1000M
Figure 35. Individual trajectories, Day 160, Site 115.
1280
1320
o
a
CD
In
UJ
a ^300
| ' | *22 ' I ' I >
DRY 160 JUKE 8 1316 SITE 115
»44
J.
L
J I 12»f
r
f I . r
J_
1-310
1-230
1-280
700 710 720 730 "HO 750 760 110
urn ZONE is IOOOM
Figure 36. Boxes generated from trajectories, Day 160, Site 115.
104
-------
4340
4330
4320
4310
4300
4290
4280
4270
4260
4250
4240
700 710 720 730 740 750 760 770 780
Figure37- Trajectory for Day 160, June 3/Sitc 103
105
-------
H320
1310 -
DRY 160 JUNE 8 13"76 SITE 103
- 1210
1260
700
720 730 710
UTM 7QNE 15 10QQM
750 160
4260
Figure 38. Individual trajectories, Day 160, Site 103.
1320
13U
1300
- 1230
Q
h-4
1280
1210
DRY I €0 JUNE 8 1416 SITE 103
440
21
14
4-8
o 42
15,
^300 =
rz
o
-z.
rn
^280
700 710 720 730 7tO
UTM 7QNE 15 1DOOM
750
160
4260
Figure 39. Boxes generated from trajectories, Day 160, Site 103.
106
-------
D159 Trajectory. Figure 31 shows the map of the derived trajectory
path for June 7. Figure 32 shows the coordinates of each backward trajec-
tory for each hour and the nearest RAPS stations for which the wind and
ambient data were derived. Figure 33 shows how the "boxes" were generated
from the trajectories. Only one box was used for the first four hours of
these trajectories.
D160/103 and D160/115 Trajectories. Two different sites were used on
June 8, one in the center of the city (102) and one just north of the city
(115). Trajectory calculations for these two sites illustrate some of the
problems with the method used. The backward trajectories from site 115
(Figure 35 ) generally maintained about a 10 Km spread showing some narrowing
from 10 am to 9 am. Most of the close wind stations were south of the
trajectory. The 103 site, however, is surrounded by wind stations (Figure 38).
After three hours of back calculations these trajectories diverge dramatically,
ultimately becoming wider than the whole city!
Figure 40 shows both sets of trajectories on the same plot.
Other Days Trajectories. Maps, trajectories (where supplied) and "boxes"
for other RAPS days used in this study are shown in Figures 41 to 54.
The trajectories and "boxes" for October 1 (Figures 52 and 53) were
not done by computer program but were done by manual methods and detailed
examination of data printouts (it was the first day to be studied by OAQPS).
UNC was supplied no supporting documentation for the shapes and locations
of the "boxes" shown in Figure 53. These boxes were used to select
emissions sources.
107
-------
UTM ZONE 15 1000M
CD
co
CO
CD
CM
CO
CD
CO
CD
CD
CO
si-
cn
CM
CD
CO
C\J
CD
r-
C\J
CD
UD
CM
cn
CD
UD
r-
CD
LO
C3
CO
CD
CM
CD
CD
CD
CO
co
CD
CM
CO
CD
<—t
CO
CD
CD
CO
CD
cn
CM
CD
CO
CM
CD
r-
CM
CD
^o
OJ
CD
CD
CD
00
O)
-M
00
0)
£ I
O)
o
to
E I
TD >
"O
c
o
l
(U
Q.
O
^1-
Ol
S-
3
cn
WOOD! ST JNOZ NIP
108
-------
4340
4330
4320
4310
4300
4290
4280
4270
4260
4250
4240
700 710 720 730 740 750 760 770 780
Figure 41. Trajectory for Day 195, July 13/Site 114
109
-------
1310
1300
1250
z:
o
o
o
l/J
UJ
Q
M
1280
1210
1260
*250
DflV 155 JULY 13 1376
+ 21
730
13*",
16
740
750
760
W10
1300
^230
O
rn
o
o
1210
*260
1250
770
UTM 20NE 15 1000M
Figure 42. Boxes generated from trajectories, Day 195,
110
-------
4340
4330
4320
4310
4300
4290
4280
4270
4260
4250
4240
700 710 720
730
740 750 760 770
780
Figure43- Column Path for Day 201, July 19
111
-------
13 n
1330
1320
1310
1300
i
§ 1290
ui
1280
1210
1250
12tQ
20
720
I ' I
211 JULY 11
122
13
19
18
44
»
15
16
17
43
730
7ft
7SO
7*0
DTM TONE 15 1IO«M
Figure 44. Individual trajectories, Day 201
112
mo
124 ft
en
1281 £
I
T70
-------
4340
4330
4320
4310
4300
4290
4280
4270
4260
4250
4240
700 710 720 730 740 750 760 770 780
Figure45- Trajectory for Day 221, August 8/Si'te 125
113
-------
UTM ZONE 15 1000M
o
CD
CD
LiJ
O
CM
CM
CO
OJ
S_
O
O
CU
to
3
T3
•r—
>
TD
c:
cu
WOOOI SI 3NOZ Win
114
-------
4340
4330
4320
4310
4300
4290
4280
4270
4260
4250
4240 ~r~
700 710 720 730 740 750 760 770 780
Figure47. Trajectory for Day 238, August 25/Site 115
115
-------
1310
f300
1290
1280
1270
4260
44
DRY 238 flUGUST 25 1976
23
+ 1
41
18
+ 3
10
17
1310
1300
1290
en
f280 g
t270
750 760 770 780
UTM ZONE 15 1000M
Figure 48. Individual trajectories, Day 238.
790
1310
1300
12SO
iu
3 1280
12"? 0
I ' I 'I ^
DRY 238 flUGUST 25 1916
» 14
_U
1300
c:
M
o
1280
740 750 760 T70
UTM "ZONE 15 1000M
160
ISO
Figure 49. Boxes generated from trajectories, Day 238.
116
-------
4340
4330
4320
4310
4300 r
4290
4280
4270
4 260
4250
4 240
"A^w^life--
700 710 720 730 740 750 760 770 780
Figure 50. Trajectory for Day 261, September 17/Site 118
117
-------
1310
1300
1230
z:
o
o
2 1280
UJ
Q
M
1270
1260
«•
20
720
I
DfiY 261 SEPTEMBER 17
*14
730
740
750
UTM 70NE 15 1000M
Figure 51. Individual trajectories, Day 261.
118
15
*
9
WOB
1230
c
M
0
rn
1280
*270
1260
760
1250
-------
4340
4330
4320
4310
4300
4290
4280
4270
4260
4250
4240
700 710 720 730 740 750 760 770 780
Figure 52 Trajectory for Day 275, October I/Site 102
119
-------
UTM ZONE 15 1000M
2:
o
o
o
LO
CL
-
-o
O)
s_
cu
c
0)
CD
l/l
O)
X
o
CO
ro
LO
QJ
S-
3
WOOOT ST 3NOZ Win
120
-------
4340
4330
4320
4310
4300
4290
4280
4270
4260
4250
4240
700 710 720 730 740 750 760 770 780
Figure 54 Trajectory for Day 276, October 2/Site 115
121
-------
7.2 AMBIENT DATA ALONG TRAJECTORIES
The ambient data alonq the trajectories were supplied by OAQPS/EPA. An
example of the ambient data printout is shown in Table 16. The table
information was generated as follows. Given the midpoint of the trajectory
path for each hour, the program searched the RAPS database for the three
nearest RAMS station (see the second and third columns of Table 16). The
hourly averaged concentration of each species and the number of minutes used to
compute the average is shown in columns 4 to 7. The hourly average in the
hour before and the hour after the calculating time were averaged to
estimate the value on the hour. That is, to calculate the 0900 LOT values
for 03, the 0800-0900 LOT average and the 0900-1000 LOT average were averaged
to give the value at 0900 LOT for each of the 3 closest stations. To
estimate the probable value at the center line of the trajectories, the
average values at the three closest stations were combined into a weighted
average using the square of the reciprical of the distance between the
trajectory location and the station location (column three, Table 16). This
weighted average is shown in the last column in Table 16.
The average values for each day were used to select the simulation
initial conditions. These values are given in Table 17.
Figures 55 to 64 show the ambient concentrations along the calculated
trajectories as a function of time. Also included are the temperature and
total solar radiation for each day. Three lines are shown in each plot. The
2
solid line connects the 1/r weighted average points. The bottom dashed
line connects the value from the station with the lowest value of the three
used to compute the weighted average and the upper dashed line connects the
value from the station with the highest value of the three used to compute
the weighted average. The hourly averaged values are centered (plotted on
122
-------
1 UJ
a
a
1 u^
1 .
t-Ll
1 0
i
, ' >
•*
1 i i
8
0
1 1
p"
^ i
^ s '•: [
o - • o ! i
-4-1 UJ !» O O
c sc ;» <*• o
il >- ,* i
« 'I
-t-> ' i
re ! i i
1
+J -*
C 1 **7
OJ i *
'.5 ! , : *~
c ' w.
< . a
ir-
r.
- ;
0
-
r*
0
_ <
•^
*ri '°
• •
o o
0
p -^
3 , -
O pl-
^rt
o
C) "&
eno
. •:
=f U'
* -T
- (V
i C
j
i
1
1
i
c
i
~0
t (X
. ".
• •
f r~
'"•':
'•** j
^ A
r ^
c. o
•• o
V 3
•f CC
"^ ^<
,_,
,*.
<
3
^
f*1
^
.,
.,«
^
•
o
• t
"V
>,
n
>
CT
r\
•^
a
(M
1 O
h^ '
1 1
;
~v r*- f-
.1 o i r, m
m tx . j
• • . • •
i • j
1
- o- ! '<>• r-
1
b o r- 'V.
13 o i .
. (.I, -. -0 ^,
!-» r^ l.i f
1
O '^C rx ^ ' -I
r pj
P Q *" 'V
O O '
r, ID- in \~- IT
o u. K ! '<" 0
1 (") i -n !**•
;:
^ icv - ' ro r- :.- ! r- 0
^ r- ^j ^c r- o ' < r- 3
0 0 0 t ^ 1C. 7iS
i ' | " i_ •..
| r.. r. ri
1 , j
•Ju I J
'" r < i >H '~ r
•-1 »-4 ^-( i / ' C\. -X C\J
1,1 + 4' V 4 V *
-. Iv -; fc ,; rj 1C ^ f"
; | i
*-< p r\ K c : ^ r~ r^
«f, .,, „-,,
-.!-.. r> r-. p (. iv .,
o p o r- ^ n ; f> i- o
•_ \T O 0 C- " "- -^
f» r- it cr- o p> O ( *-
! , 1 '
i
r\ ct ^
-------
Table 17
Simulation Initial Conditions
DAY
159
160/103
160/115
195
201
221
238
261
275/6
275/8
276
NMHC
0.49
0.26
0.10a
0.10a
0.10a
0.10a
0.22
0.20
1.10
1.89
2.98
NO
0.0082
0.0445
0.0105
0.0027
0.0052
0.0051
0.011
0.0199
0.0265
0.1466
0.1219
NO 2
0.029
0.0121
0.0037
0.0028
0.0026
0.0047
0.0096
0.0307
0.0886
0.0553
0.3644
°3
0.034
0.011
0.062
0.059
0.056
0.016
0.010
0.008
0.005
0.005
0.005
CO
0.731
0.484
0.200
0.115
0.159
0.185
0.581
0.872
1.319
3.153
9.408
assumed minimum value
124
-------
the half-hour tic mark) and the average value between the hours is plotted on
the hour tic mark. Because of the range of conditions for these 10 days,
three different scale factors had to be used for NOV, NMHC, and CO plots.
X
The 0, plots used one scale factor.
125
-------
o 20
o oo
I ' I ' I ' I ' J
JUNC 7. 1976 —
rt
2 00
1 80
1 GO
1 ^0
1 20
1 00
0 80
0 60
5 6 / 8 9 10 11 1? 13 1J 15 16 I/ 18 19
HOURS. TDT
0 20
0 00
1.00
0.90
O.BO
0.70
e 0 60
0.
^ 0.50
O
I o.io
0 30
0.20
0. 10
0.00
1 i ' I ' I ' I ' I ' I ' I ' I ' I ' I ' 1 '
JUNE 7. 1976 -
5 6
9 10 11 12 13 11 15 16 17 18
HOURS, EOT
19
1.00
0.90
0.80
0.70
0 60
0.50
0.10
0.30
0.20
0.10
0.00
--]—r—|—r
JIM. 1 19/6 _.
9 10 11 12 13 11 15 16 I/ 18 19
HOURS. EDF
120
100
096
084
0/2
060
018
036
024
012
nnn
250 r-r i , i -T-
225
-
200 -
175
150
8 125
0 I0°
075
050
-
-
-
-
-
-
025
000
" , 1 , 1 ,
-y-ry-r-pry-T
| i | i , T
JJNC 7. 19/6
56/8
10 11 12 13 14 15 16 17 18
HOURS TOT
19
250
225
200
175
ISO
125
100
075
050
025
000
9 10 11 12 13 H 15 16 II IE
HOURS TDI
19
120
108
0<36
081
072
060
048
036
021
012
000
I ' I ' I ' I
i r r^ i ^r
JUNE 7. 1976 _
I , I i I . I
10 11 12 13 11 15 16
HOURS, LOT
17 18 19
100.
90.
80.
70. _,
rn
60- 1
•yo
50. |
10. m
30. •"
20.
10.
0.
Figure 55. Ambient data, weighted mean, max and min for Day 159 trajectory.
126
-------
2 UU
1 80
1 60
1 to
; 20
1 00
r- 80
0 C 0
0 to
0 ^5
0 00
-r-n-r-n i-m
-
-
;-
-
' 1 ' 1 '
*
;'Vxi
5 6 7 8 9 10 11 12 13 1
HOURS r:
JUN
. * ,-^-<
•4 l^ [
: a
6 1
' 1 '_
1976 _
_
-
-
-i
•/\ "
-
L '_!_•
7 18 1
1 80 0 45
1 60 o 40
! "0 o 35
1 20 " r, 30
1 00 0 2r
0 80 £ Q 20
0 GO o IS
0 to Q ,0
0 20 o 05
0 00 o 00
9
_'' 1 ' I ' 1 '
-
-
-
-
-
•-
-
i 1 , 1 , 1 ,
5 6 ; a
•-»-•
,
-"
9101
1
P
S/
i
^
A
,
• , | 'I'
\^
, i
JUM
.--?-*
,
— ] —
L 8
•~~»-J
,
1 1 '
1976 _
—
—
-
-
-
—
,-Z-i -
, 1 ,
1 12 13 la 15 16 17 18 1
r HOURS
U 3D
0
-------
CJ
F
£
O
i_)
0.
0.
o
^
c.
r.
0
2 00 r-r p- | r
1 80 j-
1 60
1 ^ 0
1 20
i 00
0 BO
0 60
0 d 0
0 20
0 00
-
-
-
-
7 /
-
_
-, i , T ;
567
10«
0%
OB'l
072
000
0J8
030
024
012
000
-
-
—
-
-
"
*/*
",!,*«
5 6 7
J20 -r-, _,--,
,03 -
096 —
OB4 —
L
o;2 L
rjeo
r -5
03C
Ci2
-
>
| -, T-T-J - r-[ -r-T-r-pr- [ -i-pr-pr -pr-pr-
JUNf 8 197G -
"A"
/A
/ \ ~
/ A'
\ A
Xy A
• V-^-\^:^-'
-
i , i , i . i , i , i , i . i , i , , j.. , "
8 9 10 11 12 13 1« 15 16 17 18 1
2 00
1 80
1 00
1 "0
1 20
1 00
0 80
0 00
0 «0
0 20
0 00
9
HOURS. EDr
JUNE B, 1970 _
-
—
-
-
-
-
"*x«
* i PTf-*-»-*-?-»-<-4— *-+-<-i , 1 _j 1 , I ,
8 9 10 11 12 13 11 IS 16 17 18 1
HOURS EOT
~f n j l | l |~T— ] i p i | , | , | i | , | ,
JUNC 8 1976 _
-
-
-
-
\ ;
V "-
\^A < o-^V
* \ fi~~~-/
* . 1 , 1 . 1 V 1 , l T 1 ' 1 , 1 . 1 .
100
096
08'<
072
000
018
036
024
012
000
9
120
108
096
081
072
060
O-'B
030
024
012
nnn
r 1 r1 i ' i ~r i r~i T T i ' r n i ' ' ' ' '
0 90
0 80
0 70
0 60
0 50
0 "0
0 30
0 20
0 10
n nn
JUNE 8. 1976 _
r ~
_.
- —
A t -
l\* f
- ^J Y \** - J
m^^^* * -i \ / \ H *,&—£ -i
V"; ^\/ \
l . - V . ' . j
• , , , i , j . i , i , , i , i , i . i , . , '
i UU
0 90
0 80
0 70
0 60
0 50
0 ^0
0 30
0 20
0 10
n nn
10 11 12 13 11 IS 16 17 18
HOURS. EDr
19
, £ -J V
225
200
175
150
125
ill
075
050
025
-i | , | , | , | , | , f , | , | , | , | , | , | , | , _
JUNE 8. 1976 _
7" f ~
' sJ-V"
t-^^i -
1 T
I ' ~
JT 1.
S -
~ * ^S ~-
nnn , 1 , T ' 1 , ! • 1 , .1.1 I.I,
dO
-------
J'JL'
-,-, 0 SO
76 J 0 15
J 0 to
-| 0 3S
- o «
I
OHO
250 r,-T
225 -
200 h
100
FT' r
JJLY 13. 1976
10 11 12 13 H 15 16 17 18
HOURS FCT
19
250
225
200
175
150
125
100
0/5
050
025
000
18 19
n ' i ' i ' i ' i ' i ' i ' i ' i ' i ' i ' i
HOURS ft) I
6 1 8 9 10 U 12 13 11 15 16 17 18
HOURS. LOT
19
Figure 58. Ambient data, weighted mean, max and min for Day 195.
129
-------
\J »J V
0 IS
o to
0 35
't 0 30
Q.
D.
. 0 25
f 0 20
0 15
01 A
1 U
0 05
n on
_ ' | i | i | i
-
-
i-
—
-
-
— y
-^^»
1
",1,1,1,
' TT I ' I '
V-
Kj
« ^
,1,1,1,
1 i '
< X
A.
/
/
, ! .
1 1 '
ULY 19.
. 1 i
!
197
p
— !
6 _
-
-
—
-
-
-
~
I ,
'J J\J
0.45
o to
0 35
0 30
0 25
0 20
0 15
01 n
i U
0 Q5
n nn
5 6 7 8 9 10 H 12 13 1 '< 15 16 17 18
HOURS.
19
015
036
032
028
f 024
" 020
r?
^ 016
012
003
004
000
1 '
—
-
-
-
-
—
-
" , 1 ,
5 6
-r-
\-^^
,
7
-r | , | , | . [ r-p- -p-r-|-r-| p-r-y-r-j
JUL' 19 1976 __
-
-
-
_
• - A J
< fl! \
>-«-/ ^k, -
. 1 , 1 , 1 , I . I , I , I , I , J , | '
B 9 10 11 12 13 1" 15 16 17 18 1
HOU^S. EDT
nun ->c n
y T u . 6 j j
03b 225
032 200
028 175
024 150
020 | 125
016 0: 100
012 .075
008 050
004 025
9
-
-
-
-
-
-
~-
'
L
1
t
G
— i —
-A-,
7
i -| , )-, | ,
*•/
--A' r
-AA/
. '
a 9 10 n i
HOURS
" ' 1 '
^
/
, 1 ,
2 13 1
, CDT
1 ^ — j — i — '
JULY
• ! .
•* 15 i
!
19
,
6 1
' 1 '.
1976 _
-
-
_
,
-
:
_
, i , "
7 18 1
250
225
200
175
ISO
125
100
075
050
025
000
9
5 6 ; 8 9 10 11 12 13 1' 15 1G \1 18
0 o
£ o
I ' 1 ' I ' I ' 1 •' I ' I ' I ' I ' I ' 1 ' I
1 r1 i '
JULY IS. 1976 _
3 10 11 12 13 14 15 16 17 18 19
HOURS. LOT
100.
90.
80.
70. -H
rn
60- I
33
50. |
40. S
30. ""
20.
ID.
0.
Figure 59. Ambient data, weighted mean, max and min for Day 201.
130
-------
2.00
1 00
1 60
1 -»0
1 2')
1 00
0 SO
0 CO
0 J0
0 21
0 "i '.'
036
032
028
02"
Ir'O
110
112
,,08
;')'
136
132
VO
'124
"20
.116
"12
008
>JOJ
000
-i y-r-pT- p-| -r-rrT-T^rTT-T-r-r-T--p-r-r-r-r-r-
_ flUGUST 8 19/6 _
--
-
"..
_
_
- ' - * ~
""•X
y ~^^~^'~A"-« -*-< * * ^^^t"^'^^ *
I 1,1 l,t, ",_],!,, j !__, _[ , 1 , 1 .
S 6 7 8 9 10 i; 12 13 !•' 15 15 17 18 1
FlUGUST 8, 1976 _
-
_
-
-
_
D
" , 1 , 1 , 1 . 1 , 1 , 1 , ! , 1 , 1 . 1 , 1 , 1 , 1 .
5 G / 8 110 11 12 131' 15 16 17 181
H3UP1 for
. ' 1 ' ' 1 ' ! ' i ' 1 ' 1 ' 1 ' 1 ' 1 ' 1 ' 1 ' I ' .
_ flUGUSF 8 1976 _
-
_
-
-
" « A < " ' ~~
X * * ft
** * ' A^
J* ^-f/ ^
^T^^^^^'t^^TT^-t;^-""" ^ , ._:
' ' 1 1— ' 1 ' J-L i_l_J_!_L_l_l_J_J_Ll_l J _ iJ .i.A- 1 \
2 00
80
1 60
1 .-o
1 20
1 00
0 CO
0 Gl
0 ")
• 0 20
0 00
9
nan
036
032
028
02'
020
OlG
0 ; ^
00"
00'
9
03G
032
023
02-i
020
OlG
012
009
00'
000
225
200
175
150
125
100
075
050
025
000
/r~l ' ' '
-
-
-
-
-
-
A-
- A/^
" , 1 , 1 • » -
5678
r~ j — '"j — ' — | — i [ — I — | — i — | — i | i — j r^
RUGUSr 8 1976 -
-
-
-
' "-^
^S*^' ~
j^f*-^"*^
' ' ' ' '
-
1 , . ,1,1, ,1
. d J'V
225
200
175
150
125
100
075
050
025
nr, n
11 12 13 H 15 16 17 18 19
HOUR", for
8 9 10 11 12 13 14 15 16 17 18
HOURS FOT
19
12 13 It 15 16 17 18
HOURS, LOT
Figure 60. Ambient data, weighted mean, max and min for Day 221.
131
-------
JU
1 80
1 60
1 "0
o 1 20
a 1 00
S 0 80
0 60
0 "0
0 20
n nn
. ' ' ' ' ' I ' 1 r 1 ' 1 ' 1 ' 1 ' 1 ' "~^l [ I ' .
_ flUSUST 25 1976 _
-
-
-
X A X X
— x —
A - /\T -
\ * / \'
V— fvi-« / ^
* ' , • * • J -
" , i , i , i , i ; i , , i , i , i , i , i , i , i , •
C \J\J
1 PO
1 60
1 10
1 20
1 00
0 80
0 60
0 "0
0 20
n nn
5 6 7 8 9 10 11 12 13 1» 15 16 I/ 18
HOURS. CDT
19
flUbUSF 25 1976 _
010
036
032
028
024
020
016
012
008
004
15 16 17 18 19
p-
y-r-
RUGUST 25 19 76 _
V
-
i . ! , I , !.. I i I . I i i . .i_i_Lj-J . I
7 8 9 10 II 12 13 pi 15 16 17 18
HOURS, COT
19
036
032
028
024
020
012
008
004
000
U J U
0 ^5
0 J0
0 35
C.J
° 0 25
CJ
j 0 20
0 15
0 10
0 OS
0 00
25 0
225
200
175
150
n 125
o I™
0/5
050
025
QOO
2 00
1.80
1.60
t 1.40
f 1.20
'{=
r i.oo
o
u 0.80
° 0.60
1C
°= 0.10
I
8 0.20
n nn
T-pr-p-r— p— | ' i , , j i | , | , i | i i—
- ^ RUGUST 25 1976 _
^
r\
^A \ -
=< V \
\ / 1
\ * I \
1 . . ^-5S»J ._, J
" , 1 , 1 1 < 1 , , 1 , ,1,1,1 , 1 , , "
5 6 7 8 '1 10 11 12 13 la 15 16 17 18 1
HOURS COT
--.-pry-i'I ' | ' 1 ' 1 ' 1 ' F ' 1 ' 'I'l1 ' _
1 RUGUST 25. 19/6 _
- -
/^ -
/ -
/ -
S'* ' ~
- /
?''
-------
I r-p-r
' I ' ! ' 1 ' 1
CR 17. 1976 _
-
-
-
-
-
—
-
i 1 . 1 , 1 i "
3 60
3 20
2 GO
2 -"0
2 00
I 6"
I 20
0 £>0
0 ~G
0 00
. r i ' i : * i i ' i ' i ' i ' i ' ' i '• ' i ' i '_
A SCPTEPfCK 17 1976 _
/ \ -
:. . \
/ V
/ • --
•• / - •
/ ' '•>
"- . '-V,, -
. 1 J I • 1 . 1 . 1 , 1 1 "« 'f *-• -+ . . I , I I I , J
5 6 / e 9 i o i j i .' 13 it i s 16 i ; i a i
HOUR': nu
120 «'•>
108 225
096 200
OBJ 175
0/2 'SO
C60 R 125
r^s o 100
C.'6 0/5
Q; i 050
012 025
q
-' i ' i n~' i '' i ' ""'' ' i ' i ' f ' 'MM.
_ S£PFEM3CR 17 1976 _
-
X -
/
/
x ;^ '
^^^/^ 3
5 6 / 8 9 JO 11 12 13 H 15 16 17 IE 1
HQJRS TDT
225
200
175
150
125
100
OJ5
050
025
000
g
10 11 12 13 11 IS 16 17 18
HOURS. LOT
19
Figure 62. Ambient data weighted mean, max and min for Day 261.
133
-------
1 ' 1 '
V"
>V
V
L: 1 .
Oil!
HOURS
1
N^
-
1 i 1
-r-p-p-IT -P-P~
OCTOBER 1 1976 _
-
_
_
-
-
XyV -:
, 1 ' 1 , t , 1 , 1 , •
2 13 1-4 15 16 17 18 1
ED
10
9
8
7
6
5
4
3
2
1
0
9
r
0
0
0
0
0
0
0
0
0
0
0
4
3
3
2
I 2
° 2
CJ
F i
~Z
i
0
0
0
oo p-p-p-pnm
60
-
20 •-
80 1-
UQ [_
00
60
20
80
J0
00
- y\ -,
•
-------
OC1QRCR 2 1976 _.
10 n
9 i
8 0
7 0
6 0
5 0
•* 0
3 0
QQ
9CTC3ER 2 1976 _
i CO
3 60
3 20
2 CO
2 J0
2 00
1 60
o "0 .-
0 3t
9 10 11 12 13 It 15 16 17 18
HOURS. LOT
19
Figure 64. Ambient data, weighted mean, max and min for Day 276.
135
-------
7.3 EMISSIONS ALONG TRAJECTORIES
7.3.1 Procedure for Processing Emissions
The procedure adopted by OAQPS/EPA for estimating emissions was to
calculate an average emission rate over the entire uncertainty band of the
trajectories for each hour. Thus the coordinates of the extreme trajectory
locations at each hour were used to define the area over which the average
would be computed. These areas were shown in Figures 31-54. A computer
program was used to identify and process the emission inventory grid areas
that fell inside each of these areas each hour. A detail of the grid near
the center city (RAMS stations are plotted as *) is shown in Figure 65.
Table 18 is an example of the area source printout from the computer
program for hour 10 (LST) of day 159, June 7, 1976. Listed are the grid
squares that were totally or partially included in the trajectory area, their
area, the fraction of the area included and the proportional amount of the
emission rate for the pollutantb. Table 19 is a printout for all the point
sources that were in the area of trajectory between hour 10 (LST) and hour 11.
The emissions database used was the so-called "modellers tape" in which
the hydrocarbon specialion of each of the source types has already been
accounted for and the composition has been converted to an appropriately
magnitude weighted,five level, lumped species inventory. These lumped species
are: NR, non-reactive; PAR, paraffins; OLE, olefins; ARO, aromatics; and ALD,
aldehydes. Tables 20 and 21 give the inventory average molecular weights and
carbon numbers for the hydrocarbon species. These were obtained from
Demerjian (1980).
In their in-house study (Gipson, 1980), OAQPS had converted the emissions
to units relative to initial concentrations and initial mixing heights for
use in the OZIPP program. In addition OAQPS had treated point sources as if
136
-------
o
c
O)
-a
ai
s-
cn
CO
a_
c
(U
OJ
1/1
ai
o.
-------
O
cu
CO
i
ro
O)
OI
Q.
E
re
CO
oo
K
? C
LxJ «
-1
O
ac
o
u
I/- G
UJ
U
u.
0-
l/l
CC
o a
03
V
2
CC
c t)
f
c
t-> 1^
X ^
X.
U">
^
NOX
MS EH IS SIC
0 is
t- u-
t.
LJ
is
^ c;
LLJ
H
5
;
n
r\.
j
ru
r*-
— PJ
i~
K1
T •
^
— P.
T »
V C
c: m
x *•<
^ cc
CT
£* r-
f> P,
J .
-j
P>
n
f
Kl
ry
C)
c*. t*.
0,0
fn
f\» o>
rvj p^
PJ
- if
r?2
"
nj ,0
• «
a __,
cc or
0- a)
""» in
a IT*
co a
in
0- m
O u
-c »
x r-
«• -e
f, (*~i
PJ f\
r c.
o r-
r- f
in a
PO C\
••«
PJ ir
» o
o c\
r- -c
ft C\
m •-
h- r,
M
m 0
i"j *n
•^ oj
CO P"
f*- CV)
m in
o- cc
CJ '">
cc fv-
o cc
._, fl
iH
c> "-
cr
i j fi
-0 N
-rfl -C
p. r-
J
a- tr
CO 0
* 4
o* a
m *
r- f-
j» *-
a o
o- m
ru =*
o- c-
in tr
,0 ,0
o '-^
c . CJ
it.'
a tr
n. o-
t\ P-
«o
PJ "
COK
_« p.
(E P-
in t
U> »"
* CO
,0 0
CO r;
-0 -0
m cvj
0 PJ
tc r-
rj D
n PJ
.-» r-
,0 r-
r> O-
P. P.
C f-
(-1 U
oa
•" P
•o ^
in F.
ro c
in *
-O ,;
«-• o
^ a
CO (T
-O rv
1 CC
-c rv
* *
O o
t~> -i
* •
0 o
K> fr,
K r<-\
r- c
.,
to «
:]
2*
eo 0*
££
n ir
!C '""
O f>
«
- o
0 h-
fX r-
0 f»-
C^ O-
o r-,
. »>
P, I**
en K\
K* K
r- C
J
*
r^ ^
r»
(-1 .«
CM n
.
LT r*
a .*
w%
m *
m
fM ^
^ r*
in
•o »
a- r-.
• *
CO 0
T- »
f^i
m ,T
.'i 0
~« 0
in ->
0 ,-i
a p,
Kl ^
K Kl
*"i C.
PJ O-
o U^
f , u-
1^ ^
0 O
_j r-j
f3 0
I-1 *— »
fo a
^ -c
l»-| K\
C- r-i
J
*) fO
^ a
PJ in
o -
o> «-•
PJ
in o
-^
0- *
PJ
on •-•
o cr
~« a
r* o-
*\ PJ
* ~;
> Kl
1 »^
. .
-2
rj a
•_* i*"*
f> O1
-' t^
m -o
^i
CC fM
•C t_
0 0
ri o
0 o
ri C
-» o
ft if\
{-• c_
a o-
o* «
K1 CO
« J
in f^-
^
IT (T
» t-,
CO (O
let
u> r»
r. a
f^
IT
ft P-
•-4 a
^3 r>
ot,
in cc
PJ
K> CO
-0 0
PJ
a inJ
O PJ
M
a 4
K, CO
o* r*
Kl
••* P.
a co
PJ
r^ - 1
IT f\
f^ O
PJ
-0 0
•0 ri
CO O
IT f~-
CC O
o- -.
(•5 /-i
C C
CO Pi
in
•A
~Q
m ~*
PJ *n
co o
,0 P-
a *
r- r-
4A ^
O* r->
r* 0"
a P'
O* C
(\i PJ
0 a
1-1 P.
• '
-•:
r*. i**)
arJ
* r3
in «C
r^ r-
PJ a
r* >n
^4 00
-o **
tn ^
•C f**
et CJ
fx in
a O
a PJ
in •&
--3
» i
50 S
o- *#•
re tc
-S
ru ;»•
~2
r- tf\
r^ ro
cr
r> O-
-j 0
CO -'
.A OJ
K>
frt f
a IT
PJ PJ
in CM
K\ f*-
«1 0
a or
0 PJ
o ' *
P. K
C r
o if
r«
-. IT
^- rv
in «£
*
^1 «0
c\
in. o
**
^ CM
f- a
a
AJ *>
> IP
in cr
J -0
a
in '-j
r- ,1
•o '"
"^
K 'C
K4 K
r» PJ
i-.
CTiO
o
f*^ m
f. M
a-
o ru
f\i
v"> »-i
in
^1 M
o- PJ
•_i a
• •
U1
J» N>
O »-t
•_ o-
iC M
a
.n a-
-j -0
i a i*\
w r-
~> •-!
o> fo
f-l f
c c
-c r-
• •
-* (ft
r* a-
*- a
PJ 0-
in »i
^
> 0)
•« in
ro
M »p
KI m
o- m
m ^
in KI
o o*
"1 M
p.
r* LP
CO 0
^ rv.
*r -O
a ru
• •
a m
a a
tr t*
r; c
o
M
r-
PJ
in
XI
r~
ra
a
rvi
-C
I
»
m
in
1*1
CO
r-
^
Jl
O
PJ
o
r*
O
r*
!"*\
«H
a
(E
PJ
Jl
in
m
•^
?
CO
fo
rA
a
EC
3
0- «
^ Q
PJ *
a »o
a- ^,
C\J o
-c a-
• •
m
PJ
•» 3C
M JiH/S3TOW 3)1
(UX TS XH>£N >
- >-
* i/;
•^ ^
-i |,,
O
- O
/I ^
' w
- 1^
»- t-l
>' K-
UJ
138
-------
_!_._!
0
C
i.
Q.
£*>
O
c
Qj
C
HH
CO
O
i/l
1/1
•p-
LU
0)
I,
^3
O
CO
£
£X
0)
CL
E
ro
00
CTt
1 •
OJ
-Q
c.
i ' <
rr
T
_j or
o <
in
wn
Ul
0 -1
u- O
Q_
l/l
O
u K
< o <
•a: a
X
UJ
i CC
I 1
' Z
1
tt !
" ' 2
o a —
U1 ' l/l T
'*-t IS
h- ' UJ —
!£ '<_>
•— !
O ' 1/1
a 'o ~
/i X
»— c
U- **
1
' o
•s
y tv
'^ V
i: t"
U, *:
2
'.*-
UJ
^^_
t-
f ^.
s
i T
I
00
o o
G f^i
,-, r-
cr
0 -0
' ~
^ tr>
(*> tn
K>
8 A
0 ,1
C <_.
•\J "•'
3 ^
_ r t
3 er j
3d,
5 O C
!
0 C
CM OC
m CO
A, -
nr ^
a *
0 P*
O (^
f^ fu
r* i T
"' '"
c- r-
» 5
r
? o
/} -C
, a
0 C
,**
^ ^
tr CC
,r, «
,; ^
-<3 -C
"-1 —
LJ X
(.1 -t
Ci O
if IT
" r
cr, o
rf> -i
^ *»
"l
!c o
i
-O f^
«
CO ,0
in vi
LA r\;
•fl _-
*-• CM
' •» O
,1 „-
D Ci
•> I-'
t
r. ^
i t
D DJ
W ^
1
J 0 1
1
1
i
0 0
^'«;
3 ^
•O O
-o ^
o! j;
in f^,
nr, >~i
f^ if>
( i (_ i
T . 1
= 5
- 2'
f' C
rf o-
a 4
i
1
1
1
1
«cv
•
„ „
OJ f-
i m
0. O
S™
-. uO
-o r-
r^- a
- *
r^ ,
ff ff
E H
^ r.
1 * *
1
=r ^
» r
*-. o
". *:
^ CO
p~ '\
* r-
!•' r,
if <
? n
^ HI
^ n
r c.
tr tr.
f *
0 0
'
] i
^ — i .0 c*
•«
1
|
1
[ \
i i
i_ j,l.j
1 •-<
!
1
i ;
i
1 i
|
r- f\'
' ") o ^*
f> .-»• fv
^ CM •>
*•' u" er
1
a r\ r-
l
i
"";""
?-~ -
»- t_ r
or cr o
<; j- <:
H ^ H
>
;
^
1
|C
1 *
1 —
^
«
fM
-O
^n
ji
^
;.
">
'
J
^_
f -
a
3
^ ,
0 0
n o
1-4 -O
K> -.
f\j
o m
-0 -
rt "
^ wl
~
...
j O
j i"-i
M
M r-t
i t"
0 0
0 C
^ sfl
J, ^
:"
. rr
~ ~
". *\
•. -c
,0
*. LJ
*> rj
-> C_
3 0
N.S
- J
f ^
El
S S3
« r«
j O
0 0
f^ CO
•A
* tt
&7
^ en
CC,
s!
" *
^
~l C_J
r, ^
"» CJ
M rv,
» n
~"
>-2
£ j;
^ »
""
-j r-,
0 0
0 0
a ^
« ^
"~
O f>
*• o-
r\j ^r
M ro
.?§
t*5 .-j
3 O
"1 U,
^ r-
3 ft
-» M
-' 0
0 0
r- o
* ^
W I-*
M t>
r> r--
-O 0,1
^
lOj tC
•^
j « i
r i '_
3 ~>
= S
"~
~ ?
a a
i J>
a =
;;
0 0
^ ^
03 ;»
S"
M^
--t £
^ c-»
rr 53
nj
"V C1
" o
•j C
^ ?,
r ^
*• r^
t» a
•< ( ,
*t <
t? o-
I
»• »~l
0 0
0 0
sc
ra 0
-IS
O t"!)
D C
^ 0
nS
0" O
r* r-
__
f!
c» -J
T O
- ri
0 0
O 0
(M
"*
u^ '
~
00
r\i (->
^
•\j
* o
D C.
- «_
— (. '
-i <-;
B" Of
* 3
-"
•-*•
0 0
C CO
--
r-'o
fM (M
r-' <0
^ o
* a-
-. —
"> ec
r^ in
-t -0
— "
•> o
n ^
~ 0
3 L-_>
is
~> o
3 a
C S
' 2
:
t— ^^
„
i o
<-•
0
CO
o
i r-
^
r^*
o
cs.
J
OJ
1 K
J
r~
OJ
l
u
r* ' •
in O-
£
M 1
j
in cr
X ^
-• ^
I
1
:fc
f\j in
^r
1
rv "
*~ *j i—
(V C -I
T IT C
XX
l- C1
t/1 i/ */"
c. c o
u1 1/1 tr
L.I H UJ
— 1 —^ — '
** -* <
c c o
H- ^ »—
139
-------
Table 20. Species Molecular Weights
Species Molecular Weight
HC 14.5
CH4 16.0
NR 68.0
PAR 88.6
OLE 42.3
ARO 100.0
ALD 46.1
Table 21. Hydrocarbon Carbon Numbers
Species Molecular Height
CH4 1.0
NR 3.0
PAR 6.3
OLE 3.0
ARO 7.1
ALD 2.0
140
-------
they were area sources. That is, the summed point source emissions for each
hour (the values at the bottom of Table 19) were divided by the total area of
the "box" for the hour (the value near the bottom of Table 18). These
averaged point source emissions were then summed with the area source
emissions to give a single emissions value for each hour.
In this study, UNC decided to use absolute emission units of ppm-km/
2
km -hr in the models instead of the relative units used by OAQPS. An emission
rate of 0.25 ppm-km/hr would result in 0.25 ppm of material accumulating in
2
a box 1 km and 1 km high in one hour and would result in 0.50 ppm
2
accumulating in a box 1 km and 0.5 km high in one hour assuming no
reaction of the material. The value is simply divided by the current mixing
2
height to get the rate of change of concentration for a 1 km box.
To be consistent with the OAQPS study, UNC decided to apply the same
assumptions as OAQPS did to the point sources but to determine their
contribution to the total emissions. The OZIPPM model program allows for a
single emissions composition to be used for the entire day. To be consistent
with this approach, UNC decided to compute this compositon from each day's
trajectory's emission inventory.
To perform all the above tasks, a computer program was written. The
program accepts as input the area and point source averaged emission values
such as those at the bottom of Tables 18 and 19 in units of kg/hr or moles/hr.
These are converted by use of the area, molecular weight, and average carbon
number into ppm(C)-km/hr units.
141
-------
The basic equations are:
Ei = (1000*factor/MWi)*(Esupi/A)
Ei = Ei*CNi
ET = EEi
where Ei = emission rate in ppm-km/hr for the ith species
o
factor = conversion from moles-km ' to ppm-km (at 25°C)
factor = 24.46656 x 10~6
MWi = molecular weight for the ith species
Esupi = supplied emission rate in units of kg/hr
o
A = area covered by trajectory in one hour (km )
CNi = carbon number for ith hydrocarbon species
Ei = emission rate in ppmC-km/hr for the ith
hydrocarbon species
Ey = total emission rate in ppmC-km/hr for all
hydrocarbons and aldehydes.
In addition the program computes the carbon fraction composition for each
hours emissions and the running carbon fraction composition for point sources,
area sources, and total emissions. An example printout for the 1100 LOT
(1000 LSI) hour of June 7, Day 159, is shown in Table 22'. One printout page
is produced for every hour of each day.
142
-------
o
o.
o c
O rt
0 CN
O CO
r*>
o
o
c
o
o
o
o
o o o
coo
o o o
o o o
C -H "
o o o
o c c-
o o o
o o o
^ ri "
o c j TJ
o o o
o o o
i- ro
o •
o r.
o
000
-.' O O
o e c
o c c-
o •:< o
o c c
coo
c -• c
O -I r-
o o o
o o o
o o o
0}
O
i,
IXJ
OJ
o
u:
<:
LJ
_!
O
oc
IN
r-i
o
<
0
rj
Ch
0
o
«
o
fN
O
.
M
O-
0-
o -
O -i
o «
c
-o .-.
o •
o o-
o n
o
n
c
o
C'
o
CK
o
0
o
o
o*
CO
-c
0-
o
o
r^1
6
o
c
<•
o
o
o
r- ra
o o
o ^
o c
r- c-
T-H ro
0 Q
0 fl
0 O
IN
0
o
c
-o
fx
c-j
o
*c-
o
o
c
-0
c.
^
0
C- 0
-'"' I*
u~ lj~
o c
O 0
CN CN
OS CC
o- o
O 0
0 0
TH ri cc
r. LI -7
O x; *C
0 O O
0 O O
fi IN O
<* r. o
c K re
coo
O
Q_
s,
o
o
-------
7.3.? Results of Emissions Calculations
Table 23 summarizes the calculated distribution of emissions among point
and area sources for the 10 days; these values are illustrated in Figure 66.
For comparison, Table 24 gives the distributions for the entire St. Louis
area emissions inventory.
Figure 66 can be interpreted readily by reviewing the location of major
point sources, Figure 29, and the trajectory paths for the days. The tra-
jectories for days 159 and 160/15 were both north of the central business
district (CBD) and spent significant time near the Alton/Wood River group of
point sources. The trajectories for days 160/3, 221, and 261, which have
similar distributions in Figure 66, were all west of CBD, and day 238 was
east of CBD, areas reasonably low in major point sources. Days 195, 201, and
275, which also have similar distributions in Figure 66, had trajectory
paths near the CBD (stations 101, 103, and 109). Day 276 was northeast, of
the CBD.
Since the point sources were uniformly distributed into the entire
trajectory area for one hour, it is difficult to estimate whether their
influence has been over or under estimated. For example, a large point
source located just inside the upwind edge of a trajectory "box" contributes
for nearly an hour in advance to its being encountered. On the other hand,
it is dispersed over the entire box volume (in effect, it diffuses "upwind")
thus greatly reducing its contribution to areas near it. Sometimes point
sources located 20 km perpendicular to the trajectory could make significant
contributions to a given hour's emissions.
Table 25 gives the computed percent carbon distribution among the five
nonmethane hydrocarbon classes in the emission inventory for point sources,
area sources and the total for the ten trajectory paths. The total percen-
144
-------
Table 23
Distributions of Emissions for Selected RAPS Days
(Percent)
CO
, NO.
HC
Day
160/3
160
159
221
261
201
195
276
275
238
Date
Jun
Jim
June
Aug
8
8
7
8
Sept 17
Jul
Jul
Oct
Oct
Aug
19
13
2
1
25
Area
89
53
25
90
94
78
85
47
73
99
.8
.9
.1
.1
.5
.6
.7
.3
.7
.7
Point
10
46
74
9
5
21
14
52
26
0
.2
.1
.9
.9
.5
.4
.3
.7
.3
.3
Area
86.9
72.0
34.6
84.3
90.5
58.7
66.0
65.1
62.8
83.0
Point
13
28
65
15
9
41
34
34
37
17
.1
.0
.4
.7
.5
.3
.0
.9
.2
.0
Area
81.6
64.5
27.8
78.7
76.0
81.5
85.8
84.5
88.1
65.7
Point
18.4
35.5
72.2
21.3
24.0
18.5
14.2
15.5
11.9
34.3
TABLE 24. TOTAL EMISSIONS FOR THE ST. LOUIS AQCR (TONS PER YEAR)
Point Sources
(% of Total)
Area Sources
(% of Total)
Total
Particulates
45,224
(3%)
1,299,782
(97%)
1,345,006
so2
1,007,530
(97%)
30,813
(3%)
1,038,334
NOX
322,730
(72%)
125,567
(28%)
448,297
HC
47,610
(23%)
157,204
(77%)
204,814
CO
164,331
(11%)
1,325,556
(89%)
1,489,887
145
-------
Table 25. Carbon Distribution of NMHC from Emission Inventory for
Selected RAPS Days (percent)
Day Date
159 June 7
160/103 June 8
160/115 Jun 8
195
201
221
238
261
275
276
July 13
July 19
Aug 8
Aug 25
Sept 17
Oct 1
Oct 2
Source
Area
Point
Total
Area
Point
Total
Area
Point
Total
Area
Point
Total
Area
Point
Total
Area
Point
Total
Area
Point
Total
Area
Point
Total
Area
Point
Total
Area
Point
Total
NR
13.7
5.1
7.7
11.2
3.8
9.9
15.3
4.7
12.0
12.0
10.1
11.7
11.8
12.0
11.8
15.6
4.8
13.7
6.8
75.9
6.75
11.6
4.0
10.0
11.8
15.2
12.2
15.2
5.3
14.0
PAR
50.2
79.6
70.6
57.0
46.4
55.1
44.3
80.1
55.5
54.3
47.5
53.4
54.3
45.9
52.9
42.0
79.0
48.3
75.5
11.1
75.5
55.0
56.1
55.2
55.5
39.0
53.7
44.2
78.0
48.5
OLE
22.3
10.2
13.9
17.4
4.2
15.2
26.1
10.1
21.1
19.1
5.3
17.4
18.7
6.0
16.6
28.9
7.5
25.1
10.1
5.3
10.1
18.6
1.8
15.0
18.4
6.8
17.2
26.2
4.1
23.3
ARO
11.2
5.1
7.0
12.4
45.6
17.9
11.3
5.0
9.3
12.4
37.0
15.5
12.8
36.0
16.7
10.1
8.6
9.9
6.6
4.2
6.6
12.7
38.1
18.1
12.0
38.9
14.9
10.6
12.5
10.8
ALD
2.95
0.00
0.82
2.09
0.07
1.72
3.07
0.00
1.98
2.31
0.00
1.98
2.38
0.00
1.94
3.43
0.69
2.86
1.01
3.03
1.08
2.21
0.00
1.68
2.29
0.00
2.02
4.05
0.00
3.42
Fraction
of Total
27.8
72.2
81.6
18.4
64.5
35.5
85.8
14.2
81.5
18.5
78.7
21.3
96.4
3.6
76.0
24.0
88.1
11.9
84.5
15.5
146
-------
UJ
O-
OQ
CO
2
O
I— I
CO
CO
o
o
100
90
80
70
60
50
40
30
20
10
0
159 160/3 160/15 195 201 221 238 261 275
DISTRIBUTIONS OF EMISSIONS FOR RRPS DRYS
276
100
90
80
70
60
50
HO
30
20
10
0
Figure 66. Distribution of CO, NO , and HC emissions for trajectory
paths through the RAPS emission inventory.
LU
CJ
ce
LU
Q_
o
I—I
»—
ZD
00
t—I
ct:
i—
CO
•—i
C3
Z
O
CO
ct
cc
o
100
90
80
70
60
50
to
30
20
10
0
ALD
Figure 67.
PAR
159 160/3 160/5 195 201 221 238 261 275 276
TOTflL NMHC FOR SELECTED RflPS DflYS
100
90
80
70
60
50
to
30
20
10
0
Carbon distribution (percent) of NMHC for trajectory
paths through the RAPS emission inventory.
147
-------
tages are plotted in Figure 67. Day 238 (Aug 25), in which the trajectory
came from east of the CBD and curved north, had the highest paraffin frac-
tion, 75%. Day 221 (Aug 8), in which the trajectory carne from northeast
of the CBD, swept through the city and ended west of the CBD, had the highest
olefin carbon fraction, 25%. Day 261 (Sept 17), in which the trajectory
started slightly north of the CBD and went south, had the highest aromatic
fraction, 18%. The significance of these compositional changes will be
examined more fully in later sections (see 7.3.4 and 8.3).
The computed hourly averaged emissions rates for each selected day are
given in Table 26. These rates exhibit significant changes from hour to
hour and the solution methods in the computer programs (such as the Gear
routine in OZIPP and PKSTM) do not work well with such changes.
One method for dealing with this problem is to substitute a smoothly
changing curve for the hourly averaged values. Figure 68 shows an example
of NO and HC hourly averaged emission values and a smooth curve fitted
X
through these values. An important criterion for such a curve is that the
average emission computed from the curve each hour must be equal to the
original constant hourly value. Another way of stating this requirement is
that when the average hourly values are plotted in a histogram-like plot,
then the area under the smooth curve each hour must be equal to the area of
the histogram bar each hour.
The algorithm for producing these equal-area, smooth-curve histogram
fits is not simple. Unfortunately, the algorithm in the OZIPP program does
not always work properly and sometimes it produces negative emissions rates;
this was true for several days in this study. Appendix A discusses the
problem more fully and describes an algorithm developed by UNC to solve the
problem without giving negative rates.
148
-------
1 . -J
1.3
1.1
0.9
0.5
0.3
0.1
-0.1
-0.3
-0.5
-n ~i
. ' 1 '
i
A
i
i
1 1 '
i
1
i
- DflY/22\l fiUGUST 8, 1976 NO, HISFIT — 1
—
—
—
-
,
/
1
\
v — -
•x. -/•*
^
,-,
' \
\
^v^
" , 1 •
|
i
i
1
i
,
, i
— x
\
i
—
—
—
_
S ••>
—
—
, "
1 .
-------
CD O O O O O
O CD O O O
O O O CD O O O
10 co o r-. (
en cri 01 o
O O O — r
CM
E
_i1 -rj- I_T) ID
o o o o o o o
o o o o o o o
r^. cO en CD .— cxj n
O O O -— r- r— ,—
CXJ
roncnr
OCDO.
CDOO
OCDOCDOCDOOO
CVlCXJLO
OOO
ooo
OOO
LOuOCOtMCDCO
i — i — O t — UDO
oooooo
CDOOOOO
OOCDOOOOCDO
OOCDOOCDOOO
CD
t.
.
in i.n '— UDf^-CNJ^
CDOt'Oi— •— i— LnOOQO
CDOCsjOCDOOOOOO
CDOOOOCDOOCDCDO
CD O ro r^ *3~ ro O CT OOCDOOOOCD
CDC'OC OOOOOOOCD
o* o o cr o o o' o o o o o
v
f->rouDc\'U->OJO.— i— OOO
CDOCXI.— OOCDOOOOCD
OOOCTCDCDCDOCDOOO
OCDOCTOOCDOOOO'
uD r-. CO O". O .— cxjrota-iD
OCDOO-— r-r— ^-r— r-
i~O CT> CTv <— CM CSJ i— CD IT1
i— r— i— cxi ixj CM m ro o
onrororoo-jroi— o CD
ooooooooo
OOCDOCDOOOO
ooooooooo
OOOOOOO CDO
CD OJ r^- UD -^T O CXJ
CD O f"> 'JD i— .— CD
O O O O O CD CD
CD O O O O O O
CD O O O CD O O
CD O O O CD O O
cxicxji-OCQinrocor- tn
O co f*^ r-- cvi <^j- LD LO •—• ^D
ro o O txi i—
-------
Figure 69 illustrates, on the same scales, the emissions rates used in
the UNC simulations. The equal-area histogram fitting procedure was applied
to the values in Table 26 to produce the rates plotted in Figure 69.
151
-------
NOi. ppm»km/mm (XI IT* 3
On. ppm«km/mln CXI0^3
U-OtX) "
o o cz> o
(r-OIX) uiui/uj>|«"id l/l
fO >>
S- (O
ID
CO rO
CO
E to
Ol C
O
01 -i-
+-> 4->
13 (O
O 3
££
=t to
CTl
ID
O)
I.
C,-01X) ui
(c-OIX) u|iu/"i>l«l''d<1 '03 'OH
152
-------
N0». ppm»km/m!n
o o o a
Oi. ppm»km/mln CX10")
|«uid|«ujdd '03 'OH
s- c
O O
O) •
•!-> to
ro >,
1- rt3
T3
c:
o i—
E 01
ai c
o
O)
i-
O1
"iuj/ui>(*ujdd 'o
153
-------
NO,. ppmokm/mln CX10"1)
O>
-&
a
o
o
OJ
OJ
"o.
"D
O)
CO
13
CJ
NO,,
CXIO")
CO
O)
CO
>,
ro
-a
CO 03
CO
f= CO
a> c
o
OJ -i-
4-> 4->
Z3 ra
vo
OJ
s-
CD
C.-OIX3 uiuj/iij>|.iiidd '03 'JH
154
-------
7.3.3 Method of Treating HC Composition in Each Mechanism
Mechanisms differ in the number and characteristics of the species used
to represent the hydrocarbons present in the photochemical systems to be
simulated. Hydrocarbon composition data available from the RAPS emission
inventory (see Table 25 and Figure 66) usually needed to be converted to the
appropriate representation used in each mechanism. The actual conversion
depends on the characteristics of each mechanisms' hydrocarbon species as
defined by the mechanism developer.
In simple trajectory models, HC composition data are utilized in calcu-
lating emission composition along the path, in addition to calculation of the
initial morning HC concentrations.
The HC concentrations needed internally in the model are in ppmV. The
values for each HC specie used in the mechanism are calculated by multiplying
the measured NMHC value from RAPS (or the NMHC values calculated from emission
inventory) times the calculated carbon mole fraction for each specie, divided
by the determined carbon number for each specie. The conversion calculations
used in this study for each mechanism will be described. Table 27 lists
the HC fractions.
7.3.3.1 Dodge Mechanism. The Dodge mechanism used a mix of butane and
propylene to represent the hydrocarbon species present in a photochemical
oxidant system. Dodge had determined that a carbon mole fraction of 75%
butane and 25% propylene yielded a system which had a photochemical reactivity
equivalent to the Bureau of Mines smog chamber experiments. The Dodge
mechanism, therefore, does not utilize HC composition data.
In the standard OZIPP program the assumption has been made that aldehyde
fraction would be 5%: 2% formaldehyde and 3% acetaldehyde.
155
-------
Species Fraction Carbon No.
Prop 0.25 3.0
But 0.75 4.0
HCHO 0.02 1.0
ALD2 0.03 2.0
7.3.3.2 Demerjian Mechanism. The Demerjian mechanism uses three HC
classes (OLE, PAR, ARO) and one aldehyde species to represent all aldehydes.
The Demerjian mechanism utilizes HC composition data of carbon mole fraction
only. The carbon number for each HC and aldehyde class is defined by
Demerjian. Therefore, the RAPS carbon number data in Table 21 is not
utilized. Ethylene is included in the olefin group. These conditions allow
direct use of the RAPS emission inventory derived carbon mole fractions
(Table 25). These fractions were utilized for initial condition determina-
tions as well as for the emission fractions.
Species Fraction Carbon No.
OLE
ALK
ARO
ALD
RAPS OLE
RAPS PAR
RAPS ARO
RAPD ALD
3.0
6.0
8.1
1.0
7.3.3.3 Carbon Bond II Mechanism. The Carbon Bond II mechanism's HC
species represent carbon bond types rather than hydrocarbon classes, with
ethylene explicitly represented as the exception. The CARB specie represents
the carbonyl carbon present in aldehydes, ketones and other oxygenates. Both
carbon mole fraction and carbon number are needed from the RAPS HC composition
data. Each carbon bond type also has a defined carbon number (e.g., olefins:
C - C has 2 carbons). Portions of each HC composition class are added to the
paraffin carbon bond specie, depending on the degree that the HC composition
carbon numbers exceed the corresponding carbon bond type carbon number. The
carbon mole fraction of the olefin portion (of the HC composition data) which
156
-------
is ethylene needs to be determined (or assumed), as well as the non-ethylene-
olefin carbon number. For this study both a 50/50 or 33/66 ethylene/olefin
carbon mole fraction was used. Little effect was observed when one split
was substituted for the other. The carbon number for OLE as determined in the
St. Louis emission inventory was 3.0. Since the carbon number for ethylene
is 2.0, the eth/ole split and the inventory carbon number determine the
carbon number for ole without ethylene. The formula below is used to
calculate the Carbon Bond II's Olefin # from the RAPS total Olefin fraction
(includes ethylene) and the assumed ethylene fraction/total olefin fraction
ratio (E).
CBII Olefin # = ] " E
1
(RAPS Ole #)
Ethylene carbon. F.. „ n, „ Inventory
Olefin carbon "n ff uie ff Total Olefin #
50/50 2.0 6.0 3.0
33/66 2.0 4.0 3.0
The carbon number for Ole in the CB2 mechanism is 2.0. Therefore to
arrive at the CB2 emission fraction for olefin from the St. Louis inventory:
(1 - r— )(ole carbon frac particular day)
CB2 ole frac = - — - - • - - — — - - - x 2.0
4.0 carbons/inventory
olefin
The initial olefin value is: CB2 init olec = NMHC * CB2 ole frac
Likewise for ethylene:
CB2 eth frac = x (ole frac St. Louis)
270 ~~~ ~~ X >U
157
-------
and
NMHC x CB2 eth frac
roo • -4. +u
CB2 mit ethc -
?~0
The aromatic carbon number is 7.1 in the emission inventory and 6.0 in
the CB2 mechanism.
Therefore,
rno .. f (arom frac St. Louis) r n
CB2 aromatic frac = •* - 7— ^ - ^ x 6.0
roo • -4- NMHC x CB2 aromatic frac
CB2 init a roc = - F-~H --
o .u
The aldehyde carbon number is 2.0 in the emission inventory and 1.0 in
the CB2 mechanism: the carbonyl carbon atom (CARB). Therefore
CB2 CARB frac - -= x 1.0
x CB2 CARB frac
CB2 init CARB =
Q
The paraffin carbon number is 1.0 in the CB2 mechanism. Therefore the
St. Louis fractin for paraffin can be used directly. However, several other
species contain paraffin carbon atoms in side chains which must be added to
the paraffin class.
The number of carbon atoms over two in the average olefin molecule are
added to the paraffin fraction. The carbon atoms in the aromatic side chains
are added. The carbon atoms in excess of the carbonyl carbon in the aldehydes
are also added to the paraffin fraction.
Therefore
CB2 par frac = (par fraction St. Louis)
+ (4.0-2)(l-£~)(ole frac St. Louis)
__
158
-------
+ (7.l-6)(aro frac St. Louis)
_ . __,
+ (2.0-1.0)(aid frac St. Louis)
2.0
CB2 init par =
The non reactives were treated as follows
CB2 NR frac = NR frac St. Louis. x 3>Q
CB2 1n1t NR =
« U
7.3.3.4 California Institute of Technology (CIT) Mechanism. The CIT
mechanism utilizes four HC species: olefin, ethylene, aromatic, and alkane,
and two aldehyde species: formaldehyde and a general aldehyde species. Both
carbon mole fraction and carbon number data are required to implement this
mechanism.
Generally the RAPS emission inventory derived HC carbon mole fractions
can be used directly with the CIT mechanism. However, the fraction of the
RAPS Olefin fraction which is ethylene must be determined or assumed. A 50/50
split was assumed for the Ethyl ene/Total olefin fraction ratio for the Level
II simulations (The BOM autoexhaust analysis indicates 42% of Olefin is
ethylene). Therefore, the carbon number for the non-ethyl ene olefin fraction
must be approximated. The value used for this study was 4.0. Also the HCHO/
RCHO fraction needs to be determined or assumed, as well as the carbon number
for RCHO. A value of 33%/66% was assumed. Since the aldehyde carbon
number determined from the emission inventory was 2.0, the carbon number
for RCHO was assumed to be 2.5.
159
-------
Species Fraction Carbon No.
OLE (non-ethylene RAPS OLE
olefin factor)
used 4.0 RAPS
non-ethylene
olefin
ETH (1 - non-ethylene RAPS OLE 2.0
olefin factor)
ALK RAPS PAR RAPS PAR
ARO RAPS ARO RAPS ARO
HCHO 1/3 (RAPS ALD) 1.0
RCHO 2/3 (RAPS ALD) 2.5
160
-------
V_Q
(S,^
OJ
ID
I*-**.
> °J
Q
OO
o_
5? 00
co
«= Q
cu Q
cu
-C
^~^ I—
S— *^
4- Q
C
O
T3 z r—
o 0 O •— O
o o o o o
en o m o «*•
in m r--- co co
CM r^ co in o
ID O O i — O
o o o o o
en in co oo «d-
in o in in in
en in CM in o
r~- o o o o
o o o o o
«3- m oo r~v oo
in in CM co ^j-
r~^ CM 10 co i —
LO •— 0 O O
o o o o o
i — o Ln i — r~~
ID co r— i— en
CD co ^ «=r o
U3 O O r— O
O O O O 0 1—
1—4
o
*a- o in o en
i — i^ co § — en
i — CO «3" CO O
r— O O O
CD O O O O
r^* o CD co ID
CO ID CO i — OO
CM f-^ co in o
VD O O i— O
CD O O O O
to LO r-. CM r-
in en *a- en «d~
in ID co in o
r^ o o o O
in in i—m
r^. CM CM en ID co
co en en «d- o i—
in o o r— o o
o o o o o o
o o o o
o o o o o
ID in in i-- in
o en en o CM in
r--. us VD i—o o
o o o o o o
O O C5 O O O
C 5-
O
•e
(0
o
E
£
13
CU
o
cu
Q.
O
VD
«£
Q-
O
CO
LU
cl
r- O
00 r—
O O
o; _i
-------
7.3.4 Sensitivity of Demerjian and Carbon Bond II Mechanisms to RAPS
Ambient HC Composition in Smog Chamber Type Simulations
Figures 70 and 71 show simulations of BOM-type outdoor smog chamber
experiments at 0.2 ppm NO and 1.0 or 2.0 ppmC NMHC initially, using the
/\
Demerjian and CB2 mechanisms and four HC compositions: 1) BOM auto exhaust,
2) RAPS D261, 3) RAPS D221, and 4) RAPS D238.
The compositions of the BOM exhaust and the three RAPS days, representing
the days of highest aromatic, olefin, and paraffin fractions were:
Day NR PAR OLE ARO
261 0.100 0.552 0.150 0.181
221 0.137 0.483 0.251 0.099
238 0.0675 0.755 0.101 0.066
BOM 0.105 0.417 0.173 0.287
Each figure shows simulations performed with a single mechanism. In
each figure, columns show the same initial NMHC concentration and each row
shows a different HC composition.
Generally the BOM composition yielded the highest predicted Oq concen-
trations for a given mechanism and initial NMHC condition. Although the
two mechanisms agreed within 20% for both levels of initial HC utilizing
the BOM auto exhaust composition, the mechanisms respond differently to
changes of composition. The CB2 mechanism, for example, predicted that the
high aromatics day (D261) at 2.0 ppmC NMHC, would produce as much Oq as
the CB2 2.0 ppmC BOM-composition system, while the Demerjian mechanism shows
a substantially reduced Oq yield for the same change in composition. The
same change in composition would also have a dramatic effect at the 1.0 ppmC
initial level, with the Demerjian mechanism again showing the greatest effect,
yielding less than half the 03 than predicted by the CB2 mechanisms.
162
-------
0.50
"' I ' I ' I'' I ' I ' I "r I/1 ' HJ ' I ' I ' i ' I ' ] °-5:
Df-M - 2.0 PPMC NMHC ^ '
i | i | i p-] ' | ' I ' I ' I ' rr
I, n-r+ i I i I -t I i I i I i J
- 0.10
- 0.30 P
~ 0.20
- 0.10
0.00
l-J 0.00
6 7 8 9 10 11 12 13 11 15 16 17 18 19 20
HOURS, LDT
5 6 7 8 9 10 11 12 13 11 IS 16 17 18 19 20
HOURS. LDT
O.SO
0.10
0.30
0.20
0.10
0.00
Trn ' i • i ' i
DEM - 1.0 PPMC NMHC
- HIGH RRO - DflY 23B
6 7 8 S 10 11 1? 13 11 lli 16 17 18 Ifl 20
HOURS, LOT
0.50 0.50
0.15
0.10 0.10
e 0.35
0.30 P n 0.30
I I 0.25
o.rc j 0.20
B-»-«
0.10 0.10
o.os
0.00 0.00
I ' I ' I ' I ' I ' I ' I ' I
DEM - 2.0 PPMC NMHC
- HIGH flRO - DflY 261
NO
0.50
0.10
0.30
0.20
0,10
S 6 7 8 9 10 11 12 13 11 IS 16 17 18 19 20
HOURS. LDT
C.50
0.10
0.30
c5 0-20
.
0.10
o.oo
Di-M - 1.0 PPMC NMKC
HICi( OLE - DRY 238
PflN
0.50 0.50
0.15
0.10
0.50
9 10 11 12 13 11 IS 16 17 18 19 20
HOURS. LOT
0.00
5 6 7 8 9 10 11 12 13 11 IS 16 17 18 19 20
HOURS. LDT
0.00
V.3U
0.10
0.30
0.20
0.10
0.00
"' 1 ' i r 1 ' 1 ' 1 ' 1 ' 1 ' 1 ' 1 ' 1 ' 1 ' 1 ' 1 ' 1 '
DEM - 1.0 PPMf NMHC
- HIGH PRR - DRY 238
-
NO
~~ ^^> ~
i- HO, ______——-
,i ,i ,i , i i.i . _i_, i- ,_i .t4- .--I'-.-i-L i , i , i i
5 6 7 8 9 10 11 12 13 11 15 16 17 18 13 2
u.ou
0.10
0.30
0.20
0.10
0.00
0
HOURS. LDT
0.50
0.15
O.SO
0 00
S 6 7 8 9 10 11 12 13 11 IS 16 17 18 19 20
HOURS. LDT
0.00
Figure 70. Effect of HC composition and concentration on outdoor smog
chamber type simulations with the Demerjian Mechanism.
163
-------
I I I . I , I I I I-1 , , . I ,-J
-l7r-4-i-l-5jnrrT. I , I . I . I
- 0 30 P 0 30 -
rf 0 20 -
0. 10 -
-1 0 20 j 0 20 -
- g 10 P
0 00
5 6 1 8 S 10 11 12 13 II 15 16 11 18 19 20
HOURS LOT
0.00
5 6 1 8 9 10 II 12 13 11 IS 16 11 18 19 20
0 50
p~r pr-pr
CB2 - 1 0 PPMC l.'MHf
CB2 - 2 0 PPHC N1HC
HIGH flRO - BUY 261
HIOH flRO - DflY 261
5 6 1 8 9 10 11 12 13 II 15 16 1! 16 1« 2C
5 6 7 8 9 10 II 12 .3 M 15 16 11 18 15 ?0
0 50
0 10
d ° 2°
0 10
CB2 - 1 0 PPMC NMHC
HIGH OLE - OflY 221
0 00
5 6 7 8 9 10 11 12 13 II
HOURS IDT
=._-J^-]_i_L_i_l_i_
it n is 19 /
0 SO
0 SO i i r i | i | i | i [ i |-i~
C82 - 3 0 PPMC NM<'C
0 10
e
£ 0 30
i I
020 g°2°
0 !0 So,o
- HIGH OLE - OrtY 221
1 f.
0 00
NO
6 1 8 9 10 11 12 13 H 15 16 II
HOURS i nr
15 20
0 50
0 10
I ' I ' I '' I ' I ' I "H
CB2 - 1 0 PPM NMHC
HIGH PflR - DOY 238
0.30 -
rf 0 20
0 00
0 50 0 50
0 10 0 10
E
a
0 30 P " 0 30
2T
r, in 0 10
| I | I | ' | ' | ' I I | ' I ' | I
CB2 - 2 0 PPMC NMHC
HIGH PflR - DflY 258
HO
0 30 P
0 ?3
0 IT
Figure T\ . Effect of HC composition and concentration on outdoor smog
chamber type simulations with the Carbon Bond II Mechanism
164
-------
The switch from BOM composition to the high OLE day composition (D221)
also illustrates the difference in directional results. Both mechanisms at
the 1.0 ppmC NMHC system showed similar decreases in 03 yield, but at the
higher 2.0 ppmC level systems the Demerjian mechanism indicated that the
high OLE would be slightly less reactive system than the BOM composition
system and the CB2 would be slightly more reactive than predicted with
the BOM composition.
Similar effects are seen with the high PAR day (D238). The initial
1.0 ppmC NMHC system predictions were most dramatic for both mechanisms.
Figure 72 shows the effect of changing from a summer to fall (October)
distribution of photolytic rates, for the Demerjian mechanism at 2 ppmC
initial HC. The aldehyde photolysis rate constants change most dramatically:
decreasing about 40%. There is little effect shown for the more reactive HC
compositions (BOM and high olefin) but there is considerable effect in the
high paraffin composition. Similar studies with the Carbon Bond II mechanism
showed less of an effect and are not shown here.
165
-------
0.50
1 ' ' 1 ' I ' I ' I ' 1 ' I/1 "1-U ' I ' I ' i ' I
DEM - 2.0 PPMC NMHC/ \v
0.00
6 7 8 S 10 II 12 13 11 15 16 U 18 19 20
HOURS. LOT
o.oa
o so -,
0.10
0.30 -
0.20 -
0.10
0.00
I ' I ' I ' p T1 I' U-U ' I ' I ' I ' I '
- 0.10
6 1 8 9 10 11 12 13 11 IS 16 17 18 19 20
HOURS. LOT
0.00
0.50
0.55
0.10 -
e 0.35
S0.30 h
I 0.25
c7 c-20
•j.
-,..5
o.io
0.05
0.00
I'I'I' I' I ' I ' I ' I' I' I ' I ' I ' I' I
DEM - 2.0 PPMC NMHC
HIGH flRO - DflY 261
O.EO
NO
O.SO
- 0.10
- 0.30
- 0.20
- 0.10
6 7 8 9 10 11 12 13 11 IS 16 17 18 19 20
HOURS. LOT
5 6 7 8 9 10 11 12 13 11 IS 16 17 18 19 20
HOURS. LOT
0.00
0.50
- 0.10
- 0.30
- 0.10
10 11 12
HOURS,
13 11 IS 16 17 18 IS 20
LOT
0.00
HIGH OLE - DRY 221
PHOT RRTES FOR OCT LIGftT
O.SO
- 0.10
- 0.30 P
- 0.20
7 8
9 10 11 12 13 11 15 16 17 18 19 20
HOURS, LOT
0 00
0.50
O.SO
- 0.10
- 0.30
- 0.20
0 00
6 7 B 9 10 11 12 13 11 IS 16 17 18 IS 20
HOURS. LOT
0.00
0.51)
0.15
0 10
E 0.35
S 0.30
1 0.25
g 0.20
* 0.15
0.10
o.os
0.00
1 1 ' I ' 1 ' I ' I '
DEM - 2.0 PPMC NMHC
HIGH PHR - DflY 238
PHOT RflTES FOR OCT LIGHT
i I i I i I I I I I i
f1 I ' I
0.50
0.10
0.30 P
0.20
0.10
567
9 10 11 12 13 11 IS 16 17 18 19
HOURS. LOT
Figure 72. Effect of aldehyde photolytic rate with HC composition and
concentration in outdoor smog chamber type simulation with
the Demerjian Mechanism.
166
-------
7.4 MIXING HEIGHT PROFILES
The atmospheric boundary layer (ABL) is generally classified by its
stability. The classification systems generally involve the depth of the
ABL, h and the Obukhov length (a buoyancy length scale), L. Using this
system,the atmosphere can be divided into three general categories: unstable
or convective, neutral or near neutral, and stable. The convective
boundary layer is usually observed during the daytime over land while stable
boundary layers are generally observed during the nighttime. The neutral or
near neutral boundary layer is rarely seen over land, except for momentary
episodes during the transition from the daytime to the nighttime boundary
layer and vice versa.
In convective boundary layer regime, the buoyant production of turbu-
lence dominates over the mechanical production. The convective boundary
layer is characterized by three layers, a mixed layer, a transition layer
and a cloud layer. The mixed layer is characterized by a uniform mean velo-
city profile, with no significant vertical gradients. This is due to a large
amount of mixing. The transition layer is slightly stable due to the entrain-
ment downward of warmer and drier air from the cloud layer. The transition
layer occurs around the height of the capping inversion, in which there is
a noticeable jump in the wind speed across the inversion, with significant
wind shears in the cloud layer above. The wind velocity in the mixed layer
is nearly geostrophic (Yamada, 1976).
The simplest methods to determine the depth of the boundary, h, are
direct measurements, i.e., using Sondar (acoustic sounder) and Lidar (the
167
-------
optical equivalent of radar). The echoes detected by a monostatic acoustic
sounder are produced entirely by temperature fluctuations; with a bistatic
system the echoes are produced by both wind and temperature fluctuations.
The principal scattering sources for lidar echoes are aerosols and light
molecules. During the daytime, these direct measuring devices may not have
enough power to reach the top of the boundary layer. Other direct measure-
ments include turbulence sensors mounted on research aircraft, high towers
or on tethered balloons. These provide a means of measuring h directly.
Generally, these types of turbulence measurements are rarely made. More
often, the ABL depth is estimated from the measured temperature and humidity
profiles obtained from radiosonde observations.
If direct measurements are not available then climatic or historical
methods must be relied on. These generally involve'the use of an early
morning radiosonde observation from the nearest National Weather Service
Radiosonde station. Proposed methods include Holzworth (1972), Benkley and
Schulman (1979) and EPA (1981).
During the nighttime, the boundary layer becomes stable, and the depth
reaches a minimum. The ABL is generally identified with a surface inversion.
The height defined by the observed mean potential temperature profile may
not correspond to the actual height of the ABL due to turbulent exchanges --
thus the need for direct measurements.
The boundary layer height is highly variable in both time and space.
This is due to horizontal roll vortices and convective cells. Thus, the
average height of the interface between the turbulent ABL and the free
atmosphere should be determined by the averaging or smoothing of continuous
data; a single sounding can only give an instantaneous value (Arya, 1979).
Methods for choosing mixing height profiles given radiosonde data are
168
-------
likely to vary from investigator to investigator. One objective of this
study was to examine how sensitive the control requirements might be to some
likely-to-be-used methods for estimating the mixing height profiles.
The method that is built in to OZIPP assumes an exponential rise between
the 8 A.M.'mixing height and the afternoon mixing height. The reason is
simple: this profile results in a constant dilution rate. It is an inade-
quate representation of the ABL behavior, however, and was not used in this
study.
In the OAQPS EKMA study (Gipson, 1980), two alternative mixing height
profile methods were explored. The first method, Linear, consisted of a
simple linear interpolation between the mixing height calculated from the
near-sunrise sounding and the mixing height calculated from the late-morning
soundings and between the late-morning mixing height and that calculated
from the afternoon sounding. In the second method, Handpicked, the hourly
mixing heights were manually calculated by a meteorologist using the same
three soundings and hourly surface temperatures. It was decided to include
these two methods in this study. The data for the ten days were supplied
by OAQPS/EPA.
The third mixing height profile method used in this study was the
Characteristic curve method. The method is based upon a piecewise curve
thought to describe the characteristics of the mixing height rise. The
numerical values for the curve were derived from detailed data from about
40 days in the RAPS study by Schere and Demerjian (1980). Table 28 gives the
numerical values of the curves break points and Figure 73 shows the norma-
lized curve plotted. The curve has been normalized by fraction of total
growth in the mixing height versus fraction of the total daylight period.
Use of the curve requires the sunrise mixing height and the maximum afternoon
169
-------
Table 28. The Demerjian/Schere Characteristic Mixing
Height Growth Curve
Interval:
Fraction of
day:
Fraction of
growth:
Growth
rate:
= 0.0 0.07 0.14 0.33 0.50 0.70 0.90 1.0 2.0
= 0.0 0.02 0.10 0.588 0.85 1.00 1.00 0.5 0.5
=0.286 1.143 2.526 1.588 0.750 0.0 -5.0 6.0 0.0
0.00
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 O.d 0.9 1.0
TSR TSS
FRACTION OF DAYLIGHT
1.00
0.90
0.80
0.70
0.60
0.50
0.10
0.30
0.20
0.10
0.00
TSR2
Figure 73. The characteristic growth curve.
170
-------
mixing height. The maximum mixing height is assumed to occur at 70% of the
total daylight period. Note that 10% of the growth occurs in 14% of the day
and that 85% of the growth occurs in 50% of the day. Morning and afternoon
mixing heights for use with the curve were supplied by OAQPS/EPA.
Table 29 lists the hourly values of mixing height for all three
methods for the ten days studied.
The effect of a mixing height rise, assuming uniform mixing to the
mixing height, is three-fold: 1) the volume of the "box" is bigger and the
same mass in a bigger volume means a decrease in concentration, i.e., dilu-
tion; 2) the rate of increase of concentration due to emissions is less
because the volume is bigger; and 3) material that was above the mixing
height is entrained into the mixed layer and is diluted uniformly throughout.
As described in Table 1, the first and third effects are implemented in the
model as a dilution rate while the second effect is accounted for by dividing
the emissions rate by the current mixing height.
To calculate a dilution rate, the rate of change of the mixing height
is needed, as well as the mixing height; that is, the dilution rate is given
•y' k , = -r- -nr , where k. = dilution rate, h = mixing height
To obtain a smooth and continuous value for the mixing height and its rate
of change,rational spline functions with variable tension were used to fit
the values in Table 29. The resulting mixing height profiles must be mono-
tonically increasing functions,that is, the derivative must be zero or posi-
tive everywhere. The mixing height values were initially fitted with a low
tension which results in a cubic spline-like fit. If the derivatives any-
where were negative, the fit was repeated with increasing tension until the
derivatives were positive everywhere. The resulting curves for each of the
171
-------
i— r-~ r— O> CO LO ro r-^
ro in vj- ro r— CD cr> CD
i— CM LO co i— ro en ,— OOO
<^t- «J LO UD CO Cn i— CM
OOOOOOr- r-^
O O CT> "X> •— •^f'-OOi— CMCTiCTiCTt
OOLooov-crocTiLOCMcrirororo
r— i— .— <^-GUDCnroir>'^>r-^r---r~-
O O O O r— i— •— CMCMCMCMCMCM
OOOOCOOOOOOOO
LOLOLOOOLOLOLOLni-OLOLOI-O
OO>Or— CMCOr^r— r— r— !— .— r—
OOOOOOOronororororo
CM CM CM CM CM CM OJ
OOOOOOOO
oooooooo
r~- CO en o r— CM ro rj-
OOOr— r— r-r-,—
ooooooooooooo
ooooooooooooo
OOOOr
- i— ro in UD
- LD no ,— c\j
ix) r^. co 03
O O O r- r— .- r-
i— CDcnr-—
OOOOOOOCDOOO
ooooooooooo
OOOCDOOCDOOOO
O«— i— cxjcNjro^Lnin^or*-
ooooooooooo
O O O O O O O
O O O O O O O
i— co ro ^j- LO 1^3 r-^
ooooooooooo
OOOOOOOOOOO
CTi
C\J
a>
'QJ
-ME H
oocvioo
LOUOOOOOOOOO
O^O^CTiCTiCTlCriCTlCTiCTt
i— i— OOOOOOOO
OOC\JC\JC\JCMCXJC\JC\JC\J
OOOOCDOOOOO
ooOLncrtu~)kDooororo
Oi— i— CNJ«d-l£)Oti— ,— ,—
OOOOOOOf— i— .—
OUDCOi —
OOOOr
OOOOr— r— r— r-
COLOr— CO«^-COOr—
^O«=tCMCnr~-CMCMr—
CM- COCO
OOOi— r—'^i—^r-^r—"r-^
OLOr—OOOOOO»O
LO «^- LO <~ ~
O r— i •
O O O r
r— r— O LO LO O i - . —
OOOOr— ,— r— r—
ooo
(IT)i— CM
OOOOOOi— .—
oooooooooo
oooooooooo
cocnOr—cMm^j-LOLor--
OO' > " r— , i . r—
OOCDOO'^OO
OOOOO'OOO
OOO-— r— -—
172
-------
u
! '1H3I3H ONIXIW
'1HOI3H ONIXIW
CO
>>
to
I ' I
I ' I
LO
01
o
S-
Q.
CD
OJ
O) •
c:
X S-
•r- 3
s: o
01
(OCMCNJOJ—«—"C>
""( '1H3I3H ONIXIW
^'i 'J.HOI3H ONIXIW
173
-------
T3
01
"I '1HOI3H ONIXIW
'1HOI3H ONIXIW
o
u
to
-a
«t- O O3 OJ CD
CVJ C\J —' —' O
""I 'J.H2I3H ONIXIW
I ' I ' I ' I ' I ' I ' I
OJ
4-
O
1-
Q.
CD
O)
en
c:
O)
s-
Z3
CD
'1HOI3H ONIXIW
174
-------
""( '1HOI3H ONJXIW
•O
-------
profile methods(linear, handpicked, and characteristic curve) for each of
the days are shown in Figure 74, all on the same scale.
The different methods of estimating the mixing height profile do pro-
duce large temporal differences in the profiles.
In the solution of the model it is the dilution rate, rather than the
mixing height itself, that is of interest. As has been explained, both the
height and the rate of change determine the dilution rate. Thus, if the
rate of change is constant, then the dilution rate will decrease over time.
If this period is followed by one in which the rate changes, then the dilu-
tion rate can rapidly increase before decreasing again. To illustrate this
process, the dilution rates resulting from the three profiles for two days
are shown in Figure 75. Obviously, the different dilution patterns will
have a strong influence on the concentration profiles. Such discontinuities
in dilution probably do not occur in the real world.
176
-------
DILUTION RflTE. /win
DILUTION RflTE. /»
DILUTION RflTE.
'1HOI3H ONIXIU
""I '1HOI3N ONIXIU
to
OJ
O
S-
Q.
CD
O
O)
cn
DILUTION RrtTE.
DILUTION RflTE. /mn
I . I . I . I
DILUTION RBTE
4J
3
CD
s_
Ol
+J
fO
s_
c
O
CD
en
"1 'iHOI3H ONIXIU
"1 '1HOI3H ONIXIW
177
-------
7.5 OZONE ALOFT
The last data needed to perform the simulations are estimates of 63
aloft or ozone that is above the mixing height in the morning that would be
entrained into the mixed layer as the mixing height rises. Direct measure-
ments of this value are rare. It has been shown that there is a relationship
between the average 0^ aloft as measured by helicopters and subsequent
reading of surface 63 monitors following inversion breakup. One example of
this relationship is shown in Figure 76 (Evans, 1979). The recommended pro-
cedure for estimating the 03 aloft values is described in the EPA guidance
document (EPA, 1981). The procedure basically consists of using 03 measure-
ments from surface monitoring sites upwind of the city during the first hour
after breakup of the nocturnal inversion as an estimate of the ozone aloft.
This technique was used by OAQPS/EPA to estimate the 63 aloft values for the
days studied herein; these studies were supplied to UNC and are listed in
Table 30.
In addition to 63 aloft, all simulations assumed a natural background
value of 0.20 ppm carbon monoxide aloft.
Figure 76 AVERAGE OZONE CONCENTRATION IN EARLY VERTICAL PROFILE
3 VS UPWIND POST-BREAKUP CONCENTRATION
8.15 —| "~~~ ?
A DAY 201
O D»Y 305
a DAY 216
* OM217
O DAY 22i
V DAY 225
X DAY 228
a.05 8.10 e.15
SURfACE UPWIND HOUR AVERAGE OZONE CONCENTRATION
FOLLOWING INVERSION BREAKUP
178
-------
Table 30
Assumed Ozone Aloft Used in Models
of Selected RAPS Days
(ppm)
Day Concentration
159 - June 7 0.12
160 - June 8 0.09
195 - July 13 0.07
201 - July 19 0.08
221 - August 8 0.07
238 - August 25 0.09
261 - September 17 0.06
275 - October 1 0.06
276 - October 2 0.08
-------
8.0 RAPS SIMULATIONS WITH SIMPLE TRAJECTORY MODELS
8.1 METEOROLOGY-ONLY SIMPLE TRAJECTORY MODEL SIMULATIONS OF RAPS DAYS
8.1.1 Brief Description
The Meteorology-Only Simple Trajectory Model (MOSTM) has no chemistry.
That is, the concentrations of the species of interest are effected by the
initial concentrations, their boundary values (the concentrations transported
and entrained in the mixed layer) and their emissions; all of which are di-
luted as the mixing height rises. Thus, the rate of change in the concentra-
tion is prescribed by the equation:
dci = - 1 dh r 1 dh /r_nn „„ . + Ei
- — C-j + - — (FRA-; CA,-) —
dt h dt h dt h
where,
dC-j
— is the rate of change of the concentration of the
ith species
- — is the dilution rate
h dt
h is the mixing height
C-j is the concentration of the ith species
FRAj is the fraction aloft of the ith species at time t
CA-j is the concentration aloft of the ith species
Ej is the emission flux of the ith species
180
-------
A meteorology-only simulation gives an indication of the reasonableness
of the input data. If the precursors are not predicted"correctly, then the
Oo is not likely to be predicted correctly. Since adding chemistry to the
simulation acts as a sink for NO and HC and a source for 0.,, the predictions
X o
of NO and HC concentrations in MOSTM should be greater than the observed
/\
concentrations along the trajectory. Carbon monoxide predictions should be
close to the ambient data if the emissions and mixing height data
were accurate. The magnitude of 0. aloft entrainment can also be seen.
When the meteorology-only simulation underpredicts the observed NO and HC
X
concentrations, the input data should be reexamined.
8.1.2 Results of MOSTM Simulations
Figure 77 shows the model predictions as solid lines and the ambient
data that were shown in Figures 55 - 64 are shown here as symbols and
dashed lines. One mixing height profile was used for each row and a single
day is in each column. One page shows the NMHC and CO predictions and the
facing page shows the NO and 0.,. Remember that the 0, model predictions
X j o
are the result of initial concentrations and entrainment from aloft as the
mixing height rises and that, since there was no chemistry, there is no
reaction of the NO with the 0-.
In the June 7 (D159) model predictions, the first column of Figure 77,
the only differences in input data were the three mixing height profiles
(see Figure 74). These do result in significant differences among the simu-
lations. In the LIN simulation, the CO is nearly constant at about 0.7; in
the HP (handpicked) simulation, the CO decreases sharply then rises to a peak
greater than 1 ppm and then falls to about 0.4 ppm; in the CHAR simulation
the CO decreases slowly all day.
A comparison of the emission pattern for this day (Figure 69) with the
181
-------
mixing height profiles (Figure 74) reveals the causes of these different
predictions. The highest CO emissions were for the 8-9 A.M. hour; from
9 to 13 the emissions were almost constant and they decreased after 1300
LOT.
The linear mixing height profile was the highest at 8 and it changed
only a small amount over the 8-9 hour. The dilution caused by the rise was
almost exactly compensated for by the emissions so the concentration did not
change very much. After 9, the CO emissions decreased and likewise the
dilution rate decreased all day, resulting in little change in the CO
concentration.
The HP mixing height profile was tiie lowest at 8 and it tripled its
height in 8-9 hour, resulting in a factor of three dilution for the initial
CO concentration. This was more than compensated for by the CO emissions.
Even though these were the same magnitude as in the LIN case, they went into
a volume that was initially 15 times smaller and at 0900 LOT, was still six
times smaller. Thus, the emissions into this smaller volume were able to
rapidly increase the CO concentration. In the hour from 10 to 11, however,
the volume increased by nearly a factor of 10, resulting in a rapid and
dramatic decrease in CO concentration. After 1100 LOT, there was no change
in mixing height and therefore no dilution. The CO emissions into the large
volume were able to slightly increase the CO concentration.
Since the CHAR m'xing height profile is in between the LIN and HP it
exhibits some characteristics of both. Its initial height and rate of rise
prevents the large increase in CO concentration that was seen in the HP case.
The dilution rate, however, is larger than in the LIN case, resulting in a
rapid initial drop. Because the initial volume was smaller than in the LIN
case, less CO mass was present, even though both simulations had the same
182
-------
-o
o
O)
S-
fC
01
c:
O
oo
rd
-o
ro
o
C_3
rc
x
o
o
<+-
O i-
-------
O)
rs
c:
c:
o
o
O)
S-
184
-------
1 S I
I , I I I I I | I I I I I , I
5 S 5 S3
I ' I ' I ' I ' I ' I ' I
QJ
3
c
o
u
(U
i.
3
CD
185
-------
L_LJ__LJ
-£
i I . I i I i I 1,1,1,1,1,,
-I-
;-
f-
-o
cu
O
-------
1:
-1-
s I
: 1
-a
cu
c
o
o
-------
0,. ppm
0:
i . i . i . i . i . i
• r'-i ' i • r
o
o
-------
*-t-' I ' I ' I ' I ' I ' 1 -
if
$~\~^'
'Y i i i i i i i i i i i i i i i i i ,'
CO. ppmC
-a
O)
o
o
-------
0.. ppm
rr-p-| ' i ' i
O)
rs
o
o
190
-------
initial CO concentration. The final volumes, however, were similar, there-
fore, the concentration was less in the CHAR simulation.
To aid in these interpretations, Table 31 has been prepared. In MOSTM
simulations there are three sources of mass: initial material, emissions,
and material aloft that is entrained. There are no losses of mass, only
increase in volume which results in decrease of concentrations (mass/unit
volume). MOSTM keeps track of how much of the final concentration resulted
from which process; these values are shown in Table 31. For Day 159
(June 7), for example, 46.6% of the final CO concentration was the result
of the initial amount of CO specified. That is, if there had been no emis-
sions and no CO aloft, the final CO concentration would have been 0.304 ppm.
The amount of material that results from entrainment varies with mixing
height profile (see Table 31). This is because the ratio of the difference
between final and initial heights to" the final height is quite different
for each mixing height profile. That is, the mass of ozone at the end of
the simulation is given by:
ozone mass = [03]f* MHf= [03]A[_ * (MHf - m.) + [03]. * m.
where, [O^L - final ozone concentration in mixed region
[O-L. = aloft ozone concentration (assumed constant)
[0,]. = initial ozone concentration in mixed region
v5 1
MHf = mixing height final
MH. - mixing height initial
191
-------
Table 31. Effects of Mixing Height Profile on the Contribution of Ea..h
Source to Final Concentrations in MOSTM Simulations, All Days
Day 159
(ppm)
Mixing Height
Profile
Spec IBS
INITIAL
Amount Percentage
EMISSIONS
Amount Percentage
ENTRAPMENT
Amount Percentage
Final
Concentration
Linear
CO
MO
N02
HC
03
0.304
0.003
0.012
0.023
0.014
46.6
16.6
86.3
14.9
16.8
0.290
0.017
0.002
0.130
0.000
44.4
83.4
13.7
85.1
0.00
0.058
0.000
0.000
0.000
0.070
9.0
0.0
0.0
0.0
83.2
0.652
0.021
0.014
0.152
0.084
Handpickea
CO
NO
N02
HC
03
0.018
0.000
0.001
0.001
O.Q01
4.8
1.3
29.5
1.2
0.7
0.264
0.016
0.002
0.118
0.000
69.5
98.7
70.5
98.9
0.0
0.098
0.000
0.000
0.000
0.117
25.7
0.0
0.0
0.0
99.3
0.380
0.016
0.003
0.119
0.118
Characteristic
CO
NO
N02
HC
03
0.105
0.001
0.004
0.008
0.005
22.5
6.7
69.5
5.9
4.6
0.273
0.017
0.002
0.124
0.000
59.3
93.3
30.5
94.1
0.0
0.086
0.000
0.000
0 . 000
0.103
18.3
0.0
0.0
0.0
95.5
0.469
0.018
0.006
0.132
0.108
Day 160 SUe 103
(ppm)
Mixing Height
Profile
Sp_e_c_i_es
INITIAL
Amount Percentage
EMISSIONS
Amount Percentage
ENTRAPMENT
Amount Percentage
Final
Concentration
Linear
CO
NO
N02
HC
03
0.044
0.001
0.004
0.003
0.001
4.7
1.7
36.8
0.6
1.2
0.810
0.063
0.007
0.460
0.000
85.7
98.3
63.2
99.5
0.0
0.091
0.000
0.000
0.000
0.082
9.6
0.0
0.0
0.0
98.8
0.945
0.064
0.011
0.463
0.083
Handpicked
CO
NO
N02
HC
03
0.002
0.000
0.001
0.001
0.000
2.6
1.1
2.7
0.4
0.3
0.341
0.026
0.003
0.194
0.000
75.6
98.9
73.3
99.6
0.0
0.098
0.000
0.000
0.000
0.088
21.7
0.0
0.0
0.0
94.7
0.450
0.027
0.004
0.194
0.118
Characteristic
CO
NO
N02
HC
03
0.037
0.001
0.003
0.002
0.001
6.9
2.9
49.4
1.0
1.0
0.410
0.032
0.004
0.233
0.000
76.0
97.1
50.7
99.0
0.0
0.092
0 000
0.000
0 000
0 083
17.1
0.0
0.0
0.0
99.0
0.539
0.033
0.007
0.235
0.084
192
-------
Table 31. continued
Day 160 Site 115
Mixing Height
Profile
INITIAL
Amount Percentage
EMISSIONS
Amount Percentage
ENTRAPMENT
Amount Percentage
Final
Concentration
Linear
CO
NO
N02
HC
03
0.014
0.000
0.001
0.001
0.000
3.7
2.7
41.7
0.6
5.0
0.279
0.009
0.001
0.129
0.000
72.3
97.3
58.2
99.4
0.0
0.093
0.000
0.000
0.000
0.084
24.1
0.0
0.0
0.0
95.0
0.386
0.010
0.002
0.130
0.088
Handplcked
CO
NO
N02
HC
03
0.010
0.000
0.001
0.001
0.003
3.9
3.6
48.5
0.8
3.5
0.151
0.005
0.001
0.770
0.000
59.0
96.4
51.5
99.2
0.0
0.095
0.000
0.000
0.000
0.086
37.1
0.0
0.0
0.0
96.5
0.256
0.005
0.001
0.011
0.089
Characteristic
CO
NO
N02
HC
03
0.061
0.001
0.003
0.004
0.019
20.2
16.5
83.5
4.3
23.1
0.170
0.006
0.001
0.079
0 000
56.6
83.5
16.5
95.7
0.0
0.079
0.000
0.000
0.000
0.063
23.2
0.0
0.0
0.0
76.9
0.301
0.007
0.004
0.082
0.082
Day 195
Mixing Height
Profile
Linear
§pecies
CO
NO
N02
HC
03
INITIAL
Amount
0.027
0.001
0.001
0.003
0.014
Percentage
8.9
3.5
23.1
2.5
20.8
EMISSIONS
Amount
0.205
0.019
0.002
0.106
0.000
Percentage
66.4
96.5
77.9
97.5
0.0
ENTRAPMENT
Amount
0.076
0.000
0.000
0.000
0.053
Percentage
24.7
0.0
0.0
0.0
79.2
Final
Concentration
0.308
0.019
0.003
0.108
0.067
Handpicked
CO
NO
N02
HC
03
0.042
0.001
0.001
0.004
0.022
10.9
3.9
25.2
2.9
32.7
0.281
0.025
0.003
0.150
0.000
72.7
96.1
74.8
97.2
0.0
0.063
0.000
0.000
0.000
0.044
16.4
0.0
0.0
0.0
67.3
0.386
0.027
0.004
0.149
0.066
Characteristic
CO
NO
N02
HC
03
0.075
0.002
0.002
0.008
0.039
23.8
8.9
44.9
6.6
61.7
0.208
0.019
0.002
0. 10?
0.000
65.4
91.1
55.1
93.4
0.0
0.034
0.000
0.000
0.000
0.024
10.8
0.0
0.0
0.0
38.4
0.318
0.021
0.004
0.115
0.063
193
-------
Table 31. continued
Day 201
Mixing Height
Profile
Linear
Handpicked
Characteristic
INITIAL
Species Amount Percentage
CO
NO
H02
HC
03
0.058
0.001
0.002
0.004
0.020
16.2
3.8
41.5
3.5
28.6
0.236
0.024
0.003
0.117
0.000
66.0
96.2
58.6
96.5
0.0
0.064
0.000
0.000
0.000
0.051
17.8
0.0
0.0
0.0
71.4
0.058
0.001
0.002
0.004
0.020
0.011
0.000
0.000
0.001
0.004
0.013
0.000
0.001
0.001
0.005
16.2
3.8
41.5
3.5
28.6
3.9
1.0
15.1
0.9
4.9
4.6
1.2
17.5
1.1
5.9
EMISSIONS
Amount Percentage
ENTRAINMENT
Amount Percentage
CO
NO
N02
HC
03
0.011
0.000
0.000
0.001
0.004
3.9
1.0
15.1
0.9
4.9
0.176
0.018
0.002
0.087
0.000
62.8
99.0
84.9
99.1
0.0
0.093
0.000
0.000
0.000
0.075
33.3
0.0
0.0
0.0
95.1
CO
NO
N02
HC
03
0.013
0.000
0.001
0.001
0.005
4.6
1.2
17.5
1.1
5.9
0.176
0.018
0.002
0.087
0.000
62.7
98.8
82.5
48.9
0.0
0.092
0.000
0.000
0.000
0.074
32.7
0.0
0.0
0.0
94.2
Final
Concentration
0.357
0.025
0.005
0.121
0.0071
0.280
0.018
0.002
0.088
0.078
0.281
0.018
0.002
0.088
0.078
Day 221
Mixing Height
Profile
INITIAL
N jou nt_ Percen tn y_e
EMISSIONS
Ainou rit_ Percentage
ENTRAINMENT
o^Tt_ Percentage
Final
Concentration
Linear
CO
NO
N02
HC
03
0.068
0.002
0.002
0.004
0.006
16.6
7.4
43.8
4.2
11.7
0.278
0.022
0.002
0.097
0.000
67.9
92.6
56.2
95.8
0.0
0.063
0.000
0 000
0.000
0.044
15.5
0.0
0.0
0.0
88.3
0.409
0.023
0.004
0.102
0.050
Handpicked
CO
NO
N02
HC
03
0.006
0.000
0.000
0.000
0.001
1.7
0.8
7.1
0.4
0.8
0.249
0.019
0.002
0.087
0.000
70.8
99.2
92.9
99.6
0.0
0.097
0 000
0 000
0 000
0 068
27.5
0.0
0.0
0.0
99.2
0.352
0.020
0.002
0.088
0.068
Characteristic
CO
NO
N02
HC
03
0.027
0.001
0.001
0.002
0.002
6.9
3.1
23.8
1.9
0.037
0.272
0.021
0.002
0.095
0.000
70.8
96.9
76.2
98.3
0.0
0.086
0,000
0.000
0 000
0.060
22.3
0.0
0.0
0.0
96.3
0.304
0.022
0.003
0.097
0.062
194
-------
Table 31. continued
Day 238
Mixing Height
Profile
Linear
Species
CO
NO
NO
HC
03
INITIAL
Amount Percentage
0.081 34.6
0.001 23.0
0.002 75.5
0.003 4.7
0.001 1.8
EMISSIONS
Amount Percentage
0.067 28.6
0.005 77.0
0.001 24.5
0.068 95.3
0.000 0.0
ENTRAINMENT
Amount Percentage
0.086 36.8
0.000 0.0
0.000 0.0
0.000 0.0
0.078 98.2
Final
Concentration
0.234
0.006
0.002
0.071
0.079
Handpicked
CO
NO
N02
HC
03
0.020
0.000
0.000
0.001
0.000
11.4
7.6
45.8
1.3
0.4
0.061
0.004
0.001
0.062
0.000
34.3
92.4
54.2
98.7
0.0
0.097
0.000
0.000
0.000
0.087
54.3
0.0
0.0
0.0
99.6
0.1778
0.004
0.001
0.063
0.087
Characteristic
CO
NO
N02
HC
03
0.072
0.001
0.001
0.003
0.001
33.6
24.9
77.4
5.1
0.5
0.053
0.004
0.000
0.054
0.000
25.1
75.1
22.6
94.8
0.0
0.088
0.000
0.000
0.000
0.079
41.3
0.0
0.0
0.0
98.5
0.213
0.05
0.002
0.057
0.080
Day 261
Mixing Height
Profile
Linear
Spec ?os
CO
NO
N02
HC
OC
INITIAL
Amount Percentage
0.078 21.4
0.006 13.4
0.004 47.5
0.005 1.4
0.002 3.3
EMISSIONS
Amount Percentage
0.575 69.0
0.041 86.6
0.005 52.5
0.331 98.6
0.000 0.0
ENTRAINMENT
Amount Percentage
0.080 9.6
0.000 0.0
0.000 0.0
0.000 0.0
0.048 96.7
Final
Concentration
0.833
0.047
0.009
0.336
0.049
Handpicked
CO
NO
N02
HC
03
0.040
0.001
0.001
0.001
0.000
4.8
2.8
14.5
0.3
0.6
0.682
0.048
0.005
0.393
0.000
83.5
97.1
85.6
49.7
010
0.010
0.000
0.000
0.000
0.057
11.7
0.0
0.0
0.0
99.4
0.817
0.049
0.006
0.394
0.053
Characteristic
CO
NO
N02
HC
03
0.058
0.002
0.001
0.002
0.001
9.1
5.6
25.8
0.5
10.9
0.048
0.034
0.004
0.280
0.000
76.3
94.4
74.2
99.5
0.0
0.093
0.000
0.000
0.000
0.056
14.6
0.0
0.0
0.0
99.1
0.638
0.036
0.005
0.282
0.057
195
-------
Table 31. continued
Day 275 @ 8 am
Mixing Height
Profile
INITIAL
Species Amount Percentage
EMISSIONS
Amount Percentage
ENTRAINMENT
AmouiVt Percentage
Final
Concentration
Linear
CO
NO
N02
HC
03
0.712
0.033
0.013
0.043
0.001
38.4
25.1
53.3
10.4
2.4
1.067
0.099
0.011
0.368
0.000
57.5
74.9
46.7
89.6
0.0
O.C77
0.000
O.COO
0.000
0.045
4.2
0.0
0.0
0.0
97.6
1.856
0.132
0.0234
0.411
0.048
Handpicked
CO
NO
N02
HC
03
0.177
0.008
0.003
0.011
0.000
14.6
8.7
24.4
3.2
0.5
0.940
0.087
0.010
0.324
0.000
77.6
91.3
75.6
96.8
0.0
0.094
0.000
0.000
0.000
0.057
7.8
0.0
0.0
0.0
1.0
1.211
0.095
0.013
0.335
0.057
Characteristic
CO
NO
N02
HC
02
0.645
0.030
0.011
0.039
0.001
28.1
17.1
41.3
6.7
2.1
1.569
0.145
0.016
0.541
0.000
68.4
82.9
58.7
93.3
0.0
0.080
0.000
0.000
0.000
0.048
3.5
0.0
0.0
0.0
97.9
2.293
0.175
0.027
0.580
0.049
Day 275
Mixing Height
Profile
Linear
INITIAL
Species
CO
NO
N02
HC
03
Amount Percentage
0.108
0.007
0.002
0.006
0.000
6.8
5.4
13.4
1.3
0.7
EMISSIONS
Amount Percentage
1.387
0.126
0.014
0.493
0.000
87.4
94.6
86.6
98.7
0.0
ENTRAINMENT
Amount Percentage
0.092
0.000
0.000
0.000
0.055
5.8
0.0
0.0
0.0
99.3
Final
Concentration
1.587
0.133
0.016
0.500
0.056
Handpicked
CO
NO
N02
HC
03
0.074
0.005
0.002
0.004
0.000
5.3
4.3
10.8
1.0
0.5
1.222
0.111
0.012
0.435
0.000
87.9
95.7
89.2
99.0
0.0
0.094
0.000
0.000
0.000
0.057
6.8
0.0
0.0
0.0
99.5
1.391
0.1556
0.014
0.439
0.057
Characteristic
CO
NO
N02
HC
03
0.248
0.017
0.005
0.015
0.001
10.5
8.3
19.5
2.0
1.9
2.041
0.185
0.021
0.726
0.000
86.1
91.8
80.5
98.0
0.0
0.081
0.000
0.000
0.000
0.049
3.4
0.0
0.0
0.0
98.1
2.369
0.202
0.026
0.740
0.050
196
-------
Table 31. continued
CO
NO
N02
HC
03
0.213
0.008
0.003
0.008
0.000
40.5
36.7
63.5
17.6
0.1
0.21E
0.014
0.002
0.037
0.000
40.9
63.3
36.5
82.4
0.0
Day 276
Mixing Height INITIAL EMISSIONS ENTRAPMENT Final
Profile iP££iSi Amount Percentage Amount Percentage. Amoun_t Percentage Concentration
Linear CO 0.213 40.5 0.21E 40.9 0.010 18.6 0.526
0.000 0.0 0.0223
0.000 0.0 0.004
0.000 0.0 0.045
0.078 99.9 0.078
Handpicked CO 0.149 34.3 0.187 43.0 0.098 22.6 0.435
NO 0.006 31.8 0.012 68.2 0.000 0.0 0.018
N02 0.002 58.4 0.001 41.6 0.000 0.0 0.003
HC 0.006 14.7 0.032 85.3 0.000 0.0 0.038
03 0.000 0.0 0.000 0.0 0.079 99.9 0.079
Characteristic CO 0.344 52.4 0.215 32.9 0.096 14.7 0.655
NO 0.013 48.3 0.014 51.7 0.000 0.0 0.028
M02 0.004 73.7 0.002 26.3 0.000 0.0 0.006
HC 0.013 25.6 0.037 74.4 0.000 0.0 0.050
03 0.000 0.2 0.000 0.0 0.077 99.8 0.077
1.37
-------
Thus, for the LIN case for Day 159, the 03 concentration resulting from
entrapment was 0.070 ppm with [0,].. = 0.12 ppm and (MH, - MH.)/MH, = 0.58,
•J ML T It
while for the HP case (MHf - MH1)/MHf = 0.97, resulting in 0.117 ppm Q^ from
entrainment. In examining the plots in Figure 77, it is evident that the
detail in the ambient data is not always simulated very well by the model.
For example, the CO ambient data for Day 159 shows some very sharp spikes
in the afternoon(1330 to 1530 LOT) that do not occur in any of the models.
The NMHC data exhibits similar rapid jumps for June 8 (D 160/03) and in the
morning the NO shows a quadrupling of concentration in one hour. Particu-
A
larly bad cases are July 13 and July 19, where the trajectories presumably
went through the center of town and a dense part of the RAMS network. It is
evident from the simulations in Figure 77 that the emissions and mixing
height data for these two days are not consistent with the ambient data.
This is especially troublesome on July 19 which should be an ideal case for
a simple trajectory model approach. The July 19 case will be examined in
detail later in this report (section 8.3.3 and 8.3.4). The oversimplistic
treatment of point sources and the large amount of "noise" in the trajectory
leading to spacial averages of the area emissions may be major causes of the
poor agreement.
8.1.3 Summary of MOSTM simulations
Based on meteorology-only simulations the days expected to perform
poorly in the photochemical simulations are: June 8, D'160/15; July 13,
D195; July 19, D201; August 25, D238; LIN and CHAR for Sept 17, D261;
starting at 0600 LOT on October 1, D275; and LIN and CHAR of October 2,
D276. More than one-half of the test cases appear to be so poorly charac-
terized by input data that they are probably not suitable for examining
the effects of chemical mechanism choices on controls.
198
-------
8.2 PHOTOCHEMICAL SIMPLE TRAJECTORY MODEL SIMULATIONS OF RAPS DAYS
8.2.1 Brief Description
The Photochemical Kinetics Simple Trajectory Model (PKSTM) is a substan-
tially expanded OZIPP-like model more suitable for research use. The
equations solved by the model were given in Table 1. Special features of
the model include:
• easy entry of different chemical mechanisms
• easy entry of photolytic processes
• use of variable temperature and temperature dependent rate constants
. use of variable humidity and water vapor dependent rate constants
. allows for CO emissions
. allows for CO aloft (constant or variable)
. allows for variable 03 aloft
. use of absolute emission units
. use of smooth, equal-area emission histogram fits
. use of smooth mixing height fits
. simple, fast solution algorithms
. large amount of output information
• simple repeat functions for isopleth diagram generation.
Of course the model incorporates all of the processes included in the
meterology-only (MOSTM) model and adds to these the processes needed to
represent the chemistry.
Each of the 10 days was simulated with each of the three mixing height
profiles (linear, hand picked, or characteristic) with each of three
mechanisms (Demerjian, Dodge, or Carbon Bond). Because of its performance
on the Bureau of Mines smog chamber, the CIT mechanism was used to simulate
199
-------
June 7 (D159) and October 1 (D275) only.
8.2.2 Results of PKSTM Simulations
The comparison of predictions and ambient data for all days is shown
in Figure 79. Figure 79 shows one day per page, a single mixing height
per row, and a single mechanism per column. The top row is always the
linear (LIN) mixing height profile results; the bottom row is always CHAR.
The order from left-to-right is always the Demerjian (DEM), Dodge (DOD),
and Carbon Bond II (CB2) mechanisms results. The CIT simulations appear
on a separate page by themselves.
Only the NO, N0?, and 0^ data are shown since the CO data would be
essentially the same as the MOSTM simulations already shown. Of course,
the mechanisms do not use NMHC, but represent individual molecules of
various hydrocarbon species. These were not plotted because there would
be no ambient data for comparison.
June 7 D159. As expected from the MOSTM simulations, this day was
reasonably well simulated by all four mechanisms. Of course this day had
0.12 ppm 0.-, aloft and significant quantities of 0^ were entrained.
June 8 D160/03. Only DEM and DOD CHAR simulations were close to
ambient data.
June 8 D160/15. As expected from MOSTM results, no simulations were
satisfactory.
July 13 D195. As expected from MOSTM results, no simulations were
satisfactory.
July 19 D201. As expected from MOSTM results, no simulations were
satisfactory. Although the DEM and DOD LIN simulated 03 values were
similar to the ambient data for the first few hours, the failure to track
200
-------
0. PP.
otn»meir>ois>o
Sryof^txl(vor-irt
(M to
O >,
Ol X>
c
I—
o s-
-c o
£X <4-
CTl
-------
S S
•a
a;
o
o
a>
202
-------
1,1,1.1.1,1.1,1.1
S 8 I S S S £ = = I
•OH ON
g a: s s s s s s
•*< ""OH 'OH
- 3
Jn*-
1,1,1.1,1,1.1.1,1
ssssssissli
•« "ON 'OK
13
O)
o
o
cu
s-
3
O)
203
-------
11 ' i' n i ' i ' i ' i ' i ' i '\-
• 1-: ' d=
OJ
3
C
O
o
CTl
r-.
t
3
204
-------
0.. m
SRSSSKgio
-a
-SJ
:\ *-. "f
- ••••-•.::::'•'•: (
i .
-
'i 1 1 1 1 .1.1.1.1.1,
s s
1 1 1 1 1
f
/
s/
\\
a
-
s
m
^
"
f
=s
-
»«•«>•----£
1 1 ' 1 ' 1 ' I ' 1 • 1 'T
-a
-s> 11
2« -I
:s*--*-,. I
.••*"*•*
""***«• .
jj _, i , 1 . r , I . I . LJ
• «
""'I ' 1 '
r
s'j~
Nv
i £
.1.1.'
s
-
"
2
-
- c
" £
= 4
-
•"« "W 'OH
•" 'IW 'ON
1. „.
2SSE2"-3s5™8
Ti' i • i' i' i • i' i' i' i' =
= 2
1 • i • i •' 1111 1111, i .I,,,
SssssssssSs
"• "ON 'OH
SPSS.
IS S S
SasESSISss
rK
'\ I
K S
- s
' i -1' i * i • i
r • i' i " ' i' i'''' i"'
-8
i, 11111 11 • i • i • i . i •-'-
i s s s s s s = \ s 1
*•' "OK 'M
SZ
O
O
cu
3
a>
205
-------
-o
O)
c
o
o
s_
3
CD
206
-------
•M -ON 'ON
-o
OJ
c
c
o
o
en
a)
s_
3
207
-------
§
-------
s
Pf a
s
i ' i ' i ' i' i ' i • i' r
£ o 10 •
s •• V-1!
i i i li i i l r l• ' • ' ' i • ' •'"
•« ••on 'on
-s :
K
I • I ' I ' I ' I ' I ' I ' I ' I 'J-
£ "•
»«« "ON 'W
-o
S
c
o
o
cr>
0)
S-
CD
209
-------
•"" "ON ON
-= a
-a
a>
e:
o
o
CTl
cu
O1
•r—
u_
210
-------
•" '•on •en
S -. 3
L. I , I , I , I . I . I . |
'•!» 'OH
c
8
en
r-.
OJ
CD
211
-------
-a
O)
c
o
O
ai
s_
en
•
-------
S S
s K
•" '-CH '0,1
T3
OJ
3
•(->
o
CT>
r-
0)
cr>
213
-------
the NO and NMHC for this day suggest that the actual processes
X
for this day were not well represented.
August 8 D221. The MOSTM results suggested that this day had poten-
tial for being reasonably simulated. DEM simulations were reasonable;
DOD simulations were slightly low; and CB2 failed to remove the NCU and
form 00.
August 25 D238. The MOSTM results showed that the afternoon CO,
NMHC, and NO were poorly predicted. Since the NO and HC rise in the
X X
13-1500 LOT period was not predicted, the 0^ predictions were significantly
low.
Sept. 17 D261. As suggested by the MOSTM the hand picked mixing
height profile appears to be the most appropriate for this day. Both DEM
and DOD predicted not only the 03 but also the NO, N02, and NMHC (and the
shape for CO) very closely. CB2 did not make much 0^ in the afternoon.
Oct. 1 D275 at SAM. The early start on this day is a good test of
having to simultaneously match the mixing height and its rise with the
emissions rates. The MOSTM LIN simulation suggested that CO would not be
predicted well. The HC and NO values, considering chemical losses,
A
might be a little low. The HP MOSTM results gave large NO overpredictions
X
and the CHAR simulation also gave a somewhat smaller but still large 8-9
morning spike.
In the PKSTM simulations, only DOD for CHAR was able to produce a
reasonable simulation.
Oct.__1 D275 at 8 AM. This was the high 0~ day. Comparison of the
8 AM simulations with the 6 AM simulations shows significant differences.
A major difference is that the 8 AM initial NO is specified in this
214
-------
simulation (and is therefore correct) whereas it was computed from two
hours of meteorology, emissions, and chemistry in the 6 AM starting
simulation. DEM and DOD did reasonably well; CB2 was lower. This was
the second day simulated with the CIT mechanism. As expected from the
high NO initial conditions and the BOM simulation results, the CIT
A
mechanism performed poorly on this day.
Oct. 2 D276 at 6 AM. The MOSTM simulations suggested that the mixing
heights specified for this day resulted in too much dilution and that the
NO emissions used were inappropriate for the 8-11 LOT period. As shown
/\
in the PKSTM these problems resulted in too little mass to continue the
03 production.
8.2.3. Summary of PKSTM Simulations
Table 32 summarizes the ozone maxima predictions by the photochemical
kinetics simple trejectory model for the eleven selected days. Figures 80
and 81 are scatter diagrams organized by mechanism and by mixing height
profiles. See Section 8.4 for discussion of "fits".
215
-------
Table 32. Ozone Maxima Predicted by MOSTM and PKSTM Using Four
Mechanisms and Three Mixing Height Profiles for
RAPS Days (ppm 03)
PKSTM
159
June 7
160/03
June 8
160/15
June 8
195
July 13
201
July 19
221
Aug 8
238
Aug 25
261
Sept 17
275/6
Oct 1
275/8
276
Oct 2
Obs
0.192
0.194
0.206
0.210
0.138
0.119
0.190
0.151
0.244
0.244
0.181
MHIP
L
HP
C
L
HP
C
L
HP
C
L
HP
C
L
HP
C
L
HP
C
L
HP
C
L
HP
C
L
HP
C
L
HP
C
L
HP
C
MOSTM
0.084
0.118
0.108
0.083
0.084
0.084
0.088
0.089
0.082
0.067
0.066
0.063
0.071
0.078
0.078
0.050
0.068
0.062
0.079
0.087
0.080
0.049
0.058
0.057
0.056
0.057
0.050
0.048
0.057
0.049
0.078
0.079
0.077
DEMER
0.2519
0.1642
0.1965
0.1782
0.1338
0.1457
0.1258
0.1059
0.1245
0.0982
0.1127
0.1325
0.1127
0.0918
0.0933
0.1337
0.1002
0.1137
0.0960
0.0951
0.0943
0.1032
0.1227
0.1042
0.0783
0.0708
0.1175
0.1970
0.0772
0.1680
0.1206
0.1012
0.1399
DODGE
0.1478
0.1143
0.1279
0.1822
0.1221
0.1300
0.0872
0.0564
0.0747
0.0937
0.1022
0.0988
0.1107
0.0954
0.0962
0.0851
0.0732
0.0807
0.0888
0.0864
0.0811
0.1382
0.1425
0.1210
0.1429
0.1254
0.1820
0.2138
0.1298
0.2100
0.0990
0.0997
0.1138
CARB BOND
0.1970
0.1320
0.1540
0.0904
0.0876
0.0825
0.1046
0.0935
0.1006
0.0719
0.0756
0.0986
0.0717
0.0706
0.0709
0.0721
0.0628
0.0651
0.0820
0.0875
0.0823
0.0491
0.0580
0.0545
0.0241
0.0297
0.0270
0.1160
0.0450
0.0808
0.0709
0.0723
0.0739
CIT
0.1963
0.1348
0.1534
0.0771
0.0385
0.0546
216
-------
en
O)
-C
'E e
O
*°
-i£
O
It
=
c.
to
**
-f-l
(O
OJ
O)
to O
tO t-
o
00
Ol
3
ov
217
-------
en
CD
QJ
c
O
'-0 OH3I03HJ
-0
C
H
O
£5
<_) "3
O)
+ 1
0)
T3
S_
QJ
S-
Z3
cn
218
-------
8.3 RESPONSE OF SELECTED PHOTOCHEMICAL SIMULATIONS TO DIFFERENT INPUT
ASSUMPTIONS
8.3.1 Carbon Bond and Demerjian Mechanism's Response to Increased Aldehyde
Emissions.
In the RAPS emission inventory (Table 25), aldehydes represented from
0.8% to 3% of the carbon for the days included in this study. The OZIPP
program with the Dodge mechanism assumes that aldehydes are 5% of the
NMHC and this assumption was maintained in this study for the Dodge
mechanism. The aldehyde fraction in the emission inventory may be a highly
uncertain number; there were no systematic aldehyde measurements conducted
in RAPS. Aldehydes are significant radical sources in the chemical
mechanisms.
The Carbon Bond mechanism performed very well in the Bureau of Mines
smog chamber simulations in which Demerjian's value of -5% aldehydes was
used and it performed poorly in the RAPS simulations in which -2% aldehydes
were used. Furthermore, Demerjian had assumed an average aldehyde carbon
number of about one for the BOM data and the RAPS emission inventory average
aldehyde carbon number was reported by Demerjian as being about 2.0. In the
Demerjian mechanism this shift is not important since there is no real effort
to have carbon accountability in the Demerjian mechanism. In the Carbon Bond
mechanism, where carbon accountability is a feature, this shift could be
critical. The extra carbon on the aldehyde would have to be counted as a
paraffin carbon in CB2 and thus the actual number of photolytic carbons would
be decreased by a factor of two when compared to Demerjian's mechanism in
which the single aldehyde species represents some combination of 1-carbon,
2-carbon, and higher-carbon number aldehydes. For Carbon Bond then, the
difference in assuming that the aldehyde carbon number is one or two repre-
219
-------
ents a doubling of the amount of carbon that is carbony! and thus a photo-
lytic source of radicals.
In Figure 82, the October 1, D275/8 standard Carbon Bond simulation is
compared to a Carbon Bond simulation in which the aldehyde initial and
emission carbon fractions have been doubled. Although the simulation pre-
dictions remain under the observed CL, there is a substantial improvement in
the predictions. Other simulations, not shown here, with higher aldehyde
fractions show that large aldehyde fraction assumptions can give nearly
perfect fits, suggesting that for this day the Carbon Bond mechanism is a
little short of radicals.
A similar experiment was conducted with the Demerjian mechanism. In
Figure 83, the October 1, D275/8 standard Demerjian simulation is compared to
a Demerjian simulation in which only the initial aldehyde carbon fraction was
increased from 2.0% to 5.0%; the emission aldehyde carbon fraction remained
at 2.0%. The effect was dramatic; the 0~ maximum increased by 20%! The cause
for this large response will be discussed below.
The effect of increased aldehydes was tested on a day that was modelled
well by Carbon Bond with the standard assumptions, June 7, D159. In Figure
84, the standard Carbon Bond simulation is compared to one in which both
the initial and the emission aldehyde carbon fraction had been increased
from 0.82% to 5%, a fivefold increase. Less than a ten percent effect on 0,
maxima was observed. However, this day had 0.12 ppm Og aloft and October 1
had only 0.06 ppm 0^ aloft.
The reaction of aloft 03 with emissions can serve as a radical source. The
addition of two radical sources does not have a linear effect in mechanisms;
if there are sufficient radicals, the effect of a second source may not be
very discernible. Furthermore, a detailed analysis of the October 1 simu-
220
-------
a.. PP.
"« 'MN 'ON
0.. ppm
'-ON 'OH
0.. ppm
E*
c cu
(O T3
^: r—
o re
O)
E -o
O)
CU
•">« 'TO 'OH
o
CO
c
o
re re
en co
c c
O ••-
'•Jj c
O T-
re co
s- QJ
If- ^2
o; c
-o o
>>••-
x: +•>
o> re
T3 I—
r— rs
re E
O) i/>
j=
+-> to
D.
cncC
c S
r— IO C
J3 r~ E
3 CTi 3
O r- •—
T3 O
O 4-)
s- ^:
4-> <]j en
O^ T-
o» o s-
M- -(->
M- O c
LjJ O -i-
CNJ
00
O)
3
co
221
-------
p
E 4^
Ss
s-
o to
+J CD
o c
£'£
0.. ppin
'"ON 'ON
Oi en
T3 <1)
O)
fO -P
rtJ
•r- (/)
E
•I- CO
Q_
S« et
LO CC.
CD IO
4J O)
O J2
O) O
f- -M
4- O
LLJ O
CO
CO
CD
•I—
U-
222
-------
E
V)
I . I . I . I . I • I • I • I • I • '"•
«« 'MN 'ON
'-ON 'ON
u
O)
-o
c
o
CO
c
o
(O
o
c
o
o
(D
c
o
•f—
•i- C
E O
1,1,1,1.1,1,1,1.1
sssssssll
«« '-ON 'ON
- 3
Oi.
1,1,1.1,1,1,1,1,1
cu ••-
XJ tn
(B I/)
o.
-o <
a; a:
to
to UD
oj r*~
s- en
o i —
c
O -3
a>
<+- s_
4- o
oo
a>
s-
s s s
•« '-OH 'ON
-------
lations showed that In the Demerjian mechanism the (L + OLE reaction
(Reaction 20, Table 7) gives two radicals (one R02 and one H02) and is a
significant source of new radicals (-5% of the radicals) for this mechanism.
The Carbon Bond mechanism, on the other hand, uses newer (and presumably more
kinetically correct) 03 + OLE chemistry in which the radical yield is
only about 0.1 (compared to Demerjian's 2.0) and this process is not a signi-
ficant source of radicals (<0.1% of the radicals) in the CB2 mechanism. There
is an additional radical source in the Carbon Bond mechanism and it is
photolysis of unique carbonyl products resulting from aromatic reactions.
Demerjian has no unique carbonyl aromatic products; arornatics are treated
similar to olefins with an extra standard aldehyde being produced.
Thus, for Demerjian, once 0, production begins and olefins are present
in the hydrocarbon emissions, new radicals are abundantly available to
sustain the 0^ production process. This may not be a good representation of
the actual process.
Extra aldehydes are not always adequate to improve the Carbon Bond
mechanism's performance. Figures 85 and 86 compare standard Carbon Bond
simulations to those with increased aldehyde emissions; not much improvement
occurred for these days which were simulated better by the other
mechanisms.
For the Dodge mechanism, the aldehyde emission fraction assumption is
constant. As was shown in Table 10, however, new photolytic rate data are
available which, if used, might affect the radical production in the Dodge
simulations. Figure 87 compares the standard Dodge October 1, D275/8 simu-
lation using the original ("OLD") photolytic rates from the OZIPP program
with a simulation using the newer photolytic rates ("NEW"). The effect was
very small.
224
-------
»"« ''ON 'ON
e
to
•r~
C
ro
.C
u
OJ
e
c
o
CO
c
o
-Q
s_
to
o
s_
o
c
o
o
(O
s_
c
o
CO C
oo O
QJ (O
OJ 3
S g
'-ON 'ON
*" "ON 'ON
JT 00
OJ
•O
O> r—
S_
O «
C CO
cn
O
O>
UJ
LO
00
O)
S-
O
225
-------
LO
O
oo
o
c
o
o
(O
LO C
c: , c
-C O
>
+-» (O
<_> a
a>
4- i-
M- O
00
CD
i-
226
-------
s s s
««H ''ON 'ON
""I 'MN 'ON
O
•SJ
O)
Ol
X3
(O
0)
o c
•M O
O •<-
^: •(->
CL to
"2 i
QJ •!-
C in
CO
•O Q.
C <
(Q o:
= VO
•a r^
r— <7>
o •—
O J3
01 o
4- 4->
M- O
LU O
CO
S-
'ON OK
221
-------
8.3.2 Demerjian and Carbon Bond Mechanism Response to Hydrocarbon
Compositional Changes
The Carbon Bond mechanism did well in the Bureau of Mines smog chamber
simulations in which the aromatics carbon fraction was nearly 30%, the
aldehyde carbon fraction was assumed to be about 5%, and the aldehyde carbon
number was assumed to be one. The mechanism did poorly in the RAPS simula-
tions in which the aromatics carbon fraction was 6% to 18%, the aldehyde
carbon fraction was 0.8% to 3%, and the aldehyde carbon number was 2.0.
The effect of switching from an aldehyde carbon number of 2.0 to an
aldehyde carbon number of 1.0 has already been demonstrated and discussed
(Figure 82). The effect of using the BOM HC composition in the RAPS
October 1 simulation will now be investigated. To see the effects of com-
position changes only, all simulations assumed an aldehyde carbon number of
1.0.
The left column of Figure 88 shows the October 1 simulation using the
October 1 HC composition and an aldehyde carbon number of 1.0; this is the
same as the right column of Figure 82. The center column of Figure 88 shows
the October 1 simulation with the BOM HC composition but using the October 1
aldehyde emission carbon fraction and an aldehyde carbon number of 1.0. The
right column of Figure 88 shows the October 1 simulation with the BOM compo-
sition and the BOM aldehyde carbon fraction (5%) and an aldehyde carbon
number of 1.0. The simulations with the full BOM assumptions nearly per-
fectly fit this day.
The HC compositional change was (% carbon):
NR PAR OLE ARO ALJJ
Oct. 1 12.2 53.7 17.2 14.9 2.0
BOM 10.5 41.7 17.3 28.7 5.3
228
-------
g 8 a E 5 «"• 8 5 8 s
0*. PP«
in o v^ o v> » v>
r>mfw«>r>tnn
— — — — 000
§ I
"ON 'OH
(/I
o
Q.
- 3
•» W IM
a. PP»
sSSSlEss
o
3
(O
3
CO
o
-M
229
-------
That is, about 12% of the carbon was shifted from paraffins to aromatics
between the simulations.
The change in 03 resulting from the various assumptions were (ppm 03):
Composition
Aldehyde Fraction
Aid. Carb. No.
Mixing Heiqht:
LIN
HP
CHAR
Oct. 1
0.020
2.0
0.116
0.045
0.081
Oct. 1
0.020
1.0
0.147
0.060
0.117
BOM
0.020
1.0
0.192
0.079
0.164
BOM
0.053
1.0
0.252
0.118
0.233
That is, the Oct. 1 to BOM compositional change increased the 0-, by
0.045 ppm (31% increase) and the Oct. 1 to BOM change with BOM aldehydes
increased the 03 by 0.105 ppm (71% increase). The total change in going
from standard Oct. 1 conditions to full BOM conditions, however, increased
the On by 0.135 ppm (117%). As described above, aromatic products are sig-
nificant sources of radicals for the Carbon Bond II mechanism and it has
some difficulties in dealing with these RAPS days with low aromatic content
as determined from the emission inventory and the trajectory paths.
Another demonstration of HC compositional effects on mechanism pre-
dictions, similar to that discussed in Section 7.3.4, is shown in Figure 89.
The left column is the standard Demerjian Oct. 1, D275/8 simulation from
Figure 79. The center column shows the effect of substituting the HC compo-
sition fractions from the day with the highest olefin content (August 8,
Day 221) for the October 1 HC composition fractions. The right column shows
the effect of substituting the HC composition fractions from the day with
the highest paraffin content (August 25, Day 238) for those of October 1.
This compositional change was (% carbon):
230
-------
•" '"ON 'ON
231
-------
NR & PAR OLE ARO ALD
Oct. 1
Aug. 25
65.9
86.0
17.2
10.0
14.9
6.6
2.0
1.0
These modelling results suggest that the control strategy for St. Louis
would be to replace 15% of the olefin and aromatic carbon with paraffin carbon
and have no_ reduction in total emissions; smog chamber data, however, suggests
that this would not be true.
232
-------
8.3.3 Effects of Trajectory Assumptions for Selected Days
The purpose of this section is to show the effects of emission pattern
variations that might result from trajectory assumptions. The emission rate
pattern is determined by the column path of the air parcel. The column path
is defined as the area over which the emission rates are determined. For
this study the column path was defined by the area encompassed by five back
trajectories from the site of observed ozone maximum. Since the meteoro-
logical parameters are assumed to be uniform throughout the urban area, the
emission rate pattern variations show the effects of column path variations.
To examine the effects of different emission rate patterns two approaches
will be taken. First days 238 and 201 will be examined, since each exhibited
similar ozone profiles for the three different mixing height profiles when
modeled with the three different chemical mechanisms (see Figure 79 of
section 8.2). The other approach is to see the effects of altering the
pattern of the hourly emission rates while keeping the total amount of the
emitted mass constant.
Day 238 - August 25. The column path from the supplied information is
repeated as Figure 90. The back trajectories from the site of the highest
ozone maximum indicate that the column enters the RAPS data region from the
southeast then turns approximately 90°, just southeast of the Wood River-
Alton point source region (see Figure 29). Figure 91 shows the modelled and
ambient concentrations of NOV, Oo, CO and HC. The ambient concentrations of
X o
these species along the column path are plotted as the dashed lines. These
represent the weighted average concentrations from the three closest
monitoring sites to the centerline of the column path. The solid lines
represent the predicted concentration of the species by the Demerjian, (DEM)
233
-------
4340
4330
4320
4310
430Q
4290
4280
4270
4260
4250
4240
700 710
720
730 740 750 760 770 780
Figure 90. Trajectory for Day 238, August 25/Si te 115
234
-------
mechanism.
Since the ozone maximum predicted by all three mechanisms for the three
different mixing height profiles was considerably less than the observed
ozone for the day (0.190 ppm), and since the CO concentration is so poorly
predicted, one can assume that the emission pattern of the column path
reaching the observation station is different than that used in the model
run.
For Day 238, August 17, three alternative column paths were examined:
A. The first, Tl, uses the centerline trajectory of the supplied
column path and places a two kilometer box around it, one km radius. Figure
91 shows that the ozone predicted varies little from the supplied column
path. Table 33 contains a list of the ozone predicted for each column path.
Table 34 contains the emission rates in units of ppm-km used for each column
path, Figure 92 represents the instantaneous fit of these emission rates.
The emission rates for T3 and TH column paths were selected from the emission
rates from the supplied column path for Day 201.
Notice the rise in ambient NO concentration in Figure 91, during the
/\
last hour and a half,and the rise and fall of the hydrocarbon and CO concen-
tration (also in Figure 91), during the last three hours of the simulation.
These changes in the ambient concentrations are not predicted usiny either
the supplied or Tl column paths.
B. The T3 column path considers a discontinuous column path, during
the last three hours of the simulation. The column path considered does not
curl around from the southeast, but comes down from the northwest passing
235
-------
through the Wood River-Alton Point source region, Figure 29. The T3 emission
rates consist of the Tl rates for the first four hours and the major emissions
from the Wood River-Alton region for the last three hours of the simulation.
Using this discontinuous column path, the DEM mechanism predicts much
more ozone than either the supplied or the Tl column paths, Table 33. The
N02 peak is now present, Figure 91, and the ozone concentration profile
appears to be best predicted using the characteristic or handpicked mixing
height profiles. A hydrocarbon peak is predicted, but in an excessive
amount, while the CO concentration is not modeled well.
C. Since there was promise from the T3 results,a hypothetical column
path was determined passing through the Wood River-Alton region, TH. The
TH column path was constructed from a straight line hypothetical path through
this region using a two kilometer wide box surrounding this line. The
emission rates for the TH column path consist of the T"! emission rates for
the first four hours and these new rates for the last three hours. Figure
91 and Table 33 show that this column path is an inadequate description of
the processes occurring on this day. The NOo levels are excessive; while
the hydrocarbon and CO predictions are better, they do not agree with the
observed ambient data.
The turning of the wind back trajectories in the region of the observed
ozone maximum may indicate aconvergence zone just outside the center city.
If this is the case then the air parcels reaching thic region came from
many paths. To adequately establish if this was the case a more detailed
wind field analysis must be performed. The determination of a column path
is not a simple process, as established by the TH column path selection.
236
-------
CO. pp»C
CO. pp»C
- 3
8 5
-3
v> e> i/>
228
ca u> a irt
CO. pp»C
jut" 'DWJN
CO. ,pmC
>»« 'MM
CO. >P«C
ce r- to 10
TS§
= 3
£0
CO
(M
(O
CO
-------
-a
a>
O
u
CTi
O)
5-
:3
CD
238
-------
co,
CO. PBmC
CO. ppmC
- 3
•E
• 3
2%
OJ — —
CD r- jj
- 8
L- 3
*
1,1,1,1,1,1,1,1,1,
inainoipoinotn
-" 5
-o
0)
c.
o
o
en
a>
&.
=j
O)
239
-------
1.1.1,1.1.1.1,1,1, ... LuJ_L
"O
O)
Z3
O
O
CTl
CD
240
-------
Table 33. Effects of Trajectory on Ozone Maxima; Day 238
DAY 238
Observed ozone = 0.190 ppm
Trajectory Sup T1 T3 TH
Mixing height profile
Linear 0.095 0.092 0.137 0.068
Handpicked 0.095 0.093 0.130 0.086
Characteristic 0.095 0.092 0.127 0.079
241
-------
co
c~
O
••~
(/)
to
E
UJ
(_}
m
nr
I—
oo
1—
I
I—
Q.
* /N
V J
.
•1-3
ro
s_
h-
o
00
o
o
o
o
co
o
o
o
o
co
o
0
o
CO
oo
o
0
o
o> en
E i
•r- CO
^-
r-.
"^3~
o
0
o
r-
«sj-
o
0
o
r-^
•=3-
O
O
O
o
un
0
o
o
o
1 —
1
en
O CM
^j- en
0 O
o o
o o
O CM
«3- en
0 0
0 O
o o
O CM
«* en
0 0
0 O
0 0
CO CM
CM O
*^J~ r—
0 O
0 0
r— CM
1 — ^—
1 1
0 r-
r— i
en
oo
CM
O
0
un
CM
0
oo
o
i —
0
CM
O
o
*^~
CM
i —
0
0
1 —
1
CM
1 —
CO
00
0
o
oo
r^^
un
CM
0
un
oo
i —
o
o
CM
i —
r—
O
o
CM
I
i —
OO
un
(^
i —
o
t
f-*v
un
•—
o
en
^j-
o
o
o
0
"Cj~
o
o
o
00
i s:
CM UJ
Q
un
r—
un
o
o
en
CO
CO
i—
o
en
CO
CO
0
o
o
co O
c:
O
4->
0
ro UJ
S- _l
u. o
CO
un
CO
CO
o
un
en
CO
o
un
en
00
un
o
0
un
un
r-~
o
ry"
c£
Q.
un
*^j-
, —
CM
O
un
en
o
o
un
en
o
0
o
CO
CO
0
0
o
Qi
00 UO
r- CO
CO O
0 0
O 0
r— O
CM CM
i — CM
•— o
0 0
r- 0
CM CM
i — CM
i— O
0 0
un oo
r^. o
CO i—
O 0
o o
f~^
cy t— i
•Z. «=C
o
u
OJ
fO
O)
O)
o
en
oo
OO
CM
>
n: co
oo
X) CM
sz
fO >i
fC
CO Q
cu
•4-> -
fO CO
o; cz
O
C T-
O 4->
•i- Q.
to E
CO 3
•r- tO
E tO
UJ c£
O r— CM
CL) CTli— i— i— i— CMOO
E I I I I I I I
•i- cocnoi— cMi— CM
O i— CM
CD CTl r— i— i— r— CMOO
E
•i- co en o i— CM i— CM
oo
CD
_a
ro
242
-------
NO,. ppm*km/mJn
C^OIX) uiui/ms|»uidd '03 'D
NO,, jpm'km/min CXIO"")
-I
NO,. ppm«km/mln-
1 I ' I ' I ' I ' I ' I ' I ' I ' I '
C..OIX) uiui/cuij
-Oc 'O
NO,. ppm»km/m(n CX1CT1)
I i I i I , I i I i I i I i I i I .
o en OD
CO
co
CVJ
fO
Q
CL)
il
O
(1)
to
O
<
CO
(O
cc:
c
o
in
10
LU
o
CJ
c
c:
fO
CM
cn
CD
243
-------
Day 201 - July 19. The column path for this day from the supplied
information is shown in Figure 93. On this day, the column path runs north-
south through the city from the south. In comparison with Figure 29, note
that the supplied column path encompasses the two major clusters of point
sources in the RAPS database, the St. Louis-East St. Louis region and the
Wood River-Alton region.
Again the dotted lines in Figure 94 represent the ambient concen-
tration interpolated along the centerline of the column path, and solid
lines represent the predicted concentrations by the Demerjian mechanism.
For the supplied column path, neither the Demerjian (shown), Dodge, or
Carbon Bond II mechanisms did well in predicting the ozone maximum. For
the three different mixing height profiles the results were very similar
for each mechanism. The observed ozone maximum on this day was 0.144 ppm.
Again different emission patterns or column paths will be examined to see
their effects on the ozone prediction.
For day 201, July 19, three alternative column paths were examined.
A. The first, Tl, uses the centerline trajectory of the supplied
column path and places a two kilometer box around it, one kilometer radius.
Table 35 and Figure 94 indicate that the predicted ozone concentration was
not any better than the predicted concentration using the supplied column
path. The predicted ozone precursor concentrations (Figure 94) for
the Tl column path, do not agree with the observed ambient concentrations
for this day. In determining the emission rates for this column path, two
point sources were encountered in the region, but their locations could not
be established from the information at hand. The Tl column path does not
consider either of the sources to be within the column path. The remaining
two column paths consider both of the point sources to be within the column
244
-------
4340
4330
4320
4310
4300
4290
4280
4270
4260
4250
4240
700
710 720
730
740 750 760 770
780
Figure 93. Column Path for Day 201, July 19.
245
-------
path. Both of these point sources are hydrocarbon sources only. With
their inclusion into the column path, the hydrocarbon fractions for the
DEM mechanism are shifted. The column path TAB is the Tl column path for
emission rates, with these new hydrocarbon fractions. Table 35 and Figure
94 indicate that shifting to these new fractions has a small effect on
the predicted concentrations.
The remaining two column paths are a variation of the Tl column path.
The emission rates represent levels which were increased or decreased
slightly, based upon the comparison of the predicted concentrations with
the observed ambient concentrations.
B. In the column path TA, the emission rates for the hour between 10
and 11 were increased five fold from the Tl emission rates. Table 35 and
Figure94 show that the ozone predicted is much closer to the observed con-
centration than either the Tl or the supplied column paths. The ozone
concentration is overpredicted with the linear mixing height profile.
Figure 94 shows that the prediction of the precursor concentrations
are also overpredicted, with the greatest overprediction for N09.
C. In the column path TB, the emission rate for the oxides of nitro-
gen was decreased by approximately 15% from the TA column path, Figure 95.
This decrease causes the model to predict an increased ozone maximum. The
ozone concentration predicted using the handpicked and characteristic mixing
height profiles increases to a level which is very close to the observed
maximum; with the linear mixing height profile, the predicted ozone con-
centration also increases by 10%.
The fact that the predicted ozone concentration rose by 10% with a 15%
decrease in oxides of nitrogen emissions indicates that the timing and/or the
mass of the emission rates is important to the amount of ozone predicted.
246
-------
--5
5 8
X
I , I I I I. L-l-1 I I . I I I I I
Outdd 3HWH
CO. ppmC
CO. ppmC
= 3
s -
CO. ppmC
S J
- s
8 -
I i U I i 1 i I i I i I i I
- S
O
CvJ
0)
i.
O
O
0)
(O
S-
cr
O)
S-
0>
2:
in
c.
O
4->
03
ZJ
O>
s-
CT>
'MIH
247
-------
01
c
o
o
en
5-
ZJ
248
-------
fe»
!
-------
T3
O)
c:
O
O
0)
i.
cr>
"">" 'MN 'ON
250
-------
Table 35
Effects of Trajectory on Ozone Maxima, Day 201
DAY 201
Observed ozone = 0.144 ppm
Trajectory S TI TAB TA TB
Mixing height profile
Linear 0.113 0.086 0.084 0.154 0.172
Handpicked 0.091 0.076 0.074 0.128 0.140
Characteristic 0.093 0.077 0.076 0.128 0.141
251
-------
o
CM
ro
0
ft
C
O
CL
3
l/>
to
Trajectc
4->
SUOLSS
L§
0
re
h-
<
h-
P
Q.
CM
O
O
O
O
CM
O
o
o
o
CM
o
o
o
o
0
o
o
o
UD
o
o
o
o
UD
o
o
o
0
<
to
o
o
o
c
to
CM
o
o
o
CTl
0
0
CTl
LO
O
CD
C C
cn
0
o
cn
co
CO
o
o
cn
o
cn
LO
0
Q
cn
CO
o
o
to
o
0
r-~
0
0
r~
o
o
0
o
o
to
o
o
CO
o
o
CO
o
0
CO
o
o
o
o
CM
o
o
0
CM
r—
o
o
o
CM
0
o
o
r-.
CM
o
o
o
CD
CU
UD O t—
CO i- O
4- CM
~
O C~ ecL"
Oi— CMi—
to
c:
o
•r—
4_I
O
ra
L.
u.
CM
CO
to
f—
•
0
OQ
c-f
CM
CO
to
r~
O
-)c
CO
CM
r"^
•—
o
o
to
to
<—
o
LU
i
0
r^
<^J-
r^
LO
•
O
f-*^
«^J-
r*^
LO
o
o
to
LO
o
o
cn
CM
LO
O
o;
di
D_
CM
O
CO
i—~
•
O
CM
O
CO
•—
o
o
CO
CM
i —
O
o
(•" —
<£>
1 — "
o
o
—
o
o:
z
to
(—
o
•
o
to
r-x
r—
O
o
CO
00
o
o
*^~
CTt
r^
o
0
Q
t—H
"^
c
•r—
O
r>
C
0
^j
ro
O
O
— 1
3
0
c
\/
c
—^
0)
-I-J
• 1 —
O)
f~
to
s~
fQ
O
o
— 1
c
^
0
c:
\s
C
^
-C
o
xi
to
^1 1
tL>
•f—
I/)
c
0
o
CO
*%
"^
to
ca
o
ra
CJ)
zc
-a
to
a>
D;
10
to
o
o
a.
^
oo
l£> CM l£> LO «^" CO CO
O i— to r^ r~- r^ ^j-
o o cri «d- oo i— o
o O o to o o o
o o o o o o o
UD CM UD IT) «d~ CO CO
o i— to r~~ r- r^~ -3-
o o cr> "3- co r— o
o o o to o o o
o o o o o o o
to CM to LO «3- co co
o i— to cr> r^ t~^ •=]-
o o cri CM co i— o
o o o r— o o o
o o o o o o o
CTI LO r^. 01 §— §— LO
o LO en LO i— to co
o o to ^- co CM o
o o o i— o o o
o o o o o o o
to
c
o
to
to
O
a.
3
to
i— CM LO O tO CTl -3-
O O O O CT) O O
O O r— LO CM O O
o o o o o o o
o o o o o o o
i— CM LO LO to cr> ^t-
o o o co en o o
O O i— LO CM O O
o o o o o o o
o o o o o o o
r— CM LO (--. UD CTI ^d"
O O O r— O1 O O
O O r— ,— CM O O
o o o o o o o
o o o o o o o
CM to "vt- r- CO i— co
o o r-~. c\j <^- UD o
O O O r— O O O
o o o o o o o
o' o o o o o o
O r— CM
CO COCTli— i— i— i— CM
£ I I I I I I I
•i- r-~cocr>Oi— CM.—
01 coaii
E
i — i — CM
i
i — CM. —
252
-------
NO, ppm-km/min CX10")
NO, ppm.km/mi
NO,, ppm-km/nun-
JL— ^
S^-J\
UI«J/»>(.«<1<1 03 'OH
NO,. ppm-km/IBIn CX10-)
NO, ppm»km/mln [XICT*)
i' i ' i' i' i' i' i ' i i-q- (.' i' i' i • i' i
QJ
O
0)
en
c.
O
-------
8.3.4 Effects of Emissions Pattern for Selected Days
The purpose of this section is to examine the effects of changing the
pattern of the emission rates while keeping the total emission mass constant.
On three days, Day 159-June 7, Day 261-September 17, and Day 275-October
1 starting at 8 o'clock, three different emission patterns were examined.
The emission rate patterns were determined using a random number table.
Each hour:s emission rate was assigned a value corresponding to the hour of
its occurrence, the first hour being one, the second, two, etc. (Table 37).
Then these numbers were reordered by going down the random number table until
all the numbers had appeared. Table 37 indicates the order of the emission
rates for the three different patterns, A, B, and C, and the total emission
mass in ppm-km for the three days.
Day 159, June 7 is the day with the lowest oxides of nitrogen emissions
while day 275 starting at 8 AM has the highest. Both of these days have
similar hydrocarbon emission rates. The results are shown in Table 38.
Table 39 shows the absolute and relative deviations as percent of each
emission pattern from the supplied pattern. Day 275 shows the largest devia-
tions while day 159 has the smallest deviations.
The results suggest that the order in which the column path intersects
the emission inventory is important.
8.3.5 Effects of Ozone Aloft Assumptions
The purpose of this section is to examine the effects of two different
ozone aloft profiles on the predicted ozone maximum. Three days from the St.
Louis RAPS study will be used to examine the effects, using the Demerjian
chemical mechanism.
The two profiles are a constant profile and a variable profile. In the
constant ozone aloft profile a constant concentration of ozone is entrained
into the mixed layer for a given mixing height rise. For the variable ozone
254
-------
Table 37
Random Emissions Patterns and Total Mass of Emissions
Pattern Hour Order of Emissions
Supplied 1 23456789
A 345863719
B 536479128
C 364985217
CO NOV HC
X
Day 159, June 7 0.5248 0.0346 0.2347
Day 261, September 17 0.7535 0.0589 0.4344
Day 275, October 1 at 8 AM 0.8364 0.0859 0.2884
255
-------
Table 38
Predicted Ozone Maximum Considering Three Different
Emission Patterns
Emission Pattern
Day 159, June 7 S A B C
Mixing Height Profile ^— ———
Linear 0.252 0.248 0.243 0.239
Handpicked 0.164 0.157 0.157 0.152
Characteristic 0.197 0.104 0.189 0.184
Day 261, September 17
Mixing Height Profile
Linear 0.103 0.118 0.113 0.119
Handpicked 0.123 0.136 0.137 0.137
Characteristic 0.104 0.118 0.114 0.115
Day 275, October 1 at 8 am
Mixing Height Profile
Linear 0.261 0.233 0.228 0.238
Handpicked 0.096 0.070 0.067 0.070
Characteristic 0.234 0.186 0.179 0.189
256
-------
Table 39
Deviations of Ozone Maximum Prediction from the
Supplied Emission Pattern
Emission Pattern
Day 159, June 7
Mixing Height Profile
Linear
Handpicked
Characteristic
A
-0.004*
1.59
-0.0007
4.27
-0.093
47.21
B
-0.009
3.57
-0.0007
4.27
-0.0008
4.06
-0.0013
5.16
-0.012
7.32
-0.013
6.60
Day 261, September 17
Mixing Height Profile
Linear
Handpicked
Characteristic
0.015
14.56
0.013
10.57
0.014
13.46
0.010
9.71
0.014
11.38
0.010
9.62
0.016
15.53
0.014
11.38
0.011
10.58
Day 275, October 1 at 8 AM
Mixing Height Profile
Linear
Handpicked
Characteristic
-0.028
10.73
-0.026
27.08
-0.048
20.51
-0.033
12.64
-0.029
30.21
-0.055
23.50
-0.023
8.81
-0.026
27.08
-0.045
19.29
*The top number represents the absolute deviations, calculated
as difference in ozone maximum between the variation pattern
and the supplied pattern. The lower number represents the
relative deviation, represented as percent; computed as
absolute deviation divided by the predicted ozone maximum
from the supplied emission pattern.
257
-------
aloft profile, the concentration of ozone entrained into the mixed layer
depends on the fraction of the total rise of the mixing height for that day.
The fractional multiplier is prescribed by the equation:
FAL - 2FR3 - 3FR2 + 1
where FAL is the fraction of the aloft concentration
FR is the fractional rise in mixing height, given by
the current mixing height minus the initial mixing
height divided by the total rise in mixing height
for that day.
As one might expect, the maximum ozone concentration decreases as the
concentration of ozone entrained into the mixed layer decreases,
i. e. a constant aloft profile provides the greatest entrained concentration,
and the variable aloft profile provides a lesser concentration, Table 40.
The constant aloft profile is considered in Table 41. The first column
represents the effective concentration of the entrained ozone. This was
determined by subtracting the maximum ozone concentration with ozone aloft
from the maximum ozone concentration using the constant ozone aloft profile.
The second column represents the concentration of the ozone entrained
considering only dilution, i. e. no chemistry. The third column can be
interpreted as the effect of chemistry on the entrained ozone concentration,
expressed as a percentage. A value of zero indicates the entrained ozone
concentration is not affected by chemistry. For the three days examined,
no pattern of the affect of chemistry with respect to the different mixing
height profiles can be determined. For day 261, the ozone maximum is
enhanced by the chemistry and the entrainment ozone, with the characteristic
mixing height profile.
258
-------
Table 40
Variation of Ozone Aloft on the Prediction
of Ozone Maximum
Day 159, June
Mixing Height Profile
Linear
Handpicked
Characteristic
Aloft Profile
Constant Variable None
0.252 0.232 0.215
0.164 0.117 0.072
0.197 0.159 0.123
Day 261,September 17
Mixing Height Profile
Linear
Handpicked
Characteristic
0.103
0.123
0.104
0.081
0.096
0.072
0.067
0.070
0.058
Day 275, October 1 at 8 am
Mixing Height Profile
Linear 0.261
Handpicked 0.095
Characteristic 0.233
0.241
0.070
0.209
0.218
0.053
0.193
-------
The results from using a variable aloft profile are presented in
Table 42. The three columns contain the same information as in Table 41,
considering a variable ozone aloft profile. Again no pattern can be deter-
mined with respect to the mixing height profile.
In comparing the days for the two different aloft profiles, specifically
days 159 and 275, a trend of the effects of chemistry are similar, with the
magnitudes being close in most cases. For day 261, the enhancement of the
ozone maximum with constant ozone aloft profile was not reproduced with the
variable aloft profile.
In conclusion, the predicted ozone maximum decreases as less material
is entrained from aloft, while the effects of chemistry on this entrained
ozone vary widely between days and mixing height profiles.
260
-------
Oi. ppm
man TO 'ON
IT)
re
a
c
o
_E
to
o
(O
CD
C
o
O
ifl
O
O)
CT>
S-
261
-------
_->•. t
CT1
(O
Q
o
•r—
+J
fC
ul<1<1 'GN UN
hl-1 I I I I I J L.I 1 1 J.U ! ..I ! U
it-
CD
to
-------
Table 41
Effects of Chemistry on a Constant Ozone
Aloft Profile
Effective^"'
Entrained
Ozone Cone
Entrainedv ' Percentage of
Ozone Cone the Entrained
(dilution only) Ozone Survival
Day/Mixing Height Profile
Day 159,
Day 261,
Day 275,
June 7
Linear
Handpicked
Characteristic*
September 17
Linear
Handpicked*
Characteristic
October 1 at 8 AM
Linear*
Handpicked
Characteristic*
0.037
0.092
0.074
0.036
0.053
0.058
0.043
0.042
0.040
0.070
0.117
0.108
0.048
0.057
0.056
0.045
0.057
0.048
52.9
78.6
71.8
75.0
93.0
103.6
95.6
73.7
83.3
(C)
(A) Computed as the difference between the predicted ozone maximum with
no ozone aloft from the predicted ozone maximum using the constant
ozone aloft profile.
(B) Considers only the effect of dilution on the entrained ozone.
(C) Computed as A/B *100.
*
Good prediction of ozone maximum (See PKSTM Section).
Z63
-------
Table 42
Effects of Chemistry on a Variable
Ozone Aloft Profile
Effective'"'
Entrained
Ozone Cone
Entrained '
Ozone
Cone
Percentage of '
the Entrained
Ozone Survival
Day/Mixing Height Profile
Day 159,
Day 261,
Day 275,
June 7
Linear
Handpicked
Characteristic*
September 17
Linear
Handpicked*
Characteristic
October 1 at 8 AM
Linear*
Handpicked
Characteristic*
0.017
0.045
0.036
0.014
0.026
0.014
0.023
0.017
0.016
0.035
0.058
0.051
0.024
0.029
0.028
0.023
0.028
0.024
48.6
77.6
70.6
58.3
89.7
50.0
100.0
60.7
66.7
(A) Computed as the difference between the predicted ozone maximum with
no ozone aloft from the predicted ozone maximum using the variable
ozone aloft profile.
(B) Considers only the effect of dilution on the entrained ozone.
(C) Computed as A/B *100.
Good prediction of ozone maximum (See PKSTM Section).
264
-------
8.4 SUMMARY
In analyzing the RAPS data and in generating input data and reference
data for photochemical simple trajectory models, significant uncertainties
were encountered. Although used in this study, the simple inverse distance
weighing of hourly wind data at the closest RAPS stations was judged not
adequate for defining the origin and history of air arriving at a particular
station. In some cases this procedure resulted in variations of origin
that encompassed the whole city of St. Louis; thus divergence and convergence
problems were significant. A mass conservative wind field model should be
developed for the RAPS database and trajectories determined from this model.
The three closest RAPS stations to the assumed trajectory location at
each hour often showed large differences in measured concentrations resulting
in uncertainties in starting concentrations and in values along the trajec-
tories.
The uncertainties in the trajectories and the conceptual basis of the
simple trajectory model lead to an averaging technique for the emissions
inventory that resulted in a large uncertainty for the contributions of
point sources. Hence a significant amount of "tuning" could be necessary
for any particular day. A sub-model in the above proposed wind field model
could provide better treatment of the impact of significant point sources.
The calculated composition of hydrocarbons along the trajectories had
the following ranges (in percent carbon): 6.7 to 14.0, nonreactive; 4.8 -
75.5, paraffin; 10.1 to 25.1, olefin; 6.6 to 18.1, aromatic; 1.08 to 2.86,
aldehydes. There are no smog chamber databases in which the hydrocarbon
composition has been systematically varied in a complex mixture to test or
develop chemical mechanisms. The Dodge and Demerjian mechanisms were both
developed on the BOM auto exhaust composition. The CB? and CIT mechanism
were developed on synthetic mixture smog chamber experiments.
265
-------
For the Demerjian and Carbon Bond II mechanisms, in static diurnal
light simulations, changing the hydrocarbon composition over the ranges
given above without changing the hydrocarbon concentration resulted in 03
decreases of more than a factor of 5 (80% reduction). A few UNC outdoor
smog chamber experiments indicate that the models greatly overrespond to
compositional changes in complex mixtures.
Significant uncertainties existed in calculating the mixing height
profiles. Typically, soundings were taken at 0500, 1100, and 1700 LOT. In
this study, three techniques for converting these soundings to temporal
variation in mixing height, and thus dilution rates, were compared. The
first method was a linear interpolation between the soundings. The second
was to have a meteorologist examine each day and generate a "handpicked"
mixing height profile. The third method was the application of Demerjian
and Schere's Characteristic Curve developed for their photochemical box
model. Each of these profiles were used for each mechanism for each day.
A meteorology-only simple trajectory model (MOSTM), that is, a model
incorporating initial conditions, emissions, mixing height rise, and entrain-
ment from aloft but excluding any chemistry, was used to examine the rea-
sonableness of the non-chemistry assumptions, especially mixing heights
and emissions patterns.
MOSTM simulation results were that: a) the "handpicked" mixing height
profile generally gave poor results for most days in that it was usually
too low, too long, resulting in too high early concentrations and then rose
too fast, resulting in too much dilution and low final concentrations;
b) the probable parcel pathways were poorly represented by the "trajectory
boxes" used and therefore the emission input figures are inadequate for
June 8, Site 15, July 13, July 19, and the last 2 hours of August 25. A mass
266
-------
balance for each species in MOSTM simulations showed that: a) 4-47% of CO
mass, 2-25% of NOV mass, and 1-15% of HC mass was the result of material
A
present initially, b) 44-86% of CO mass, 83-98% of NOV mass, and 85-99% of
X
HC mass was due to emissions, and c) 9-24% of the CO mass was entrained
from aloft. This suggests that the chemistry should be emission dominated.
A photochemical simple trajectory model (PKSTM), equivalent to the
OZIPP "calculate" mode but having more flexibility, was used to simulate the
ten RAPS days. There were 9 simulations for each day: three mechanisms at
each of three mixing height profiles. The mechanisms were Demerjian, Dodge,
and CB2. Only June 7 and October 1
were simulated with the CIT mechanism because of its previous
poor performance on the BOM simulations.
PKSTM Results. Different chemistries dp_ give large differences in
results for the same conditions of emissions and meteorology. The CB2
mechanism generally predicted the lowest 03 values, the DODGE mechanism
generally predicted a middle value, and the DEM : mechanism generally
predicted the highest value.
For four days, 03 was underpredicted by more than 25% by all mechanisms
for all mixing height profiles. On a fifth day such underprediction also
occurred for all mechanisms for the handpicked and characteristic mixing
height profiles. These underpredictions are associated with very poor
predictions of observed NO and HC concentrations. The MOSTM simulations
/\
had already suggested that these days had problems with the trajectories
and possibly the large averaging process applied to the emissions.
For three days, including the days with the highest and lowest observed
0^, Oq was overpredicted from 2 to 23% by the DEM mechanism for the linear
267
-------
mixing height profile. These same three days were predicted within +5% for
the DEM mechanism and the characteristic mixing height profile. The other
two days were underpredicted by 10-15% by the DEM mechanism for the linear
and characteristic mixing height profile.
On different days different combinations of mechaniams and mixing
height profiles gave the "best fits" using the standard assumptions.
Slight modification in the standard assumptions can dramatically improve the
fits. Below is a summary of "best fits" based on examination of CO, NMHC,
NO and Oo profiles:
A 3
Using Standard Assumptions
Mechanism Mixing Height Comment
Day
159 DEMER CHAR
160/03 DEMER CHAR
160/15
195
201 DEMER,DODGE : IN
221 DEMER CHAR
238
261 DODGE,DEM HP
275/6am DODGE CHAR
275/8am DEMER LIN
276/6am
CB2 LIN also good;
LIN better in 03, but N02 bad; CO bad
None good
None good
03 18% low and N02 bad; HC bad at end
N02 probably within measurement error
(lowest day); missed HC peak
None good
DEM low on 0^, but better shape
Oo 25% low but good shape until last hr.;
CO good
Oq 19%, low but good shape until last hr.;
CO bad
None good
268
-------
Day Mechanism
275/8 DEMER
275/8 CARB BOND
238
201
Using Modified Assumptions
Mixing Height Comments
CHAR
CHAR
LIN
HP
5% initial aid gives
excellent fit.
BOM HC composition
including aldehyde gives
excellent fit
Modification of
trajectory/emissions still
25% low only DEM tested
Modification of
trajectory/emissions only
DEM used
269
-------
9.0 ISOPLETHS AMD CONTROL CALCULATIONS
9.1 OZONE ISOPLETHS GENERATED FROM PHOTOCHEMICAL SIMPLE. TRAJECTORY
MODEL SIMULATIONS
The purpose of generating an 03 isopleth diagram is to illustrate how
the simple trajectory model's maximum predicted 0~ varies as a function of
increases and decreases in HC and NO initial conditions and emission values.
A
To accomplish this, the trajectory models in the previous chapter were run
repeatedly; in each run, the HC and NO initial concentrations and the
X
emissions of the standard run were multipled by scale factors. Figure 98
shows an example of the locations of the run points as a function of initial
condition of HC and NO . These initial conditions also represent the emis-
A
sions since the same factor was used for each. That is, ".hen the initial
conditions were halved, so were the emission values.
When all of the simulations had been run, a bicubic spline surface was
fitted to the maximum 0^ and initial conditions. The locations of points of
constant Oo were found by interpolation over this surface. These isopleths
are shown as a "level set" by viewing the surface from above ((A) in Figure
99). To aid in comparison, the surface was also "sliced" and its cross-
section shown for various conditions (the dotted lines in (A) of Figure 99 ).
Figures will be presented for 03 as a function of HC 1.0, 0.8, 0.6, 0.4, and
0.2 of maximum MO and as a function of NO at the same values of maximum
X X
HC ((B) and (C) of Figure 99). The location of the ambient data point,
that is, the initial ambient HC, initial ambient NO , and maximum ambient 0^,
A ^
is shown on the diagrams as a "+" ((D) and arrows of Figure 99).
Ten isopleth surfaces were done for each of two days: the Demerjian,
270
-------
OOE'O
' 0.300
c:
03
Ol
U
0.200
0.100
NOX.PPM
-O
0)
to
c
O
(O
r-~
=3
•r~
OO
i-
M- Cl)
O
(/) to
C 4-
o s-
O .C
O •(->
_l O)
r—
4- CL
O O
l/>
.O) i— i
Q. O)
E C
fO O
X N
uj o
CO
CD
OJ
271
-------
E
3
E
o co
•r- O
.«4_)
^^ CO T3
CO .— E
--- 3 O) 4->
C "O -i-
(D O "O
+-> ET'c
^C_>
O i-
O O i—
-t-> re
-M O T-
fO OJ -4->
•1-3-1—
T3 ro E
O) S- i— t
i— O) O
OO r—
D- l/l
(U E CD
U T- +•>
na oo O)
O) E -r-
O ro <4-
(T3 •»-> -r-
<4- tO -M
S- E E
Z3 O O)
CO C_) -O
OJ
Q. O)
O O •"
E d) >•
O O ra
NJ n3 Q
E 3 O
eC CO M-
CTt
OJ
272
-------
Dodge, and CB2 mechanism at each of three mixing height profiles and the CIT
mechanism at the characteristic curve mixing height profile, for October 1,
1976, Day 275 at 8 am and for June 7, 1976, Day 159. October 1 was the high
0- day for the year (0.244 ppm 0~). The initial HC-to-NOv ratio was 9.4 and
O O X
the emissions HC-to-NO ratio (determined for MOSTM simulation) was 3.35.
A
Of the ten days studied, June 7, 1976 had the highest initial HC-to-NO ratio,
X
13.2,and the highest 0., aloft (0.12 ppm). It was also the only day simulated
well by all mechanisms. The isopleth levels used were: 0.08, 0.12, 0.20,
0.24, 0.28, 0.30, 0.32, 0.34 and 0.40.
The results for October 1, Day 275, are shown in Figure 100 A,B,C,D, E.
Recall that this day wasunderpredicted by DEM2758.LIN simulation, was under-
predicted slightly by DEM2758.CHAR simulation, was underestimated about 10%
by both the DOD2758.LIN and .CHAR simulations, and was underpredicted greatly
by all other simulations (all .HP, all CB2, and CIT.CHAR).
In Figure 100B, there are obvious large difference in the isopleths for
different mechanisms-mixing height combinations, even where there were less
obvious differences among the PKSTM simulations. The Demerjian, Carbon Bond,
and CIT mechanisms have "steeper" sides than the Dodge mechanism. That is,
the Oo formation is more responsive to HC and NO in these mechanisms than in
*5 X
the Dodge mechanism. The response to NO is so great in the CB2 and CIT
A
mechanisms that they cannot simulate this day. At low NO , they exhibit a
X
higher 0., dependence on HC than Dodge and therefore have more, and more
closely spaced, isopleths than Dodge.
The "flatness" in response to HC and NO gives the Dodge isopleths a
X
characteristics "U" shape. The other mechanisms have more "V" shaped iso-
pleths. These shapes have important implications for EKMA control calcula-
tions which will be discussed in the next section. The U-shaped isopleths
273
-------
o 600 o son e 100 a 300
Di.rrr
a 200 4 voo o 0 toe t
I ( 111 I «0. D IOC I HE • Bit (
in
c
o
03
cu
-0
O
•(->
O
O
I 0 ooE'o oo» o oat a ooe 0 oot o 0
ooe 0 oot I oot 0 ooc 0 OOE 0 ool 0 0 " 001 0 las o 00* 0 Oot 0 00: 0 OOt 0
0 too 0 MO 0 *00 0 >00 D 100 0 tod
o
CD
CO
CO
o
i.
CJ
CJ
a>
c:
o
N
o
o
o
ai
i_
3
O)
D tOO 0 500 0 II
0 300 0 200
09 ffH
274
-------
got o coz
in
C
O
ro
I.
Ol
-D
O
275
-------
got 0 D
if)
C
o
+j
IB
s_
Ol
o
o
co
c
o
-
CJ
CO
to
O
X
o
O)
£=
o
tsl
o
CJ
o
CD
CD
3
CD
276
-------
l/l
c
o
CO
a>
i.
O
o
CD
-t->
O
n3
rC
flat o is: I III I lit I oil o lita a
101 I IDS I lot 0
to: o lot I
S-
a>
o
o
•
o
a>
1/1
O
i,
o
O)
Q.
O
cu
c
o
M
O
CD
O
O
a>
s_
3
en
277
-------
o
•I—
(O
i.
o>
.0
o
4J
o
o
s-
0)
o
T3
O
DO
S-
10
CJ
s.
o
Ol
CD
c
rt)
s_
s_
Ol
Dl
S-
i_
0)
o
to
.C
4->
Ol
CL
O
•r-
Ol
O
IM
O
LU
o
o
O)
S-
cn
278
-------
generally "spread" as the HC and NO inputs are increased, that is, the iso-
X
pleth location moves to lower and higher HC-to-NO ratios with increasing
A
HC and NO values. The V-shaped isopleths, however, tend to parallel a par-
X
ticular HC-to-NO ratio as the HC and NO inputs are increased, that is a
X X
particular 0, value may never occur at a particular HC-to-NO ratio no matter
%5 X
how much HC and NO input is used. Figure 100E illustrates CB2 isopleths
A
over a larger range of HC and NO than those of Figure 100B. It is evident
A
that the fifth isopleth (0.24 ppm Oo) will never intercept a line drawn
through the origin and the "+" which marks the HC and NO for October 1.
X
Therefore, mechanisms which exhibit V-shaped isopleths must predict the
observed 03 maximum for present emission conditions (the PKSTM simulation)
within reasonably narrow limits (maybe +10 or 15%) if a standard EKMA pro-
cedure is going to be used.
The results for June 7, 1976, Day 159, are shown in Figure 101A,B,C,D.
This day was overpredicted by DEM159.LIN simulation, predicted within 5% by
DEM159.CHAR, CB2159.LIN, and CIT159.LIN simulations, and was 12% underpre-
dicted by DEM159.HP, was 25% underpredicted by the DOD159.LIN, CB2159.CHAR,
and CIT159.CHAR simulations. The rest were more than 25% underpredicted.
The first observation to be made about the June 7th isopleths is that
the Demerjian handpicked isopleth is missing. There was apparently a closed
isopleth for this condition and the computer code to find isopleths on the
On response surface was not designed to deal with closed isopleths, therefore,
no surface was depicted.
The second observation is that the 03 value at the origin was nearly
0.10 ppm Oo for most surfaces. This is because of the very large value of
Oo aloft on this day. Since the upper left and lower right simulations
(zero HC, 0.07 NO and 1.0 HC and zero NO ) interacts with the entrained 0,
A A O
279
-------
DOC 0 OPE t DP! 0
g 400 0 30C
t 400 o loo
oot o oorc BOC o
0 200
M.rrn
00i D OOC 0 001 0
0 100 D 300
100 0 SOU 0 300 0 10(
0) rrn
C
O
(O
Z3
E
CL)
c
13
tut o ooc 0 ooz 0 oat c
o
QJ
to
CO
I/)
O
s_
o
C_3
n:
i
cu
c
o
N
O
O)
5-
Z3
CD
280
-------
•ow o eoso a oe*o o ooco o oozo o so
oow o one g »to 0 «io-o o«o o 0010 o o •
0 MOO 0 MOO 0 0100 0 WOO 0 0200 I 0100 9
not rrN
CO
C
O
(O
I
•I—
(/)
r-»
0 0600 0 OSOI
281
-------
•ot • tie «
tOt I OBI « 001 I
eot o oat a eo; • out e
.
is
t g BBC «
oot o DO; a oai o
tot i oct o eoi a oot <
a 209 0 100
V I
(O
r—
3
0)
c
^
o
l/l
o
• 0 »BO 0 301
o
ai
I/)
CO
co
o
o
I
CD
c
o
N
O
O
i—
O
a>
s-
282
-------
o
•r—
4->
CO
•r—
i~
OJ
-l->
o
(O
S_
(O
.c:
o
r^.
O)
c
o
o •
M- C
o
QJ (T3
'o-'o
o E
CO •!-
•I— CO
0)
o s_
N 3
O
-------
differently for each mechanism, the CL values in the corners are different.
This same isopleth surfaces have a 0.08 ppm 0, isopleth that intercepts the
edge and some have no 0.08 ppm 03 isopleth at all. Because of a high 03
photolysis reaction (R23 in Table 9 ) in the CIT mechanism, it consumes
03 near the origin (Figure 101D), resulting in a 0.08 ppm 0~ isopleth that
only touches the HC axis.
The Dodge surfaces exhibit almost no response to HC and only a low
response to NO . That is, the 0^ vs HC at constant NO cross-section plots
X -J A
(Figure 101A),the lines are nearly horizontal but do show increasing Oo
(only slightly for HP and CHAR), increasing constant NO levels. As in the
X
isopleths for Day 275, the Demerjian and CB2 surfaces show a much stronger
dependence upon both HC and NO than the Dodge surfaces.
A
On Day 275, the LIN surfaces are very similar in appearance to the CHAR
surfaces. On Day 159, the LIN surfaces and the CHAR surfaces differ substan-
tially in appearance. These similarities and dissimilarities will be
examined further in the EKMA control calculation discussion.
Although somewhat affected by the high 0,, aloft values, the U-shaped
versus the V-shaped character of the Dodge versus the Demerjian and CB2 iso-
pleths is still evident. It is expected, therefore, that, if the upper right
corner values of DOD159.HP and DOD159.CHAR were made high enough, the corner
03 value would exceed 0.20 ppm 03, the value needed for an EKMA type calcu-
lation for Day 159. These high concentration simulations were not done.
284
-------
9.2 OZONE CONTROL CALCULATIONS FROM PKSTM ISOPLETHS AND EKMA PROCEDURE
The isopleths in the previous section show the effects of emission
changes on simple trajectory model's 0,, predictions. The Emperical Kinetics
Modelling Approach (EKMA) is a method for translating the models' response
to emission changes into atmospheric control requirements.
Stated in its simplest form, the EKMA procedure is (Figure 102):
a) to find the point of interception between a radial line at the initial
HC-to-NO ratio and the isopleth for the ambient 0? maximum for this day;
X v
b) divide the values on the NO axis by the NO values at the point of inter-
X X
cept; c) draw a line at constant NO through the intersection point, draw a
X
line on which NO increased 20% for 50% HC decrease, and draw a line on which
/\
NO decreased 20% for 50% HC decrease; d) the values of 0,, along these lines
/\ O
as a function of HC fraction are assumed to represent the 0- values in the
atmosphere as a function of ambient HC emission reductions.
Note that in the example in Figure 102, the model was not perfect. That
is, the DOD2758.LIN PKSTM simulation, with our best estimates of initial
concentrations (0.202 ppm NO and 1.89 ppmC HC) and HC NO emissions, only
X X
produced 0.214 ppm 03 instead of the observed 0.244 ppm Oo (observed-to-pre-
dicted ratio 1.14). The EKMA solution point occurred at 2.38 ppmC HC and
0.254 ppm NO , a ratio of calculated-to-observed initial conditions and
X
emissions of 1.26. In other words, the model's emissions had to be increased
by 26% to get 14% more Oo- These increased emissions become the starting
point for the control calculations. Following down the 0,-HC fraction line
of Figure 102D shows that 0.12 ppm 0~ is reached at a HC fraction of 0.62,
that is, to achieve the Oq standard with no change in NO , a 38% reduction
O A
in HC is required by this model.
The EKMA solution technique was incorporated into the isopleth generation
285
-------
NOX
Solution
Actual 8 AM HC and NOV Values
X
The EKMA Solution
Point
The Isopleth for
the Observed [Ojmax
(A)
(B) and (C)
(D)
Standard
.20% NOV increase
A
—No NO change
20% NO decrease
A
Figure 102. The EKMA Procedure
286
-------
Q
2:
0
'JO
O
CD
•Si
in
o
LO
5
UJ
CM
X
0
1 1
LO CO
1 1
:r
\—
GO
CL.
• —
*"
O i —
U3 «*
ro r~-»
CO
o
C5O CO
^jO ro
0 0
0 0
O
<0
*^
E
CO
Ol
_
,-_
cu
-d UJ
4-> 10 kf Q
U m ro OJ O
O QJ O • O
t- i —
xp
in •»-> o OJ
+-> c z: •
C QJ lO
O -0
a. E s- o
i — ro
CM i—
0 0
+-*
^_
o
CO
UJ
ro
^T*
cu z:
_Q >— .
ro *~D
h— a:
UJ
E:
UJ
_Q
•z.
—I
o
CO
rf*
*s<
UJ
0
C\J
ro
t-H
X)
X
en i
CO 00
OJ Z
1
o
o
31
ro
o
i —
OJ
3£
LO
i«i
CL
u
,_- .
en
ro «at>
OJ i— 1
^ CO
o
r^H ^~-
<-> o
o o
CU
C JZ T-
•r- Q1H-
X T- O
— (D i-
s; :c o-
S-
(T3
OJ
C
•r- 0.
— i 31
i
CO
1
in
Cn
IT)
,
•y
o
d
o
en
og
p™
lO
OJ
Q
cn
CO
•3-
OJ
OJ
,
0
OJ
o
o
OJ
•a-
OJ
"I
d
^o
Ol
ro
OJ
in
«^
r-H
00
lO
o
s-
o
l/l
p_
1/1
OJ
CO
E T3
O- O
0- E
CU >i O
'O O O
*J C
w o a.
•4-> OJ CL
C t)
+-> 'cu
0 OJ 0
"O CL
c E in
X -l-J
O i — OJ
U 'r-
r~~ E
IO O) •—
f. U U
OL O OJ
0- JC Q.
Q- E
OJ UJ
i- OJ
fa .£: oj
•*-> c:
to -*->
•*-> to
•i- c tn
C 0) C
Z3 OJ 03
E OJ
0 £
3; s:
i — «f
I— CO S
5 CX UJ
ro -Q
in
QJ
r—
-M
C
CU
JQ
E
CO
-o
OJ m
+-> c
o o
•a •*-»
s- "a
o. c
t—
CO !—
D. •-—
5 I
o -a
cu
OJ •>->
3 TO
> U
C U
QJ
•S §
E 2
(O UJ
M- *4-~
0 O
o o
4-J 4 >
n3 ry
!_ t.
CU •*->
in i/)
i— OJ
s- s~
o
-a
c
13
O
c
c
o
•*-*
,H
o
t/1
c
QJ
E
I
u_
i
• «•
QJ
JO
In
0
u.
c
0
•f-*
o
o
c
to
c
aj
OJ
e
i
i
-o
287
-------
o
4->
-------
ggf a oo* a oot a eoc o DDE o ooi o
3 5:0 9 'GO
009 S ODE 0 OOi 0 OOt 0 001 0 DD1
D too e 500 e HO a 100 0200 o i
to
c
o
E
c/1
0)
.a
o
-i->
o
o
to
4->
O
"a.
c;
o
o
13
0)
c
o
N
O
CO
o
O)
i.
289
-------
* I 601 I 001 0 001 0
tot o ooc a oo; o oo
13
E
s_
o
1/5
O
'o.
c
o
+J
o
Ol
c
o
N
O
•=}•
O
O)
s_
3
en
290
-------
program, so that the EKMA solution point could be found and Oo-HC fraction
plots automatically produced. The procedure was applied to the ten isopleths
for October 1 (Day 275) and the nine isopleths for June 7 (Day 159) and the
results are shown in Table 43 and 44 and Figures 103and 104. As had been
indicated in the isopleth discussion of the previous chapter, not all iso-
pleths produced solutions, and some could no^, even if 100 times the initial
conditions were used.
The data in Figures 103and 104 have been converted into control diagrams,
that is, plots of percentage 0~ reduction versus percentage HC reduction, and
these are shown in Figures 105to 106- The HC reduction necessary to achieve
the 0.12 ppm 03 value are summarized in Table 45 and in Figure 107,
From Figure 107 it is evident that the most important factor in the
amount of HC reduction needed was the choice of day. Day 159 (June 7) was
not the day of highest 03; it was the day with maximum 03 aloft (0.12 ppm).
This Oo aloft is not reduced by emissions reductions in the same manner that
Oo produced from emissions is reduced by emissions reductions. This resulted
in "flatter" ozone control diagrams (Figure 106).
The next most important factor in the amount of HC reduction needed was
the chemical mechanism used. When it could give a solution, the Carbon Bond
mechanism gave the least HC reduction needed (other conditions the same).
The Demerjian mechanism, when it could give a solution, always required less
HC reduction than Dodge (other conditions the same).
The third most important factor was the choice of mixing height profiles.
The characteristic curve profile, in general, required less HC reduction than
the linear profile. Except for the Dodge mechanism, solutions were not
possible with the handpicked profile. For the Dodge mechanism, the hand-
291
-------
>}
_Q
T3
CU
CJ
•r—
"O
CU
S-
D-
°C?i
Q.1
Q- «
r^.
CM
r— CU
• c:
O 3
cu
> T3
CU E
•t- ro
-c
O UD
*-C I*"**
Ch
0 i—
-l_)
*
£ZZ r—
O
•r™ S*.
-^ CU
O _Q
3 O
•0 4J
CU U
a; o
O i-
31 O
^
4-3
C~ (/)
CU r-
O CU
i- -0
0) O
Q- s:
•
LO
^f-
cu
1 —
t~}
CO
1—
-o
c
0
CO
c
o
.a
^_
re
o
X
0
z
0
o
CM
LO CO
OO CM
X
o
^"
^i
o
III 1
CO CO CO i — CO LO
z: -z. -ZL *& -z. oo
iii i
X
o
z
&5
o
CM
1
O LO
LO *J-
O)
ID
O
Q
X
o
*sj~ CM ^i" r —
C
•r-
*' — D
s-
cu
e
Q
X
o
"^
0^
0
CM
-f-
r~~ CNJ o o
1— 1 i — LO LO
X
o
z:
cH5
O
1 1
CM co LO CD co r^
CM 2: i — LO sr LO
i i
X
o
TZL
^s>
0
CM
1
O CM LO (^
OO CM l£> CO
0)
C7)4-> r—
•r- O>M-
X -r- 0
•i— CU S-
21 nr ex.
r—
S_ S-
s- to res
CU CU n- r~-. CU rv
o •>- D- x: cu -r- a. re
-u—jnrcj c_i3co
U 3
O 0
0)
CD
rO
C
O
•r—
4->
S_
I ^
C
cu
CJ
o
0) C
3 """
•r- -O
to c
LO !3
0 0
Q. H-
c a
0 C
**~ »i —
4-> -t->
3 13
i — i — -
O O
CO CO
0 O
c c:
c/i to
c c
(O res
cu cu
e e
i i
CO Lu
Z 2:
1 I
n3
292
-------
NOI10n03M "0 I
o
to
S-
o
-o
CD
O)
O)
o
•i —
•t->
o
3
-o
CD
a>
sr
•a
O)
fO
Q
O
O
cn
03
S 3
293
-------
CO
o
a.
a.
o
ITS
a>
s-
OJ
-o
a>
O)
c
E
o
o
-a
CD
OJ
c
O)
^:
CO
-------
D CD CD C
3 cn oo r
-1
_' 1 ' 1 '
—
—
4—G 1
_
~, 1 , 1 :
• * •
3 O CD C
D cn co r
-»
:> CD CD C
a? in =t
. , . | .
^_
1 fi i
h
i n I
^.
4.
.1,1.
* • •
3 CD CD C
\T> LO =1
3 C
a
i
Q
O-l-
1
3 C
r c
3 0 CD C
:> c\j -r-4
1 1 ' 1 '_
—
—
—
J. _„
». Q -I-
«•! • 1
O4-
.1,1,"
• • «
D CD CD C
0 C\J *-•
3
n
O
C_5
a
o
i~}
*^j
D
Q_
~T~
cC
1C
2T
i—)
_1
Q.
_L
cn
II C
CO ••-
X
CTi OJ E
i — -a o
O -i-
ro II ai
Q Q -r-
s_
^ QJ
r-- c: 4->
ro <_>
a> -1—03
c -i— > s-
zs s- ro
•-3 CD -C
E= 0
CD II
Q C£
II <
•-5 in
E -
01 -a
•r~ O!
C -il=C
(a
LD -a -
rv s_
CM >, 03
jQ , C
Q C r-
O II
i — -4-> i— (
O 1
S- •<-
QJ -a
X3 CJ
O S- l/>
4-> Q. £ •
U (/)(/>
o • — -r- at
S_ ra -r-
4-> -C 4—
O CU t-
cj 2: cx
r-x
0
-------
picked profile gave lower control requirements than the other profile. These
solution points, however, were often at very large HC and NO concentrations.
/\
296
-------
10.0 RAPS SIMULATION WITH LEVEL III MODEL
A Level III analysis has much less stringent data requirements than a
Level II analysis because of simplifying assumptions inherent in the Level
III model. The Level III model assumes that the air parcel containing the
measured 0^ maximum originated in mid-city at 8:00 am, traveled in a
straight path at a uniform speed such that the parcel would arrive at the
monitoring station at the time that the ozone maximum was measured.
Therefore a back trajectory need not be calculated and an
extensive wind direction-speed monitoring network is not required. Wind
direction data is only required to insure positive identification of the
correct downwind monitoring station. A gridded emission inventory is not
required. Only city or county wide average emissions are utilized, since
the actual trajectory is not determined.
The air quality data requirements are also much less for Level III.
Only three 0^ monitoring sites and one or two THC/CH. and NO monitoring
sites are recommended, as compared to the 7-11 03 and 4-6 THC/CH. and NOX
sites recommended for Level II modeling.
Control requirements are estimated using procedures described in
EPA 1981. These are estimated using the ozone design value
and prevailing 6-9 am NMHC/NOV ratio to identify a starting point on the
A
isopleth diagram.
10.1 LEVEL III INPUTS AND CONDITIONS
Three of the ten days previously simulated with the Level II approach
were picked to be simulated with the Level III approach: D159, D195, and
297
-------
D275. Only the Demerjian and Carbon Bond II mechanisms were used. Level III
isopleths utilizing the Dodge mechanism for these days were supplied by
OAQPS/EPA.
As explained in the EPA EKMA guidance document (EPA 1981),
Level III assumptions include an initial morning 8:00 am mixing height of
250 meters. Additional simulations and isopleths were performed with the
Demerjian mechanism with assumed initial mixing heights of 350 meters for
comparison. Only the characteristic curve for mixing height profile was
utilized. In each case, the measured final mixing heights were used in the
characteristic curve calculation. The input data were supplied by OAQPS/EPA.
The initial conditions of NMHC and NO. the observed Oo maxima, the
X «j
initial and final mixing heights, and the assumed county/hr path for each
day are given in Table 46. The county emission densities are given in
Table 47. The NMHC distribution and carbon fractions are shown in Table 48.
The latter values were recommended by Demerjian (personal communication,
1981).
For comparison with Level II simulations (Figure 79), Figures 108 and
109 show the plots of NO, N02» and Oo profiles along the assumed trajectories
for each simulation. Note that two mixing heights (250 meters and 350 meters)
were used for the Demerjian simulations.
Table 49 lists the 0^ maxima predicted for each simulation, the observed
Oo maxima for the day, and the ratio of the two values. Also included in
Table 49 are the Level II simulations results for comparison with the Level
III results.
An important point is that although the Level II simulations end at the
time (hour) of the observed 0,, maximum the guidance suggests that Level III
simulations be commonly performed for 10 hours regardless of the time of
298
-------
Table 46. Initial Conditions and Observed [0^] Max
for Level III Simulations
Item
[NMHC]Q ,
[NOx]o '
[03] max ,
Max Time
[03] aloft
D159
ppm 1.79
ppm 0.205
ppm 0.192
1700
, ppm 0.12
Init Mix Ht. , m 250,350
Final Mix
Trajectory
Hour
800
900
1000
1100
1200
1300
1400
1500
1600
Ht. , m 1900
St. Louis City
11
11
St. Louis Co.
St. Charles Co.
11
Madison Co.
11
Jersey Co.
D195 D275
0.26 1.90
0.048 0.236
0.210 0.244
1700 1600
0.074 0.06
250,350 250,350
1800 900
St. Louis City St. 1
II
II
II
M
II
M
n
St. Louis Co.
299
-------
Table 47. Area Wide Emission Densities for
Level III Simulations
County
St. Louis City
St. Louis County
Madison
St. Charles
Jersey
Kg-mole/(Kmz-hr)
ppm(C)-Km/hr
HC
2.716
0.533
0.163
0.070
0.014
NOV
X
0.465
0.148
0.044
0.092
0.003
HC
0.06641
0.01305
0.00406
0.00174
0.00029
NOV
X
0.011362
0.003634
0.002254
0.001058
0.000092
Table 48. NflHC Compositional Carbon Fraction3
for Morning Initial Conditions and
Emissions for Level III Simulations
(from Demerjian 1981)
Species
Initial
Emissions
Avg. Carbon Number
NR
OLE
PAR
ARO
ALD
0.12186
0.18195
0.43596
0.18549
0.07478
0.12791
0.19089
0.47965
0.17054
0.03101
3.0
3.0
6.3
7.1
2.0
aTo obtain ppmV for each class, multiply total NMHC in ppmC
units by fraction and divide by average carbon number.
300
-------
0.. PPM
">« 'MN 'OH
o
LO
c\j
en
c
•r—
CO
3
LT)
t^
CM •
-oi
c to
03 T-
LT>
to o
>>DQ
(O
•o c
o
CO -Q
Q- S-
eT (O
c2 o
oi -o
cu c
S- fO
4-> C
to
S- -i-
O T-)
M- t.
CU
to E
c cu
O Q
to en
CD
r— C
O) -r-
> X
CO
o
cu
3
CO
'"ON 'ON
301
-------
(C
o
tn
CO
cr>
c
to
3
Lf>
-o
c
(T3
un
01
IT)
T3
CO CO
O- T-
to
^:
u
o -o
co E
c cu
-P S-
res o
CO CD
i-H -E
I — i
cn
r- C
(1) -1-
> X
en
o
cu
3
'•OH 'ON
302
-------
Table 49. Ozone Maxima Predicted in Level III Simulations of
Selected RAPS Days. Comparison with Level II Simulations
Item June 7 (159) July 13 (195) Oct 1 (275)
ACTUAL DATA
[0,] max, ppm 0.192 0.210 0.244
Time max 1700 1700 1600
LEVEL III SIMS.
DEM/250 Ma
[Oq] pred 0.302 0.216 0.226 0.418 0.418
Time pred 1700 1700 1800 1600 1800
Rlc 0.636 0.972 0.929 0.584 0.584
CB2/250 Ma
[0,] pred 0.260 0.262 0.143 0.156 0.250 0.282
Time pred 1700 1800 1700 1800 1600 1800
Rl 0.738 0.733 1.469 1.346 0.976 0.865
DEM/350 Ma
[OJ pred 0.332 0.210 0.458
Time pred 1700 1700 1600
Rl 0.578 1.00 0.533
LEVEL II SIMS.
DEM/CHARb
[Oo] pred 0.197 0.133 0.168
Time pred 1700 1700 1600
Rl 0.975 1.579 1.452
CB2/CHAR
[0-1 pred 0.154 0.097 0.081
Time pred 1700 1700 1600
Rl 1.247 2.165 3.012
Code is DEM = Demerjian Mechanism; CB2 = Carbon Bond II Mechanism
number after / is initial mixing height in meters; characteristic
curve used in all level III simulations.
Code is same as in a; CHAR = characteristic curve using individual
day's initial mixing height.
p
Rl is ratio of actual day's Oo to simulation predicted 0-,.
303
-------
the observed 03 maximum. Therefore, Table 49 lists two sets of values for
some simulations to indicate the effects of the extra time.
10.2 DISCUSSION OF DAILY SIMULATIONS
Level III simulations predicted much higher [0-j] max than the Level II
counterparts (see Table 49). Only one simulation, CB2/D195 underpredicted
the observed [0^] maximum. For the Level II simulations of these three days,
only one (D159, a day dominated by high 03 aloft) was overpredicted (the
Demerjian mechanism^by -2%).
There are three major reasons for these results:
1) Level III simulations start in center city with 250 meter mixing
heights and maximum center-city concentrations resulting in
more mass initially present than in Level II;
2) the average hourly emissions of HC and NOV are often higher in
A
Level III simulations than average emission values in Level II
simulations, because the assumed trajectories start in a high
emissions density region; and
3) differences occurred in assumptions of emission inventory
composition in Level III as compared to Level II.
Initial concentrations in Level III simulation are commonly higher
than Level II since it is assumed that the trajectory originates in the
middle of the city. The morning mixing height assumed (250 m) is often
higher than the morning mixing height used in Level II. These conditions
tend to result in greater amounts of mass initially present in the Level III
air parcel. This can be illustrated by conversion of the initial conditions
for HC and NO and the assumed mixing height into equivalent emission terms
A
304
-------
(i.e. ppm-km units) and comparison of the values with the post 8 am emissions.
This has been done in Table 50. The table shows that two of the Level III
simulations are clearly initial condition dominated (D159 and D275).
Since the Level III approach assumes that the trajectory travels from
mid-city straight to the downwind station which measured the [Ck] max, the
average hourly emissions of HC and NO can be higher than average emissions
A
in Level II simulations (D195 in Table 50). This would be especially true
when the trajectory in the Level II simulations traversed the edge of the
city as was the case for D195. Consequently, Level III simulations can have
more mass input from emissions into the parcel than Level II simulations.
The third reason for differences between these Level III and Level II
simulations is related to the emission inventory assumptions of these two
versions. Level II simulations require a rather extensive emission inven-
tory. Level III does not. It has been shown that the aldehyde fraction of
both the initial HC concentrations and consequent emissions is a key
parameter in the photochemistry represented in this study. The Level II
inventory provided aldehyde fractions which were used in the Level II
simulations. The fraction ranged from -1-3% for the 10 days studied. These
may be rather low values. The fact is, however, that the inventory was
conducted and these values were reported. A Level III simulation does not
require an extensive inventory, but rather uses county wide average data.
It is not probable that an inventory of this type would include satisfactory
aldehyde data. Therefore this value would need to be approximated.
Based on his extensive knowledge of the emissions inventory for St. Louis,
Demerjian estimated the values given in Table 48. He used morning, ambient,
detailed hydrocarbon analysis to aid in estimating the initial conditions.
305
-------
to
CZ
o
•r—
4->
fO
r—
Z3
•r—
00
t—t I
Q.
i— Q-
0)
> "
CU X
_J O
-O
fO
i — i
i — i
r—
0)
c
^- 0
fO -r-
o to
i •rn>
j— i
UJ
O) O
C7)-r-
03 00
S- oo
CU -r-
> E
•i —
CT
1— H
c:
i— O
fO •!-
4-) OO
O OO
I— T-
E
UJ
^- ^J-
ro oo
CM CM
• •
o o
c
cu o
CD •' —
f<3 00
S- oo
CU T-
> E
cC ui
to LO
CM C\J
0 0
o o
2!
-)c
U
C
o
o
oo op r^
OO «T CM
r— t* tO
• • •
O O CD
OO LO
CO «*•
oo cr>
0 O
0 O
«3- LO
to o
o <—
O CD
. ,
o o
tO CM
O r-
o o
O 0
LO ^3-
CO «3-
i— LO
o o
•— o
oo to
0 O
o o
O LO ,—
CM tO CT)
r- O O
000
to cr»
LO O
co cr>
0 O
0 O
r— i — ->
t— t—
O 0
0 O
CM cr>
CM LO
o o
0 0
CO ,—
00 OO
CM LO
o o
to to
oo to
0 0
o o
tO LO LO
O r*-^ to
CM -vf tO
• • •
000
O
oo
S-
tV
Q.
E
O
o
LO
Ol
-Q
(O
E
•r~
OO
rtJ
CM O O
r-. LO LO
CM CM OO
o
o o o
i— LO LO
r— C\J OO
O O
O LO LO
r— CM 00
O1
LO
LO
CT>
LO
1^
CM
306
-------
The morning initial aldehyde fraction is an estimate made by Demerjian.
These values were used in the Level III simulations herein.
D159
Discussion for June 7. The observed [03] max was 0.192 ppm. Table 49
gives the Level II and III simulation 03 maxima. Both Carbon Bond II and
Demerjian mechanisms were used with a 250 m initial morning mixing height
and the Demerjian mechanism was also used with a 350 m mixing height. All
the Level III simulations overpredicted (36-73%) while only Demerjian
overpredicted at Level II (~2%); and CB2 underpredicted in the Level II
simulation by 20%. The reasons for these differences in simulations are
readily explained. The overriding factor for D159 was the difference in
the assumed initial location and initial conditions of HC and NO . The
A
initial mixing height in both cases were almost identical: 250 m in Level III
and 272 m in Level II (characteristic curve). The initial HC, however, was
0.49 ppmC for the Level II simulation and 1.79 ppmC for the Level III
because of the differences in the assumed starting point of the trajectories
(see Figure 31). The NO initial conditions were also higher by a factor
/\
of -5.5 The emissions in both simulations were almost identical (because
the trajectory traveled very far). Therefore the Level III simulation
predicted more 03 than the Level II simulation because of more mass initially
in the system. The same effect can be seen between the two DEM simulations.
The 350 m simulation made 0.332 ppm 03 as opposed to 0.302 ppm at the 250 m
initial mixing height. These simulations also show that for the Level III
approach DEM produces more 03 than CB2 (e.g. 0.302 ppm vs 0.260 ppm at 250 m
initial height).
307
-------
D195
Discussion for July 13. This day illustrates the range of assumptions
possible in switching between Level II and Level III approaches. This range
makes it difficult to compare. The Level II trajectory analysis for this
day (Figure 41) shows that the parcel originated in the far south east of
the city, so much so that the earliest hour that can be assigned a position
is 1100 LOT (with an initial mixing height - 1198 m). The Level III
assumption, however, originates the parcel in the center of St. Louis with
250 m at 0800 LOT. The initial conditions (concentration and mixing height)
were such that in either case, the mass they contributed was low. Thus,
this day is dominated by emissions, both for Level II and Level III approaches.
But since the Level III trajectory originates in the city, the emission's
contribution to the Level III simulation is much greater than the emission's
contribution to the Level II simulation. Hence, for the Level III simulations,
the Demerjian mechanism predicted 1.03 times the observed ozone maxima;
the CB2 mechanism predicted 0.68 times the observed. For the Level II
simulations, the Demerjian mechanism predicted 0.63 times the observed;
the CB2 mechanism predicted 0.46 times the observed ozone maxima.
D275
Discussion for October 1. Both effects, greater emissions and greater
initial mass, played a part in Day 275 DEM and CB2 simulations. Both of
the simulations predicted greater [03] maxima in the Level III than Level II
simulations. Initial concentrations were nearly the same at the two
different levels. The Level II initial mixing height, however, was 109 m
(for Characteristic curve). Therefore the initial condition has -2.5 times
3 OR
-------
the mass in the Level III simulation as in the Level II simulation. Also
in the Level II simulation the trajectory was on the eastern border area of
the city. Therefore the Level III emissions resulted in more than twice
the mass input into the parcel.
The resulting difference between Level II and Level III in [CL] max
prediction is most dramatic for this day. The observed was 0.244 ppm. In
the Level II simulation with characteristic curve, the Demerjian simulation
yielded 0.168 and the CB2 simulation 0.081 ppm 0^. The CU maxima for Level
III simulations were: (DEM/250 m) 0.418 ppm; (DEM/350 m) 0.45 pom; and
(CB2/250 m) 0.250 ppm.
A basic difference therefore between the Level II and Level III
approaches, is that the Level III assumptions often result in more total
mass being injected into the system than Level II assumptions, resulting
in more ozone predicted in Level III than in Level II simulations. The
increased aldehyde initial condition and emissions factor also has a major
impact on these simulations.
10.3 ISOPLETH DIAGRAMS AND EKMA CONTROL CALCULATIONS FOR
LEVEL III SIMULATIONS
Isopleth diagrams for the Level III simulations for use with the EKMA
procedure were obtained in the same manner as for the Level II simulations.
Figures 110 to 113 show the isopleth diagrams and Oo as a function of
relative HC for three levels of NOV control for the three days simulated
/\
with the Level III approach. Table 51 lists the %HC reduction estimates
obtained by EKMA procedure. Finally, plots of %03 reduction vs %HC
reduction estimates for these Level III simulations are presented
as Figures 114 and 115.
309
-------
cr>
LT>
-P
r- O
3 -4->
-O -C
en
S- -r-
o
E C
fO -r-
i- X
CD-r-
to E
•1^
TD r-
tO
OOt 0 00£ 0 002 0 001 0
-(-> C
CJ -i-
fO
s- s_ .
f- CU E
o o
o c\j S-
•o to re
r— C_5
J= =5
-P £ T3
QJ T- C
i — (/) (O
ex
o •-* c
C CU OJ
O > E
M OJ O)
O _J Q
a>
rs
en
3>0
-------
oot o oot o ooz'o
in
CTl
O)
CO o
i — -Q
s_ ^:
O en
<+- •!-
ai
CO J^
E
(t5 CD
S_ C
CD-r-
(O X
•r— *r—
-a E
C t—
O IB
one o on: a oot o
0 200
03.ffft
O -i-
i"O ^—
i- -r- •
oEl
-c. o c
i un to
o> CM .£:
c o
o re cu
o o>
C I—(
-a -r- i->
ta =s T3
CO d
CO CO O
E S_
r— <_5
^: 3
-M E "O
Q) -i— c
I—• CO fO
ex
o >--< c
CO I—I tO
CD •— $_
C 0} CD
O > E
N O) O)
O —I O
CU
S-
C7I
311
-------
ooc'O oos'O ooT'O
IT)
1^.
CVJ
IO
OJ
s- o
(Li ja
jD
O S-
4-> O
o f-
o
4J
S- -C
O Dl
4- T-
O)
CO -C
(t> O>
s- c:
cn-r-
(O X
•r" *r--
•O E
C r—
o to
O •!-
fO C
S- -r-
OOfc 0 QQC 0 OQZ 0 001'C
O 'r-
rr: o c
I LT> (O
cu CM -c:
c o
C3 fO QJ
tsl £
O cn
C H-I
c i I~H
(B 3 "D
10 C
S-
r— C_>
CU -I- C
i— oo ro
CL
o i—< c
to I—l E
[si Ol CU
O —I Q
cu
S-
0.300 0 200 0 100
oa.rrn
312
-------
oot o oos o ooz a
c:
•r-
CO
O «3
•r- I —
+J CT1
3 S_
E a;
•r- -Q
oo O
-t->
i— i o
i— t O
cu ro
>
CU A
—I CO
ocz o ooi a
0 300 0 200
OOt 0 OOE 0
oat o oot 0 OOE o ooi 0
i. r--
C31
rtj (1)
•i- C
-o rj
O S-
•r- O
o
ro
en
l
CU CD
c c:
O -r-
N X
O -r-
nf co
CO +J £
F •"- «/)
fO C •>-
i- T- C
O> O3
ca i- .c:
•i- tu o
T3 +J (V
0) E
-c E
•M C
CU O ro
i — IT) -r-
Q-CO -r-j
O t.
co en CO
CD £= Q
C 3
O <^ CL>
M to -d
o to -t->
9 100 0 300 a to I
313
-------
10.3.1 Discussion of June 7, Dl59
Figure 110 shows the isopleth diagrams and 03-HC fraction diagrams for
the Demerjian mechanism and the Carbon Bond II mechanism simulations
starting at 250 m initial mixing height. Figure 113 shows the Demerjian
mechanism simulation starting at 350 m initial mixing height. It should be
recalled that all three simulations overpredicted the observed ambient 03
maximum. The EKMA solution points were lower than the measured initial
HC and NO for all three isopleths: the DEM simulations; were about l/3>and
/\
CB2 were about 1/2 of initial HC and NO. D159 has the highest 0., aloft
X O
values: 0.120 ppm and results with diagrams showing only a portion of the
0.08 ppm isopleth line. The 0~ value at the origin was greater than 0.080
ppm.
Although the DEM/250M and 350M isopleth diagrams do appear different,
the EKMA procedure when applied to each resulted in remarkably similar
HC control estimates, 79.6 and 77.6% for a -20% change in NOV (Table 52).
A
The CB2/250M isopleth diagram resulted in a much lower control estimate:
57.3% (-20% NOV). For the Level II simulations the HC emission reduction
X
for the corresponding conditions were much less. DEM Level II was 67%
reduction and Level III was 79.6% reduction; CB2 Level II was 45% reduction
and Level III was 57.3% reduction.
10.3.2 Discussion of July 13 , D195
Figure 111 shows the Demerjian and the Carbon Bond II 250 m isopleth
diagrams prepared for this day. This day can be compared with D275 in that
the absolute emissions rates were the same except for the last hour.
314
-------
Table 51. Level III %HC Control Estimates
NO Control
X
Day Mech/Mix Ht I2Q%_
159 DEM/250 79 71 64
DEM/350 78 69 64
CB2/250 57 47 41
Dodge/250 74 72 67
275 DEM/250 47 38 32
DEM/350 53 42 34
CB2/250 38 29 23
Dodge/250 56 48 38
195 DEM/250 45 35 29
DEM/350 43 34 29
CB2/250 32 24 19
Dodge/250 59 55 46
315
-------
The initial HC values, however, were 7.5 times less than Day 275. Also
the final mixing height for D195 (1800 m) is twice that of Day 275 (900 M).
The Oo aloft was similar for the two days.
The isopleth diagrams for these days were plotted on axis with the same
magnitudes used for D159 and D275. The isopleths suggest that this day
would produce ozone dramatically. What must be remembered is that the
emissions as well as the initial conditions are multiplied in generating
the diagrams. Therefore at the point on the D195 isopleth diagram which
matches the initial conditions of D275 (1.9 HC, 0.236 NO ) the emissions
/\
are 7.5 times as great as the design point emission rate (observed initial
conditions and emissions). This is a great amount of material compared to
the XI emission rate which is of course at the design point of D275.
Of further interest is that the control estimates for the DEM/250M
day 195 (44.6%) closely match those of DEM/250M day 275 (47.4%) while the
350 m control estimates differ: DEM 195/305M was 42.6% and DEM 275/350 M
was 52.5% reduction.
The isopleth diagram generated with the CB2 mechanism (250 m), which
appears to be similar to the DEM isopleth diagram for this day, yielded a
significantly lower control estimate (32.2% vs 44.6% for -20% NOV).
A
10.3.3 Discussion of October 1 , D 275
Figure 11 shows the isopleth diagram and the O^-HC fraction diagrams
for the Demerjian and Carbon Bond II simulations using 250 m initial
mixing height. This day contrasts with D159 in that the initial conditions
on this day were similar to those of D159, yet the emissions were much
larger on D275. This was true in the absolute sense and the effect was
316
-------
further enhanced by the 900 m final height on D275 compared to the 1900 m
final mixing height on D159. These factors resulted in Day 275 having more
Oo producing ability for both DEM and CB2 mechanisms than on Day 159.
Since the DEM/250M simulation predicted much more than the CB2/250M
simulation the isopleth diagrams also show similar trends.
On the DEM/250M isopleth diagram, the EKMA solution point was
found at 52% less_ than the day's initial condition and on the CB2/250M
isopleth diagram the EKMA solution point was found at 8% less thanttie
day's initial conditions. In spite of these differences in absolute pre-
dictions, the EKMA HC control estimates are 38.2% for CB2 and 47.4% for
DEM (at 250 m and -20% NO reduction). The DEM/350M isopleth diagram,
A
like the day's simulation,predicts the most ozone formation for this
day's conditions. The EKMA control estimate using this isopleth diagram
is only 5% more (52.5%) than that using the DEM 250 M isopleth diagram.
10.3.4 Discussion of Oo Reduction Diagrams
Figures 114 and 115 show the percent Oo reduction as a function of the
percent HC reduction for the Demerjian and Carbon Bond II mechanisms for
the three days simulated. Except for June 7, Day 159, which had 0.12 ppm
03 aloft, the results suggest a greater than 1:1 control effect. That is,
a given %HC reduction results in more than that % reduction in Oo (e.g. a
25% HC reduction gives a 50% 03 reduction).
The ratio of % 03 reduction to %HC reduction is about 2:1 for Carbon
Bond II and about 1.3:1 for Demerjian (independent of initial mixing
height) in the Level III simulations with small amounts of 0-, aloft. For
Day 159 with 0.12 ppm 0^ aloft, the ratios are less than one (i.e. 0.8:1 for
Carbon Bond II and 0.5:1 for Demerjian). Recall that in the Level II
317
-------
-H 01 at
- 5 5
_ o O
O
LO CU
CM >
05 CT>
c
•i- O
in 4->
3
T3
CO d)
c: -a
O O)
•,-
CO O
i—i "O
>—( OJ
"O
c
O -r-
(O tO
+J
c c:
ro o
•r- N
•1-5 .,-
s^ s-
cu o
-o
O)
o co
co O.
E Q-
t> •
01-C i—
03 CD •
•1- T- O
-£3 qj *
c -a
o en s-
•i- C (O
qj
(O
+j
CO
O ••— 03
N c ^:
O -r- -t->
a>
i.
=3
O1
318
-------
-Is I
4-
CT> ra
C. -O
-i- c
x re
•i- +J
E to
i — CO
m o
•r—
-l-> OJ
O)
E >
oo o
+J
CT)
C "O
•i- -J
S- to
01 -i-
Q c
S- i
o
(D
re o
S. N
ro S-
•r- o
-t-> to
o re
3 Q '
•O
(U
S-
01
c:
o -
Q.
c\j
O) O
O)
S-
01
319
-------
simulations for Day 275, the ratio was about 2:1 for Demerjian characteristic
curve and that the Carbon Bond mechanism gave no solution because of under
prediction (the Dodge simulation ratio was about 1.6:1). For Level II
simulations of Day 159, the ratio was about 0.7:1 for Demerjian and 1:1 for
Carbon Bond II. Thus it appears that the Carbon Bond II mechanism con-
sistently suggests greater sensitivity of ozone to HC control than either
Demerjian or Dodge and that this sensitivity is more than (up to twice)
that of simple linear roll back.
10.3.5 Effect of Initial Aldehyde Fraction
It was stated earlier that the Level ill approach does not require more
than a county-wide emission inventory. Aldehydes are generally not well
characterized in these inventories and therefore must be approximated by
the modeler.
Based on suggestions by Demerjian, values of 3% Carbon and 8% Carbon
were used for the ALD emissions fractions and ALD initial conditions fractions
in this study. A value of 3% for the aldehyde carbon fraction marks the
upper limit of the values determined from detailed emission inventory
calculations for the 10 days studied in the Level II section. It is
a reasonable value.
In the Level II discussion, the effects of variation of the aldehyde
fraction on the simulations of certain days were demonstrated. To examine
the effect for the simpler Level III approach, two level III simulations
with the Demerjian mechanism were redone with an aldehyde fraction for the
iniital conditions equal to the emissions aldehyde fraction: a change
from 8% to 3%. The results are shown in Table 52.
320
-------
3% Initial Aid.,
ppm
0.359
0.208
Ratio
3% to 8%
0.859
0.963
Table 52. Effect of Initial Aldehyde Fraction on (k Maximum
for Level III Simulations with Demerjian Mechanism
[CU] max for [Oo] max for
8% Initial Aid.,
Day ppm
275 0.418
195 0.216
The magnitude of the effect of the variation on [03] maxima depends
on the relative contribution of the initial conditions to the total mass
of pollutants emitted into the parcel, hence a sizable effect for Day 275
and a small effect for Day 195,which was dominated by emissions.
The aldehyde fraction effects the 03 maximum concentration mainly on the
left-hand side of the ridge onanisopleth diagram. For those days in which
the EKMA solution points are located on this left-hand side, the effects
of aldehyde initial carbon fraction on the HC reduction are probably
minimal. For those days in which the solution points are on the right-hand
side (which is not much effected by aldehyde fraction) the crossing of the
ridgeline in a control calculation will probably result in a significant
effect on the HC reduction required.
10.3.6 Effect of Simulation Length on Control Requirements
The guidance supplied by EPA implies that Level III simulations (for
isopleth generation) would commonly be performed for a 10 hour day,
regardless of the time that the [Oo] max was measured. For example, the
days simulated by the Level III approach in this study ( D159, D195, D275 )
had maximum ozone concentrations measured during the 9th or 8th hour
(after 0800 LST). Emissions data were supplied only for the 8 or 9 hours it
321
-------
took to reach the maximum measured (see Table 46). The Dodge isopleths
generated by OAQPS for these days used 10 hours of simulation time and
8 or 9 hours of emissions data and used zero emissions for the extra one
or two hours until the end of the 10-hour simulation period. Isopleth
diagrams generated in this report for the Demerjian or Carbon Bond II
mechanisms used the actual time to [O.J max, 8 or 9 hours, depending on the
O
RAPS day in question. Concern for the possible effects from these
additional assumptions (10 hour default run time and zero emissions past
supplied data) suggested that simulations should be compared with these
different conditions. Five additional simulations and an additional isopleth
were obtained. These additional simulations are reported in Table 49 as the
second column under each day.
For D159 and D195, simulations were performed for an additional hour
(nine to ten hours). The one simulation for D159 using the CB2 mechanism
showed only a 1% increase in 03 maximum indicating that the 03 profile
had almost peaked at 1700.
Two simulations were performed for D195 using both the DEM and CB2
mechanisms. DEM showed a 5% increase and CB2 showed a 9% increase.
Two simulations showing the most dramatic effects were performed for
D275. The DEM mechanism shows that the 03 peak is closer to 1600, so that
the effect of simulating an additional two hours actually yields a lower
final Oo maximum. The CB2 mechanism, however, having not reached 03
maximum at 1600, yields a 13% higher 03 maximum after simulation for an
additional two hours (0.250 to 0.282 ppm).
For Level III, the effect of simulation time is important not for
absolute [03] predictions but for the HC reduction estimates obtained
322
-------
from different isopleths generated for a day using different simulation times.
To obtain the greatest possible effect, one isopleth diagram was generated
with the conditions showing the largest change in 0^ maximum in the tests
just described: C82 for D725, simulating 10 rather than 8-hours.
The isopleth and 03 reduction diagrams for these two conditions are
shown in Figure 116. The isopleths on the left-side of the ridge have been
shifted to the left (the initial condition point occurs between 0.28 and
0.30 instead of between 0.24 and 0.28) for the longer simulation. The
control diagrams, however, are essentially identical. For no change in NOV,
.A
the required HC reductions are 29.0 and 29.8% for 8 and 10 hours of simu-
lation. The EKMA solution points, however, are on the left-hand side of the
ridgelines and an effect might occur if the EKMA solution points were on
the right-hand side and the required HC reduction crossed the ridgeline.
10.4 SUMMARY OF LEVEL III SIMULATION RESULTS
The Level III simulations predicted higher 03 maxima than the Level
II simulations. This is due to the assumptions in
the Level III approach; a trajectory originating in mid-city has higher
initial conditions, higher emissions and hence more mass in the air parcel.
Assumptions of initial mixing height and the aldehyde fraction are also
parameters that effect the magnitude of the ozone maxima in the Level III
approach.
Figure 117 is a graphic summary of the control requirements predicted
by three mechanisms (Demerjian, Carbon Bond II, and Dodge) for three RAPS
days (159, 195, and 275) using a Level III modeling approach as described
in the recent EPA draft guidance (EPA, 1981). Included in Figure 117 are
323
-------
I 200 0.100
NOX.PPM
Nonano3» •o i
rtS
c:.
o i
>>
10
Q
c: to
-C en
+-> •—
2 •
to
O -O
-d O
o o
i— O
O 4-
4-> O
oo
E
O •!-
s- c:
f- to
CD O
E ^
+J
o
•r- T3
+J c:
rc( O
r— DQ
Z3
E C
•r- O
c/1 -Q
s_
cn tc
c: c_3
s_
O i-
c: o
1- in
O c:
O
+J >r-
CJ CO
a> co
a>
s_
ZJ
0 200 0.100
HOX.PPM
324
-------
the control requirements predicted in the Level II modeling study (see Chapter
9 for comparison).
The Carbon Bond II mechanism always gives the lowest control require-
ment by a substantial margin (ratios of CB2 to DEM are 0.66:1, 0.67:1, and
0.70:1, and ratios of CB2 to Dodge are 0.65:1, 0.43:1, and 0.56:1 for days
159, 195, and 275). This mechanism is the newest, most complex, and most
validated mechanism of the four studied. Except for Day 159 in which the
Demerjian and Dodge control requirements were similar, Demerjian always gave
lower control requirements than Dodge at both Level III and II. For the
days studied, Dodge always gives the highest control requirement.
Secondly, regardless of mechanism, the Level III control requirement
predicitons are significantly higher than Level II predictions (ratios of
Level III to Level II requirements are 2.4:1, 1.5:1, and 1.3:1 for Demerjian,
Dodge, and CB2).
Level III, like Level II, gives the maximum control requirement not for
the day with the highest observed 0^ maximum, but for the day with the
maximum O aloft and a lower contribution from local emissions.
325
-------
OCDCDCDCDCDCDCDOCDCD
OO")OOr^LD=fr-09OJ»—'
CD
1
«M>^
— ••
1 I ' 1
1 1 ' 1
• « J
' 1 ' 1
i ri
-j- Q
el
1 |
-
QJ X
1 1 1
__
— —
— —
In .4.
|
4-0-
11^
" O
, 1 ,
t
— f
».
•
, 1 ,
i n •
1
4-
e,i
, 1 ,
1
1
i-
, i r
\^ ^j i
c c E
*° j~ O)
- u >
Q
CJ
0
LO
i —
C\J
Q
O
,—J
en
tn
•~
r^ (/)
« •!—
n! c s_ s-
03 OJ i 3 O
-Q 0 r-
11
-a en .«
o Ln x
-C r- 0
> c _c cn
5 0 U C
J] •!- _OJ (C
u ^ *"
•r- -a s_ r
-o c: (U ;
ai o a.1*
s- ca Q.
'o -Q >d
<_ r <^\ J
-M ro Z:
C (_)
one
ej t_> -r- J
r-.
IDCDCDOOCDOOCDOO ^
:D CD co r
- vO LT> =3- CO CM •— • =>
en
GH %
326
-------
References
1. Uses, Limitations, and Technical Basis of Procedures for Quantifying
Relationships Between Photochemical Oxidants and Precursors,
EPA-450/2-77-021a, U. S. Environmental Protection Agency, Research
Triangle Park, North Carolina, November 1977.
2. Arya, S.P.S. (1979) Atmospheric Boundary Layers, to be published in
Engineering Meteorology, Elsevier.
3. Benkley, C.W., L. L. Schulman (1979) "Estimating Hourly Mixing
Depths from Historical Meteorological Data," JAM, Vol. 18,
pp. 772-78.
4. Demerjian, K. L., K. L. Schere (1979) Applications of a Photochemical
Box Model for Ozone Air Quality in Houston, Texas, In Proceedings:
Ozone/oxidants Interactions with the Total Environment II,
Oct. 14-17, Houston, Texas, pp. 414-21.
5. Demerjian, K. L. and K. L. Schere (1976) personal communication.
6. Demerjian, K. L. (1980) personal communications.
7. Demerjian, K. L., K. L. Schere and J. Peterson (1980) "Theoretical
Estimates of Actinic (Spherically Integrated) Flux and Photolytic
Rate Constants of Atmospheric Species in the Lower Troposphere,"
Advances in Environmental Science and Technology, Vol. 10,
John Wiley and Sons, New York, New York.
8. Demerjian, K. L. (1981) personal communication.
9. Dickerson, M. H. (1978) MASCON - A Mass Consistent Atmospheric Flux
Model for Regions with Complex Terrain, JAM, Vol. 17, pp. 241-53.
10. Dodge, M. C. (1977) Combined Use of Modeling Techniques and Smog
Chamber Data to Derive Ozone-Precursor Relationships
EPA-6007^77^001 b.
11. Dodge, M. C. Effect of Selected Parameters on Predictions of a
Photochemical Model EPA-600/3-77-048, U. S. Environmental
Protection Agency, Research Triangle Park, North Carolina,, June 1977,
12. Evans, R. B. (1979) The Contribution of Ozone Aloft to Surface
Ozone Maxima, Ph.D. Thesis, University of North Carolina, Chapel
Hill, NC.
13. Federal Register, "Data Collection for 1982 Ozone Implementation Plan
Submittals" Nov. 14, 1979, 44, (221) 65669-65670.
327
-------
14. Gipson, G. L. (1980) personal communication: OAQPS study.
15. Gipson, G. L., W. Freas, R. Kelly, E. Meyer (1981) Guideline for
Use of City-Specific EKMA in Preparing Ozone SIPs,
EPA-450/4-80-027, U. S. Environmental Protection Agency, Research
Triangle Park, N.C., March 1981.
16. Goodin, McRae, Seinfeld (1980) An Objective Analysis for
Construction Three Dimensional Urban Scale Wind Fields,
JAM, Vol. 19, pp. 98-108.
17. Holzworth, G. C. (1972) Mixing Heights, Wind Speeds, and Potential
for Urban Air Pollution throughout the Contiguous United States,
AP-101, USEPA Research Triangle Park, NC.
18. Jeffries, H. E. and K. G. Sexton (1981) Modeling Aspects of Nitrogen
oxides Using Smog chamber data, in workshop proceedings on
Formation and Fate of Atmospheric Nitrates, EPA-600/9-81-025,
U. S. Environmental Protection Agency, Research Triangle Park, NC.
19. Jeffries, H. E., J. E. Sickles, M. Saeger, and M. A. Carpenter
"Experimental Determination of the Specific Photolysis Rate of
Nitrogen Dioxide," Draft Final Report, Bruce Gay, Jr., Project
Officer, Environmental Protection Agency, Research Triangle
Park, NC, Feb. 1981.
20. Littman, F. E. (1979) "Regional Air Pollution Study: Emission
Inventory Summarization", EPA-600/4-79-004, Rockwell International
Creve Coeur, Mo.
21. Liu, M. K,, J. H. Seinfeld (1975) On the Validity of Grod and
Trajectory Models of Urban Air Pollution, Atm. Env., Vol. 9,
pp. 555-74.
22. Schere, K. L. and K. L. Demerjian (1980) personal communication:
characteristic curve.
23. Shreffler, J. H. (1978) Detection of Centripetal Heat Island
Circulation from Tower Data in St. Louis, BLM, Vol. 15,
pp. 229-42.
24. Spath, H. Translated by Hoskins, W. D., H. W. Sager (1974)
Spline Algorithms for Curves and Surfaces, UTILTAS Mathematica
Publishing Incorporates, Winnipge.
25. Whitten, G. Z. and H. Hugo User's Manual for Kinetics Model and
Ozone Isopleth Plotting Package, EPA-600/8-78-014a, U. S.
Environmental Protection Agency, Research Triangle Park, NC,
July 1978.
328
-------
26. Whitten, G. A. and H. Hugo U. S. Environmental Protection Agency,
EPA-600/8-78-014b, Research Triangle Park, NC, July 1, 1978.
27. Whitten, G. Z., J. P. Killus, and H. Hugo Modeling of Simulated
Photochemical Smog with Kinetic Mechanisms, EPA-600/3-80-028a,
U. S. Environmental Protection Agency, Research Triangle Park,
NC, February 1980.
28. Yamad, T. (1976) On the Similarity Functions A, B, and C of the
Planetary Boundary Layer, JAS, Vol. 33, pp. 781-93.
329
-------
APPENDIX A
THE HISFIT ALGORITHM: AN AREA TRUE HISTOGRAM FUNCTION
330
-------
Appendix A
The HISFIT Algorithm: An Area-True Histogram Function
It is frequently desirable to represent a series of interval histogram
values as a smooth function having the area-true property. Consider a
series of n intervals and associated histogram values h-j, \\2 •• h . We
define CQ, e-| .. e to be the abscissa values (x) at the edges of the intervals
Thus e-j .. e I are common edges, eg and e-| are the edges of the first
interval, e-j and 62 are the edges of the second interval and so on. For any
abscissa value x we assume the function produces an ordinate value y. Then
the function is defined to be smooth (differentiate):
lim fP(e. ) = lim fP(e. ) = y[ (p = 0,1 and k = l..n-l)
e-K) K c e->0 K"£- k
The area true property may be expressed as:
/k f (x)dx - hk(ek - ek ) (k = 1 .. n)
ek-l
An algorithm for finding the coefficients of a fourth degree spline
with these properties is presented in Spath (1974).In the present situation,
negative function values are physically meaningless and so we must ensure that
the function nas no negative values. Cubic splines (as originally used in
OZIPP) have fewer degrees of freedom and when a cubic spline is constrained to
have no negative values, very poor fits often result. Oscillating histogram
values also lead to poor fits with excessive over- and under- shoot.
Comparative examples are presented in Figures Al and A 2 below.
331
-------
The STAIR algorithm (Spath, 1974) solves the equation system over a
domain given the ordinate values at the edges and boundary values for the
first derivative. Thus a solution domain consists of a contiguous series of
intervals a, a+l,...,b (l^a^b^n), the given ordinate values at the edges
y ,, y , .., y, and the two boundary values dy -, and dy, .
u~i a o a~ i D
It was discovered that some histograms resulted in spline functions
having negative values when the solution domain is the entire histogram.
In particular when either of the ratios:
or
hk+l
is too large or too small, negative function values result. In either case,
we control the maximum (minimum) value of the first derivative at the
various edges. Examining the histogram, we apply a heuristic to find those
edges which are likely to have an excessively large (or small) first
derivative, and we break the solution domain at those points. Each such
edge then becomes a common boundary for two solution domains, and we can
define the edge first derivative as a boundary condition such that the
resulting function has no negative values. Any h.= 0 must always be a
boundary of at least one solution domain.
MISFIT finds the minimum set of solution domains and defines the ordinate
and boundary values as follows:
332
-------
§ 0.
I . (-
1.0
0. 8
0.6
0.1
0.2
0.2
0.1
n ft
i
i
i
i
i
_ DRY 160
—
—
—
, 1
f
, 1
^-
i
/--"
X
i
-\
V
i
i
1 '
i
i
JUNE 8. 1376 m _2
^_^
,
/
/
^_^/
/
\ —
I "•
i
i
\J
—
1
1 . ^
1.0
0.8
0.6
0.1
0.2
. 0
-0.2
-0.1
_n F,
1 5
INTERVflLS
o
X
to
CO
z:
UJ
1 . d
1.0
0.8
0.6
0.1
0.2
0.0
0.2
0.1
n P,
I
i
i
i
I
DRY
P—^
i
i
160
y
i
i
JUNE 8
i
' ' i
. 1976
v ^
i
I
NOX
/
i i
H/sVlT J
' \
—
^:
\-
—
i
, I ,
1 . C
1.0
0.8
0.6
0.2
0.0
-0.2
-0.1
-n f,
15
INTERVflLS
Figure A.I. Comparison of equal area histogram fit as produced
by OZIPP algorithm giving negative area (top) and by
HISFIT algorithm (bottom).
333
-------
t
o
t—I
X
co
CO
LU
O
I I I I 1 I I I I I I I I I I I I I I
Y 221 flUGUST 8, 1976 HC
t _ i . i i i i i i
0123456789 10 11
CO
CO
UJ
o . v
5.0
n n
H. U
3.0
2.0
1.0
On
. U
-1.0
-2.0
T n
1 1 '
i
i
i
1
I
1 A DflY 221 flUGUST 8, 1976 HC HISFIT _I
—
_
1
\s*—-J\
^__
^ — ^
^ /-\
\
\
\ j
^-^
/ \
\
\_ J
"• — "
—
—
r , i ,
,
L i _j
i
1
,
—
,
A-
V
—
—
!
U . U
5.0
3.0
2.0
1.0
n n
-1.0
-2.0
-3 n
5 6 7
INTERVRLS
10 11
Figure A.2. Comparison of equal area histogram fit as produced
by OZIPP algorithm showing strong oscillation (top)
and by HISFIT algorithm (bottom).
334
-------
hk = 0
or
k+l
dy =0
5.01 > nk+] > !____
Fk"" 5.01
h, 0.5
\>hk+i
'k+1
> 5.01
or
> 5.01
'k+1
2.5 hk.
5.25 h
k
-5.25 h,
hk+l > hk
It Is assumed that h , = 0. There are two alternative assumptions about
the value of h«. In one case (mode = 0) we assume h« = 0. Otherwise
(mode = 1), we assume h0>h, and we set yQ = 2 h-, and dy^ = -yQ/(e,-6Q).
33 R
-------
APPENDIX B
SPECIAL EKMA COMPARISONS
336
-------
APPENDIX B
Special EKMA Comparisons.
EKMA is a technique which is used to obtain HC control estimates from
03 isopleth diagrams which do not necessarily agree in the absolute sense to
the observed atmospheric conditions to which the isopleth diagram is being
applied. EKMA is a scaling procedure; it is not needed for isopleth diagrams
which agree with the observed precursors and 63 values. In simple models or
models with gross assumptions, the absolute agreement is generally poor and
EKMA provides a means of estimating control requirements for these cases.
This study has shown that assumptions made by modelers can have an effect
on the calculated control requirements. The validity of EKMA and the effect
of incomplete and assumed data has been considered by the EPA (1980):
"...Sensitivity studies have shown that differences resulting
from incomplete input data or gross assumptions employed by
the model tend to exhibit proportional impacts on the positions
of various ozone isopleths on an isopleth diagram. Thus, the
model should perform well when applied in a relative sense..."
Some results from this study tend to disagree with the above conclusion.
Figure B.I shows the results of two applications of the Level II approach
to D275 October 1, 1976 (RAPS) with the Demerjian mechanism and the char-
acteristic curve mixing height. The carbon mole fraction for aldehydes
needs to be determined for both the emissions into the air parcel along the
trajectory and the initial conditions of NMHC. Each row of Figure B.I shows
the simulation and observed profiles for NO, N02> and 03 for the day, the iso-
oleth generated, and the final %Q^ reduction-%HC reduction plot obtained with
the EKMA procedure (the horizontal dashed line shows the % Oo reduction
needed to obtain the 0.12 ppm level). Row A used the emission inventory
value of 2.02% carbon mole fraction for aldehydes for both emissions and
337
-------
0,. ppm
iO «- OJ
1 !
0, pom
E
o
o o ••-
rr: -i- oj
s: r- -c
z: Q.
O_ CD
i— (O C
(O T-
•1- >-H X
4-> •—I •!-
•'- E
•r- O +•>
O 4— oo
S_ C i-
4- O CU
•i- +->
O) •*-> O
'o 3 £
E O fD
u c:
O E
C
J^: ro
-!-> C -r-
o ••->
en S-
C r • CU
••- CO B
cn»— 0)
C {^
-C C3
O • ^O
ur> r~^
4- CTl
O O r—
+J
1 < ^
O '—-i—
QJ eX
4 i.
4- CU
CD ^S -Q
CM O
O) O 4->
-C • O
I— CO O
CQ
Q)
s_
0>
CO
338
-------
o
C\J
c
o
O LD
r— 1
1
"~ "0
« O)
"~~ 0
S-
O)
-Q O
O 3T
O
0
•r—
cn -^t- -a
cn ID r^ a)
00 cn r- j_
Q-
i — CO CM
c:
O
s- •<-
O -I-1
M- • fO
• — -• » —
co E
+- ' t/)
C •!••• C^
•i- c: i —
O fO S-
o r~
O
Z3
CM E
1 LT) -^J- ••-
i o
<: E
2: CD
X^ Q
D- 3
^r co «*• . i —
^ UD CO
CM i — CM
• • •
CD CD O
1 i 1 «.
, —
•
CO
O)
, —
-Q
(O C
\— 0
•1 —
•J-)
•1 —
-a
cr
o
o
CO
a> c: c
3 O O
n3 4-> 4->
> fD fO
! ,
4-> rs r:
£= E E
CD -i- T-
•c— CO to
_Q
E ^5; ^V
rd CM IT)
(O
4-)
to c
•4-J CD
C -Q
^3 E
(O
CO
0 c,-
O
T3
c o
rt3 •' —
^J
^ rt3
X S-
O
2: cu
<~
" 4-*
C_)
3: co
•i —
^H ,
fO &—
ns XI
CO
cu
(O
>
4-)
CZ
CU
•r~
E
n3
0
+-)
CO
c
o
*r-
4->
•i —
"O
C
0
o
, —
to
•i —
4->
•i —
C
T3
O)
4->
fO
r —
23
(_j
i —
fO
O
2:
UJ
(4-
o
0
4-}
ro
S-
ai
f~
+j
CO
••-
CM
S-
o
339
-------
initial NMHC. Row B used the emission inventory value for the emissions,
but used a value of 5.0% for the initial conditions of NMHC. It can be
seen that not only the simulation profiles and isopleth diagrams changed
considerably, but that the EKMA calculated control requirements changed
considerably as well . The change in scale on the isoplel.i diagram was
necessary to obtain an EKMA solution for the upper set of conditions. The
numeral values are shown in Table B.I. The solution point moved from 15%
more initial and emitted HC and NO to 110% more initial and emitted HC and
X
NO , and the control requirement went from 15% to 22%, a ratio of 1 .47.
Figure B.2 shows again two approaches to estimating control requirements
to D275 October 1, 1976. Row A is a Lavel II application and Row B is a
Level III application. The different assumptions for each Level has been
described in this report. Approximately the same mechanism (CB2), emissions,
initial conditions, 0^ aloft, and final mixing height were used on both
applications for this day. The basic difference lies in the initial mixing
height assumed (Level II 109 m, Level III 250 m) and the NMHC carbon mole
fractions assumed for emissions and initial conditions (see report for dis-
cussion). Not only are the simulation profiles and isopleth diagrams
drastically different, but while an EKMA solution was obtained from the
Level III approach, no solution at all was possible with the Level II
approach.
340
-------
o
LO
(J
-c: to
O (O
4-> O
(O
o -o
••- E
r— CO
O)
us
^l--»
CTl
QJ
> i-
O) O)
_J -Q
o
(O -4->
O
E O
O
i. 5-
M- O
<4-
CT1
C C •
•r- o *~~*>
CD-I- +->
C +-> JC
ra 3 OI
JT i — -r-
O O O)
in -C
o c:
+j i, -i-
o +-> x
at s= •<-
M- o s
t- o
OJ O)
•a: >
-------
APPENDIX C
SIMULATIONS OF INDIVIDUAL BUREAU OF MINES EXPERIMENTS
USING CARBON BOND, DEMERJIAN, DODGE, AND CIT MECHANISMS
342
-------
NO. NO,. PflN, ppm
l/y o tj~) o 10 ° LO
^- «*- co co ex CM •-*
'ON 'ON
i, ppm
pr,
r- 10 U7 «/- o^ CJ --* T>
c^ r> (—. o O cj <~i c^ri
-rj-. rr-rn-n-'-T < '
_!.Li J
O O CD O O
':ON 'ON
'NOd ':OfJ 'ON
343
-------
0-. ppm
0,. ppm
L1__U_l_i_l_uLj_L1_LL_UJ_i_L
'NUd ''ON 'ON
uldd 'NUd ''CH
'OM
0,. ppn
O CJ O O O O CJ CIO
'Nttd <:ON 'ON
'ON
344
-------
L
u - y .
U.: -i :_4 L4-!_i-L.4J_i-U i i.i_i._
''ON 'ON
3. ppm
), ppm
in et- Of.or3i/>c3i/>c»lOO
CD O O
C3OOOC>C»C;>OC3O
""" 'NUd ''ON 'ON
'Nb'.) ''ON 'ON
345
-------
.I ..J_. Li\>L,l_,L-
i, ppm
ujdd 'f
346
-------
''ON 'ON
0,, |,| m
'NUd ''CM'ON
347
-------
'.NUd '!OM 'ON
'NBJ ''ON 'DM
0,. ppm
lucid -f
348
-------
0«. PFitl
ppm
Oj, ppi
UJdd 'N'l'd '
-------
Oi. ppm
'Jdd 'NUd ''ON 'ON
'fjyj -ION 'ON
Oi. ppm
>^dd 'NUd ''ON'ON
-------
0,. PPM
pptn
Wdd 'NUd ''ON 'ON
0,. PPM
CO CO
'WON
WJd 'NUd '!ON 'ON
351
-------
Oi. ppm
oooo
'NBd ''ON 'CM
'NUd ''ON 'ON
Oi. ppm
'NUJ ''OM 'ON
352
-------
1_4_L_1.J 4_J L.J 1—J |_J_4—1--1—.1 I
0,. ppm
'N:Ud ''ON 'ON
ujdd -Nud ''OH 'ON
Os. ppm
!CN 'CN
353
-------
~rn~rrrr
Q», p pm
LoJ_l-l-I_i_lJ_lJ_l-l J.l-l-i.4-1
—• *— o
-Ofj
u-"'d 'NUd ''ON 'ON
354
-------
Oj, ppm
Oi. ppm
'!DM 'ON
UJdd 'l!Ud '"ON 'ON
ppm
""I" 'NBd ''ON 'ON
'NUJ ''ON 'ON
355
-------
_LLj_J , I • LJ : •. U__.x.:
Oi. ppm
O IO O
QOOcnOOOClOO C3O
OC3OOOOOC3CDCDO
-1OM 'OM
"^d 'NUd ''ON 'ON
OCj
-------
>. ppm
e> to o 1/7 o
O O 00
. I .'
uO «*- OJ C3
O CD O O
CD Q CT O O
'NUd '"ON 'ON
1 ^ ,
t + U
j '. J ; j:
OOC3C3C?C_3C3O
'NUd ''ON 'ON
357
-------
0,, pptn
C* CJ Ci
-'o,M 'OM
'NUd ''ON 'ON
uidd 'f,y
''ON 'ON
358
-------
0.. ppm
'Ndd ''ON 'OH
0,. p[,rn
Oj. ppm
''CN 'ON
'ON
359
-------
ppm
CO CO CM OJ
'NUJ
'ON
03. ppm
CT, LQ
-------
Oi, ppm
'NUJ ''CM 'OH
t"
-------
ppm
O CJ O O
'NUd ''ON 'ON
ujdd
''ON 'ON
0>, ppm
''ON 'OM
'NUJ '!OM 'ON
362
-------
Oi. ppn
Oi, ppm
CO CO OJ CNJ *-t f-*
uidd, '
"ON 'ON
0>, ppm
O C3 O C_ O Q i^> CD
-r-prpr rrj ,- p"t i
cS>
OOOOC3OO
d 'NbM "ON 'ON
-'ON 'ON
363
-------
0,. pf.ii
ppm
,-T-r
I . I . I . I . I . I . I , I
NHd ''ON 'ON
'NUd ''ON 'ON
ppm
«ii 'NUd '-DM 'ON
'NUd ''ON 'ON
364
-------
Oi. ppm
CO O C J OJ
l/j 0> LO
1JON 'ON
0>. ppm
Oi. ppm
'»CfJ 'ON
'NUd ''ON 'ON
365
-------
0%. ppm
''ON 'ON
'NUd ''ON 'ON
ppm
''ON' 'ON
'NUd ''ON 'ON
36fi
-------
0.. ppm
CO CO CNJ OJ
uiad 'Ndd ''ON 'CM
'NHd ''ON 'ON
'NtlJ ''ON 'ON
"ON 'ON
3P7
-------
0). ppm
». ppm
CV O O3
_ o
' I ' 1 '
'Ndd ''ON 'ON
''ON -CN
Os.
CO C\J CM CM
'!C.M 'CN
'NUd ''ON 'ON
-------
s. ppm
Ot, ppm
'NUJ ''ON 'UN
ujdd
''ON 'ON
Os. ppm
CD C^> O O O
'NUd ''ON 'DM
3*9
-------
0,. PPM
CM —• —•
o o o o o
i hh I oil 1.1.1 ill h I i h iLJ
•*- CO OJ •-• O C7) CO |— vO \J~) *f CO
-JOQOOOC3OOOO
Wdd 'Ndd "ON 'ON
'Ndd ''ON 'ON
0,. PPM
Wdd 'NUd 'BON 'ON
LoJ-d-l LLJ J.ljJ_Ll-LLLLd_l_L J-lJ-L.1
Wdd 'KUd '!ON 'ON
370
-------
0,. PPM
CO ti 'ON
371
-------
0,. PPM
0,, ppm
Wdd 'NUd '"ON 'ON
''ON 'ON
0,. PPM
0,. PPM
-..J-L.L.L.I I I |J_I I I
VJdd 'NUd MON 'ON
Wdd 'NUd ''ON 'ON
372
-------
Ot, ppm
a '!ON 'ON
'NUd '"ON 'ON
ppm
'Ndd ''ON 'ON
-SON 'ON
373
-------
TECHNICAL REPORT DATA
(Please read Instructions on the "'verse before completing!
1. REPORT NO.
EPA-450/4-81-034
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Effects of Chemistry and Meteorology on Ozone Control
Calculations Using Simple Trajectory Models and the
EKMA Procedure
5. REPOR r DATE
November 1981
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8 PERFORMING ORGANIZATION REPORT NO.
E. Jeffries, K. G. Sexton, Cc N. Salmi
9. PERFORMING ORGANIZATION NAME AND ADDRESS
University of North Carolina
Department cf Environmental Sciences and Engineering
School of Public Health
Chapel Hill, North Carolina 27514
12 SPONSORING AGENCY NAMF AND ADDRESS
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
10. PROGRAM ELEMENT NO
11 CONTRACT/GRAN"1" NO.
68-02-3523
13. TYPi. OF BFPOHT AND PERIOD COVERED
!14 faPOMSORlNG AGENCV CODE
15 SUPPLLMT-NT ARY NO") t£
EPA Project Officer: Edwin L. Meyer
16 ASS r
Three chemical kinetics mechanisms (Carbon Bond II, lumped species mechanism used
in Demerjian and Schere's photochemical box model and the Cal Tech mechanism developed
by McFae) were used in a modified version of the OZIPP model to replicate smog chamber
data in which automotive exhaust was irradiated (Bureau of Mines data). Two of the
mechanisms (CBII and Demerjian) agreed with the data, as well as the existing mechanism]
in OZIPP (Dodge propylene/butane). The OZIPP model was next used to simulate several
days of observations made during the St. Louis RAPS. In several cases, it was found
that the meteorological input and/or assumptions did not allow an adequate basis to
compare the mechanisms. Three days in which the meteorological input appeard appro-
priate were simulated using OZIPP with the Dodge, Demerjian and CBII mechanisms.
Differing peak ozone and control estimates were obtained, depending on the day simulated
and choice of mechanism. Sensitivity tests were employed to explore observed dif-
ferences among the mechanisms tested.
17.
a
DESCRIPTORS
KEY WORDS AND DOCUMENT ANALYSIS
I-1 IDENTIFIERS/OPEN ENDED TtHMS
C. COSATI 1 Icld/Utoup
Ozone
Photochemical models
Chemical kinetics mechanisms
OZIPP
EKMA
Sensitivity studies
18 DISTRIBUTION STATEMENT
19 SECURITY CLASS (This h'epori.
Unlimited
21 NO. OF PAGES
393
20 SECURITY CLASS (This page)
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDIT'
F- nvironmental rroiecnon gency.
Region V, Library * '
230 South Dearborn Street
Chicago, Illinois 60604
-------