United States
Environmental Protection
Agency
Environmental Sciences Research
Laboratory
Research Triangle Park NC 27711
EPA 600 3-79-050
May 1979
Research and Development
Smog Chamber
Validation Using
Lagranglan
Atmospheric Data
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ECOLOGICAL RESEARCH series. This series
describes research on the effects of pollution on humans, plant and animal spe-
cies, and materials. Problems are assessed for their long- and short-term influ-
ences. Investigations include formation, transport, and pathway studies to deter-
mine the fate of pollutants and their effects. This work provides the technical basis
for setting standards to minimize undesirable changes in living organisms in the
aquatic, terrestrial, and atmospheric environments.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/3-79-050
May 1979
SMOG CHAMBER VALIDATION
USING LAGRANGIAN ATMOSPHERIC DATA
by i
Charles Eugene Feigley
Harvey E. Jeffries, and
Myra A. Carpenter
Department of Environmental Sciences and Engineering
University of North Carolina at Chapel Hill
Chapel Hill, North Carolina 27514
Grant Number 800916
Project Officer
Bruce W. Gay, Jr.
Atmospheric Chemistry and Physics Division
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the U.S. Environmental Protection
Agency, and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the U.S. Environmental
Protection Agency, nor does mention of trade names or commercial products
constitute endorsement or recommendations for use.
ii
-------
ABSTRACT
A method was developed for validating outdoor smog chamber experiments
as a means of determining the relationships between oxidant concentrations
and its precursors - hydrocarbons and nitrogen oxides. When chamber
experiments were performed in a manner that simulated relevant meteorological
processes and precursor concentrations, the validation method showed that
photochemical smog reactions observed in the smog chamber generally agreed
with data from the Los Angeles Reactive Pollutant Program (LARPP). The LARPP
data consist of detailed airborne and ground level pollutant and meteorological
measurements.
iii
-------
iv
-------
CONTENTS
Abstract ill
Figures vil
Tables , xiv
Abbreviations and Symbols xvi
Acknowledgments xvii
1. Introduction 1
Photochemical smog chemistry 3
Trends of photochemical pollutants in the
atmosphere 11
2. Conclusions 16
3. Recommendations 22
Precursor central requirements 22
Further experimental simulations of Operation 33. . 22
Numerical modeling 23
Field studies 23
4. Methods of Relating Oxidant to Precursors 24
The EPA assessment of control requirements .... 25
Other methods of assessing the effects of
controlling oxidant precursor emissions 32
Conclusions 45
5. The Los Angeles Reactive Pollutant Program 47
General description 47
Analysis of Operation 33 48
Method of comparison of smog chamber results with
LARPP data 93
6. Experimental Methods and Results 123
Methods 123
Results 133
7. Discussion 181
Solar radiation 181
Atmospheric inhomogeneity ... 184
Temperature 196
Discussion of Comparisons 199
References 211
-------
Appendices
A. Flag field messages for Operation 33 220
B. Source reconciliation and associated calculations . . . 223
C. Numerical computation of vertical flux along a LARPP
trajectory 226
D. Calculation of k. from theory 233
E. The interactive effects of inhomogeneity and
atmospheric processes on photostationary state income . 237
F. Correction of data for the dark reaction of NO and 0»
during sampling 250
vi
-------
FIGURES
Number
1 Important reaction cycles In photochemical smog 10
2 A typical diurnal pattern of photochemical smog
constituents at a stationary monitoring site 11
3 Freeway and non-freeway emissions 14
4 Annual trends of photochemical oxidant in three U.S.
cities 15
5 Maximum smog chamber oxidant corresponding to observed
and projected NMHC, 1960-1990 35
6 Oxidant/0,. isopleths derived from combined use of smog
chamber and photochemical modeling techniques 43
7 Typical LARPP helicopter pattern from (a) Lagrangian
(moving with mean wind) viewpoint, and from (b) Eulerian
(fixed point and ground) viewpoint, in a 1.5 m. sec
mean wind 49
8 Tetroon trajectories for LARPP Operation 33 50
9 Tetroon altitude above ground level versus time 52
10 Example of concentration-position figure for carbon
monoxide 61
11 Concentration-position figures for Operation 33 carbon
monoxide before 1030 PST 63
12 Concentration-position figure for Operation 33 carbon
monoxide after 1030 PST 64
13 Concentration-position figures for Operation 33 nitrogen
oxides before 1030 PST 65
14 Concentration-position figures for Operation 33 nitrogen
oxides after 1030 PST 66
vii
-------
Number Page
15 Concentration-position figures for Operation 33 nitric
oxide before 1030 PST 67
16 Concentration-position figures for Operation 33 nitric
oxide after 1030 PST 68
17 Concentration-position figures for Operation 33 ozone
before 1030 PST 69
18 Concentration-position figures for Operation 33 ozone
after 1030 PST 70
19 Helicopter pattern average nitric oxide concentrations
for Operation 33 73
20 Helicopter pattern average nitrogen oxides concentra-
tions for Operation 33 74
21 Helicopter pattern average scattering coefficients for
Operation 33 75
22 Helicopter pattern average carbon monoxide concentra-
tion for Operation 33 76
23 Helicopter pattern average temperatures for Operation
33 77
24 Helicopter pattern average dew points for Operation 33 78
25 Helicopter pattern average nonmethane hydrocarbon con-
centrations for Operation 33 79
26 Helicopter pattern average methane concentrations for
Operation 33 80
27 Helicopter pattern average ozone concentrations for
Operation 33 81
28 Helicopter pattern standard deviation of nitric oxide
concentration 84
29 Helicopter pattern relative standard deviation of
nitrogen oxides concentrations 85
30 Helicopter pattern relative standard deviations of
scattering coefficients 86
31 Helicopter pattern relative standard deviations of
carbon monoxide concentrations 87
viii
-------
Number Page
32 Helicopter pattern relative standard deviations of
temperatures 88
33 Helicopter pattern relative standard deviations of dew
points 89
34 Helicopter pattern relative standard deviations of
nonmethane hydrocarbon concentrations 90
35 Helicopter pattern relative standard deviations of
methane concentrations 91
36 Helicopter pattern relative standard deviations of
ozone concentrations 92
37 Mixing depth estimated for the LAKPP Operation 33
trajectory 105
38 Carbon monoxide vertical concentration profiles for
Operation 33 at one-hour intervals estimated by
numerical solution 106
39 Estimated carbon monoxide flux at altitudes of 150
feet and half the mixing depths from numerical
solution 107
40 Zero gradient nonmethane hydrocarbon concentrations in
the absence of reaction as estimated from source re-
conciliation results and regression line 110
41 Comparison of nonmethane hydrocarbon. Concentration
line from source reconciliation and nonmethane hydro-
carbon concentration as measured by environmental
chromatograph Ill
42 Nonmethane hydrocarbon vertical concentration profiles
for Operation 33 at one-hour intervals estimated by
numerical solution 113
43 Estimated nonmethane hydrocarbon flux at altitudes of
150 feet and half the mixing depth from numerical
solution 114
44 Logic for calculating injection and dilution rates
from flux and mixing height estimates 117
45 Nonmethane hydrocarbon injection rates for simulation
ofLARPP Operation 33 119
ix
-------
Number
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
Chamber dilution rates for simulation of LARPP
Cumulative nonmethane hydrocarbon injection rates for
simulation of LARPP Operation 33
Cumulative dilution volume for simulation of LARPP
Operation 33
Schematic diagram of UNC outdoor smog chamber dilution
NO, NO^, and 0_ profiles from static experiment of
September 1, 1976
Solar radiation and ultraviolet radiation for
September 1, 1976
NO, N0», and 0« profiles from simulation experiment of
September 6, 1976
NO, NO-, and 0., profiles from static experiment of
September 6, 1976
Solar radiation and ultraviolet radiation from
September 6, 1976
NO, N02, and 0., profiles from simulation experiment of
September 25, 1976
NO, NO, and 0, profiles from static experiment of
September 25, 1976
Solar radiation and ultraviolet radiation from
September 25, 1976
NO, N0?, and 0» profiles from static experiment of
October 4, 1976
NO, N0?, and 0_ profiles from static experiment of
October 4, 1976
Solar radiation, ultraviolet radiation, and NO
NO, NO-, and 0_ profiles from simulation experiment of
October 6, 1976
NO, NO, and 0_ profiles from static experiment of
October 6, 1976
Page
120
121
122
129
135
136
138
139
140
142
143
144
146
147
148
150
151
X
-------
Number Page
63 Solar radiation, ultraviolet radiation, and N0_
photolysis rate constant from October 6, 1976 152
64 NO, NO-, and 0- profiles from static experiment of
October 11, 1976 154
65 Solar radiation, ultraviolet radiation, and NO-
photolysis rate constant from October 11, 1976 .... 155
66
j ~ * " -
157
NO, NO-, and 0- profiles from simulation experiment of
October 12, 1976
NO, NO-, and 0_ profiles from static experiment of
October 12, 1976
68 Solar radiation, ultraviolet radiation, and NO-
photolysis rate constant from October 12, 1976 .... 159
NO, N0_, and 0- profiles from simulation experiment of
October 13, 1976
69
* j * <
161
70 NO, N0«, and 0, profiles from static experiment of
October 13, 1976 162
71 Solar radiation, ultraviolet radiation, and N0_
photolysis rate constant from October 13, 1976 .... 163
72 NO, NO-, and 0- profiles from static experiments of
October 15, 1976 165
NO, NO-, and 0- profiles for simulation experiment of
October 16, 1976
73
' i ~ * ~ ~
167
74 NO, NO-, and 0- profiles for October 16, 1976, red
side . . . . . 168
75 Solar radiation and ultraviolet radiation for October
16, 1976 169
76 Comparisons of NO, N0_, and 0- profiles for October 12
simulation and LARPP zero-gradient region 175
77 Comparison of NO, NO-, and 0- profiles for October 13
simulation and LARPP zero-gradient region 176
78 Comparison of NMHC source reconciliation regression
line with NMHC mix concentrations estimated from
n-pentane peak for simulations of October 12 and 13. . 177
xi
-------
Number Page
79 0_ concentrations at 1400 hours and maximum produced
versus initial NMHC concentration during static exper-
iments on clear days 180
80 Comparison of Mt. Disappointment UV radiometer meas-
urements with theoretical calculation of UV radiation
from 290 to 390 nm above the boundary layer in Los
Angeles on November 5, 1973 183
81 Theoretical estimates of k-, the rate constant of N02
photolysis at four altitudes above ground level in Los
Angeles on November 5, 1973 188
82 Theoretical estimates of k-/k«, the ratio of the NO-
photolysis rate constant to the rate constant of the
N0-0_ reaction at four altitudes above ground level in
Los Angeles on November 5, 1973 186
83 Two hypothetical parcels of air at photostationary
state at ^/k = 0.016 ppm 188
84 Estimated helicopter pattern average covariance of NO
and 0~ from Operation 33 193
85 Estimated helicopter pattern average ratio of covari-
ance of NO and 0, to the NO concentration 195
86 Comparison of NO, NO , and 0 profiles from static
propylene experiments in UNC outdoor chamber 198
87 PKSS model of 0.22 ppm NO , 2.00 ppmC HC mix for three
temperature profiles . . * 200
88 NO, NO-, and 0., helicopter pattern average concentra-
tions from LARPP Operation 33 202
89 Comparison of ratio of NO. photolysis rate constant to
NO-0- reaction rate constant 204
90 Comparison of the chamber temperatures on October 12
and 13 with range of helicopter pattern average temp-
eratures from LARPP Operation 33 207
C-l Region of solution 226
C-2 Solution-space indexing 228
E-l Data for example E-l and summary 238
xii
-------
Number Page
F-l Measured and corrected ozone for October 12 simulation 252
F-2 Measured and corrected NO for simulation of October 12 253
F-3 Measured and corrected N00 for October 12 simulation . 254
xiii
-------
TABLES
Number Page
1 Reactions and Rate Constants for the System
NO-N02-N2-02 in Sunlight 5
2 Selected Reactions of the NO -H-0-CO-Air Photo-
chemical System 6
3 Other Methods of Relating Oxidants to Precursors and
the Sources of Data for Each Method 33
4 Files Produced by Program UNMERGE and the Number of
Records Transferred from the Operation 33 Archive to
Each File Type 54
5 Some Programs Developed for Examination of LARPP Data . 56
6 Contents of Helicopter Data Records and Sources of Data 57
7 Weight Fractions of Tracer Species in Major Hydrocarbon
Sources in the Los Angeles Area 98
8 Nonmethane Hydrocarbon Distributions for Automotive
Exhaust, Gasoline, Gasoline Vapor, Commercial Natural
Gas, and Geogenic Natural Gas 99
9 Bag Sample Time and Altitude Estimated Weight Fraction
of NMHC from Each Source Category and R from Source
Reconciliation 101
10 Comparison of CO Flux Estimates from Panofsky and from
Numerical Flux Computation 108
11 Pollutant Injection Data 127
12 Hydrocarbon Mix Composition Concentration of Compound,
PpmC 128
13 Summary of Conditions and Results of Static Runs. . . . 178
xiv
-------
Number Page
14 Estimates of N02/N0 Ratio for Simulations of October
12 and 13 and the Effective N(>2/NO Ratio for the LARPP
Data Corrected and Uncorrected for Inhomogeneity. . . . 209
A-l SMOG 2 Data Flag Messages for Operation 33 221
A-2 SMOG 3 Data Flag Messages for Operation 33 222
C-l Comparison of Flux Calculations Using Different Time
and Altitude Grid Spacing 232
E-l Results of Various Approaches to Estimate Photo-
stationary State 0. and k-j/ko 242
xv
-------
LIST OF ABBREVIATIONS AND SYMBOLS
ABBREVIATIONS
A angstroms
CAMP Continuous Air Monitoring Program
hr hour
in. inch
1pm liter(s) per minute
LAAPCD Los Angeles Air Pollution Control District
LARPP Los Angeles Reactive Pollutant Program
m meter(s)
min minute(s)
ml milliliter(s)
mm millimeter(s)
nm nanometer(s)
0(u) single activated state of atomic oxygen
0(3P)
PAN
ppb
pphm
ppm
ppmC
sec
SYMBOLS
CH2°
H
HC
HCO
HC002
HN0
HO
H02
HONO
HONO,
triplet activated state of atomic oxygen
peroxyacetyl nitrate
parts per billion
parts per hundred million
parts per million
parts per million carbon
second(s)
methylene radical
methane
formaldehyde
hydrogen radical
generalized
hydrocarbon
formyl radical
peroxyformyl radical
nitrous acid
nitric acid
hydroxyl radical
hydroperoxy radical
nitrous acid
pernitric acid
HOONO,
4
H2°
hv
NMHC
NO
N03
NO
x
°2
°3
RC03
RCHO
pernitric acid
water
Planck's constant times
the wave length of light
nonmethane hydrocarbon
nitric oxide
nitrogen dioxide
nitrogen trioxide
oxides of nitrogen
nitrogen pentoxide
oxygen
ozone
generalized peroxyacyl
radical
generalized aldehyde
xvi
-------
ACKNOWLEDGEMENTS
Thanks are extended to Dr. Donald Fox for helping define this
project initially and for his support through the final stages of this
research. Special thanks also are expressed to Dr. Basil Dimitriades,
Dr. Larry Kupper, Dr. Parker Reist, and Dr. Lyman Ripperton, and to
Richard Kamens for his help with the experimental work.
Also acknowledged is the assistance of the staff of the School of
Public Health, in particular, Mr. Frank Malcolm and his shop staff;
Dr. Robin Baker for his programming help; Ms. Laura Alexander; Ms. Cynthia
Crossen for her keypunching; and to Ms. Shirley Milton for typing the
final copy.
xvii
-------
SECTION 1
INTRODUCTION
Until the late 1940s, most polluted air was characterized by a
chemical reducing power because of its sulfur dioxide content. The
sulfur dioxide was released to the air mainly as a consequence of burn-
ing sulfur-bearing fossil fuels. It was indeed surprising then that Los
Angeles air was found to contain materials with oxidizing ability. The
first hint of this property was the rapid cracking of rubber products
such as automobile tires. Ozone (0_) was known to cause a similar
cracking of stressed rubber. Ozone concentrations in ambient air were
measured by rubber cracking ability, but there was no known source of
ozone which might account for the concentrations found.
In 1953, Haagen-Smit et al. (1) demonstrated that synthetic mix-
tures of hydrocarbons and nitrogen dioxide in oxygen could crack rubber
upon irradiation with natural and simulated solar radiation. This was
the initial step toward understanding the mechanism of formation of
oxidants in polluted air. Soon after, Stephens et al. in 1956 (2)
confirmed the presence of ozone in irradiated mixtures of nitrogen
dioxide (N0_) and hydrocarbons in air, as well as in ambient southern
California air, by long-path infrared spectroscopy. Since then the
mechanisms of oxidant synthesis in polluted atmospheres have been dis-
cussed in detail: Leighton (3); Wayne (4); Haagen-Smit and Wayne (5);
Pitts (6); Altshuller and Bufalini (7); and Demerjian, Kerr and Calvert
(8).
In the past three decades, certain very obvious effects of smog
have been noted: (1) eye irritation and lacrimation probably due to the
formation of species such as acrolein, peroxyacetyl nitrate (PAN), and
peroxybenzoyl nitrate; (2) the phytotoxicity of ozone, N0_, PAN, and
-------
aldehydes; (3) rubber cracking caused by 0_; (4) reduced visibility due
to NO- and photochemically produced aerosols.
The Clean Air Act, as amended in 1967, required the Administrator
of the Environmental Protection Agency to prepare air quality criteria
documents for certain pollutants. These documents were intended to
explore the relationship between the presence of a pollutant in the air
and any identifiable effects on health and welfare, based on the latest
scientific knowledge. The Administrator was allowed up to one year and
30 days after enactment (of the Act) to prepare the criteria documents.
Then, based on these criteria, national ambient air quality standards
were promulgated (9). Primary standards "are those which, in the judg-
ment of the Administrator, based on the air quality criteria and allow-
ing an adequate margin of safety, are requisite to protect the public
welfare from any known or anticipated adverse effects associated with
the presence of air pollutants in the ambient air" (9).
The air quality standards for photochemical oxidants, both primary
and secondary were set at 0.08 parts per million (ppm): a 1-hr average
concentration not to be exceeded more than once per year. This is
essentially a standard for 0, since the "reference" method for measuring
photochemical oxidants is the ethylene chemiluminescent technique which
is specific for 0-. The standard for non-methane hydrocarbons (NMHC)
was set at 0.24 ppm: an average from 6:00 to 9:00 a.m. not to be ex-
ceeded more than once per year. The hydrocarbon standard is not based
on direct effects on health or welfare, but it is intended "for use as a
guide in devising implementation plans to achieve oxidant standards"
(9).
The methods used to arrive at the NMHC standard and the emission
reductions required to meet the standard have been criticized from many
quarters. It has been suggested that other methods or sources of data
be utilized to reevaluate the standard. One such source of data is
laboratory experimentation (smog chamber work) which simulates the
formation of smog. Use of this data base for developing a quantitative
relationship between oxidant and its precursors for control purposes is
just beginning since the extent to which these data were influenced by
-------
the artificial experimental conditions was not known. Nor could the
magnitude of these influences be determined because, until recently,
there was no data from the ambient air with which to compare smog cham-
ber data.
The purpose of this research was to study the suitability of data
from the Los Angeles Reactive Pollutant Program (LARPP) for use in
validating smog chambers. More specifically, the purpose was to perform
chamber validation experiments in the UNC outdoor smog chamber.
PHOTOCHEMICAL SMOG CHEMISTRY
General Discussion
Tropospheric 0_ is formed by a reaction cycle involving mainly
nitrogen oxides, hydrocarbons, and atmospheric oxygen. The chemical
species emitted directly to the atmosphere from a source are known as
primary pollutants, while those formed by reaction in the atmosphere are
known as secondary pollutants. Relative to other tropospheric con-
stituents, NO- is an extremely efficient absorber of light at wave-
lengths reaching the lower troposphere (290 to 415 nm). One possible
consequence of the absorption of a photon is the rupture of the NO- mole-
cule to form nitric oxide (NO) and atomic oxygen (0.). Most of the
extremely reactive oxygen atoms combine with molecular oxygen to form 0-
because of the high concentration of oxygen compared with the other
possible reactants such as NO, NO-, 0-, or hydrocarbons. Most of the 0,
produced reacts with NO to regenerate N0_.
N02 + hv -> NO + 0 (1)
0 + 02 + M -»- 03 + M (2)
03 + NO -> N02 + 02 (3)
The specific rate of reaction 1, k.., depends mainly on the inten-
sity and wavelength of light. Without reactions 2 and 3 to regenerate
NO-, Leighton (3) calculated that the half-life of NO- in typical sun-
light would be about 2 mins. While NO- absorbs light over most of the
ultraviolet and visible portions of the solar spectrum, not all absorbed
-------
photons are sufficiently energetic to photodissociate the molecule. The
quantum yield of photolysis is near one at wavelengths less than 3660
angstroms (A), continuously decreasing between 3800 A and 4047 A to near
zero at wavelengths greater than 4358 A. Non-dissociative absorption by
N0? results mainly in thermal quenching back to the ground state, but
some singlet molecular oxygen may be created by energy transfer during
collision between excited N02 and ground state oxygen molecules.
Reaction 2 may be thought of as a two-step process: the formation
of a high-energy 0 molecule and the collisional loss of energy which
stabilizes the 0 molecule. "M" is any third body which absorbs the
energy of reaction. This reaction is extremely fast; the half -life of
the oxygen atom in air is about 10 sec. (10).
Reaction 3 is also rather rapid (k_ is 24.2 ppm~ min~ at 298°K)
such that 0_ reaches a steady-state concentration in several minutes
under constant light intensity. When this steady state is reached in
the absence of other slower reactions there is a continuous conversion
of light into heat with no change in concentrations.
Of course a great many more reactions actually take place.
Demerjian, Kerr and Calvert (8) listed 101 reactions which involve only
inorganic species. Nevertheless, reactions 1, 2, and 3 are so rapid
that they control the 0» concentration. Some of the more important
reactions of the NO - NO - air system are shown in Table 1 (11) .
The calculation of a steady state 0_ concentration based on the
rates of reaction and the nitrogen oxides emitted into ambient air
yields 0 levels well below those observed. The enhanced 0, concentra-
-^ j
tions of the troposphere are primarily the result of hydrocarbon reac-
tions which oxidize NO to NO- without consuming 0_. The steady-state 0
concentration is given by
The effect of oxidizing NO to N0» without removing 0, is to raise the
NO^-to-NO ratio and to increase the steady-state 0,, concentration.
Thus, to explain high observed 0, levels in an active photochemical
-------
TABLE 1. REACTIONS AND RATE CONSTANTS
FOR THE SYSTEM NO-N02~N2-02 IN SUNLIGHT (11)
Reaction
Number
Reaction
Rate Constant at 298 K
1
2
3
4
5
6
7
8
9
10
11
N02 + hv -> NO + 0
NO
0 + N02 -» NO + 02
0 + NO, + M -* NO, + M
£ -J
NO
0 + NO + M -» N02 + M
2NO
2N0
depends on light
-5 -2 -1
2.33 x 10 ppm min
2.95 x 101 ppm"1 min"1
1.38 x 104 ppm"1 min"1
3 2 1
4.50 x 10 ppm min
1.48 x 10 ppm min
-3 -2 -1
2.34 x 10 ppm min
7.62 x 10~10 ppm"2 min"1
4.43 x 103 ppm"1 min"1
1.38 x 101 min"1
4.6 x 10 ppm min
system, one must explain the conversion of NO to NO which is necessary
for 0- buildup.
The Air-H^O-CO System
If, in addition to NO and N02, the air contains water (H_0) vapor
and carbon monoxide (CO), other reactions become possible which may
partially explain the oxidation of NO. While many reaction pathways are
possible, the most important have four steps in common: (1) the forma-
tion of hydroxyl radical, HO; (2) reaction of HO to form H09, the hydro-
peroxy radical; (3) oxidation of NO to N02 with regeneration of HO; and
(4) recombination and other termination reactions.
HO, unlike many atoms, hydrocarbon and alkoxy radicals, does not
react with molecular oxygen; and thus its concentration may rise high
-------
TABLE 2. SELECTION REACTIONS OF THE NO -H_0-CO-AIR
PHOTOCHEMICAL SYSTEM (8) X
Reaction Number
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Reaction
Initial Formation of HO
0_ + hv (2900 A° 20H
z
N02 + NO + H20 -> 2HONO
HONO + hv (X<4000A°) -* HO + NO
N205 + H20 j 2HON02
HONO., + hv (A<3300A°) -> HO + N02
Initial Formation of HO
HO + HO + M + H202 + M
HO + H202 > H20 + H02
N03 + H202 -* HON02 + H02
HO + CO ^ C02 + H
H + 02 + M -» HO + M
Oxidation of NO
H02 + NO -> NO + HO
Recombination and Termination Reactions
HO + N02 + M -» HON02 + M
HO + HO + M -> H202 + M
HO + HO ^ H20 + 0
H02 + N03 -> 02 + HON03
H°2 + °3 "*" H° + 2°2
H02 + H02 -» H202 + 02
H02 + OH -» H20 + 02
-------
enough to react with other trace gases. The rates of reactions involv-
ing nitrous and nitric acid, reactions 14 and 16, may be affected by a
significant heterogeneous contribution, thus the rate constants used for
these reactions in modeling smog chamber data are often considered to be
adjustable parameters.
Demerjian, Kerr, and Calvert (8) ran simulations at various initial
concentrations of H~0 and CO. In a simulated system initially contain-
ing 5 parts per hundred million (pphm) of NO and 5 pphm NO- the major
reactions of HO after 60 min are with NO (62.9 percent) and NO- (34.2
percent) to form HONO and HONO-. When CO is added, the significance of
reaction 21 increases drastically over a range of initial CO concentra-
tions from 1 to 100 ppm. The 0_ produced after 151.6 min by these
systems also increases: from 0.40 pphm to 7.49 pphm with 100 percent
relative humidity and from 0.42 pphm to 6.88 pphm at 50 percent relative
humidity. Without CO, the 100 percent and 50 percent simulations pro-
duced 0.17 pphm and 0.24 pphm of 0,. The concentration of various
intermediates was also greatly influenced by increasing initial CO.
Over the range considered, [0( P)] and [0( D)] increased by an order of
magnitude; [NO.,] increased by a factor of 500; [H0~] increased by a
7
factor of 10 and [HO] was reduced by a factor of 10. Thus, in inor-
ganic systems, carbon monoxide's ability to serve as an electron donor
to the hydroxyl oxygen leads to the formation of H00, the oxidation of
NO, and the increased photostationary-state 0- concentration. In the
atmosphere, however, CO concentrations are not sufficient for reactions
21 and 22 to account for the 0- concentrations observed. It is nec-
essary to consider the effect of hydrocarbon reactions for such an
explanation.
The Role of Hydrocarbons
From examination of the structure of hydrocarbons commonly found in
urban air, it may be inferred that in general they are more vulnerable
to electrophilic attack than CO. The rate constant for the reaction of
HO with CO is about 250 ppm min , the rate constants of studied
reactions of HO with atmospheric hydrocarbons are all higher than the CO
-------
rate constant with the exception of the methane rate constant. The rate
constants for some typical alkanes, olefins, aldehydes, and acetylene
were listed by Seinfeld (12). The reactivity of alkanes with OH gen-
erally increases with molecular weight and complexity: methane, 16.4
ppnf1 min ; ethane, 443 ppnf min~ : n-butane, 1500 ppm min ; cyclo-
hexane, 12,000 ppm min~ . Most rate constants for alkenes and alde-
hydes are about 2 x 10 to 3 x 10 ppm" min" . The rate constant for
1 1
acetylene is 1470 ppm" min . Hampson and Garvin (13) cite rate con-
stants for the reaction of HO with two aromatics, benzene and toluene;
3 -1 -1
they are greater than 10 ppm min . Thus, in numerical simulations
of urban atmospheres most of the HO reacts with hydrocarbons rather than
CO or other organic species. The most significant result of the oxida-
tion of hydrocarbons by HO is the formation of organic radicals which
may directly oxidize NO to N0?, and/or produce H02 radicals. H0_ may
then, in turn, oxidize NO and regenerate the hydroxyl via reaction 23.
Such a sequence of cyclical reactions is known as a free radical chain
reaction. HO is called a chain carrier. Individual reactions involved
in chain reactions may be grouped according to their function in the
overall system. The three typical functions in a chain reaction are
initiation, propagation, and termination.
The hydrocarbon free radical chain may be initiated by reactions
which form HO or HO , such as reactions 12 through 22 in Table 2. In
the presence of organics the following reactions are among those that
may initiate the chain:
03 + olefins -> nH02 + other products (31)
CH20 + hv -» H + HCO (32)
HCO + 02 + CO + H02 (33)
H + 02 + M -> H02 + M (34)
RCHO + hv -> R* + HCO (35)
where R represents an arbitrary organic carbon-hydrogen group.
After the cycle has been initiated, propagation reactions oxidize
hydrocarbons and NO. The major reaction cycles in photochemical smog
-------
are shown in Figure 1 (14). A characteristic of propagation reactions
is that for each HO molecule reacting with hydrocarbons more than one
molecule of H02 may be produced. As shown in Figure 1, H0_ may then
oxidize NO and regenerate HO. The concentrations of HO- and HO increase
very rapidly after initiation. They would increase to rather high
levels if not for consumption by termination reactions such as 24 through
30 in Table 2 that result in the formation of more stable chemical
species. In addition to mediating the propagation of the chain, termin-
ation reactions form several of the secondary species which are char-
acteristic of photochemical smog such as peroxyacyl nitrates (PAN) and
nitric acid.
RC03 + N02 * PAN (36)
OH + N02 -»- HN03 (37)
Gas-to-Particle Conversion
Another aspect of smog chemistry is the formation and growth of
aerosol in photochemical smog. This area is not as well understood as
the gas-phase reactions discussed above; however, much has been learned
in recent years. About one half the particulate mass during periods of
moderate and heavy smog in Los Angeles is secondary material, i.e., the
result of gas-to-particle conversion processes (15). The particle mass
loading increases during the day, and distribution of material on a
number basis shifts toward larger particle sizes in the ambient air (15)
and in experimental studies (16). Hasar et al. (17) have demonstrated
that coagulation cannot account for observed distribution shifts from
smaller to larger particle sizes. Thus, particle growth by gas-to-
particle conversion must be postulated (15, 17, 18).
In the work of Hasar et al. (17), homogeneous nucleation was ob-
served in filtered Pasadena air exposed to sunlight, indicating the
formation of secondary, condensable substances. Heisler (18) found that
the growth of existing aerosol in a system of ambient Pasadena air with
added NO, N0~, and an organic could only be explained by a mechanism of
formation of secondary materials in the gas phase and subsequent dif-
fusion to the aerosol surface. The aerosol growth was found to be
-------
MlN''p+nV »Kli^
0-+ H00+ hv <-OH N
[N0]
2 . .
:>+ olefin y-ald + y H00 -,
02 12,
0 ^ 2
iij, 4 .
W>2+NQ-*"N02+OH-N
/NO\
hv f "~ *" >i fMOl
^NO, »-NO + 0, l
2 02 J , v[NO].
T
. . °3* ^
**OH+ olefin ».RO Y +
x 1 , 1 +2 12
, i
>OH+ paraffins-*.R'Ox t
i
1 i
-»OH+ a 1^ » R"0 f '
0 »-R"0 HajNO-vajNOg+ald'+e-jHO.-
HNO _».| T 1
^ un T nUrt^^^nliU- ^! 1 1
it 3
Figure 1. Important reaction cycles in photochemical smog (14).
-------
controlled by the diffusion of a number of organic species to the parti-
cles. The organics identified in the condensed phase from olefin
experiments were carbon-hydrogen chains oxygenated on both ends of the
molecules. The functional groups found were acids, aldehydes, alcohols,
and nitrates. When dienes were studied, molecules were found in the
aerosol with the oxygenated functional groups on one end and a double
bond on the other end.
TRENDS OF PHOTOCHEMICAL POLLUTANTS IN THE ATMOSPHERE
Interpreting the Diurnal Pattern of a Fixed Monitoring Site
The diurnal pattern of the concentrations of both primary and
secondary pollutants as observed in polluted air is a readily recog-
nizable characteristic of photochemical smog. It is the result of the
combined trends of primary pollutant emissions and meteorological
factors, such as sunlight, dispersion, and advection. An example of
this diurnal pattern is shown in Figure 2 as observed from a stationary
monitoring site in downtown Los Angeles.
70
Figure 2. A typical diurnal pattern of photochemical smog constituents
at a stationary monitoring site.
11
-------
In order to interpret data from a fixed location, it must be recog-
nized that air arriving at the station at any particular time has a
unique history of emission, dilution, and extent of reaction. Attempts
to explain fixed-station data as if the air surrounding the station were
more or less homogeneous can lead to misconceptions about the processes
at work. At a fixed location in downtown Los Angeles it is known that
the mixing depth generally increases with time by turbulent entrainment
during daylight hours. The boundary layer turbulence is a result of
buoyancy which is, in turn, due to a positive surface heat flux. The
rate of entrainment is dependent upon the intensity of turbulent con-
vection (19). Also, emissions are greatest during the morning rush
hour. Given these two basic facts, it is tempting to accept them plus
some loss by reaction as a total explanation of the behavior of hydro-
carbons and nitrous oxides in Figure 2. Both these primary pollutants
reach a peak early in the morning and taper off rapidly afterward. From
a Lagrangian viewpoint the concentration of a pollutant observed at a
fixed station is some complex function of the emission rates along the
trajectory, the windspeed as a function of time, the dilution due to
increase in mixing depth, and chemical processes.
Meteorological Trends in the Los Angeles Basin
During the warmer monthsroughly April to Octoberthe climate of
the California coast is dominated by a major high pressure system cen-
tered several hundred miles offshore. The anticyclonic air movements
result in prevailing south-easterly winds along the southern California
coast. The subsidence associated with the high, and cooling at the
ocean's surface by upwelling currents near the shore produces very
stable lapse rates within the planetary boundary layer. In addition,
turbulence near the air-sea interface humidifies the air, forming a
"marine layer" which is about 1800 m deep near Honolulu and tapers to an
average of 450 m at the California coast. Thus, air moving onshore
generally consists of a humid lower layer capped by a temperature in-
version, with extremely dry air above the inversion base.
Effects of the local topography are also quite important in the Los
Angeles basin. In the night and early morning, cooling on the slopes of
12
-------
the Santa Monica, San Gabriel, and Santa Ana Mountains and the Palos
Verdes and Puente Hills sets up a flow generally toward the southwestern
quadrant; wind speeds are quite low and direction is variable. A con-
vergence zone is created at coastal areas since the nocturnal mountain-
valley and land-sea breezes are in opposition to the frontal circula-
tion. Roughly two hours after sunrise the flow regime is reversed with
marine air moving in over land. The marine layer is deepened and di-
luted with dry air along its land trajectory. Wind speeds increase
until late afternoon reaching an average of 2-3 m sec" .
Edinger (20) studied the modification of this air as it is affected
by wind speed, heat input, and position in the Los Angeles basin. His
observations are summarized as follows: the marine layer during day-
light hours (1) is shallow at the coast and deep inland, at any time;
(2) increases in depth early in the day and then becomes progressively
shallower during the afternoon at a fixed location; and (3) reaches its
maximum depth first along the coast and later inland. At any location
and time, the mixing depth and thus the depth of the marine layer is a
function of three variables: (1) convergence or divergence of the
horizontal wind averaged throughout the marine layer and over the coas-
tal plain; (2) convective mixing which modifies the temperature profile;
and (3) advection which determines how long the air has been over land.
(The mixing depth may not correspond exactly with the depth of the
marine layer, although they are usually nearly equal. Edinger (21)
described the interaction of marine air and the temperature inversion in
a developing boundary layer above the Santa Clara valley,)
Roth et al. (22) studied the temporal trends of motor vehicle
emissions. The percentage of the total daily traffic for each hour is
shown in Figure 3 (23) . Note that the emissions at noon are only about
20 percent less than the rush-hour peaks.
Data from a fixed station in downtown Los Angeles as in Figure 2,
shows the air reaching the station 2 to 3 hours after sunrise probably
contains mostly local emissions and the mixing depth is quite low since
surface heat flux is not large. As the heat flux goes up, the sea
breeze reverses direction bringing clean air onshore. Nevertheless, the
13
-------
tO
o
O 8
x
ee
UJ
0.
U
It 6
<
O
U.
O
i-
z
FREEWAYS
NON-FREEWAYS
2400 0400
0800 1200 1600 2000 '2400
TIME OF DAY
Figure 3. Freeway and non-freeway emissions.
low wind speed provides substantial time over source areas before the
air reaches the station. In the later morning, wind speed is increas-
ing, source strengths decreasing, and the mixing height nearing its
maximum for that location; thus, primary pollutant concentrations are
dropping off. For reactive species, the chemical consumption should
also be significant by this time. After 1200, the concentrations con-
tinue to drop due to the increasing wind speed and consumption by
reaction. As wind speed increases the amount of time spent over source
areas decreases, as does the amount of time for photochemical reaction.
Thus, in late afternoon the 0_ concentration is almost always higher at
the eastern reaches of the basin than in the downtown area.
Qxidants in polluted atmospheres often show annual trends that are
directly related to the seasonal variability of meteorological param-
eters. Figure 4 shows typical annual cycles for Los Angeles, Denver,
and Phoenix (24). Maximum concentrations are found in the summer and
fall months.
14
-------
0,20
0.16
OL
Q.
I °'12
Si
(T
ui 0.08
o
z
o
o
0.04
LOS ANGELES
1964-1965
JAN 1967 -JUNE 1969^
F M
0 N
A M J J A S
MONTH
Figure 4. Annual trends of photochemical oxidant in three U.S. cities.
-------
SECTION 2
CONCLUSIONS
Photochemical smog results from an extremely complex system of
reactions in the ambient air. Primary pollutants, mainly nonmethane
hydrocarbons and nitrogen oxides, in the presence of adequate sunlight
react to form secondary pollutant species. Oxidant (measured as 0« only)
and N0? are two secondary species for which EPA has promulgated national
ambient air quality standards. The control of 0~ and NO- must be based
on reducing emissions of primary pollutants. One of the most difficult
and controversial problems in implementing controls for these pollutants
has been the determination of the degree of emissions reduction of
primary pollutants required to assure compliance with the air quality
standards for oxidant. The many techniques employed to relate precursor
emissions to the resulting oxidant concentrations fall, in general, into
three categories: cause-and-effect, correlative (or statistical), and
mechanistic. Correlative techniques suffer from the reliance on present-
day data; for determining control requirements, such techniques must be
extrapolated beyond their data base. The mechanistic, or modeling,
approaches have great promise for control strategy applications. Their
major difficulty has been the great complexity of the chemical and
meteorological processes, forcing the use of many simplifying assump-
tions. Also, kinetic mechanisms require validation and the only source
of data for validation is cause-and-effect type studies. Cause-and-
effect relationships are those that result from observation of reaction
systems from the initial introduction of precursors to the formation of
oxidant or other products, and they are the only conceptually valid
means of relating oxidant to precursors at this time.
Two sources of cause-and-effect data are Lagrangian monitoring
studies and smog chamber experiments. Both of these sources have
16
-------
certain practical difficulties. Lagrangian monitoring is expensive and
the turbulent diffusion of material must be accounted for. Smog chamber
experiments may be affected by the processes occurring on chamber walls,
by the spectral distribution of the light source, and by the absence of
a diurnal light profile, in some cases. One method of relating oxidants
to precursors, that of Dimitriades, uses cause-and-effect smog chamber
data but avoids some of the problems by assuming that the atmosphere's
ozone isopleths on the NO -NMHC plane are proportional to 0, isopleths
X j
from the chamber modified to reflect conditions of the ambient air.
Both modeling and smog chamber methods require that the reactions in the
smog chamber are not greatly different from those in the atmosphere.
Validation of kinetic mechanisms for use in smog models depends in a
quantitative fashion on the ability of the smog chamber to simulate
ambient reactions. Modelers often include processes in modeling chamber
results, such as the formation of nitrous and nitric acid and hydro-
carbon desorption from walls, which are either eliminated or greatly
decreased for application to the atmosphere. There is a great deal of
uncertainty, however, concerning these adjustments.
Given the importance of smog chamber data in oxidant control deci-
sions, it was decided to explore available atmospheric data in order to
develop a data base for smog chamber validation and to attempt to validate
the UNC outdoor smog chamber by simulating observed atmospheric data.
Lagrangian monitoring is the only means of obtaining cause-and-effect
data from the atmosphere for validation, and the Los Angeles Reactive
Pollutant Program (LARPP) was judged to be the best available collection
of Lagrangian data on reactive pollutants. Operation 33 (November 5,
1973) was selected for study on the recommendation of those who col-
lected the data and because of the large number of complete helicopter
patterns.
All the available data for operation 33 were examined from list-
ings. The helicopter data were then analyzed in detail to gain an
understanding of the processes affecting the concentrations of oxidant
and precursors. A graphic technique was developed for displaying large
numbers of individual concentration measurements and their spatial
1?
-------
positions on a helicopter flight pattern. A three-dimensional figure
was constructed for each helicopter pattern: the bottom outline of a
figure represents the helicopter's flight path in geographical coor-
dinates and the height of a figure represents the scaled pollutant
concentration at the geographical position. It was possible using this
technique to assess the extent of inhomogeneity in a qualitative fash-
ion, to identify patterns with erroneous data, and to note the temporal
trends of concentration at four altitude levels. It may be concluded
that this representation of data is excellent for both screening data
and for elucidating meteorological and chemical processes. The effects
of mixing height elevation and 0 generation in the morning inversion
were noted.
While the concentration-position figures emphasize the inhomogen-
eity and random variability, consideration of helicopter pattern average
concentrations revealed an overlying order to what on a small scale
appeared chaotic. A region of the atmosphere was found with zero
average vertical gradient for pollutants emitted at the ground. During
the early morning, when concentration measurements extended to the
inversion base, it was noted (as pointed out by Panofsky (25)) that this
region extends from an altitude of 50 m to one-half the mixing depth.
On theoretical grounds, the ratio between the upper limit of this region
and the mixing depth should not have changed greatly later in the day.
This region represents a much greater mass of air than does the surface
layer (i.e., below 50 m for operation 33). Indeed, ground level con-
centrations differ from the concentrations in the zero-gradient region
only to the extent that they are altered by local sources. Simulation
of the zero-gradient region concentrations is probably more important
than simulation of the concentration profiles at a single ground sta-
tion, since the data at the ground may be influenced to a large extent
by local sources. Thus, it was decided to analyze the zero-gradient
region to obtain information necessary for simulation.
Since hydrocarbons are not a conservative species in the ambient
air, it was necessary to calculate the total NMHC concentration which
18
-------
would be present in the absence of reaction based on the observed con-
centrations of relatively unreactive species.
Next an algorithm for determining vertical flux at the top and
bottom of the zero-gradient region was developed and tested using CO
data. The general agreement of the estimates of CO flux in the morning
with estimates of Panofsky (25) and the greater sophistication of
calculations used in this study lends credence to the early flux esti-
mates. The validity of later flux estimates is based primarily on
boundary layer theory. It should be noted that zero-gradient region
concentration changes were assumed to be due to vertical transport;
changes which occurred as a result of horizontal diffusion were incor-
porated in the vertical flux estimates. Horizontal diffusion was not
treated explicitly because oxidant precursors near the operation 33
trajectory are primarily emitted from area sources, and there was not
enough cross-wind data for an accurate estimate of this effect.
The fluxes of NMHC were then determined using the regression line
through source reconciliation points as the lower boundary condition for
the numerical solution. An analogy was drawn between the zero-gradient
region and a variable volume batch reactor by means of Equation 1 below:
dC _ ^b ft jC dH m
dt ~ H ~ H ~ H dt U;
where C is the concentration of a conservative species, F, is the verti-
cal flux at the bottom of the zero-gradient region, F is the vertical
flux at the top of the zero-gradient region, and H is the vertical ex-
tent of the zero-gradient region. The terms of this equation were
equated with the terms in the mass balance for the smog chamber (a com-
pletely mixed flow reactor) by a logical scheme that assured that the
NMHC concentration in the smog chamber would follow the concentration in
the zero-gradient region if no reactions affected NMHC concentrations in
either regime. Injections of NO were kept at a constant ratio with
X
NMHC injections based on the early morning LARPP NMHC/NO ratio. For
j\.
reactive species, differences in the LARPP and chamber concentrations
can result from several sources.
19
-------
(1) Failure to correctly represent emissions and meteorology is one
source of discrepancy. While great effort was expended to simulate
these processes, there are nevertheless some simplifications which may
affect the comparison, for example, (a) the use of constant NMHC/NO in-
X
jections; (b) the use of a constant composition hydrocarbon mix; (c) the
use of six source categories to represent the myriad hydrocarbon sources
in Los Angeles; and (d) numerical computation of flux at altitudes for
which no data exist.
(2) Differences between LARPP and the smog chamber in factors which
are not under control in this experimental setting also cause differ-
ences in results. These factors include solar radiation, atmospheric
inhomogeneity, and temperature. The magnitude of these effects has been
discussed, as well as the effects of instrumental error.
(3) Basic differences between the chemical processes in the atmos-
phere and in the smog chamber are another source of discrepancy. The
principal aim of this work was to explore the extent of differences
between chamber and atmospheric chemical characteristics.
Simulations were performed during September and October, 1976. The
two best experiments with respect to execution of the experimental pro-
cedure and weather conditions were those of October 12 and 13. After
consideration of the factors which may cause differences in the pollu-
tant profiles, it may be concluded that the atmospheric reaction pro-
cesses were well simulated by the chamber reaction processes. The
differences in 0. between the LARPP trajectory and the experiments seem
to be primarily due to very small differences in the NO concentrations.
The chamber NO concentration did not reach concentrations as low as the
atmosphere. This difference in NO could be explained as either too
great an injection of NO or lower oxidizing radical concentrations in
the chamber. Excess NO could have resulted from the simplification of
maintaining a constant NMHC/NO injection ratio. A lower oxidizing
X
capacity of the chamber reaction system could easily be due to the use
of acetaldehyde to represent all exhaust aldehyde emissions, or the
down-mixing of second-day photochemical systems from above the inversion
base in operation 33.
20
-------
Examination of the factors affecting photostationary state 0 in the
atmosphere revealed much interesting information. A disagreement
between k-i/k- estimated at the four altitude levels along the LASPP
trajectory with [0-][NO]/[N021 was attributed to atmospheric inhomo-
geneities as Calvert (26) suggested. Beyond Calvert's analysis, the
equation for photostationary state 0,, in an inhomogeneous medium was
derived, for single measurement values and for helicopter pattern aver-
ages. Using the estimated k./k, and the pattern average N02/N0 ratios,
it was possible to obtain estimates of the covariance of NO and 0_. This
is the first time this value has been estimated for the atmosphere.
Inhomogeneity contributed an average of 2 pphm to the spatial
average atmospheric 0, concentration in the zero-gradient region on a
day that produced only 15 pphm at 1400 PST. In the morning, near and
above the inversion base, the contribution of inhomogeneity was as great
as 6 pphm. The effects of some atmospheric processes on the covariance
and on photostationary state 0_ were also discussed.
21
-------
SECTION 3
RECOMMENDATIONS
The recommendations for work beyond that presented here may be
divided into several categories, but they are to some extent inter-
related.
PRECURSOR CENTRAL REQUIREMENTS
Of the methods discussed in Chapter II, the Dimitriades method is
the only one to use cause-and-effect data to relate oxidant to precursor
levels in a manner that takes into account the differences between
static chamber experiments and sample atmospheric processes. There are,
however, several assumptions which require further examination. The
principal one is that the 0_ isopleths in the atmosphere are propor-
tional to the 0_ isopleths from a smog chamber adjusted by kinetic
modeling to represent the atmosphere. One use of the data from the
simulation experiments may be to validate the adjustment of chamber
isopleths by modeling. A kinetic mechanism which has been validated
against static hydrocarbon mix experiments may be run under the condi-
tions of the simulation and the results compared with that of the smog
chamber. Also, analysis of other LARPP trajectories in a cursory fash-
ion could lead to the establishment of a set of 0--NMHC-NO points
3 x r
against which the assumption of proportionality may be checked.
FURTHER EXPERIMENTAL SIMULATIONS OF OPERATION 33
Several refinements in the simulation of operation 33 may lead to
even greater agreement with the LARPP concentration profiles. It is
believed that major improvement would result by injection of formalde-
hyde in the appropriate quantities to simulate exhaust emissions.
22
-------
Beyond this a more representative hydrocarbon mix could be developed
although the mix employed was not greatly different from that predicted
by source reconciliation. Another refinement would be to vary the
NMHC/NO injection ratio and the N00/N0 injection ratio. Information
x / x
on the NMHC and NO emissions along the trajectory could be used to
X
determine the proper ratio-time profile.
NUMERICAL MODELING
The modeling of chamber simulation results would have been a more
sophisticated method of investigating the effect of light and temper-
ature differences between the chamber and operation 33. Also, modeling
of operation 33 itself would be an excellent test for a trajectory
model, especially if the flux estimates of this work are used to specify
emissions.
FIELD STUDIES
The information obtained in LARPP may be used to design future
field programs. Because the concentration of precursors was not known
between the zero-gradient region and the inversion base, it was neces-
sary in this analysis to develop a numerical method for estimating the
vertical concentration profile of NMHC. Future field programs should
obtain data up to and into the inversion.
This analysis assumed that horizontal dispersion was negligible.
Future field studies should obtain data for estimating cross-wind dis-
persion, as well as vertical dispersion.
23
-------
SECTION 4
METHODS OF RELATING OXIDANT TO PRECURSORS
In order to control a secondary pollutant such as 0«, it is neces-
sary to know quantitatively how precursor emissions affect the amount of
0 formed. The methods for obtaining an oxidant-precursor relationship
may be divided into three categories:
1. Cause-and-effeet relationships are those that result from
observation of reaction systems from the initial emission or
injection of primary pollutants to the formation of oxidant
and other products, for example, smog chambers, or Lagrangian
air monitoring data.
2. Correlative or statistical relationships are those obtained
from observation of ambient air with no assurance that the
precursors observed gave rise to the oxidant measured at some
later time. The EPA upper limit curve (to be discussed later)
is an example of this type.
3. Mechanistic relationships are those obtained from knowledge of
the individual reactions in the photochemical smog complex
such as any of the numerical kinetic models, e.g., Hecht and
Seinfeld (27) or Wayne et al. (28).
There are problems with all approaches for use as the basis of a
control strategy, or for standard setting. Correlative, or statistical
approaches are derived from data collected at present-day emission
levels. The effect of emissions controls on the relationship cannot be
known when extrapolating beyond the emission ranges giving rise to the
aerometric data base. The mechanistic approaches must be "validated,"
i.e., its predictions must be compared with observations in the ambient
air. Also, it is not known how far the model may be extrapolated beyond
24
-------
the conditions of validation because arbitrary adjustments of rate
constants are often necessary to get a "good fit." No model has yet
been developed which provides good prediction of oxidant levels from a
hydrocarbon mixture typical of urban air. Cause-and-effect relation-
ships from smog chambers are questionable due to the possible effects of
surface characteristics, background reactivity, and simulated radiation.
Thus, smog chambers must also be validated. Cause-and-effect relation-
ships derived from monitoring ambient air along a trajectory are very
expensive and conditions cannot be controlled. Nevertheless, Lagrangian
monitoring is the only theoretically perfect source of data for vali-
dation of a smog chamber or a kinetic mechanism. The precursor con-
centrations measured in the morning along a trajectory are the cause of
the oxidant formation later in the day. Of course, it is necessary to
collect sufficient information to evaluate the effects of meteorology
along the trajectory.
Some of the individual methods of obtaining oxidant-precursor
relationships will now be described and evaluated. Mechanistic methods
(models) will not be considered because of their great complexity and
their continuous development and improvement.
THE EPA ASSESSMENT OF CONTROL REQUIREMENTS
The promulgation of a hydrocarbon standard represents a consider-
able simplification of the complex relationship between hydrocarbon
emissions and oxidant level. Essentially, a day is divided into a time
for build-up of primary pollutants and a time for reaction, although in
reality emissions and reactions occur throughout the day. It implicitly
asserts that hydrocarbon may be treated as an inert pollutant until 9:00
a.m.; emissions after 9:00 a.m. are assumed to have negligible effect on
oxidant.
Since the Federal method of assessing the effects of hydrocarbon
emission, and thus the efficacy of hydrocarbon control, has caused such
controversy it will be considered first.
25
-------
The Upper Limit Curve
The relationship between the morning concentration of nonmethane
hydrocarbons and the maximum 1-hr oxidant concentration was obtained by
analysis of air monitoring data from CAMP (Continuous Air Monitoring
Program) stations. The daily maximum 1-hr average oxidant concentra-
tions were plotted versus the corresponding 6:00 to 9:00 a.m. average
nonmethane hydrocarbon concentrations. Then, a curve was drawn such
that, at almost all hydrocarbon concentrations, the largest observed
oxidant was somewhere below the curve (29). It was reasoned that oxi-
dant read from this curve represented the highest oxidant which could be
expected from the corresponding nonmethane hydrocarbon concentration.
Thus, if oxidant was to be prevented from exceeding 0.08 ppm more than
once per year, the 6:00 to 9:00 a.m. nonmethane hydrocarbon concentra-
tion must not exceed the corresponding value, 0.24 parts per million
carbon (ppmC), more than once per year. This is known as the "upper
limit" or "umbrella" curve.
Heuss, Nebel, and Colucci (30) presented six criticisms of the
upper limit curve. First, they contended that this model includes an
excessively large probability factor. That is, they state that maximum
oxidant potential is reached less than one percent of the time. (They
give no reference.) Thus, that on the single allowable day each year
which exceeds the 0.24 ppmC hydrocarbon standard, the probability of
exceeding the oxidant standard is less than 0.01. Further, they con-
clude that "the combined probability of exceeding the oxidant standard
is only once in 100 years." This assertion may be considered as a
simple statement of probability. If event A is defined as exceeding
0.24 ppmC 6-9 am NMHC, and event B is defined as reaching maximum oxi-
dant forming potential, then the probability of exceeding the oxidant
standard is P(AnB), the probability of A intersection B. The proba-
bility statement of Heuss, Nebel, and Colucci is then:
P(AAB) = P(A)-P(B) (2)
where P(A) and P(B) are the probability of events A and B respectively.
This equality is only true when events A and B are stochastically inde-
pendent. Since it is likely that the meteorological factors which cause
26
-------
the NMHC standard to be exceeded, viz., poor atmospheric dispersion, are
also a prerequisite for maximum oxidant forming potential, the assump-
tion of independence cannot be made. Therefore, P(AnB) is probably much
greater tha P(A)«P(B); this criticism is then a great exaggeration.
The second criticism was that the EPA model does not account for
the great change in the ratio of hydrocarbon to nitrogen oxides which
should result from control of these pollutants. This is certainly a
valid objection. There was only a small quantity of data available with
which to assess the effects of NO and NMHC/NO ratio on oxidant from
x x
the CAMP stations (31) when the NMHC standard was promulgated.
The third criticism was that the often large error in nonmethane
hydrocarbon measurements may have caused considerable underestimation of
the actual NMHC of the data point defining the upper limit curve. This
would, of course, lead to an unrealistically high assessment of control
required to achieve compliance with the oxidant standard. It is not
clear how closely this was considered in the analysis by EPA, but in
general the data were very meticulously selected for producing the upper
limit curve.
Heuss, Nebel, and Colucci (30) pointed out in their fourth objec-
tion that the upper-limit curve relies heavily on weekend data. This is
suspect because emissions on weekends are only 5 to 10 percent of the
weekday emissions from 6:00 to 9:00 a.m., but nearly equal to weekday
emissions from 9:00 to 12:00 a.m. Thus, the selection of the 6:00 to
9:00 a.m. average NMHC to characterize the chemical potential for oxi^
dant formation is seriously questioned.
Their fifth criticism was that the model does not properly account
for situations of increasing stability during the day. Again, they con-
clude that the 6:00 to 9:00 a.m. average is not a good indicator of the
potential to form oxidant, since the NMHC maximum concentration may
occur after 9:00 a.m. under increasingly stable conditions. There may
be certain days on which this is valid; however, the meteorological
factors which cause the lower atmosphere to become more stable after
sunrise do not favor the formation of oxidants. Two such meteorological
27
-------
factors are: (1) very cloudy weather and (2) frontal inversion (advec-
tion of a cold front below a warmer air mass).
The sixth objection was that sampling location factors are not
adequately treated in the model. Heuss, Nebel, and Colucci (30) main-
tain that the oxidant precursor concentrations measured at a particular
monitoring site are determined by local sources, while the oxidant
measured later that day is characteristic of a much larger area since
there has been sufficient time for mixing in an air basin. This is
perhaps a much more important point than was thought: not only is the
influence of local sources unquestionable in measuring precursors, but
there is recent evidence that mixing within an air basin during the day
may be much less than was originally supposed (29). Angell, Dickson,
and Hoecker (32) reported the results of following a triad of simultan-
eously-launched constant-density tetroons in the Los Angeles area.
Their work, performed as part of the Los Angeles Reactive Pollutant
Program (LARPP), showed limited horizontal mixing.
The National Academy of Sciences Committee on Motor Vehicle Emis-
sions (33) based its criticisms of the NMHC standard primarily on the
smog chamber work of Dimitriades (34). Since it was not possible to
state that the smog chamber duplicates the behavior of the atmosphere,
it was not possible to validate this criticism. Dimitriades (34)
states, "Considering the difference between the experimental system of
this study and the natural atmosphere, it would be of interest to com-
pare smog levels observed in this study with smog levels in the atmos-
phere for similar mixtures of pollutants. For such comparison, atmos-
pheric data not readily attainable must be used; for example, HC and NO
X
levels in an air parcel during early morning hours and levels of smog in
the same air parcel after several hours of sun irradiation."
The Los Angeles Air Pollution Control District (LAAPCD) also criti-
cized the NMHC standard because it does not consider the role of nitro-
gen oxides (35-36). Their alternative approach, which will be discussed
in more detail later, is based on smog chamber experimentation and an
empirical equation for relating chamber oxidant to atmospheric oxidant.
28
-------
Since the initial flurry of criticism, other problems with the
upper-limit curve have been identified. One principal difficulty with
this approach is that most of the variation in hydrocarbon concentration
resulted from meteorological variation rather than changes in emissions.
There is some experimental evidence that systems which are diluted with
clean air may produce more 0» than an undiluted system under certain
conditions, even though the initial hydrocarbon concentrations were the
same (37). The data were not available when the standard was promul-
gated, and thus the significance of this problem was not known. Con-
sidering this effect alone, it indicates that reduction of emissions may
be more effective in controlling oxidants than is indicated by the
upper-limit curve.
A second problem in applying the upper-limit curve is that it
assumes that the stationary-site monitoring data are representative of
the concentrations over a large urban area. That is, it assumes that
the parcels of air which contain the maximum 1-hr oxidant at a fixed
station had approximately the same NMHC as was measured at the station
several hours before (from 6:00 to 9:00 a.m.). This approach was justi-
fied in the Hydrocarbon Criteria document by pointing to the good cor-
relation between various stations in the Los Angeles area (38).
Regardless of such correlations, it would be desirable to know the
concentration of hydrocarbons (and NO ) which truly gave rise to the
X
maximum oxidant. This is necessary to obtain a valid cause-and-effect
relationship, as opposed to a relationship based on correlation. Also,
more recent studies (39-41) emphasize and attempt to explain the spatial
variation of photochemical pollutants in the Los Angeles basin. In
order to obtain a causative relationship in the ambient air, it is
necessary to tag and follow parcels of air as they traverse an air
basin, measuring the relevant chemical species. Of course, experimental
techniques which employ a captive air parcel, smog chamber experiments,
produce relationships which are valid with respect to their cause-and-
effect character. There may be other factors which cause differences
between the ambient and the experimental air parcels.
29
-------
The Rollback Model
In the two-part EPA model for relating emissions of precursors to
oxidants, the first step was to establish a NMHC standard which would
assure compliance with the oxidant standard. The remaining step was to
determine what degree of emission reduction would be needed to comply
with the NMHC standard. This was done by means of the "proportional,"
or "rollback" model (42). This approach is based on the assumption that
a particular percentage reduction in emissions of a pollutant will
result in equal percentage reduction in the ambient concentration cor-
rected for natural background.
Obviously, the only circumstances under which concentration dis-
plays a nearly linear relationship with emissions are on rather long
averaging times, such as a year. On shorter time intervals, the meteor-
ological factors which affect pollutant dispersion are likely to intro-
duce considerable variation around the straight concentration-emission
line. The variation of pollutant concentration for a wide range of
averaging times has been shown to be approximated by a log-normal dis-
tribution (43-44). (That is, the logs of concentrations appear to
exhibit normal distributions.)
The distribution results from variations in both emissions and
meteorological factors. For nonmethane hydrocarbons between 6:00 and
9:00 a.m., it is felt that the major part of concentration variations is
due to changes in meteorological conditions, although emissions at this
time of day do show some day-of-the-week effects. One study found that
only 20 percent of the variation in NMHC could be explained by changes
in emissions; the total range of hydrocarbons and oxidant values varied
as much as 200 percent (29). Thus, the Federal model for control is a
straight-line, concentration-emission relationship with the residuals
about the line exhibiting a log-normal distribution at any particular
emission level.
Also, the calculations of the required percentage reduction of
hydrocarbons failed to take into account a very important point. It was
assumed that emission reduction has the effect of simply translating the
30
-------
distribution of concentrations to produce a lower geometric mean value
without changing the spread of the distribution. This is not reasonable
on common-sense grounds. In simple dispersion models, concentration is
directly proportional to source strength. Thus, if the source strength
is reduced and the distribution of meteorological variables remains
unchanged, the standard deviation of the distribution would also be
reduced. The significance of this is uncertain now, but it bears fur-
ther consideration.
Another problem with the rollback model is that the concentration-
emission line may not be straight if atmospheric reactions are import-
ant. There may be interactions that have severe effects on the control
requirements. The NO- air quality standard and the emissions standard
for nitrogen oxides are based on the avoidance of adverse health effects
from NO- with no consideration of the effects on the production of
oxidant. There is evidence that reduction of NO concentrations can
produce increased maximum oxidant at the concentration ranges of pre-
cursors projected for Los Angeles air (34, 45). Thus, this should be
considered in setting emission standards for NO .
X
Conversely, control by hydrocarbons may affect the dosage of NO-.
In chamber experiments, the rate of N0_ formation up to the N0? peak is
proportional or a little less than directly proportional to the initial
hydrocarbon concentration under constant light intensity (34, 46). In
outdoor smog chamber experiments, Kamens et al. (47) found that reduc-
tions of initial NMHC concentrations of 50 percent to 85 percent from a
concentration of 2.9 ppmC (at the same initial NO ) resulted in high
«n-
nighttime NO- concentrations. In these experiments it was noted that
combinations of hydrocarbons and meteorological variables which were
sufficient to convert NO to N0_, yet slow enough that a low maximum
daily 0- concentration was produced, gave rise to high NO- concentrations
during the night. The observed effect of NMHC concentration reduction
was to reduce oxidant concentrations, and maximum NO- concentrations,
while increasing NO- average concentration over a 22-hr period.
31
-------
The possibility of such interactions, as described above, points up
the need to consider the hydrocarbon-NO -oxidant system as opposed to
X
separate handling of N02 when setting standards.
OTHER METHODS OF ASSESSING THE EFFECTS OF CONTROLLING OXIDANT PRECURSOR
EMISSIONS
There are numerous approaches to relate oxidants to precursors,
other than the method used by EPA. These approaches have several things
in common. They are, for the most part, well-thought-out uses of avail-
able data and of present scientific understanding of the problems.
Also, they all are attempts to use data from present-day air to predict
the effects of future events on future air quality. Thus, they are all,
in this respect, extrapolations, and contain various degrees of uncer-
tainty. The present approaches may be categorized in many ways; one way
of interest is the source or type of information utilized. The types of
information utilized by most of the published methods are presented in
Table 3. It should be noted that most of these methods depend to some
extent on stationary monitoring data to define the oxidant-precursor
relationship. Such approaches cannot yield valid cause-and-effect
relationships between precursors and products because an air parcel
passing by a monitoring site has a different trajectory and emission
history from the air that passed that site sometime before. The methods
enumerated in Table 3 will now be summarized and critically evaluated,
except the upper-limit curve which has already been considered. The
method developed by the Los Angeles Air Pollution Control District
(LAAPCD) is considered first and in considerable detail because it has
caused much controversy.
The LAAPCD Method
LAAPCD (35) has criticized the EPA method of assessing the effects
of precursor emissions on the following grounds. (1) The very important
ratio between NMHC and NO has not been considered. (2) Reduction of
X
stationary-source emissions is not necessary since the air into which
these pollutants are emitted lacks either the NMHC or the NO to form
3£
smog. That is, these emissions are not mixed with emissions from
32
-------
TABLE 3. OTHER METHODS OF RELATING OXIDANTS TO PRECURSORS AND THE SOURCES OF DATA FOR EACH METHOD
Type/
Approaches
Stationary
Monitoring Data
Emission
Inventory
Smog
Chamber
Deterministic
Models
Meteorology
u>
U)
1. Upper limit
curve
2. Kent Wilson
3. Texaco
4. Trijonis
5. Hamming
LAAPCD
6. Chevron
7. Stephens
8. Dimitriades
CAMP data
17 L.A. and S.D.
monitoring sites
86 station years
from 15 cities
(includes CAMP)
APCD data
LAAPCD data
his own
only to establish
present oxidant
levels
Geocoded emission
a 2000'x2000'
gird (at nodes)
.(RAND & SAI)
APCD Profile
(1971)
Emissions
directly
determined
not necessary
Dimitriades
(BOM)
LAAPCD
chamber data
Dimitriades
(BOM)
Model of
chamber data
surface wind
direction for
monitoring data
(gross
assumptions
very basic
worst case
conditions
assumed in
model run
-------
automobiles to any appreciable extent. LAAPCD contends that only auto-
mobile emissions are important in forming high oxidant levels. (3) The
reductions of emissions by transportation controls suggested by EPA will
have minimal effect since it cannot affect the NMHC/NO ratio.
X
The alternative method of assessment is based on several assump-
tions. (1) Maximum emissions occur in downtown Los Angeles and maximum
oxidant concentrations are observed in outlying areas, such as Pasadena
and Azusa, thus the emission in the downtown area shows its effects in
outlying areas. (2) "Wind-flow patterns show that characteristic pat-
terns of transport for related source area-effect area combinations are
separate and parallel, not mixing or joining. Each source area has its
own "reaction pipeline" to its own effect area or areas. Generally wind
flow follows the same route on all smoggy days." (3) Since the emis-
sions of a particular area are assumed to remain segregated, the reac-
tions occur in a batch mode similar to smog chamber operation. (4) The
effects of NMHC and NO control cannot be seen in the ambient air, it
x
must be studied in a smog chamber and the chamber results related to the
ambient air.
The maximum smog chamber oxidant corresponding to observed and pro-
jected NMHC and NO levels for the years 1960 to 1990 is shown in Fig-
2C
ure 5 (35). The maximum atmospheric 0~ can then be calculated from
maximum chamber ozone by the "Hamming transformation" equation.
This equation deserves further discussion. It is an equation whose
form is completely empirically determined, but it is presented in a
reply to EPA questions (36) as if it were a physical or deterministic
model. The derivation begins with the equation:
k- Cone. 0 + k_ Cone. 0
Cone. 03 - £^ S^ (3)
air f
where k- is a "constant intended to represent the constant rate of loss
of pollutants associated with test chamber experiments due to both
instrument sampling and leakage from the chamber during six hours or
more of constant irradiation." The term k9 is related "to the addi-
tional ozone that is formed from additional HC and NO introduced into
X
34
-------
CHAMBER OZONE MAXIMUM - PPM
.50
Figure 5. Maximum smog chamber oxidant corresponding to observed and
projected NMHC, 1960-1990 (35).
35
-------
the 'pipeline' by sources, predominantly traffic, along the path trav-
eled by the air moving from a central source area in the morning...,"
"D, defines the minimum diffusion and dilution coefficient; that which
produces the maximum ozone." This equation then assumes that minimum
dilution produces maximum ozone. As shown by Fox, Kamens, and Jeffries
(37), this is not necessarily true with low reactivity systems such as
at hydrocarbon concentrations near the NMHC standard. The term D^ is not a
constant, but is some function of the time over which dilution processes
take place. The form
Df = k4 (N0x)m (Tr)n (4)
was presented without corroborating data concerning goodness-of-fit.
NO is "the amount of NO in the air parcel" and T is the time between
x x r
the NO peak and the N0? peak. It is difficult to examine this expres-
sion since V. is not measurable, and no data is presented. "Cone.
0., " is the maximum Oq concentration which would result from a chamber
ch
irradiation of NMHC and NO levels observed in the atmosphere. "Cone.
0- " is the concentration 0« would reach under the very bad meteor-
ological conditions.
Equations 3 and 4 were then combined:
Conc' °3 . k. + k9
_ air _ 1 2 K
- Conc. 0 = Df = m (T )tt
ch. x r
The exponents m and n were estimated from data on the years 1960, 1965,
1969, 1970, 1972, and 1973. The "Conc. 0- " used to calculate R for
ch
each of these years was the 0, corresponding*to the maximum NMHC and NO
observed during the year. The source of NO and T used in the estima-
X JT
tion is not clear, nor is the estimation algorithm described. These
particular years were selected because they were the years which cluster
at the highest level of R; 1971, the year with the greatest R, was
excluded from the estimation procedure. The exponent estimates were
m = -0.222 and n = 0.106.
After the exponents were obtained, the values for K were calculated
for each year. (It was not explained how m and n could be estimated
36
-------
without some value of K either known or assumed.) 1971 had the highest
K value, or 1.50 was used; no explanation of the 80 percent figure was
given.
Further adjustment of maximum 0, is made since the oxidant
standard allows the 0.08 ppm level to be exceeded once per year. Thus,
the mean ratio between highest oxidant value each year and the second
highest oxidant for several years (1.094) is divided into the oxidant
predicted from Equation 5.
This approach is similar to the EPA method in that it treated pre-
cursors as inert material which accumulate early in the morning to some
starting concentrations analogous to the initial concentrations in a
smog chamber. The EPA approach assumes that the air basin is fairly
homogeneous, and thus a single CAMP station can characterize the entire
air basin; LAAPCD rejects this but assumes that a stationary monitoring
site in downtown Los Angeles will measure the highest levels of pre-
cursors emitted and that the stations in Pasadena and Azusa will measure
the highest resulting oxidant. Both these assumptions are rather simp-
listic and greatly limit the reliability of the methods.
The LAAPCD method has numerous other difficulties. One problem is
the use of the maximum observed NMHC and NO concentrations during any
2C
year to calculate "Max. 0_ " seems to violate their own strong asser-
tion that the ratio HC/NO is very important. They do this because they
2.
believe that only automotive emissions are important in oxidant pro-
duction, and, thus, the ratio is nearly fixed in any year by the extent
of HC and NO control in the vehicle population. There is evidence that
the ratio is not fixed. Trijonis (48) presented data on summer morning
NO and HC concentrations (730 - 930 PST) measured at the downtown LA
2t
APCD station from 1966-70. Most of the HC-NO pairs have ratios which
x *
fall between 40 and 9; within this range their distribution is fairly
uniform (with respect to ratio). Even if the fixed-ratio assumption is
accepted, there is a possibility that the maximum NO during a year may
!X
have been the result of stationary source emissions, while the highest
HC occurring at some other time was the result of mobile-source emis-
sions. One must also be rather skeptical concerning the "Hamming
37
-------
transformation" operation. The form of this model has never been justi-
fied either from basic physical and chemical principles or from its
ability to fit data (35-36, 45, 49).
The good points of the LAACCD method are that it makes an attempt
to account for the role of NO in smog formation and it recognizes that
X
transport of pollutants must be considered.
The Wilson Method
Wilson (50) has developed a statistical model which is purported to
assess the effects hydrocarbon emissions have on an index of photochemi-
cal air quality. The index chosen was the number of days per year that
a given region exceeds the oxidant air quality standard of 0.08 ppm.
Input to the model during its calibration included reactive hydrocarbon
emission inventory at each node of a 2000 by 2000 foot grid, most prob-
able wind paths for Los Angeles and San Diego, and the number of days
per year that the oxidant standard is exceeded for seventeen monitoring
stations. In the Los Angeles region, air quality index data 1968, 1969,
and 1970 were used with emissions data from 1970. For each monitoring
station the emissions in a 4.5 mile-wide strip along the most probable
wind path were integrated. These values were then plotted versus the
air quality index (the mean index for the three year span, if data were
available for three years) for each monitoring station. Also, at each
station a "roll-back" analysis was performed to determine the emission
level at which the expected highest 1-hr 0, concentration would equal
0.08 ppm. The average emission level for compliance was used as the
intercept with the horizontal axis. A second-order least-squares fit to
this data was obtained. The curve appears to have been forced through
the emission axis intercept. From inspection of the plot, it is diffi-
cult to believe that the second-order coefficient in such a fit could be
statistically significant since the scatter around the line is consid-
erable; in fact the statistical significance of the first-order coef-
ficient is in doubt.
The Wilson approach is quite good in its spatial resolution, but
the treatment of smog chemistry is extremely primitive. The role of NO is
x
ignored, emissions at different times of the day are treated as though
38
-------
they have the same effect, and the "roll-back" method is used on a non-
linear process. The use of "most probable paths" to approximate the
often meandering trajectories (32) of the South Coast air basin is not
detailed enough for many purposes. The equation relating emissions to
air quality index does not fit the data well at all, thus its predictive
capabilities must be called into question.
The Trijonis Method
Trijonis (48, 51) developed a means of assessing the effects of
reduced NMHC and NO emissions as part of his economic analysis of
photochemical air pollution control in Los Angeles County- Like Wilson,
Trijonis used an index of air quality as opposed to actual concentra-
tion. The following assumptions are made. (1) Emission reduction of
NMHC and NO occur homogeneously in space and time. (2) Emissions of
3t
precursors accumulate in the morning without reacting to produce final
morning concentrations and, thus, precursor concentrations are pro-
portional to their emissions. (3) Accumulation stops and meteorological
factors act on precursors to produce the maximum oxidant level. (4) The
weather variables which affect the oxidant level produced are statis-
tically independent of the distribution of maximum oxidant later in the
day. These last two assumptions are rather difficult to justify, but
steps have been taken in selection of data to increase their validity.
The actual analysis proceeded by obtaining the joint frequency dis-
tribution for NMHC and NO from five years of data (1966-1970) at the
ji
downtown Los Angeles APCD station. The average maximum 1-hr oxidant
values between 11:00 a.m. and 1:00 p.m. for the downtown, Burbank, and
Pasadena stations, weighted according to wind speed and direction, were
used to develop the probability for exceeding the oxidant standard as a
function of NMHC and NO concentrations. By assumptions (1) and (2),
j£
the probability distribution of precursors can be determined for any
emission level if it is known for a fixed level. And, by assumptions
(3) and (4), the oxidant probability as a function of precursor concen-
trations does not change with emissions. Thus, the number of days per
year expected to exceed the oxidant standard is plotted as isopleths on
a graph of NO emissions versus NMHC emissions.
x
39
-------
This method has a great deal of appeal since a direct relationship
is obtained between emissions and air quality. Also, it is presented
clearly with nearly all assumptions stated and defended. Nevertheless,
it is marred by the same flaw as the "upper limit curve:" the rela-
tionship between oxidant and precursor concentrations reflects the
effect of variable atmospheric dilution on precursors, not the effect of
reduced emissions. If assumption (4) were absolutely correct, this
would not matter, but it is by no means completely true. As mentioned
earlier, lowering precursor concentration by dilution may not have the
same effect on oxidant as lowering by cutting emissions, especially at
low concentrations. This is intended as an air basin model, thus spa-
tial resolution is not great. Assumption (3) is another weak spot:
emissions after 9:30 a.m., while not as great as the rush-hour emis-
sions, can have significant effects on oxidant maxima as indicated by
mathematical modeling efforts at UNC (52).
Stephens Method
The work of Stephens (53) at the University of California at Riverside
is not a well-developed method for assessing the effect of changes in
emission levels as the methods discussed above. Only a very small
amount of data has been analyzed by Stephens' technique. The concepts
developed by Stephens could be used for such a purpose if its present
problems can be resolved.
By measuring tracer compounds in air with high oxidant concentra-
tions, Stephens has attempted to estimate the total emissions of hydro-
carbons into the air parcel in which the oxidant was measured. Acet-
ylene was used as an indication of automobile emissions, and ethane and
propane, as indicators of natural gas emissions. The ratio of propane
to ethene is used as a measure of the extent of reaction. This ratio
measured in the air parcel was compared with the ratio calculated for
both batch and stirred-tank reactors using the first-order removal rates
of the two olefins as observed in a smog chamber. A major problem with
this approach is that the batch and stirred-tank regimes do not repre-
sent the full range of reactant contacting schemesit is a definite
mistake to assume that the ratio in the air must fall between the lines
40
-------
representing these regimes on the ratio versus time axes. In addition,
the equation used by Stephens to compute the concentrations in a con-
tinuous-flow, stirred-tank reactor as a function of time is in error.
Also, it is not possible using this method to obtain a good estimate of
NO emissions (23). Thus, much work is needed prior to application of
X
this technique.
Texaco and Chevron Models
Both the Texaco (54) and the Chevron (51) models are based on
regression analysis of stationary-site monitoring data. Texaco used
data from fifteen U.S. cities, including eighty-six station-years of
data. The second highest annual oxidant measurements were related to
the means and standard deviations of the yearly distributions of oxi-
dant, hydrocarbon, and nitrogen oxides. The Chevron technique uses
step-wise regression to relate the 1-hr maximum oxidant to NMHX and NO
X
concentrations at a single, stationary, monitoring site in Los Angeles.
Both these methods rely on single-site data in each city examined, thus
transport is not considered. Another drawback similar to the "upper
limit curve" approach is that the range of hydrocarbon and NO values
X
considered mainly results from meteorological variations rather than
reduced emissions. Thus, these approaches are not applicable to the
assessment of control requirements.
Dimitriades Method
Recently, a method was proposed by Dimitriades (34) as an alter-
native to the current Federal method of relating oxidants to precursors.
It is similar to the current EPA method (Appendix J approach) in that it
treats the emission of precursors to produce early morning concentra-
tions as separate from the dependence of oxidant on morning NMHC and NO
X
concentrations. The method produces a much more sophisticated relation-
ship between oxidant and 6:00 to 9:00 a.m. precursor concentrations,
while the equivalent of linear rollback is used to relate the desired
and present NO levels to the required emission reductions.
The relationship between morning precursors and maximum 1-hr
oxidant was first derived from the smog chamber data of Dimitriades
41
-------
(57). These chamber runs were modeled numerically by Dodge (58). After
reasonable fits to the chamber data were obtained, the model was altered
to make the simulation conditions closer to those of an urban environment.
Initial HN02 was set to zero, chamber contamination effects were
removed from the mechanism, initial N02 was set to 25 percent of initial
NO , diurnal variation of photolytic rate constants was employed, and a
X
period of from 7:00 a.m. to 4:00 p.m. was simulated with the assumption
1
of 3 percent hr~ dilution during this time. The model computations
were repeated over a range of NMHC from 0.2 to 5.0 ppmC and a range of
NO from 0.08 to 1.4 ppm. The results were used to plot oxidant iso-
x
pleths as shown in Figure 6.
This diagram is now used to calculate the desired levels of 6:00
to 9:00 a.m. NMHC and NO by the following steps:
X
(1) For a given geographical region, the maximum 1-hr oxidant
concentration and the average 6:00 to 9:00 a.m. NMHC/NO ratio and its
X
90 percent confidence interval during the smog season are determined for
a base year. The average ratio and maximum oxidant define a "chamber
counterpart" as shown by point e in Figure 6 at the intersection of the
0.4 ppm 0 isopleth and the 8 NMHC/NO ratio line.
3 X
(2) The percent reduction of annual mean N0? to achieve the NO-
air quality standard is calculated. Assuming that the reduction of the
6:00 to 9:00 a.m. NO will be proportional to the reduction in the
X
annual mean, point e' is determined by applying that percentage reduc-
tion to point e.
(3) To achieve the oxidant air quality standard a reduction of
NMHC along line e'f is required. It is assumed that the required per-
centage reduction of NMHC in the atmosphere is equal to the percentage
reduction from Figure 6 even though the resulting NMHC and NO in the
atmosphere is not necessarily equal to the concentrations at point f in
the diagram.
(4) Similar calculations are performed using the 90 percent con-
fidence interval values of NMHC/NO ratio, and the greatest percentage
X
42
-------
u>
OXIDANT/03, Win-
.08 .20 .30 .40 .50 .5S .30 .65
2.0 3.0
NMKC. ppm C
4.0
5.0
Figure 6. Oxidant/0- isopleths derived from combined use of smog chamber
and photochemical modeling techniques (56).
-------
reduction of NMHC from the three NMHC/NO ratios considered is used as
X
the control requirement.
In relating the worst case NMHC/NO ratio with the maximum observed
"X.
oxidant, assurance of adequate control is maintained even under condi-
tions of the simultaneous occurrence of ratios favorable to oxidant
production in the early morning and meteorological conditions favorable
to oxidant formation later in the day. The advantages of this technique
over the present EPA method are undeniable. Dimitriades (56) pointed
out these advantages and several limitations in relation to the Appen-
dix J approach. This approach rests primarily on a cause-and-effect
relationship derived from smog chamber data. Also, the role of NO in
X
oxidant formation is considered. On the negative side, the chamber and
chamber model do not exactly simulate the atmosphere. Also, the rela-
tionship obtained by Figure 6 does not contain information concerning
frequency of occurrence of oxidant levels, while the air quality stand-
ard is specified in terms of frequency of occurrence.
One major assumption of this method should be pointed out. Al-
though the actual oxidant level in Figure 6 at any NMHC-NO point may
Jt
not be the same as the atmospheric oxidant for the same NMHC and NO , it
is assumed that a given oxidant reduction may be accomplished in Fig-
ure 6 and in the atmosphere by the same percentage reduction of NMHC.
At this point in the computation the desired NO level has been deter-
j£
mined, thus this assumption is concerned with the relation of oxidant to
NMHC at constant NO both in the atmosphere and in the adjusted chamber
X
data (Figure 6). For this assumption to be correct, a figure similar to
Figure 6 plotted for atmospheric oxidant levels would be exactly the
same shape as Figure 6. That is equivalent to saying that the oxidant
in Figure 6 may be related to atmospheric oxidant by a multiplicative
factor which is constant over the entire hydrocarbon-NO plane. There
is no atmospheric data available to support or refute such an assump-
tion. Nevertheless, recent efforts to assess the validity of this
technique have indicated that changes in the shape of Figure 6 which
result from different light intensity, dilution rates, emission pat-
terns, and hydrocarbon reactivity have little impact on the predicted
44
-------
control requirements (59). This also implies that small differences in
shape between Figure 6 isopleths and atmospheric isopleths will have
only small effects on the outcome of the implementation of controls
based on this method. Comparison of predicted and observed maximum 0_
in Los Angeles from 1968 to 1975 showed excellent agreement, but some
discrepancies were noted in the observed and predicted decrease in
reactive hydrocarbon emissions (mobile sources) from the 1968-1971
period to the 1972-1975 period (59).
The criticisms of the rollback methods are also applicable to the
manner in which NO control requirements are calculated. Nevertheless,
2t
the method proposed by Dimitriades is a great improvement over the
present EPA method of relating oxidant to precursor emissions.
CONCLUSIONS
All the methods discussed above with the exception of the LAAPCD
method and the Dimitriades method rely on ambient monitoring data to
relate oxidants to precursors. Thus, the NMHC and NO ranges explored
JL
by these relationships are mainly the result of meteorological varia-
tion, not emission variation. The LAAPCD approach, while it uses smog
chamber data to relate oxidant to precursors, employs the highly suspect
"Hamming transformation" equation to relate chamber results to the
outdoor air.
Most of the methods discussed make use of stationary monitoring
data. Such data cannot reflect transport of pollutants in the air
basin, thus they cannot yield cause-and-effect relationships. A true
cause-and-effect relationship can only be obtained by following ambient
air, captive air experiments, or using some technique to determine
precursors in the same air in which elevated oxidant is present. Data
obtained by following ambient air are quite expensive, requiring instru-
mented aircraft and tracking equipment. Certainly an entire approach to
relating oxidant to precursors cannot be based on such data because of
the extreme cost. Smog chamber experiments, while much less expensive,
have not been shown to simulate the behavior of ambient air. Methods
45
-------
for determining precursors in the same sample in which oxidants are
measured have not been developed.
46
-------
SECTION 5
THE LOS ANGELES REACTIVE POLLUTANT PROGRAM
GENERAL DESCRIPTION
The Los Angeles Reactive Pollutant Program (LARPP) was a Lagrangian
air monitoring study funded by the Coordinating Research Council and the
Environmental Protection Agency. The purpose of the project was to
assemble a data package for studying the processes of transport, dif-
fusion, and chemical reaction associated with photochemically reactive
air pollutants. The field work phase of LARPP was carried out from
early September to mid-November, 1973.
On a day of an operation, three tetroons, ballasted to float below
the inversion base, were released from a ground station in the early
morning. Transponders were suspended beneath the tetroons so that they
could be identified and their positions could be determined by the radar
unit located on Flint Peak; a position determination could be made each
minute taking each tetroon in turn. As soon as the tetroons reached a
more or less stable altitude, the centroid of the triangle formed by
projection of the tetroon positions onto the earth's surface was deter-
mined. A helicopter, flying a one-half mile square pattern around the
centroid at an altitude approximately midway between the ground and the
inversion base, released fluorescent tracer particles. After the tracer
release, an instrumented Bell 212 helicopter provided by EPA-Las Vegas
began flying square, constant altitude patterns around the centroid at
constant air speed (60 knots). The helicopters used were also equipped
with transponders as well as radar altimeters. A second radar unit was
used to track the helicopters and to position them at the beginning of a
pattern. Patterns were flown at four altitudes between the ground and
the inversion base; the lowest level used was 200 feet above the ground.
47
-------
Usually, a helicopter started at the upper level flying patterns at
descending altitudes down to the lowest level, then completed another
ascending and descending series of patterns before returning to base.
Prior to completion of its last low pattern, it was often joined by the
other helicopter flying an upper level pattern at the commencement of
its own series of pattern levels. A typical pattern flown at constant
air speed is shown in Figure 7. In addition to the helicopters, LARPP
also utilized a mobile van which attempted to remain close to the tet-
roon centroid.
The helicopters and van were equipped to measure 0_, NO, NO , CO,
j X
CH,, non-methane hydrocarbons, fluorescent tracer, air temperature, and
dew point and to take samples in Mylar bags for later gas chromato-
graphic analysis in the El Monte laboratory of the California Air
Resources Board (ARE).
Other information collected during the operations included: total
solar and UV radiation at five stations, lidar data, and monitoring data
from the Battelle and GM vans. After the data were collected, they were
brought together with stationary-site monitoring data, meteorological
data and traffic data from state and nearby county agencies for the days
on which LARPP operations were carried out. All the data, with the
exception of the lidar measurements, were merged into a huge time series
to produce archive tapes. Thus, data records of different types and
lengths were intermingled in a fixed block format. Archive tapes are
available in 7-track and 9-track versions from NTIS.
ANALYSIS OF OPERATION 33
General Description of Run
LARPP operation 33 was carried out on November 5, 1973. The tet-
roon triad was launched at 0715 PST. The data record from the heli-
copters begins at 0748 PST and runs through to 1427 PST. During this
time the helicopters flew 49 complete patterns: more than any other
LARPP operation. The tetroon trajectories are shown in Figure 8.
During the early morning the tetroons were carried southward on very
48
-------
WIND- 1,5 m/s
vo
£
N
(0
*
OJ
(a) Lagrangian
(b) Eulerian
Figure 7.
Typical LAKPP helicopter pattern from (a) Lagrangian (moving with mean wind)
viewpoint,_and from (b) Eulerian (fixed point and ground) viewpoint in a
1.5 m. sec mean wind. *
-------
Ul
o
ANGELEJ
s'TORANGE
30
35 40 45 50
LARPP COORDINATE, KM
Figure 8. Tetroon trajectories for LARPP Operation 33.
-------
light winds typical of the nocturnal mountain winds in Los Angeles. At
0920 PST, the sea breeze regime was established and the tetroons, now
over Paramount, veered to an easterly heading at an increasing speed.
At this time, tetroon number 3 was carried along a more northerly tra-
jectory. While tetroons 1 and 2 remained relatively near each other for
the entire run, the trajectory of number 3 continued to diverge, as
would be expected after the initial separation. The location of the
centroid was based on tetroons 1 and 2 only, after the separation of 3
became significant. At 1100 PST, tetroons 1 and 2 took a northeasterly
route at an increasing speed. Just after 1300, they encountered the
Puente Hills. Convection currents flowing up the warm hillsides carried
the tetroons to much higher altitudes as may be seen in Figure 9. Only
4 of the 49 helicopter patterns did not enclose the centroid.
Because of the number of patterns flown and the length of the oper-
ation, 33 was chosen as the run to be studied by five consultants, whose
task was to evaluate the data and to suggest ways in which the data
might be used. Although much useful information and ideas for further
analysis are contained in the consultant reports, only the material
pertinent to simulation of the smog system of operation 33 will be
referenced here.
Initial Analysis
Prior to receiving the LARPP data, background material was col-
lected. The coordinate locations of all stationary air monitoring and
meteorological stations in the Los Angeles air basin were obtained from
numerous sources. These stations are listed in the LABPP archive docu-
ment (60) in Tables 32 through 35, 38, 41 through 43, 45 and 46 (of
above document). The coordinates of these stations were estimated in
universal transverse mercator (UTM) coordinates. Also, during this
period, topographical, relief and road maps of the Los Angeles basin
were procured. Stationary source information for Los Angeles and Orange
Counties was obtained on tape from EPA's National Emission Data System
on a magnetic tape.
The LARPP data archive was received in July 1975. The original
data retrieval program developed by Parker and Martinez (60) could not
51
-------
Ul
15.0
10.0
CM
CD
Lul
Q
ID
f
It
5.0
0.0
»»'
15.0
10.0
5.0
7.0 8.0
9.0 10.0 11.0 12.0 13.0
PRCIFIC STflNDflRD TIME
Figure 9. Tetroon altitude above ground level versus time.
H.O 15.0
0.0
-------
be employed because it is machine-dependent and had some limitations for
the work at hand; thus, the first stage of data anlaysis was to develop
programs for the examination of the data. It was decided to sort the 25
different record types which were potentially present within the oper-
ation 33 data into separate data files on an off-line, 2314 disk.
Program UNMERGE was written to perform this sorting operation. Table 4
shows the types of data files produced and indicates the number of
records of each type which were present. After briefly examining the
listings, more specific tasks were undertaken. The programs developed
for the initial examination of LARPP data are listed and briefly de-
scribed in Table 5.
Helicopter Data
From a chemical point of view, the most important data in the
archive is that from the two instrumented helicopters. The quantity of
information in the helicopter data file exceeds that of all the other
files combined. Each helicopter record contains values of the variables
listed in Table 6; the sources of these values are also presented. In
the archive obtained, corrections for the effects of temperature and
altitude on the instrument readings had been made, but obviously incor-
rect data, such as negative concentrations and electronically produced
glitches, still remained in the data. At the LARPP Symposium (35), it
was mentioned that helicopter radio transmissions often caused erroneous
spikes in the data, especially the ozone data.
The first look at the helicopter data was obtained from program
HELIST which produced a formatted listing of the helicopter file. This
listing is about 200 pages long. Since all the information would not fit
across a computer page, the listing was divided into two portions: one
concerning instrumental measurements and the other one concerning the
auxiliary information such as position of the helicopter, bag sample
number during the pattern, and others. The flag field in each heli-
copter record is a string of 640 bits. Many of these bits refer to a
list of conditions, each of which relate to some aspect of how the data
were collected. In preparing the tape obtained for this work, the
archivists had ignored this bit field in translating the 7-track BCD
53
-------
TABLE 4. FILES PRODUCED BY PROGRAM UNMERGE
AND THE NUMBER OF RECORDS TRANSFERRED FROM
THE OPERATION 33 ARCHIVE TO EACH FILE TYPE
Record Type No. No. of Records
Description
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
(Continued)
5610
300
405
432
30
48
24
330
24
161
24
14
68
2
Helicopter data (automatic)
ARE van data (automatic)
Helicopter radar data
Tetroon radar data
Aerometric data from Los Angeles
Air Pollution Control District
Gas chromatographic data
Aerometric data collected by
Battelle Memorial Institute
Aerometric data collected at
El Monte
(Unused)
Meteorological data collected by
California Dept. of Transportation
Manual helicopter data
Manual ARE van data
Traffic counts
Airport weather data
Aerometric data from Community
Health Air Monitoring Program
Temperature and relative humidity
from Los Angeles APCD
Radiometer data
Soundings from EMSU
HN03 and PAN data
54
-------
Table 4 (Continued) Files Produced by Program Unmerge and the Number
of Records Transferred from the Operation 33
Archive to Each File Type
Record Type No. No. of Records Description
20 24 Meteorological data from Los
Angeles Air Pollution Control Dist.
21 24 Aerometric data collected by ARE
for surrounding counties
22 36 CO bag samples collected by
California Dept. of Transportation
23 Aerometric data from California
Department of Transportation van
24 9 Radiosonde data
25 Winds aloft at specified height
levels
26 4 Winds aloft at constant pressure
levels
55
-------
TABLE 5. SOME PROGRAMS DEVELOPED
FOR EXAMINATION OF LARPP DATA
Program
Description
CDCTRAN Translates data from CDC internal display code to (IBM-
compatible EBCDIC code (not needed)
EXAMINE Prints data from a single LARPP operation in raw form
UNMERG Sorts data from a single operation into 26 separate disk
files
HELIST Prints a formatted listing of helicopter data
DUMPxx Prints raw listing of specific disk files
RADBOND Prints radiosonde data and converts dew point to water
concentration in ppm
BSDATA Scans helicopter data file and computes the average alti-
tudes at which hydrocarbon bag samples were collected
TETLOOK Computes running averages and standard deviations of
altitudes for each tetroon on a variable number of
sequential data values
56
-------
TABLE 6. CONTENTS OF HELICOPTER DATA
RECORDS AND SOURCES OF DATA
Items
Source
Record Type (01 or 11)
Local Time (HHMMSS)
Turn Number
Data Quality (1-bit per flag)*
Julian Date
Bag Sample Time
Helicopter ID (20, 30)
Selected Channel Value
Time (HHMMSS)
Position N-S (XXX.X)
Position E-W (XXX.X)
Bag Sample No.
Fluor. Part Count
NO Concentration (ppm)
NO Concentration (ppm)
X
BSCT (visibility)
CO Concentration (ppm)
Temp (°F)
Dewpt (°F)"
NMHC Concentration (ppm)
CH4 Concentration (ppm)
03 Concentration (ppm)
X-POS (radar) (XXXX.X)
Y-POS (radar) (XXXX.X)
Altitude MSL (XXXX.X feet)
Altitude AGL (XXXX.X feet)
Manual
Manual
Manual
Manual
Manual
Manual
Manual
Mie 110 Flourescence Device
TECO 14B Chemiluminescent Instrument
TECO 14B Chemiluminescent Instrument
MRI 1550 Nephelometer
Andros 7000 non-dispersive IR
Cambridge CS137
Cambridge CS137
MSA 11-2 Env. Chromatograph
MSA 11-2 Env. Chromatograph
REM 612 Chemiluminescent Instrument
Radar
Radar
Barometric Altimeter
Radar Altimeter
57
-------
tape to a 9-track EBCDIC tape. Since the translation took the field to
be coded characters, not bits, the field became rearranged. Thus, a
program was written to translate the field to the original BCD repre-
sentation and to list the messages corresponding to the flags present in
field. These messages are given in Appendix A.
A quick scan of the listings revealed that variation of concen-
trations from one data point to the next made direct interpretation of
the listing very difficult. The records from the two helicopters were
often interspersed since the helicopter file was still a strict time
series. This added to the problem of visually scanning the data. Some
general trends were noted however. It appeared that concentration vari-
ation was greater in the horizontal legs of the patterns than on the
vertical portions of the flight. Also, it was seen that general trends
were present in the concentrations, with both altitude and time, al-
though the significance of these trends was difficult to assess. Thus,
it was determined that other methods of displaying the data must be
explored. Because the quantity of data was so great, computerized
techniques were preferable.
SYMAP
One method which seemed to promise a quick means of viewing all the
data for a particular pollutant was the SYMAP program, a line printer
mapping program distributed by the Harvard Laboratory for Computer
Graphics and Spatial Analysis. The program has the ability to reduce
three-dimensional data to a two-dimensional form. It is used most often
for displaying data of a dependent variable (such as population density,
mean income, number of buildings per square kilometer) as a function of
position in geographical coordinates. After the program has read the
input data, it proceeds to interpolate the values of the dependent
variable at all points in a regularly-spaced, two-dimensional grid via a
complex algorithm (61). These points correspond to print positions in
the output map. Next, the numerical values at the grid points are
reduced to a categorical representation according to the level groups
supplied by the user. Thus, a particular print symbol is assigned to
58
-------
each print position in the output grid. In general, the higher the
level of the dependent variable, the more dense (or darker) the symbol
used. This is accomplished by the use of standard print train char-
acters and overprinting for the denser symbols. SYMAP was to be used to
display concentrations as a function of altitude and time.
There were many problems to be overcome before SYMAP could be
applied to the LAKPP data. Because SYMAP was designed to work with
geographical coordinates, the scaling from independent variable units to
map units (inches on the page) is fixed by the vertical axis. That is,
scaling of the map ordinate and abscissa are not independent. Thus, it
was necessary to prescale the time values for each reading so that the
scaled values came out similar in the range and magnitude to the alti-
tude values. The next problem encountered was that SYMAP, being written
in FORTRAN, has fixed data storage limitations. It is limited to ap-
proximately 1000 points; however, the helicopter record contains over
5000 points. After contacting the programmer at Harvard responsible for
SYMAP, it was decided that the needed alterations to SYMAP were pos-
sible. Lacking access to and familiarity with SYMAP source code, the
UNC Computation Center's Contract Services group was employed to obtain
a load module of the expanded version of SYMAP and the JCL (Job Control
Language) procedures for executing the load module, and linking a user-
supplied subroutine with the load module. Another major problem was
that the interpolation algorithm produced "zero divide" errors because
of the close spacing in the altitude-time plane of the data points.
While the system action on zero divide was the proper action, the error
messages were printed out in the middle of the map and eventually caused
the job to terminate abnormally. Thus, the error messages and termina-
tion on zero divide error were suppressed.
Although great effort was expended in using SYMAP, the results were
not entirely satisfactory. The interpolation algorithm produced some
strange effects that seemed unwarranted based on the data. Also, no
attempt had been made to screen out erroneous values except that nega-
tive concentrations were not used in interpolation. Dissatisfaction
59
-------
with SYMAP results led to an effort to develop other graphic techniques
for examining the data.
It was decided to look at each horizontal helicopter pattern as a
unit. Since there had been an effort to fly patterns at more or less
fixed altitude levels, the data were stratified according to level.
Program HELBRAK was written to segregate records from the large heli-
copter data file into five files for the following altitude ranges:
100-300 feet, 300-500 feet, 500-700 feet, 700-900 feet, and > 900 feet.
All altitudes referred to are elevation with respect to ground level
(altitude AGL). Very few patterns crossed these altitude range bound-
aries except quite late in the run. Formatted lists of each level file
were produced with program LEVLIST. From these listings it was easier
to see erroneous data since the data had been broken up into distinct
patterns and the concurrent records from the two helicopters were no
longer interspersed since the helicopters were generally at different
levels. Then, the files were edited to remove all large strings of
records with obviously erroneous data, or data for which the helicop-
ter's location was in question. Program CORCT was written to remove
specific records by selective copying of the files.
Concentration-Position Figures
The next data display technique employed consisted of producing
three-dimensional figures from the individual pattern data. Two dimen-
sions of the figures were the helicopter position vectors in the LARPP
coordinate system; and the height of the figures was the concentration
value at that position. The three-dimensional figures were then pro-
jected onto a two-dimensional plane. The two dimensional matrix of
points defining the projection was then processed by the Broomall X-Y
plotter at the UNC Computation Center to produce figures like that shown
in Figure 10. The line to the right of the figure is a scale line. The
tic marks are at 1 pphm intervals, except for the ozone figures which
have 5 pphm. Two versions of the plotting program PLOT 3D were written.
The first version used a vector 3-D to 2-D projection algorithm; the
second version used a more general transformation matrix approach in
which the transformation matrix is constructed analytically prior to
60
-------
OPERRTION 33
flLTITUDE *K)0 FT.
TIME 091500 TO 092220
Figure 10. Example of concentration-position figure for carbon monoxide.
61
-------
programming, and evaluated only once at the beginning of each execution.
Both versions give identical pictures, but the matrix approach is
slightly faster.
After the projections were completed for all patterns for CO, N0x>
NO, and 0_, the figures were trimmed and composited on pieces of poster
board. The photographic reductions of the projections are presented in
Figures 11 through 18. Each pattern is shown as if the viewer were at
a point 10 km south of, 2 km west of and 7 km above the center of the
figure's base. Touch-up was needed on some of the patterns since some
noise spikes still remained in the data.
Examination of the CO concentration in the earlier patterns, shown
in Figure 11, leads to an understanding of some of the meteorological
processes occurring along the trajectory. As the helicopter flew the
descending series of patterns beginning just before 0800 PST and ending
at the 200-foot level at 0822 PST, the high CO concentration from morn-
ing traffic had not reached the 400-foot level. The next pattern, flown
at the 400-foot level from 0822 to 0829, shows that the mixing depth had
just reached this level at the northwest corner of the pattern and to a
lesser extent at the southeast corner. It is tempting to assume that
this pattern intersected warm, rising air parcels with high concentra-
tions of mobile source pollutants. The NO and NO bear out this inter-
2t
pretation; also, the 0_ was almost totally absent in the northwest
pattern corner. Although they have not been plotted, the NMHC, CH,,
and b-scat also show a trend similar to CO for this pattern. There is
no significant difference between the temperatures at the northwest
corner and the temperature values in other parts of the pattern. This
would indicate that the high concentration air parcels were not rapidly
rising when intercepted by the helicopter, but they had, at this alti-
tude, cooled to the temperature of the surrounding air.
Referring to the CO patterns, it may be seen that the turbulent
entrainment has raised the mixing depth to the 600-foot level between
0836 and 0852 PST. And, similarly, the mixing depth reached the 800-
foot level between 0923 and 1001 PST. The pattern between 0912 and 0915
PST, at the 600-foot level is incomplete. If it is compared with the
62
-------
07S1S1 TO 075927
OM61I 10 091130
091508 10 09?308
IOOIH3 TO 101036
U)
075933 TO OBOJ06 08JBSI TO 083635 TO52J7 TO C900IJ 09I1M TO 091501 09*315 TO 093010 09SJ57 TO 100037
090712 TO 08H50
TO 012816 090010 TO 090753 091500 TO 092220 093017 TO 093017 091520 TO 095250 105707 TO I031U
081501 TO 012208
090759 TO 091155
093821 TO 091S1J
Figure 11. Concentration-position figures for Operation 33 carbon monoxide before 1030 PST.
-------
lino: TO II2H1
I21M1 TO 125126
101856 TO 102700 10S709 TD 110116 112117 TO 112931 1IS236 TO 115912 123005 TO 123601 I23«10 TO 121611 125133 TO 1502I7
I035J3 TO 101231
101933 TO 10510' 112936 TO 113701 111119 TO 11S107 115919 TO 120751 1207SI TO 121157 122211 TO 122951
101237 TO I0t027
113709 TO 111111
121501 TO 122297 130125 TO 131039 131016 TO 131920
Figure 12. Concentration-position figures for Operation 33 carbon monoxide after 1030 PST.
-------
0751)1 TO 015321
misu TO nan
100013 TO IDIOM
07S933 TO 080706 OM8SI TO 083635 085227 TO 090012 091131 TO 091501 092315 TO 099010 N5257 TO 100037
080712 TO M115B 082211 TQ 062B16 090010 TO 090753 091500 TO 092220 093011 TO MM 17 091S20 TO 09S250 102707 TO 1033(6
001501 TO 012201
090159 TO 0911SS
093121 TO 091513
Figure 13. Concentration-position figures for Operation 33 nitrogen oxides before 1030 PST.
-------
1H302 TO 1I2HI
IJ165I 10 |JS«»
10I8S6 TO IOJ700 10ST09 TO IIOH6 112111 TO 112931 115236 TO IISM3 I2JOOS TO 123801 I2UIO TO 124M1
103523 TO 101231 101933 TO 10570* 112996 TO 113701 111119 TO 115107 115919 TO 120751 1J07M TO 121157 122211 TO 1229SB
101237 TO 101927
113/09 TO 111111
121501 TO 122207 130125 TO 131039 131016 TO 131920
Figure 14. Concentration-position figures for Operation 33 nitrogen oxides after 1030 PST.
-------
015IM TO 01992)
083611 TO OM1SO
09I5M TO 092308
100013 TO ItlOH
C75933 TO 080706 082811 TO 083635 0«S2!7 TO 090012 091138 TO 091SOI 092315 TO 093010 095^7 TO 100091
080712 TO OBMS8 062Z11 TO 082816 090011 TO 0901S3 M1500 TO 099220 093011 TO 093817 0515JO TO 09S2SO 102701 TO I03J16
0(1501 fO 0)220)
090759 TO 091155
093821 TO 091513
Figure 15. Concentration-position figures for Operation 33 nitric oxide before 1030 PST.
-------
oo
iii3o: TO mm
121651 TO 125*26
:iS5t ': 132700 105109 TO 1IO**6 I12H7 TO II 293 1 115236 TO 1159*2 123005 TO 12310* 123010 "0
125*33 TO 130217
103523 TO 104231 10*933 TO 105701 112936 TO 113701 l|«19 TO 115107 1159IS TO WOTS I 120759 TO |21«7 1SJJ11 TO 122951
IOT237 TO 10f927
113709 TO U«M
O' O1
121SO* TO 122207 130*25 TO 111019 1310*6 TO 131920
Figure 16. Concentration-position figures for Operation 33 nitric oxide after 1030 PST.
-------
075131 TO 015927
003MI TO OI1WO
091508 TO 09230S
100013 TO I010J6
015933 10 030106 012051 TO 003635 005221 TO 090012 091130 TO 091501 092315 TO M3010 095251 TO 100031
000112 TO OtilSI 062*11 TO 002016 090011 TO 090153 091500 TO 092220 093011 TO 093311 091520 TO 095250 102707 TO 103316
90ISOT TO 002206
090759 TO 091155
093021 TO 091513
Figure 17. Concentration-position figures for Operation 33 ozone before 1030 PST.
-------
111302 TO I1J11I
12HSI TO 12S1J6
-vl
o
TO 10U1S I01>56 TO 102700 105709 TO 110146 112117 TO 112931 1IS236 TO 115912 1231105 TO 12380* I?5O3 TO 1)02}'
103521 TO 101231 101933 TO IOS701 112931, TO 113701 111119 TO I IS 107 115919 TO 120751 120751 TO 121157
122211 TO 1229SI
101237 TO 101927
113709 TO 111111
121501 TO 122207 U0125 TO 131039 ,(, ,
Figure 18. Concentration-position figures for Operation 33 ozone after 1030 PST
-------
previous pattern at this altitude, it appears that only the air on the
low concentration side of the pattern was sampled. Thus, it probably
does not reflect the average pollutant concentrations at this level and
time.
The variability of concentrations with horizontal position is
striking. The inhomogeneity regarding CO, NO , and NO is most notice-
X
able in the morning. Later in the day, as the concentrations at all
levels get lower, it becomes more difficult to see this inhomogeneity.
Certainly the absolute variability around the patterns has decreased,
but it is difficult to determine if there is any trend in the varia-
bility relative to the mean concentrations. Ozone shows the opposite
trend. Its concentration is increasing during the day, and its inhomo-
geneity becomes more prominent during the later portion of the operation.
Ozone exhibits some other interesting trends. The first three
patterns at the 800-foot level show a distinct increase in 0_ concen-
trations prior to the arrival at this altitude of fresh pollutants from
surface sources. This indicates the presence of an active photochemical
system above the inversion base. This is not an unusual situation in
the Los Angeles air basin and has been described by Lea (62), Edinger et
al. (41), Edinger (63) and more recently by Johnson and Singh (64).
Edinger (65) in his analysis of operation 33 pointed out that the levels
of ozone found above the inversion in the morning hours could be left
from the previous day's reactions. Ozone could be trapped aloft as the
nocturnal inversion propagates into the mixed layer, or it may have been
injected into the inversion of the previous day along the heated slopes
of the mountains which surround the Los Angeles basin. This second
mechanism comes into effect when heating of air passing up the mountain
slopes produces air parcels whose density is identical with that of the
air within the inversion. Tracer experiments with liquids easily show
that when a vertical density gradient exists in a fluid medium, a fluid
of intermediate density injected into the medium tends to seek its level
and then to spread rapidly in the horizontal. Thus, polluted air may be
injected into the inversion as laminae of high concentration. Edinger,
nevertheless, points out that these same processes affect CO as well as
71
-------
0_ and that the CO concentration above the inversion in operation 33 is
much lower than that within the mixed layer. This indicated that air
entering the inversion is greatly diluted in the process. Edinger felt
the fact that ozone reached elevated levels in the inversion under these
circumstances was evidence for photochemical formation above the inver-
sion and ozone destructive processes in the mixed region.
It seems probable that the mountain injection mechanism is not
responsible for the presence of 0 or 0 precursors at these low alti-
tudes. It is more likely that this material was the result of the
deepening nocturnal inversion capturing () and precursors from the
previous day. The increase of ozone at the 600-foot and 800-foot levels
prior to the arrival of the fresh pollutant at these altitudes strongly
suggests photochemical synthesis from second day materials. Smog cham-
ber experiments at Research Triangle Institute and at the University of
North Carolina indicate the ability of such dilute, second-day systems
to produce ozone early in the day (88). Ozone destruction in the mixed
layer is demonstrated by the strong negative correlation of 0_ and NO,
as one would expect.
Pattern Averages and Relative Standard Deviations
After having examined Figures 11 through 18, which emphasize the
inhomogeneity of the air, the question naturally arises: Is there
anything here that can be modeled, either experimentally or mathemat-
ically, or are the processes which affect concentration simply too
variable spatially for any sense to be made of the data? Also, are
there any underlying trends which would suggest a model? To answer
these questions it was decided to look at the averages and the standard
deviations of data around each helicopter pattern. Program PATAVE was
written to compute these statistics.
The pattern averages for all the concentrations, temperature and
dew point are shown in Figures 19 through 27. Patterns were eliminated
from consideration if they were not complete or if there were not a
sufficient number of points within the altitude range for each level.
These data are striking in that, for most of the variables plotted, the
averages at all four levels are all nearly equal after about 0930 PST.
72
-------
.^00
I -300
O
.200
LU
O
z
O
O
.100
000
1'F
Altitude
e 200 ft.
A 400 ft.
+ 600 ft.
n 800 ft.
.300
.200
.100
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 H.O
PflCIFIC STflNDflRD TIME
000
Figure 19. Helicopter pattern average nitric oxide concentrations for Operation 33.
-------
.600
.500
o.
o.
.300
UJ
o
g .200
x
o
~
.100
000
. 1,1,1,1
I , I
1
.600
.500
.^00
.300
.200
.100
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 H.O
PflCIFIC STflNDflRD TIME
000
Figure 20. Helicopter pattern average nitrogen oxides concentrations for Operation 33.
-------
-J
Ui
TE
J-3
« i
i
r
cc
o
CO
1
1
IU . U
9.0
8.0
7.0
6.0
5.0
1.0
3.0
2.0
1.0
0.0
i i | i i j i j i j i j i
_
- Altitude
__ ^_
o 200 ft.
« 400 ft.
* 600 ft.
« 800 ft.
~ __-^>*<^ ""
1 °~~/ 1^^^ _^^^-\ S~
- / 1 / ^^^cx/^ ^^^
/ / /
r A / s* "i
+ * ^r
i ill i 1 i i i i i 1 r
JL U . V
9.0
8.0
7.0
6.0
5.0
1.0
3.0
2.0
1.0
0.0
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 H.O
PRCIFIC STflNDflRD TIME
Figure 21. Helicopter pattern average scattering coefficients for Operation 33.
-------
10.0
9.0
8.0
~ 7.0
»
I 6.0
t
£ 5.0
6
o.
O
o
o
o
o
3.0
2.0
1.0
0.0
Altitude
200 ft.
400 ft.
600 ft.
800 ft.
10.0
9.0
8.0
7.0
6.0
5.0
6.0 7.0
8.0 9.0 10.0 11.0 12.0
PflCIFIC STflNDflRD TIME
13.0 H.O
3.0
2.0
1.0
0.0
Figure 22. Helicopter pattern average carbon monoxide concentration for Operation 33.
-------
LU
ID
OI
LU
51
LLJ
100.
90.
80.
70.
60.
50.
30.
20.
10.
Altitude
200 ft.
400 ft.
600 ft.
800 ft.
1
1
1
1
1
1
100
90
80,
70.
60.
50.
6.0 7.0
8.0 9.0 10.0 11.0 12.0 13.0 H.O
PflCIFIC STflNDRRD TIME
30
20,
10.
0.
Figure 23. Helicopter pattern average temperatures for Operation 33.
-------
-J
oo
U_
o
Q_
3:
LoJ
J» \f \f
90.
80.
70.
60.
50.
40.
30.
20.
10.
n
_ . | . , , | , j , | , | , I , _
Altitude
o 200 ft.
A. 400 ft.
+ 600 ft.
o 800 ft.
~ .0. »"» «^D + «. at~
~ ° " «-*'"*** *-"**-.. -]
_
",1,1,1,1,1,1,1,"
JK VX >^
90.
80.
70.
60.
50.
40.
30.
20.
10.
n
6.0 7.0
8.0 9.0 10.0 11.0 12.0 13.0
PflCIFIC STaNDflRD TIME
H.O
Figure 24. Helicopter pattern average dew points for Operation 33.
-------
T. UU
O
S3. 00
2
O
n
h-
i 2.00
LU
O
O
o
» 1.00
2:
z
n nn
1 1 ' 1 ' 1 ' 1 ' 1 ' 1 ' 1 '
^"""^^ Altitude
* o 200 ft.
/ A 400 ft.
y » 600 ft.
x^ a 800 ft.
1 1
1 1 ^ ^^
- / / ^* ~
/ + / +!!i^
/^n * ^X0
a
. 1 , 1 , 1 , 1 , 1 , I , 1 ,
f.OO
- 3.00
2.00
-1.00
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 H.O
PflCIFIC STflNDflRD TIME
0.00
Figure 25. Helicopter pattern average nonmethane hydrocarbon concentrations for Operation 33.
-------
c»
o
5.00
t.OO
E
a.
a.
3.00
en
o
"Z.
o
o
2.00
1.00
0.00
I
I
1
Altitude
200 ft.
* 400 ft.
4 600 ft.
o 800 ft.
1
1
^O-
1
5.00
4.00
3.00
2.00
1.00
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 H.O
PflCIFIC STflNDflRD'TIME
0.00
Figure 26. Helicopter pattern average methane concentrations for Operation 33.
-------
.200
GO
E
o.
Q.
O
.100
UJ
O
O
o
000
Altitude
o 200 ft.
A 400,ft.
+ 600 ft.
n 800 ft.
.200
.100
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 H.O
PflCIFIC STflNDflRD TIME
000
Figure 27. Helicopter pattern average ozone concentrations for Operation 33.
-------
This implies, although individual measurements display a definite var-
iability, that after a certain amount of spatial averaging the concen-
trations appear homogeneous with height during the late morning and
afternoon. Thus, there is an overlying order to what seems chaotic on a
smaller scale.
This is consistent with the basic characteristics of atmospheric
turbulence and transport. Lumley and Panofsky (67) state: "The time
and length scales of the turbulent motion that serves to transport
properties are quite large, often of the same order as the time and
length scales characterizing the distribution of properties being trans-
ported, and never greatly smaller." The turbulent regime present
during LARPP operation 33 may be classified as "free convection" because
of the very low wind speeds. Under these conditions, the vertical
transfer of energy, momentum, and pollutants by mechanical turbulence
can be neglected compared to that by heat convection, except in some
small portions of the boundary layer. Mechanical eddies have smaller
wavelengths and are more nearly isotropic than eddies from heating so
that they are much less important in vertical transport than are the
larger, convectional eddies. (An eddy may be thought of as a "glob" of
fluid within a larger fluid mass which has a certain integrity and life
history of its own, the activities and properties of the bulk fluid
being the net result of the activities and properties of the eddies.)
Thus, one would expect to find large eddies, with vertical scale lengths
on the order of hundreds of meters, carrying high concentrations of
pollutants aloft and cleaner air back down. Because the driving force
for convective motion is the unstable vertical temperature structure
created by heat input at the ground, the principle direction of motion
is that which will most rapidly diminish the driving force, i.e., the
vertical direction. Thus, convective turbulence is expected to be
anisotropic, displaying a greater intensity in the vertical than the
horizontal.
The helicopters, flying at an air speed of 60 knots (69.1 mph, or
30.9 m/s) and producing a new measurement every six seconds, have indi-
vidual concentration values spaced approximately 185 m apart (608
82
-------
feet). The values clearly do not represent point concentrations because
the instruments have some characteristic averaging time. Since the
helicopters are in constant motion, these averaging times represent
spatial averaging as well as temporal. The variability of individual
measurements about the pattern mean represents the variation of con-
centration from eddy to eddy, but this variation is underestimated
because of spatial averaging.
Figures 28 through 36 show the standard deviation of measured
variables for individual patterns relative to the pattern means. Ozone
and nitric oxide concentrations exhibit relative standard deviations
greater than the other measurements, and, like the other concentration
measurements, show the greatest relative deviations in the morning.
This is to be expected since the air just below the inversion base is a
collection of air parcels from above the inversion with elevated ozone
concentrations and parcels originating near the earth's surface with
high concentrations of CO, NO , and hydrocarbons. The temperature meas-
J^
urements show the least relative deviations about the pattern means and
dew point, the next lowest. The low deviation of temperature is pre-
dictable since it is self-equalizing; that is, air parcels rise until
their density equals the density of the surrounding air and ambient air
density is principally a function of temperature. Deviations of temp-
erature from the mean for a given pattern level result partially from
interception of air parcels just passing through that level. The rela-
tive standard deviations of precursors in the zero gradient region is
rather low. The deviations are less than 0.3 for NO and less than 0.2
A
for nonmethane hydrocarbons.
The primary concept to be drawn from the foregoing discussion is
that there was a deep layer in the atmosphere which may be characterized
at a particular time, for many purposes, by the mean concentrations of
pollutants and that these means were constant with height within the
layer. This is a consequence of vertical mixing within the layer which
was rapid with respect to advection, and horizontal mixing in the free
convection regime of operation 33. Individual measurements within this
83
-------
2.00
00
.p-
S 1-50
LU
Q
Q
0£
1.00
CO
LU
P 0.50
02
-J
LU
0.00
T | ' | ' | r
NITRIC OXIDE
Altitude
« 2 no ft
A Ann ft
+ 6nn ft
a pno ft
I
2.00
1.50
1.00
0.50
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 H.O
PRCIFIC STflNDflRD TIME
0.00
Figure 28. Helicopter pattern standard deviation of nitric oxide concentrations.
-------
1.00
0.80
0.60
00
Ul
cc
LU
Q
en
a
z:
en
LU
II
5 0.20
LU
0.00
1 I
I I
I I '
NITROGEK OXIDES
200 ft.
400 ft.
600 ft.
o 800 ft.
1.00
0.80
0.60
O.fO
0.20
f I r
r I i I r \ r 1 , '
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0
PflCIFIC STflNDfiRD TIME
00
Figure 29. Helicopter pattern relative standard deviation of nitrogen oxides concentrations.
-------
1.00
0.80
0.60
oo
en
ii
>
LU
Q
O
CC
z
cc
LU
5 0.20
LU
0.00
1 ' I ' I ' I ' I 'T
B-SCflT -
Altitude
200 ft.
400 ft.
600 ft.
800 ft.
1.00
0.80
0.60
O.^fO
0.20
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 1/KO
PflCIFIC STflNDflRD TIME
0.00
Figure 30. Helicopter pattern relative standard deviations of scattering coefficients.
-------
1.00
00
CE
LU
O
o
or
a
2
LU
a:
i
LU
a:
0.80
0.60
0.20
I ' I '
CflRBON MONOXIDE
200 ft.
400 ft.
600 ft.
o 800 ft.
0.00
1.00
0.80
0.60
O.fO
0.20
r 1
r 1
r 1 . 1 , 1 t I
, 1
r
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0
PRCIFIC STfiNDflRD TIME
0.00
Figure 31. Helicopter pattern relative standard deviations of carbon monoxide concentrations.
-------
.150
00
00
o
cc
g .100
Q
CfL
CC
CC
CO
LU .050
f-H
f
cc
LU
000
Altitude
o 200 ft.
A 400 ft.
* 600 ft.
a 800 ft.
1 ' I r
TEMPERflTURE
6.0 7.0
8.0 9.0 10.0 11.0 12.0 13.0
PRCIFIC STRNDflRD TIME
- .100
.150
.050
.000
Figure 32. Helicopter pattern relative standard deviations of temperatures.
-------
CO
.500
en
LU
Q
cc
Q
CO
UJ
.300
.200
5 .ioo
UJ
000
DEW POINT
.500
Altitude
o 200 ft.
A 400 ft.
1-600 ft.
a 800 ft.
.300
.200
.100
7.0
8.0 9.0 10.0 11.0 12.0
PflCIFIC STRNDflRD TIME
13.0 H.O
.000
Figure 33. Helicopter pattern relative standard deviations of dew points.
-------
1.00
o
en
UJ
a
a
cr
a
CD
i
o CO
0.80
0.60
(MO
0.20
UJ
a:
0.00
Altitude
o 200 ft.
* 400 ft.
* 600 ft.
o 800 ft.
NONMETHRNE HYDROCflRBONS _
1.00
0.80
0.60
O.tO
0.20
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 H.O
PflCIFIC STflNDflRD TIME
0.00
Figure 34. Helicopter pattern relative standard deviations of nonmethane hydrocarbon
concentrations.
-------
1.00
1.00
6.0
8.0 . 9.0 10.0 11.0 12.0 13.0
PRCIFIC STBNDflRD TIME
- 0.80
200 ft.
400 ft.
600 ft.
800 ft.
0.20
0.00
Figure 35. Helicopter pattern relative standard deviations of methane concentrations.
-------
vo
S3
2.00
o
P= 1.50
UJ
a
a
CO
1.00
0.50
UJ
a;
0.00
1 ' T
1 'I r
OZONE
Altitude
o 200 ft.
A 400 ft.
* 600 ft.
o 800 ft.
r I r I f I
I i 1 r I
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0
PflCIFTC STflNDflRD TIME
2.00
1.50
1.00
0.50
0.00
Figure 36. Helicopter pattern relative standard deviations of ozone concentrations.
-------
region of zero vertical gradient varied about the mean concentrations.
Thus, the region is not homogeneous on a molecular level.
METHOD OF COMPARISON OF SMOG CHAMBER RESULTS WITH LARPP DATA
Before a smog chamber can be validated, one must first decide what
constitutes a legitimate validation experiment; more specifically, what
injection and dilution rates should be used to simulate the emission and
dilution occurring along a LARPP trajectory so that an honest comparison
may be made. Since it is actually the chemical characteristics of the
chamber which are to be validated, a basic assertion of this work is
that if the emission and dilution profiles in the air parcel are faith-
fully reproduced, then differences in the time profiles of the par-
ticipating chemical species result only from differences in the way
chemical reactions are affected by the chamber and differences in in-
homogeneity, light, and temperature. The effects of differences between
chamber and atmospheric inhomogeneity, light, and temperature may be
estimated; the remaining discrepancies are, then, the chamber effects.
Errors in estimating dilution and emissions may also be significant.
The means of relating the dilution and emission in the atmosphere to the
dilution and injection in the chamber may be examined by analyzing both
causes as chemical reactors.
Variable-Volume Reactor Model
A simplified model of air moving along a LARPP trajectory is a
reactor with a fixed base area and a height, corresponding to the depth
of the mixed portion of the atmosphere, which is variable. For a con-
servative substance, the reactor mass balance equation is the following:
- t) (6)
where V is the volume of the reactor, C is the average concentration of
the species in mass/volume units and their subscripts indicate different
times at which the variables are evaluated. Ffe is the flux of pollutant
(averaged from t.. to t-) into the bottom of reactor, Ft is the flux of
pollutant (averaged from t to t out the top of the reactor, A is the
93
-------
base area, and t is time. In general, it will be assumed that hori-
zontal diffusion may be neglected since most of the emissions are from
area sources. Roth et al. (22) stated that 99 percent of the CO, 71
percent of the reactive hydrocarbons and 58 percent of the NO emissions
X
in the Los Angeles area are from mobile sources. Trijonis and Arledge
(68) found that three-fourths of the reactive hydrocarbon emissions in
the Metropolitan Los Angeles Air Quality Region are accounted for by
mobile sources; the remaining quarter is equally divided between sta-
tionary source organic fuel processes and organic chemical processes.
The major fixed sources for hydrocarbons and NO combined are airports
X
and refineries, while power plants are the major fixed sources for NO
3v
alone. The trajectory studied in operation 33 is close to several oil
refineries, but no power plants or airports.
Dividing through both sides of Equation 6 by t_ - t1 and then
taking the limit as t» approaches t1 yields the differential form:
or
where H is the variable reactor height, equal to V/A. Solving for
dC/dt:
F F
d£ = _£. _! _ C. dH
dt H ~ H H dt
Panofsky (69) observed that for LARPP operation 33 the vertical
concentration gradients vanish within the region between about 50 m and
an altitude approximately equal to one half the mixing depth. While
direct evidence of the height of the zero gradient region with respect
to mixing depth is only available until 0915 PST, recent theories con-
cerning the parameterization of the boundary layer suggest that the
vertical eddy diffusivity should be a function of the ratio of altitude,
z, to mixing depth, h (112). Thus, the vertical extent of the zero
gradient region is likely to remain at the same z/h value later in the
94
-------
day (25, 69). Also, data from Edinger (21, 63) show vertical concen-
tration isopleths for much of the atmosphere below the inversion base.
Thus, there is a region of the atmosphere observed in operation 33
with zero pollutant gradient after a certain amount of horizontal aver-
aging. While this region is not completely mixed on a molecular level
as may be seen from the relative standard deviations of pollutant con-
centrations around helicopter patterns, the inhomogeneity decreases
considerably between 0900 and 1200 PST. These inhomogeneities may cause
the photochemical smog system to behave somewhat differently than would
be expected based on average precursor concentrations. Several dif-
ferent approaches have been applied to this problem (71-74, 72). While
it has been concluded that it may be necessary to account for this
effect in numerical models under certain circumstances, the atmospheric
smog models generally have not been able to treat this problem. The
most promising approach seems to be the recent work of Lamb (74). The
effects of inhomogeneities will be considered in greater detail in
SECTION 7, but for the purposes of simulation, mean values of precursors
were used.
Traditionally, modeling of photochemical processes along a tra-
jectory has been concerned with matching ground level concentrations.
Indeed, the only data available for model validation has been surface
data. Receptors are exposed to ground level concentrations. These
concentrations, however, may be greatly affected by local sources. On
the other hand, the zero-gradient region represents a much greater mass
of air laying above the surface layer. Surface concentrations may be
thought of as mere deviations from the mean zero-gradient region con-
centrations, deviations produced by local emissions before they are
completely diffused into the boundary layer. Also, the grid scales of
current emission inventories are too coarse to fully represent the
importance of sources very close to monitoring stations. Thus, it may
be that simulation of the processes in the zero-gradient region is of
greater general interest than simulation of the concentration profiles
at a single, ground-level monitoring station.
95
-------
Hydrocarbon Emission Estimates Using Source Reconciliation
To quantitate the processes affecting pollutant concentrations in
the region of zero gradient, the terms on the right of Equation 1 must
be evaluated. This is easily done for a conservative substance whose
vertical concentration profile is known from the bottom of the zero-
gradient region to the inversion base at many points along the tra-
jectory. Hydrocarbons, however, are: (1) not, as a class, conser-
vative; (2) nor are their concentrations known up to the inversion base
after 0920 PST. The first of these difficulties was averted by use of
the data from bag samples which were obtained on many of the helicopter
patterns. Hydrocarbons with low reactivity toward oxidative species
were used as tracers to estimate the total hydrocarbon which would have
been present in the air in the absence of consuming chemical reactions.
These estimates were obtained using a source reconciliation approach
similar to that of Mayrsohn and Crabtree (75). Such concentration
estimates behave as conservative substances to the extent that the
assumption of no reaction of the selected tracer species is upheld;
reaction of tracer species may lead to underestimation of hydrocarbon
emissions.
The relative amounts of selected tracer species in both atmospheric
samples and in major hydrocarbon source emissions must be known for
source reconciliation. The reconciliation algorithm, then, estimates a
weighting coefficient for each source by least-squares regression of
sample concentration on source concentrations. The major hydrocarbon
source categories for the Los Angeles area are auto exhaust, gasoline
vapor, gasoline, commercial natural gas, geogenic natural gas and liq-
uefied petroleum gas. Gasoline is distinct from gasoline vapor in that
all the liquid gasoline components were converted to a gas, while for
gasoline vapor, gasoline components are partitioned according to their
vapor pressures. Gasoline vapor is rich in the lighter gasoline com-
ponents. Carburetor "hot soak" emissions would be classed as gasoline,
while emissions produced by filling cars at a service station would be
gasoline vapor.
96
-------
Mayrsohn and Crabtree (75) employed ethane, acetylene, propane,
isobutane, butane, isopentane, and pentane as tracers, leaving only one
degree of freedom in the solution of the regression equations (degrees
of freedom = number of tracer species - number of coefficients to be
estimated), if all sources are deemed significant and included in the
regressions. The present study used an additional two tracers, hexane
and methane (minus background). Table 7 shows the source weight frac-
tions of tracers used in source reconciliation. These values were taken
from Mayrsohn and Crabtree (76) except where indicated. Once the best
fitting coefficients were determined, it was possible to calculate the
total nonmethane hydrocarbons which would have been emitted to produce
the observed tracer concentrations by using average source compositions.
The compositions used were taken from Mayrsohn (77). These are shown in
Table 8. The calculations performed by the source reconciliation pro-
gram are discussed in Appendix B. The following items were calculated
2
for each bag sample: source weighting coefficients; R for the re-
2
gression fit; estimated micrograms/m , ppm by volume, and ppm as carbon
for each substance from each source category and total for all sources;
estimated weight fraction of carbon for each compound in the sample;
residuals for all tracer species. Due to the large amount of data, only
selected portions will be included here.
Table 9 contains the estimated weight fraction of NMHC from each
2
source and the R values.
Mayrsohn and Crabtree (75) performed source reconciliation calcu-
lations on hydrocarbon samples from downtown Los Angeles, Azusa, and a
rural site in Redlands, California during June and July, 1973. They
analyzed samples for three time periods during the day: early morning,
morning traffic peak, and early afternoon. All samples were taken
relatively close to the earth's surface. They found that only about
50 percent of the non-methane hydrocarbons (weight basis) in the air
came from auto exhaust. About 25 percent resulted from emissions of
gasoline and gasoline vapor, and about 25 percent from commercial and
geogenic natural gas. Further works in 1974 by Mayrsohn et al. (79)
confirmed these findings at other sites.
97
-------
TABLE 7. WEIGHT FRACTIONS OF TRACER SPECIES IN .,
MAJOR HYDROCARBON SOURCES IN THE LOS ANGELES AREA
Tracer
Auto
Exhaust
Gasoline
Evap.
Gasoline
Vapo.
Contm.
Nat. Gas
Geogenic
Nat. Gas
Liq. Pet.
Gas
Ethane
Acetylene
Propane
Isobutane
N-butane
Isopentane
Pentane
Hexane
Methane
0.
5.
0
0.
2.
5.
2.
1.
5.
57
13
85
56
50
85
61
09*
0
0
0.1
0.5
2.2
8.0
4.2
3.6
0#
0
0
1
6
18
29
10
2
0
.1
.6
.1
.8
.3
.6
.5
#
11.
0
3.
0.
0.
0.
0.
0.
83.
30
47
34
50
11
11
07
**
72
3.
0
5.
1.
3.
1.
1.
0.
78.
38
40
26
13
33
33
39
58**
4.
0
88.
2.
0.
0
0
0
0.
04
50
38
03
04
k
Data directly from Mayrsohn and Crabtree (76) unless noted other-
wise.
Y
R. W. Hurn (78) presented a table with some compounds in auto ex-
haust. The ratio of methane to ethane was 0.726 and methane to propane
was 1.516. Based on these ratios and the ethene and propene measure-
ments of Mayrsohn and Crabtree (76), methane weight percentage is 5.23
and 4.90. Also, Stephens and Burleson (1969) state that "methane,
ethene and acetylene are present in auto exhaust in roughly equal
amounts."
Mayrsohn and Crabtree (76) found no ethane, ethylene or propylene
in liquid gasolines, thus, no methane would be present since gasoline is
produced by distillation.
**
Mayrsohn and Crabtree (76) give data indicating that the ratio of
methane to ethane in commercial natural gas as 7.41.
II
Mayrsohn and Crabtree (76) give data indicating that the methane
to ethane and methane to propane ratios are 24.6 and 13.7 which indi-
cates methane weight fractions of 79.5 and 77.5.
98
-------
TABLE 8. NONMETHANE HYDROCARBON DISTRIBUTIONS FOR
AUTOMOTIVE EXHAUST, GASOLINE, GASOLINE VAPOR,
COMMERCIAL NATURAL GAS, AND GEOGENIC NATURAL GAS (77)
Compound
Ethane
Ethylene
Acetylene
Propane
Propylene
Methylacetylene
Isobutane
Butane
Butenes
Isopentane
Pentane
Pentenes
Isoprene
2-methylpentane
3-methylpentane
Hexane
Hexenes
Methylcyclopentane
Benzene
2, 3-dimethylpentane
3-methylhexane
2 , 2, 4-trimethylpentane
Heptane
Methylcyclohexane
Dimethylhexane
Toluene
Iso-octane
Octane
Isononane
m & p-Xylene
o-Xylene
Weight Percent
A. EX
0.6
7.6
5.4
3.4
1.0
0.9
2.7
4.1
5.8
3.0
2.3
1.3
3.0
1.5
1.7
1.5
1.7
2.8
1.7
1.2
2.0
1.3
0.8
0.9
7.8
1.3
1.6
1.1
8.1
3.5
GAS
__ fc
0.1
0.5
2.2
0.4
8.0
4.2
2.5
5.8
2.9
3.6
1.7
3.5
2.4
3.3
2.4
3.5
2.5
1.2
1.6
9.5
3.5
1.2
1.2
10.1
3.5
VAP
0.1
1.6
»_
6.1
18.8
1.8
29.3
10.6
6.2
0.1
6.1
2.5
2.5
1.4
2.0
1.1
1.1
0.8
1.0
0.6
0.3
0.3
2.0
0.6
0.1
0.2
1.3
Oc
.5
CNG
69.4
21.3
2.1
3.1
0.7
0.7
0.3
0.1
0.4
0.1
0.3
** f\
0.2
OA
.3
OA
.2
0.1
0.2
0.4
"*"
GNG
15.8
25.2
~~
5.9
14.6
6.2
6.2
"*
23
1.5
1.8
2.6
0.8
4p
.5
Of\
.9
In
.2
2.2
1.8
3.3
1^\
.0
10
.8
LPG
4.1
90.4
5.1
"""~i
0.2
~~
^
~
(continued)
99
-------
Table 8 (Continued) Nonmethane Hydrocarbon Distributions for Auto-
motive Exhaust, Gasoline, Gasoline Vapor,
Commercial Natural Gas, and Geogenic Natural
Gas (77)
Compound
A
Weight
.EX
GAS
Percent
VAP
CNG
GNG
LPG
Nonane 0.9 0.6 0.1
Propylbenzene 0.6 0.4
2,2,3,3-tetramethyl-
hexane 0.3
3-ethyltoluene 5.9 6.3 0.6
1,2,4-trimethylbenzene 4.4 3.7 0.3
1,2,3-trimethylbenzene 1.3 1.3 0.1
Dimethylbenzene 0.9 2.6
Butylbenzene 2.1 1.7
C4-benzenes 2.1 2.1
Less than 0.1%.
100
-------
TABLE 9. BAG SAMPLE TIME AND ALTITUDE
ESTIMATED WEIGHT FRACTION OF NMHC FROM EACH SOURCE CATEGORY
AND R FROM SOURCE RECONCILIATION
Time, PST
0745
0752
0759
0807
0815
0823
0923
0931
0939
0945
0946
0954
1012
1020
1027
1035
1043
1045
1145
1200
1208
1215
1223
1231
1245
1320
1327
1335
1345
1354
Altitude, ft
0
845
666
493
337
500
682
499
337
0
558
600
900
881
700
516
320
0
0
516
517
342
552
700
0
833
1052
1267
1655
1900
Weight Fraction
AEX
0.510
0.390
0.410
0.457
0.000
0.000
0.412
0.000
0.350
0.523
0.724
0.420
0.460
0.712
0.493
0.413
0.716
0.474
0.464
0.476
0.695
0.652
0.000
0.641
0.000
0.000
0.000
0.497
0.462
0.464
GAS
0.470
0.564
0.518
0.475
0.964
0.950
0.528
0.928
0.582
0.413
0.000
0.497
0.477
0.000
0.439
0.532
0.000
0.472
0.469
0.454
0.000
0.000
0.905
0.000
0.901
0.960
0.930
0.440
0.473
0.474
VAP
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.171
0.000
0.000
0.203
0.000
0.000
0.186
0.000
0.000
0.000
0.207
0.258
0.000
0.268
0.000
0.000
0.000
0.000
0.000
0.000
CNG
0.021
0.019
0.044
0.374
0.357
0.000
0.307
0.050
0.043
0.034
0.057
0.058
0.035
0.038
0.040
0.031
0.052
0.030
0.034
0.037
0.052
0.040
0.050
0.041
0.060
0.045
0.046
0.035
0.036
0.033
GNG
0.000
0.000
0.000
0.000
0.000
0.051
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
LPG
0.000
0.027
0.029
0.031
0.000
0.000
0.029
0.022
0.026
0.030
0.047.
0.027
0.028
0.048
0.028
0.025
0.046
0.024
0.033
0.034
0.046
0.050
0.046
0.050
0.041
0.000
0.031
0.030
0.029
0.029
R2
0.863
0.875
0.972
0.950
0.820
0.831
0.896
0.951
0.951
0.982
0.983
0.974
0.945
0.894
0.962
0.950
0.951
0.980
0.944
0.960
0.950
0.882
0.930
0.900
0.960
0.885
0.898
0.941
0.930
0.922
101
-------
Source reconciliation results from LAKPP bag sample data showed
that a somewhat greater weight fraction of NMHC were believed to have
resulted from the emission of auto exhaust plus liquid gasoline plus
gasoline vapor. These combined sources were estimated to contribute an
average of 93 percent of the total NMHC. The sample standard deviation
of this combined contribution was only 2 percent. The variability of
the estimated AEX, GAS, and VAP weight fractions individually is some-
what greater than that of the sum of these three sources. For a major-
ity of samples, the best fits were obtained for source configurations
excluding gasoline vapor; for these cases, the estimated weight frac-
tions of AEX and GAS were often nearly equal. For six samples, GAS was
the only auto-related source to be included in best fitting model,
accounting for 90 percent or more of the total NMHC. Commercial natural
gas was estimated to account for an average of 4 percent and liquified
petroleum gas, another 3 percent of the total NMHC observed, for a
combined average of 7 percent. This was much less than the 25 percent
from CNG and GNG estimated by Mayrsohn and Crabtree.
To explain the differences between the results of this work and
those of Mayrsohn and Crabtree (75), source reconciliation was performed
on the LARPP data employing the same source compositions and tracer
species as Mayrsohn and Crabtree. The results of this yielded estimated
NMHC weight fractions quite a bit like those of Mayrsohn and Crabtree.
Auto exhaust contributed an average of 54 percent, gasoline plus gaso-
line vapor averaged 21 percent and CNG plus GNG averaged 25 percent.
Thus, it must be concluded that the differences between the findings of
this study and those of Mayrsohn and Crabtree were caused by the use of
the additional tracer species, methane above background and hexane. The
use of more tracer species is believed to improve the weight fraction
estimates, just as an increasing number of cases improves the estimation
of coefficients in other regression applications, by increasing the
number of degrees of freedom.
Estimates of Flux in the Boundary Layer
The lack of data between 1000 feet and the inversion base after
0920 PST represents a severe limitation in that the flux of pollutants
102
-------
as a function of altitude cannot be determined directly from the data.
For the period up to 0915 PST, Panofsky (69) calculated the vertical
flux of CO by the expression:
m
where p is air density, FZ is the vertical flux at altitude z, h is the
height of the inversion base and c is the mixing ratio of CO. He then
was able to calculate vertical exchange coefficient. K :
During the morning period, K between the top of the zero-gradient
Z
region of the inversion base approximately follows the relationship
Kz * ~A 10%0 (^f - X) (9)
where z is altitude, h is the altitude of the inversion base and A is an
2 1
empirical constant. Panofsky calculated an A value of about 5 m gee for
operation 33. Based on the LARPP data it is not possible to determine
whether this relationship holds all day long; however, work by Deardorff
and Willis (70) suggests that K should scale as some function of z/h
Z
(Panofsky, 25) . It was assumed for this work that the relationship
holds for the remainder of operation 33.
To determine flux as a function of altitude, a computer program was
written to solve numerically the equation
where K(z,t) is the vertical eddy diffusivity which is a function of
altitude and time. The details of this solution are given in Appen-
dix C. Input to the program consists of initial conditions of con-
centration at all altitudes, the average concentrations in the zero-
gradient region throughout the operation, and the mixing depth through-
out the day.
103
-------
The mixing depth used for flux calculations is shown in Figure 37.
The mixing depth prior to 0924 PST was taken from Panofsky (69). The
mixing depth at 1130 PST was estimated by Edinger (65), and at 1400 PST,
by Keith (80). Earlier estimates based on ground temperature and radio-
sonde data yielded afternoon mixing depths approximately twice as large.
This is probably due to the extreme sensitivity of this graphical method
to errors in the vertical temperature profile.
The initial conditions of CO concentration were determined from
helicopter data up to 800 feet. As an initial guess, the CO concen-
tration above 1000 feet was assumed to be equal to that above the in-
version base early in the day, i.e., about 1 ppm. This is probably a
good upper limit value; concentrations have been measured in non-urban
air masses over continental California from 0.05 to 1.0 ppm and 1 ppm
seems to represent an upper limit for other background measurements
(29). The lower boundary condition for the solution of Equation 10 is
the concentration in the zero-gradient region; the values used for the
CO calculation are shown in Figure 38 as well as the calculated CO
profiles at one hour intervals. The estimated CO fluxes for 150 feet
and for h/2 are given in Figure 39.
Panofsky (69) estimated the flux of CO at six altitudes for the
early morning when concentration data is available up to the inversion
base. Panofsky's computational approach, while not described in any
detail, seems to have been to first establish concentration profiles at
various altitude levels, and then evaluate dc/dt graphically between 0
and 250 m at steps of 50 m. The integration in Equation 7 was performed
numerically to estimate the fluxes. Table 10 compares these flux esti-
mates with the estimates obtained from the numerical solution of Equa-
tion 10. The greatest discrepancies occur at 0815 PST; the numerical
solution estimated a much greater vertical flux. The principal differ-
ences in the two methods are the data utilized. While Panofsky's direct
calculations use concentration data at all levels, the numerical approach
uses only the data in the zero-gradient region except on the initial
step. Also, the numerical solution depends on the validity of Equa-
tion 9. This equation generally fits the data used but the fit is not
104
-------
3.00
LU
LU
o
«I
Z
n
nr
o_
LU
I-H
X
2.00
1.00
0.00
1 ' f
3.00
I , I , 1 r I r I ,
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0
PflCIFIC STflNDflRD TIME
2.00
1.00
0.00
Figure 37. Mixing depth estimated for the LAHPP Operation 33 trajectory.
-------
3.00
2.50
g2.00
in
Q
-------
s.o
3.0
r* 2.0 -
o
-------
TABLE 10. COMPARISON OF CO FLUX ESTIMATES FROM
PANOFSKY (69) AND FROM NUMERICAL FLUX COMPUTATION
Altitude
Time, PST
0815
*
P
N*
0830
P
N
0845
P
N
0900
P
N
0915
P
N
50 m
(167
100
(333
150
(500
200
(667
250
(833
ft)
ft)
ft)
ft)
ft)
79
64
12
0
0
400
446
132
0
0
154
179
66
2
0
232
264
257
10
0
139
176
121
41
2
141
178
211
168
2
97
135
141
70
4
59
105
147
174
70
59
102
127
71
18
-10
35
84
127
126
I.
P - estimates by Panofsky (69).
N - numerical solution of Equation 4.
perfect. Examination of estimated concentration profiles at 0815 and
0830 PST revealed that the transport to the upper altitude levels was
overpredicted, thus so was the flux at all lower levels. This is the
result of the inexact fit of the equation for the eddy diffusivity.
The estimates of flux at 0845 and 0900 agree quite well. There are
some differences at the upper levels, however. These disagreements are
probably due to the combined effects of Panofsky's large vertical spac-
ing and his use of trapezoid rule integration. The use of trapezoid
rule assumes that dc/dt may be approximated by a straight line between
the altitudes for which dc/dt was evaluated. The calculations of dc/dt
on the finer grid spacing used in the numerical solution indicates that
such an assumption could result in considerable error. Thus, it may be
108
-------
concluded that reasonable agreement was obtained when Equation 9 gives
good estimates of the eddy diffusivities. In addition, the numerical
solution offers the advantage of finer grid spacing so that errors in
integration are greatly reduced. The assumption that Equation 9 accur-
ately estimates the diffusivities later in the day may not be checked
against the LARPP data, but must be accepted on theoretical justification.
After the computation of the CO fluxes, total nonmethane hydro-
carbon concentrations from source reconciliation were considered. The
concentrations, based on bag samples from the zero-gradient region, were
plotted versus time as shown in Figure 40. Since one would expect this
measure to behave similarly to conservative pollutants emitted from the
ground such as CO, the considerable scatter of these data must be
assumed to be the result of: (1) the simplicity of the source recon-
ciliation concept, and (2) the errors in measuring tracer concentra-
tions. The first source of error mentioned above is that which results
from assuming that all ambient air hydrocarbons emanate from just six
sources of fixed composition. In reality, nearly each automobile has
its own emission characteristics which vary with time; each source of
natural and geogenic gas has composition slightly different from other
sources. The second source of error above encompasses all the factors
which could cause errors in measurement. Loss of material during col-
lection, or during storage as suggested by Calvert (81), and errors in
gas chromatographic analysis are included.
Since it was believed that the scatter was the result of error
rather than a genuine characteristic of the data, these concentrations
were smoothed by fitting with a least-squares, straight line. This line
was then compared with nonmethane hydrocarbon data from the helicopters.
During the morning, the agreement between the concentration measured
aboard helicopter Smog 2 and the concentration estimated by source
reconciliation is excellent, as can be seen from Figure 41. Helicopter
Smog 3 is measuring considerably lower concentrations; it may be that
this instrument was not calibrated correctly, or that the calibration
changed for some reason. This would not be surprising since many envir-
onmental chromatographs are so noisy that calibration can be greatly
109
-------
t.OO
o
- 3.00
- 2.00
3- i.oo
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 H.O
PflCIFIC STflNDflRD TIME
0.00
Figure 40. Zero gradient nonmethane hydrocarbon concentrations in the absence of reactions as
estimated from source reconciliation results and regression line.
-------
o
S
z
o
2:
LU
O
Z
o
o
z
3. 00
2.00
1-00
0.00
SMOG
i '
SMOG 2
3.00
Altitude
0 200 ft
* 400 ft
+ 600 ft
a 800 ft
2.00
1
1 ,
1 ,
I.I.
1 ,
1
, 1
f
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 1
-------
affected, even under laboratory conditions. In the early afternoon,
NMHC measurements drop below the source reconciliation values. This is
quite understandable since the hydrocarbons in the air are, at this
time, undergoing chemical conversion to compounds such as acids, alde-
hydes, and nitrates which give a lower response per carbon atom in a
flame ionization detector. Given the excellent agreement of the two
sources of hydrocarbon data, it was decided to use the straight-line
approximation of the source reconciliation data as the lower boundary
condition in the solution for the NMHC fluxes. The initial conditions
were taken from early morning source reconciliation results and the
concentration above an altitude of 1000 feet was assumed to be the same
as that just below 1000 feet. The initial conditions are shown in
Figure 42 with the calculated vertical NMHC concentration profiles for!
later times. The estimated NMHC fluxes at 150 feet and h/2 are given iin
Figure 43.
Specifying Injection and Dilution Rates for Simulation
Now that estimates of nonmethane hydrocarbon fluxes in Equation 1
are available, it is necessary to consider their use in specifying
conditions of the simulation. The flux through the bottom and top of
the variable-volume reactor does not take into account the extent to
which the hydrocarbons have reacted. For instance, material emitted at
the ground is in the air for some amount of time before reaching the
bottom of the reactor at 150 feet. It is, nevertheless, reacting during
this time. Thus, the time available for reaction must be estimated.
The lower limit of the zero-gradient region was specified as 50 m
(167 feet) by Panofsky because this is the upper extent of the "surface
boundary layer" estimated for the conditions of wind and irradiation on
that day. The "surface boundary layer" or simply the "surface layer" is
defined as the height through which the wind stress magnitude varies by
less than 20 percent. Thus, h', the height of the surface layer is:
h' = 2000 T (11)
where T is the magnitude of the surface stress in dynes cm~ and h1 is
_2
in cm. The range of T is typically from 1 to 10 dynes cm , thus h'
112
-------
3.00
NONMETHflNE HYDROCflRBONS
cc
0.00
2 3
COMCENTRRTION, ppm
3.00
- 2.50
- 2.00
- 1.50
- 1.00
0.50
0.00
Figure 42. Nonmethane hydrocarbon vertical concentration profiles for Operation 33 at one-hour
intervals estimated by numerical solution.
-------
NONMETKflNE KYDROCflRBONS
10 11 12
HOURS. PST
14-
Figure 43. Estimated nonmethane hydrocarbon flux at altitudes of 150 feet and half the mixing
depth from numerical solution.
-------
ranges from 20 to 200 m. On a typical clear day, h1 calculated from
heat flux is about 50 m (Lumley and Panofsky, 1964). A typical value of
the exchange coefficient in this region is 100 gm cm~1sec~1 for a clear
afternoon at 15 m (82). This is equivalent to an eddy diffusivity of
2-1
about 10 m sec . The average time of transport is then calculated by
dividing the eddy diffusivity into the distance squared. Thus, the mean
time of transport of hydrocarbon species from the ground to 50 m is 4 to
5 minutes. Thus, the extent of reaction would be quite small even at
rates characteristic of the most reactive hydrocarbon species. Earlier
in the day, of course, transport to 50 m is not as rapid. Based on
Panofsky's (69) calculation of CO flux and the gradient of CO in the
lowest portion of the atmosphere, the average time of transport in the
early morning (0815-0830 PST) was on the order of 20 minutes. At this
time, the concentrations of 0. and OH are probably lower and, thus, the
rate of reaction of hydrocarbons are slower as is the rate of transport.
Thus, it was concluded that little difficulty would be encountered by
assuming that the NMHC flux at 150 feet represents fresh pollutant
input.
The vertical hydrocarbon flux at h/2 is assumed to represent re-
moval of material from the control volume with an extent of reaction
which is equal to the mean extent of reaction within the zero-gradient
region. This may be justified simply by pointing to the absence of
pollutant gradients up to this altitude.
The simulation experiment was to be carried out in the UNC outdoor
smog chamber which, when operated with continuous injection and dilu-
tion, has the characteristics of a nearly ideal continuous-flow, stirred
tank reactor (CFSTR) . A pollutant mass balance for such a system is the
following:
d£ = T _ °£ (12)
dt V
where I is the injection rate in ppm min~ , c is the concentration in
3 -1
ppm, f is the flowrate of clean dilution air in ft min and V is the
3
chamber volume in ft .
115
-------
The logic used in equating flux and mixing height terms in Equa-
tion 1 to the injection and flowrate terms is outlined in Figure 44.
When the flux into the bottom of the control is positive, then the
injection rate is based on this value. If the remaining terms in Equa-
tion 1 are then less than zero, their combined effect represents loss of
material from the control volume. This must be simulated by selecting
the proper flowrate of dilution air. Equation 12 may be solved for the
concentration at time t, to yield:
1 t
IV . , IV\ V n~.
Ct = - + (Co - -f > * (13)
where c is the concentration at some initial time and c is the concen-
o t
tration at time t. Since I and f are to be evaluated for separate 6 min
intervals corresponding to the intervals for which the flux was esti-
mated, the initial time and concentration correspond to the time and
zero-gradient region concentration at the beginning of the interval, t
equals 6 min, and c is the zero-gradient region concentration at the
end of the interval. The term I is already known, having been set equal
to F, /H. Equation 13 may not be solved algebraically for f, thus an
iterative method was employed (Newton-Raphson method).
If (-F /H - -r) is positive or equal to zero, then it represents
a condition in which the negative vertical flux at h/2 is increasing the
control volume concentration more rapidly than it is being diminished by
the increasing size of the control volume. Under these conditions, f is
set to zero, and I is equated to the total rate of change of the zero-
gradient region concentration. This situation is never encountered for
NMHC calculations since dc/dt is always negative.
If the vertical flux at the bottom of the control volume is nega-
tive but dc/dt is positive, I is set equal to dc/dt and f is zero.
Again, for NMHC dc/dt is never positive. If F and dc/dt are negative,
this would seem to represent situations in which ground-based sinks of
NMHC exceed ground-based sources. This is probably never the case, but
negative estimates of F coupled with negative dc/dt could arise when
the assumption of no horizontal dispersion is violated. Horizontal
116
-------
dt atm
Fb
IT
F
"H"
C dH
H dt
7
H
dt / chamber
IF F, > 0
b
THEN do the following:
I
Fb/H
IF
ELSE IF
t
H
<] _dH
H dt
< 0
L
THEN Solve for f in non-linear expression
ELSE do the following:
f = 0
\ dt / atm
dC \ > 0 / \ /
dt ) atm = / * _!/
7 Y
THEN do the following:
(\
dt ) atm
1
'
i
^
f
i
^
/
f = 0
.ELSE do the following:
1 = 0
V " In
ZLT7T
f -
At
»
i
Figure 44. Logic for calculating injection and dilution rates from
flux and mixing height estimates.
117
-------
removal of material from the imaginary reactor volume could lead to
negative estimates of F, . This is not a serious concern with respect to
simulation since either case is simply treated as removal of material.
The value of I is zero and f may be determined directly from the simpli-
fied form of Equation 13.
The calculated NMHC injection rates and dilution rates are shown in
Figures 45 and 46, and the cumulative injection and cumulative volume
are shown in Figures 47 and 48. Having completed the calculations of
the required injection and dilution rates, it was then possible to
commence the experimental phase of this project.
118
-------
.200
o
o.
o.
. .100
LLJ
F-
cn
000
r I r I
oooooooootoooooooooctoooooocoootooao
.200
.100
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 1
-------
N3
O
300.
U.
o
co
UJ
O
M
Q
LU
cn
cr
nr
o
200
100
0
1
300
200
100
6.0 7.0
8.0 9.0 10.0 11.0 12.0 13.0 14.0
PflCIFIC STflNDflRD TIME
0
Figure 46. Chamber dilution rates for simulation of LAKPP Operation 33.
-------
CL
CL
o
LxJ
Ir
O
m
s:
z
LU
o
8.00
7.00
6.00
5.00
f.OO
3.00
2.00
1.00
0.00
8.00
7.00
6.00
5.00
1-.00
3.00
2.00
1.00
1
1
5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 H.O
PflCIFIC STflNDflRD TIME
0.00
Figure 47. Cumulative nonmethane hydrocarbon injection rates for simulation of LARPP Operation 33.
-------
I
_l
O
20.
19.
18.
17.
16.
15.
13.
12.
11.
10.
9.
8.
7.
6.
5.
3 s:
2.
1.
0.
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
_ I
1
1,1,
1
I , I
5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0
PflCIFIC STflNDflRD TIME
Figure 48. Cumulative dilution volume for simulation of LAKPP Operation 33.
-------
SECTION 6
EXPERIMENTAL METHODS AND RESULTS
METHODS
The simulation of LARPP operation 33 was performed in the TJNC
outdoor smog chamber which has been described by Jeffries, Fox and
Kamens (83 and 84). Only a brief discussion of the facility is pre-
sented here. The chamber is located in a rural area, 17 miles southwest
of Chapel Hill, North Carolina. The background NO concentration is
X
less than 0.015 ppm and the nonmethane concentration is less than 0.20
ppmC, thus ambient air is used for chamber experiments. The chamber
consists of an external wooden A-frame which, supports a covering of 5-
mil (FEP) Teflon film. A center panel divides the chamber into two
"sides" or "halves;" each side has a volume of 148 m3 (5230 ft3).
Measurement
The instrumentation used for the simulation experiments include
the following:
1. Carle automated gas chromatograph (FID);
2. Bendix NO-NO ;
X.
3. Bendix ozone;
4. Cambridge dew point hygrometer;
5. Eppley total solar pyranometer;
6. Eppley UV radiation pyranometer;
7. Continuous chemical actinometric measurement of the
photolysis rate of N03 (85).
123
-------
Data from these instruments, except the gas chromatograph, are collected
by a computer-based data acquisition system. This computer also per-
forms much of the experimental control functions required for complex
simulations.
The NO and NO channels of the NO-NO analyzer were calibrated
X X
against a known concentration of NO and the N0» channel was calibrated
by gas phase titration. The high concentration NO stream was diluted
down to 0.7 to 0.8 ppm in air dynamically. Laboratory air was drawn
through an 0- generator at a moderate setting so that traces of NO in
the air would be converted to N0?. The air was then passed through a
Purafil bed to remove N0_ and 0_. The instrument was then zeroed. The
flowrate at the outlet of the Purafil was measured with a 448 cc soap
bubble flowmeter and the flow delivered from the NO bottle through a
stainless steel needle valve was measured with a 50 cc soap bubble
flowmeter. Both these streams were then blended in a Pyrex mixing bulb.
The NO concentration was monitored on the outlet of the bulb. The NO
<&
concentration in the tank used agreed with those measured in two other
tanks from different manufacturers within 3 percent.
After the NO and NO channel outputs on the digital voltmeter were
X
adjusted to match the calculated value, the 0 production was cut off,
the Purafil trap removed, and the flow through the 0_ generator measured
again (since it is greater with the Purafil removed). The N0_ and NO
were then measured at a new, slightly lower value. The 0_ generator was
set to deliver roughly 0.4 to 0.5 ppm of 0«. After allowing several
minutes for the system to reach steady-state, the NO- channel output was
adjusted to equal the NO decrease.
The 0_ was then measured at the outlet of the mixing bulb. With
the NO flow turned off the increase in ozone was used to check the span
of the 0. instrument. While the NO instrument required calibration
3 X
about once each week during the experimental period, the 0_ instrument
was extremely stable and required no adjustment other than occasionally
setting the ethylene flow.
124
-------
Hydrocarbon Measurements
Hydrocarbons were measured with an automated Carle gas chromato-
graph with a flame ionization detector (85). The column switching and
bypassing valves are controlled by an electromechanical valve minder.
Using a 15 min cycle time, it was possible to separate the CL through
Cg alkanes, and the C^ through C^ alkenes in the mix. The instrument
uses three columns: (1) 6' x 1/8" 5% Carbowax 1540 and 5% Apeizon L on
Supelcoport 80/100 mesh; (2) 6f x 1/8" N-octane/Poracil C 100/120 mesh
and 2' x 1/8" 20% Apeizon L on Supelcoport 80/100 mesh; (3) 2' x 1/8" FL
alumina treated with 5% Nad. Column 1 separates Cfi-CR paraffins,
olefins and aromatics; column 2 separates C, and C_ olefins and some Cfi
paraffins; and column 3 separates C..-C.,.
During the experimental period, total mix concentrations were meas-
ured based on two complete calibrations performed on August 24, 1976.
Two samples of diluted hydrocarbon mix were introduced into the chromato-
graph and the instrument response was recorded on a stripchart recorder
at a chart speed of 2 in. per min. A sample from a tank containing 3.00
ppmC of propylene was also sampled. The recorded peaks were Xeroxed,
cut out, and weighed. The areas were then calculated correcting for
paper density. Using the ratio between the propylene concentration and
peak area, it was possible to determine the total concentration in the
dilute mix sample and the relative concentrations of all species re-
solved in the chromatogram. For rapid calculation during an experiment,
the height of the n-pentane peak, the largest peak in the hydrocarbon
mix chromatogram, was used as a measure of the amount of mix injected in
the morning. On August 24, the ratio of mix concentration to n-pentane
peak height was 0.311 ppmC/inch. To use this relationship on other
days, it is necessary to correct for variation in instrument response
per molecule of carbon input due to changing carrier gas flow, elec-
tronic drift, etc. This was done by measuring a known concentration of
propylene early in each run. The mix concentration, x, was then calcu-
lated as the height of the n-pentane peak, P, times 0.622 ppmC/inch
times the ratio of the height of the propylene cal gas peak on August 24
to the height of the propylene cal gas peak on the day of the experi-
ment, H. Thus,
125
-------
X, ppmC - (P. In ) (0.622 *E5k) C^, ±* } (14)
given that the propylene calibration peak was measured on the X8 scale
and the n-pentane peak was measured on the X2 scale. Combining the
constants:
(15)
This relationship was used for all the LARPP experiments. Similar
relationships derived for the isopentane and ethylene from August 24
data were used to check the calculations. Agreement between the n-
pentane, isopentane, and ethylene calculations was excellent indicating
that the mix composition did not change during the experimental period
(at least with respect to these three compounds) .
Data collected on October 4 and 6, 1976 were analyzed after the
experimental work was completed. These data agree well with one another
and yield the relationship
X = P (^|% . (16)
This latter value yields closer agreement in predicting the height of
peaks of individual mix constituents. Thus, it was used to recalculate
the quantities of mix injected during the simulations.
Injections
Pollutants were injected into the chamber sides via the return side
of the glass sampling manifolds. These manifolds remove air from points
near the center of the chamber halves and return the unused portion to
the chambers just under a mixing fan. The flow is in excess of 60 1pm.
The return manifolds passing through the injection house are fitted with
SVL connectors through which the injections are made.
Pollutants are kept in gas cylinders at high concentration and
delivered to the injection system through two-stage, stainless steel
regulators. The injection system consists of on-off solenoid valves,
followed by either capillary flow restrictors or precision needle valves
with vernier handles. In the hydrocarbon and NO injection systems,
126
-------
either of two needle valves may be selected by a manual three way valve.
By setting these two valves at greatly different flow rates, and making
use of the computer to turn the solenoid on and off, it was possible to
produce a wide range of injection rates. The tank concentrations,
flowrates, and maximum injection rates are shown in Table 11.
TABLE 11. POLLUTANT INJECTION DATA
Pollutant
HC mix
NO
N02
acetaldehyde
^W^MMI^^IP^^M^M^-^PB^VV^V^^^W
Tank Concentration F
(ppm) Set tin
5920 ppmC 10
1
9127 6
0
2664 10
2200 ppmC 2
0
low Chamber Injection
g, 1pm Rate, ppm/min
0.4 ppmC/min
0.04 ppmC/min
0.37
.3 0.018
0.18
.96 0.044 ppmC/min
.30 0.0044 ppmC/min
.1 in . i - * i .^^» ^^^^
The hydrocarbon mixture used in these simulations is nearly the
same as the mixtures employed in the UNC chamber for other experiments
(83). The exact composition is given in Table 12.
The injection flowrates greater than 1 liter per minute (1pm) were
set and routinely checked with a Matheson 10-1pm mass flowmeter. The
mass flowmeter was calibrated against a 2-liter soap bubble flowmeter.
The soap bubble flowmeter had been carefully compared against a spiro-
meter, and its volume measured using a volumetric flask and a burette.
Occasionally the soap bubble flowmeter was also used directly. The
flowrates from 0.1 1pm to 1 1pm were measured with both a 1-lpm mass
flowmeter and a 48-ml soap bubble flowmeter. An initial comparison of
these two meters (considering barometric pressure, gas temperature, and
127
-------
TABLE 12. HYDROCARBON MIX COMPOSITION
CONCENTRATION OF COMPOUND, ppmC
Manufacturer ' s
Compound Analysis
2-methyl pentane
2,4-dimethyl pentane
2,2,4-trimethyl pentane
ethylene
1-butene
Cis-2-butene
isopentane
n-pentane
2-methyl- 1-butene
2-methyl- 2- butene
propylene
acetylene
total
690
504
712
736
164
168
725
1230
285
185
303
492
6194
Aug. 24, 1976*
494
481
636
738
147
193
807
1428
128
134
Oct. 8, 1976
491
623
663
568
180
241
805
1405
147
122
4920
Based on total tank concentration of October 8, 1976, and relative
component concentrations of August 24, 1976.
pressure at critical points in the system, and evaporation of water from
the soap bubble solution) revealed that the factory calibration of the
mass flowmeter was 10 percent low and the appropriate adjustment was
made. The lower acetaldehyde flowrate was set using a 50-ml soap bubble
flowmeter. Once set, these flowrates varied very little. Temperature
fluctuations of the injection house were reduced by a small, thermo-
statically controlled, electric heater. This was necessary for some of
the experiments in October because of the large diurnal temperature
ranges at this time of year.
Dilution
The dilution and exhaust systems are shown schematically in Fig-
ure 49. Prior to most experiments both sides of chamber were exhausted
for six to eight hours. Ambient air was drawn down the intake stacks
from an altitude of about 25 feet above the ground. It entered the
128
-------
10
n &> '
V3
a
B
V2
VI
Blast
Gate
BLUE
ii
Stack
J&.
V4
RED
4
Stack
0
KEY
I9k- Intake Doors
2,3- Exhaust Doors
Vl-VU- Ball Valves
A- Exhaust Fans
B- Dilution Fan
Figure 49. Schematic diagram of UNO outdoor smog chamber dilution and exhaust system.
-------
chamber halves via intake doors 1 and 4 and was drawn out by exhaust
blower A through exhaust doors 2 and 3. The chamber halves are mixed by
Teflon-coated impellers mounted on Teflon-coated, vertical shafts. When
dilution was required in the blue side for the simulations, door 1 and
valves 2 and 4 were closed; door 2 and valves 1 and 3 were open. Fab B
pulled air from the intake stack and blew it into the chamber, forcing
an equal quantity of air out the exhaust door. The dilution flowrate
was controlled by a blastgate upstream of the fan which was manually
adjusted every 6 min during the portion of the experiments requiring
dilution.
The blast gate was calibrated by measuring duct velocities in four-
point horizontal and vertical traverses of the 5-in. diameter duct at
various blast gate settings. Points in the traverse were selected as
representative of two concentric, equal-area zones (87). Measurements
were performed with a hot-wire aenomometer, at a point in the duct about
4 duct diameters upstream of the blast gate and over 10 duct diameters
downstream of the branch coming from valve 4. The range of flows in-
vestigated ran from 23 CFM, when the blast gate was completely closed,
to 210 CFM, corresponding to the highest velocity measurable with the
hot-wire aenomometer. Above 30 CFM, the blast gate flow control is not
very sensitive, i.e., small changes in the blast gate setting result in
large changes in flowrate. Thus, a great deal of care was exercised in
positioning the blast gate. It was found that even at flows near 200
CFM the suction just upstream of the blast gate was 0.2 to 0.3 in of
water, regardless of upstream valve configuration so long as some flow
pathway remained open from outside the chamber. Thus, the blast gate
calibration is valid whether simply drawing in outside air or diluting
with material from the red chamber (as in the final simulation experiment).
Experimental Procedures for Simulations
Because it was possible to estimate dilution and emission along the
LARPP trajectory for the period from 0815 PST to 1400 PST only, it was
necessary to bring the concentrations of precursors in the chamber at
0815 up to those observed in the zero-gradient region in operation 33 at
0815. Two techniques were employed for this purpose. In the earlier
130
-------
experiments, "slug" injections were made before sunrise so that con-
centrations of precursors were immediately brought up to their 0815
levels. In the latter runs, precursors were "ramped" into the chamber,
that is, injected at a constant, slower rate so that the proper con-
centrations will be reached by 0815. This last technique is a more
realistic, although quite simplified, representation of emissions in Los
Angeles (23). Prior to 0815, the chamber contents were not diluted.
After 0815 PST, both injection and dilution proceeded at the rates
determined from the LARPP data for the zero-gradient region. After
1400 PST, the reactions were allowed to proceed without additional
injection or dilution.
Preparation for a typical simulation experiment began the night
before the experiment was to take place. At this time, the computer was
set up for the data collection and control functions which were to be
performed, and the facility check out procedure was followed. The
computer is set up by running the data collection program and reading in
command files. Command files contain information about all the auto-
matic control functions to be performed. Some commands specify action
that is to occur once each time the command is given, while others
specify a repeating sequence of actions which are to occur until the
command is cancelled by another command, or until repeated the specified
number of times, Thus, the command files contain a list of commands,
each command specifies some of the following items: the action to be
taken, the priority of the command relative to other commands, the
specific computer channel on which the command is to be performed, cycle
time, "on" time during cycle, and computer time at which command should
be performed or initiated. In addition to action commands, there are
commands for saving a text string entered from the keyboard in a comment
file, altering the value of constants used in converting voltage to
physical units terminating the program and altering the computer soft-
ware clock.
The following functions were under computer control during a typi-
cal simulation run:
131
-------
1. Purge of chamber;
2. Shut off exhaust blower and close doors;
3. Commencement of scanning data channels, digitizing and storing
data on the disk;
4. Start up of automated GC and strip charts;
5. Slug or ramp injection of pollutants prior to 0815 in the
morning to establish initial concentrations;
6. After 0815, control variable rate, profile injection;
7. At the end of the experiment, open doors, exhaust chamber;
8. Turn off GC and strip charts;
9. Terminate program.
The following functions were controlled manually:
1. Unplugging red exhaust door and opening blue exhaust door;
2. Blast gate setting and turning dilution blower on and off;
3. Switching from large flowrate injection needle valve to small
flowrate valve at the proper time;
4. Calibrating GC;
5. Resynchronizing exhaust doors after the dilution was no longer
required;
6. Entering command files into computer at the proper time.
Two people were required to perform a simulation principally because
changing dilution settings consumed most of one person's time. The
sixth manual task above was eliminated in the later experiments because
the addition of more core enabled storage of all command files at once.
132
-------
RESULTS
During the period September 1, 1976 through October 16, 1976, 10
simulation experiments were performed. Conditions of these runs are
given on pages 134-169. In all but one of the experiments, the blue
side was operated in a dynamic mode to reproduce the processes of emis-
sion and dilution along the trajectory, while the red side was operated
in the static mode (as a batch reactor). In run number 9, both sides
were static in an attempt to determine what initial hydrocarbon concen-
tration would produce 0- equal to that observed on the LARPP trajectory
at the same initial HC/NO ratio as LARPP. In the first eight static
X
runs, attempts were made to produce initial concentrations of precursors
equivalent to the total moles injected in the dynamic chamber during the
day divided by the total volume into which these injections were made
(chamber volume plus total dilution volume). The results of these
experiments are shown in Figures 50 through 75. The results from each
day are preceded by a summary of experimental conditions.
N02 Injection
In the earlier experiments, hydrocarbon mix, NO, and N0_ were in-
jected early in the day and later in the simulation, the injection
consisted of hydrocarbons and NO only. In order to match the LARPP NO
and N0_ profiles, it was necessary to modify this approach. The desired
initial NO- was raised from 20 percent to 40 percent of the total NO
/ *
injected after the second run. Beginning with run number 7 it was
decided to add 35 percent N02 during the profile portion of the run as
well as the 40 percent initial NO- injection. There are two justifica-
tions for this increased proportion of N0_: (1) the uncertainty of the
extent of oxidation of NO that occurs in auto exhaust just after emis-
sion, and (2) the conversion of NO to N02 by the 03 above the inversion
base in the LARPP operation.
Nitrogen oxides have presented considerable problems in the mod-
eling of photochemical smog in the atmosphere. Eschenroeder and Martinez
(23) observed that the ratios of acetylene and carbon monoxide to N0x were
nearly always higher than the ratios based on these components in auto
133
-------
Run #1 September 1, 1976
Time: EDT
Blue Chamber; slug injections 0530 - 0600 EDT
initial concentrations:
HC - 2.89 ppmC
NO - 0.56 ppm
x
N02 - 20%
Simulation profile injection and dilution began at 0815 EDT.
All NO was NO.
x
Manual switch from high to low value for NO injection was not
made soon enough. NO shot up to 1.44 ppm. Thus, simulation
was cancelled.
Red Chamber; slug injections 0530 - 0600 EDT
Static all day
initial concentrations:
HC - 2.96 ppmC
NO - 0.56 ppm
X
N02 - 20%
134
-------
(Jl
1.00
0.90
0.80
0.70
e 0.60
OL
°i 0.50
o" 0.30
0.20
0.10
0.00
5.0
I ' I ' I ' I ' I ' I jl I M ' I ' I ' I I I '
SEPTEMBER 1,1976, RED CHflMBER -J
7.0 9.0 11.0 13.0
HOURS, EOT
15.0 17.0
1.00
0.90
0.80'
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
19.0
Figure 50. NO, N02, and 0^ profiles from static experiment of September 1, 1976.
-------
c
e
2.00
1.80
1.60
MO
1.20
E
? i.oo
S 0.80
Q
or
ce
ix:
01 0.60
(MO h-
° 0.20
0.00
SEPTEMBER 1,1976
Solar Radiation
Ultraviolet Radiation
70.0
60.0
50.0
O
O
D
30.0 E
20.0
10.0
5.0 7.0
9.0 11.0 13.0 15.0 17.0 19.0
HOURS, EOT
0.0
Figure 51. Solar radiation and ultraviolet radiation for September 1, 1976.
-------
Run #2 September 6, 1976
Time: EDT
Weather: Sky had been clear in the east very early in morning. By
0930 EDT, it was almost totally overcast and remained mostly
cloudy until noon.
Blue Chamber: slug injections 0530 - 0600 EDT
initial conditions:
HC - 3.55 ppmC
NO - 0.598 ppm
x r
N02 - 29%
Simulation profile injection and dilution began at 0815 EDT.
All NO input as NO during dynamic portion of run.
Ozone max of only 0.045 ppm occurred rather late (1538 EDT)
and was light limited. This was partially the result of morning
cloudiness.
Red Chamber; slug injections 0530 - 0600 EDT
initial conditions:
HC - 3.90 ppmC
NO - 0.531 ppm
X
NO 2 - 28%
The static side, while initial NO oxidation was slow, eventually
got going. Reached 0 max of 0.327 at 1458 EDT about twice the
0 produced in the LAKPP run.
137
-------
u>
00
, I . I . I . I , I .
SEPTEMBER 6,1976, BLUE CHflMBER
0.00
11.0 13.0
HOURS, EOT
19
0.00
.0
Figure 52. NO, N0~, and 0, profiles from simulation experiment of September 6, 1976.
-------
1.00
u>
1 I ' I ' I ' I ' I ' I ' I '
SEPTEMBER 6,1976, RED CHRMBER -I n.90
5.0
9.0
11.0 13.0
HOURS,EOT
15.0 17.0
0.00
19.0
Figure 53. NO, N02, and 03 profiles from static experiment of September 6, 1976.
-------
2.00
1.80
1.60
f 1.20
| 1.00
S 0.80
g 0.60
C£
_J
° 0.20 h-
0.00
5.0
T
T
Solar Radiation
Ultraviolet Radiation
SEPTEMBER 6,1976
70.0
60.0 <
50.0
fO.O
30.0
20.0
10.0
o
Q
7.0 9.0
11.0 13.0
HOURS, EOT
15.0 17.0 19.0
0,0
Figure 54. Solar radiation and ultraviolet radiation from September 6, 1976.
-------
Run #3 September 25, 1976
Time: EDT
Weather: Sky was totally obscured by fog in the early morning. By
0818, a light mist was falling. The fog started to break
up around 1000 and was clear by 1030 EDT.
Blue Chamber; slug injections 0530 - 0600 EDT
initial conditions:
HC - 4.85 ppmC
NO - 0.527 ppm
x rr
N02 - 45%
Simulation profile injection and dilution began at 0815 EDT.
Decided to abort at 0907 EDT because of weather and buildup
of NO. Chamber run in static mode for remainder of day.
At 1400 EDT, 0 was about half that observed in LARPP. Even-
tually made 0.20 ppm at 1630 EDT.
Red Chamber; slug injections 0530 - 0600 EDT
initial conditions:
HC - 4.85 ppmC
NO - 0.549 ppm
x
NO, - 40%
Static side mode 0.60 ppm of 0, peaking at 1630 EDT also.
141
-------
CD
1.00
0.90
0.80
0.70
0.60
- 0.50
_ ' I ' I
o
z:
. 0.30
0.20
0.10
0.00
I ' I ' I ' I ' I ' I ' I ' I ' I '
SEPTEMBER 25. 1976. BLUE CHflMBER
8 9 10 11 12 13
HOURS. EOT
1.00
0.90
0.80
0.70
0.60
0.50
15 16 17 18 19
0.30
0.20
0.10
0.00
Figure 55. NO, N0_, and 0 profiles from simulation experiment of September 25, 1976.
-------
CO
, I I I , I . I , , , I , I , , ,
SEPTEMBER 25, 1976, RED CHflMBER
8
9 10 11 12 13 11- 15 16 17 18 19
HOURS. EOT
Figure 56. NO, N07, and 0_ profiles from static experiment of September 25, 1976.
-------
E
i
o
a
o
»
a
en
2.00
1.80
1.60
1.1-0
1.20
1.00
0.80
0.60
£ O.fO
o
CO
0.20 -
0.00
,1,1
Ml1!1
i i i i
5.0 7.0
SEPTEMBER 25,1976
Solar Radiation
Ultraviolet Radiation
70.0
60.0
50.0
o
O
D
30.0
20.0
10.0
9.0 11.0 13.0 15.0 17.0 19.0
HOURS, EOT
0.0
Figure 57. Solar radiation and ultraviolet radiation from September 25, 1976.
-------
Run #4 October 4, 1976
Time: EDT
Weather: At 0826 EDT, sky was partially cloudy but bright. By 0932,
sky was totally clear but at 0945 became very cloudy again,
and remained so for the rest of the day.
Blue Chamber; ramp injection 0530 - 0815 EDT
concentrations at 0815:
HC - 3.24 ppmC
NO - 0.503 ppm
NO 2 - 40%
acetaldehyde - 0.36 ppmC
Simulation profile injection and dilution began 0815 EDT.
All NO injected as NO.
3C
System was slow and produced little 0- due to cloudy conditions.
Red Chamber; slug injection 0515 - 0530 EDT
initial conditions:
HC - 3.44 ppmC
NO - .495 ppm
2v
N02 - 20%
acetaldehyde - 0.38 ppmC
145
-------
.500
I .300
o.
- .200
o
.100
.000
1 I ' I ' I ' I ' I ' I ' I ' I ' I ' I ' I ' 1 ' I '
OCTOBER '''[ t \ r I r I r I _j_.1 .^hVV^H-^-i^L^-i-i-
.500
.300
.200
.100
5 6 7 8 9 10 11 12 13 14-
HOURS, EOT
15 16 17 18 19
.000
Figure 58. NO, N09, and 0^ profiles from simulation experiment of October 4, 1976.
-------
.500
*00
.300
o
« .200
o
o
.100
000
I ' I ' I ' I
1976, RED CHflMBER
.500
8
9 10 11 12 13
HOURS. EOT
- .400
- .300
.200
- .100
15 16 17 18 19
000
Figure 59. NO, m^ and 0^ profiles from static experiment of October 4, 1976.
-------
2.00
1.80
Y 1.60
"c
i 1.40
T§ 1.20
j_
a 1.00
o
M a 0.80
S cc
*0.60
cc
c! o.fo
CO
0.20
0.00
Solar Radiation
Ultraviolet Radiation
I
ni i M^I] i
OCTOBER 4, 1976
I i. I i I i_l
6 1 8 9 10 11 12 13 14 15 16 17 18 i9 20
HOURS EOT
.910
.819
.728
.637
.546 2*
.455 E
3
.364 *
.273
.182
.091
.000
Figure 60. Solar radiation, ultraviolet radiation, and N02 photolysis rate constant from
October 4, 1976. (UV scale is 74.57 mLy/min. maximum).
-------
Run #5 October 6, 1976
Time: EDT
Weather: Clear until noon; partly cloudy afternoon
Blue Chamber; initial injection command file failed to execute, thus
pollutants were slug injected manually at 0630 - 0700 EDT
initial conditions:
HC - 3.31 ppmC
NO - 0.500 ppm
x Vf
N02 - 40%
acetaldehyde - 0.37 ppmC
Simulation profile injections started at 0815 EDT.
Results here bear a much closer resemblance to the LARPP data.
The main difference is that events occurred later in the day.
The 0» produced was equal to that of the LARPP run, but it reached
this level 1 1/2 to 2 hours later in the day. This difference
was seen to be mainly due to the differences in solar time. The
solar time at any time during the morning in EDT in early October
is equivalent to the solar time on the LARPP run 90 min later in
PST. The subsequent simulations were performed by matching solar
times at which events occurred.
Red Chamber; slug injections 0630 - 0645 EDT
initial conditions:
HC - 3.27 ppmC
NO - 0.496 ppm
X
N02 - 19%
acetaldehyde - 0.36
Static side produced a 0.28 ppm 0 max at 1556 EDT.
149
-------
500
400
I. .300
o.
V)
O
« .200
.100
000
III I I
I ' I r I
.-...-'.
i ' I ' I ' I ' I ' I ' I ' I
OCTOBER 6, 1976, BLUE CHRMBER
r I '{ I r.l.r 1 r.J-^Kr I V'K^-L^. J.. r I., i r I ,
500
.^00
.300
.200
.100
5 6 7 8 9 10 11 12 13 1* 15 16 17 18 19
HOURS,EOT
000
Figure 61. NO, N02> and 03 profiles from simulation experiment of October 6, 1976.
-------
500
too -
i. .300
o.
o
N .200
o
o
z.
.100
.000
OCTOBER 6. 1976. RED CHflMBER .
.500
.300
-.200
.100
5 6 7 8 9 10 11 12 13
HOURS, EOT
15 16 17 18 19
.000
Figure 62. NO, N02, and 03 profiles from static experiment of October 6, 1976.
-------
Ui
2.00
1.80
T 1.60
i 1.40
*§ 1.20
g
. ' I ' I ' I ' I ' I ' I ' 1 ' I ' I ' I
o
-------
Run #6 October 11, 1976
Time: Corrected to PST
Weather: Clear until 1130 PST, then partially cloudy
Blue Chamber; ramp injection began at 0530 PST, but computer failure
caused this simulation to be aborted at 0602 PST
Red Chamber; slug injection 0500 - 0530 PST
initial conditions:
HC - 4.33 ppmC (based on injections)
NO - 0.506 ppm
X
N02 - 29%
acetaldehyde - 10% of total hydrocarbons
as carbon
Static side had 0 max of 0.257 ppm at 1506 PST.
153
-------
.500
.400 -
E .300
Q.
Q.
.500
HOURS, CORRECTED TO PST
- .400
- .300
I ' I ' I ' I ' I
OCTOBER 11,1976, RED CHflMBER
- .200
Figure 64. NO, N02> and 03 profiles from static experiment of October 11, 1976.
-------
£ . UU
1.80
- 1.60
i 1.10
i
M
1
g 1.20
i
~a 1.00
0
a 0.80
en
* 0.60
ce
cr
c! 0.10
CO
0.20
n nn
_ ' 1 ' 1 l 1 ' ! ' i ' 1 ' I ' i '
1 1
, j , j . , .
OCTOBER 11, 1976 _
Solar Radiation
_
Ultraviolet Radiation*
^
* 4.
^^
"~~" » K jl .
*00* <
""
1 * n *
. .i Is 1
v ii 1 1
"if t
fl
!r il
H
n
MI
+
j\
*
_ If
*
^.
^
'
" r ! , ! . i . I , 1 , 1 , 1 . 1 ,
, i
i
^MMH
0
MMH0I
^
«PVH»
^
__
__
I
r A ~
T fc
IHJ 1
yf|A «.
i I r %NV 1 i "
f^ ' ^r
.783
.696
.609
.522 f
.135 E
~^
.318 "
.261
.171
.087
nnn
8 9 10 11 12 13 11 15 16 17 18
HOURS EDI
19 20
Figure 65. Solar radiation, ultraviolet radiation, and NO. photolysis rate constant from
October 11, 1976. (UV scale is 64.90 mLy/min. maximum.)
-------
Run #7 October 12, 1976
Time: Corrected to PST
Weather: Clear all day
Blue Chamber; Ramp injection began at 0530 PST, but was stopped by
computer failure at 0534. Computer was not operable
until 0702 PST. At this time, slug injections were
made.
initial conditions:
HC - 3.76 ppmC
NO - 0.517 ppm
2C
N02 - 39%
acetaldehyde - 0.42 ppmC
Acetaldehyde added with profiled HC injection so that it made
up 10% of total HC as carbon.
35% NO in profile injection was NO..
X t-
Simulation profile injection and dilution started at 0815.
This simulation produced morning NO and N0» profiles which match
the LARPP profiles quite well. In the early afternoon however,
the NO - is somewhat lower than the LARPP values. This run pro-
duced about half the ozone observed in the LARPP run at 1400 PST
but the general shape of the ozone curve is the same.
Red Chamber; slug injection 0500 - 0530 PST
initial conditions:
HC - 4.38 ppmC
NO - 0.499 ppm
x
N02 - 21%
acetaldehyde - 0.49 ppmC
Static run made about 0.33 ppm of 0» at 1500 PST.
156
-------
U1
500
Q.
n
O
.300
« .200
o
o
21
.100
000
i I i I r i r
' I ' I ' I ' I ' I ' I
OCTOBER 12, 1976. BLUE CHRM3ER
,
i /
11
>>
tf
it
H
,-J - L ef - I
-»~
.500
.300
.200
.100
5 6 7 8
9 10 11 12 13 U- 15 16 17 18 19
HOURS, CORRECTED TO PST
000
Figure 66. NO, N02, and 03 profiles from simulation experiment of October 12, 1976.
-------
.500
500
V/i
00
- .4-00
.300
1 I 1 I I I
OCTOBER 12, 1976, RED CHflMBER -
E .300
- .100
8 9 10 11 12 13 1* 15 16 17 18 19
- .200
HOURS.CORRECTED TO PST
Figure 67. NO, N02> and 03 profiles from static experiment of October 12, 1976.
-------
2.00
1.80 -
T 1.60 h-
c
i 1.10 |-
7g 1.20 -
j_
a 1.00
o
r^ 0.80
01
CC
_J
O
CO
0.60
O.HO
0.20
0.00
I I ' I I I T
OCTOBER 12, 1976 -
Solar Radiation
Ultraviolet Radiation
i . i
t i t . i
t . I . I
8 9 10 II 12 13 14 15 16 17
HOUR,PST
.890
.801
.712
.623
18
.356
.267
.178
.089
.000
Figure 68. Solar radiation, ultraviolet radiation, and N0» photolysis rate constant from
October 12, 1976. (UV scale is 66.00 mLy/min maximum.)
-------
Run #8 October 13, 1976
Time: Corrected to PST
Weather: The sky was clear with a few scattered clouds in mid-morning.
The day was warmer than the two previous simulations, with
chamber temperatures in the lower 80s by early afternoon.
This is about 10 warmer than the two previous runs.
Blue Chamber: ramp injections 0530 - 0815 PST
initial conditions at 0815 PST:
HC - 4.01 ppmC
NO - 0.472 ppm
X.
N02 - 47%
acetaldehyde - 0.45 ppmC
Simulation profile Injection and dilution began at 0815 PST.
Dilution profile terminated at 1200 PST.
35% of N02 was as NO .
Acetaldehyde was injected such that it equalled 10% (as carbon)
of the total hydrocarbon injected.
This simulation came closest to the pollutant profiles observed
in the LARPP operation. N02 is slightly higher in the afternoon
than the N0« in LARPP. NO and 0- agree quite well with the LARPP
profiles.
Red Chamber; slug injections were made 0500 - 0530 PST
initial conditions:
HC - 3.84 ppmC
NO - 0.50 ppm
x
NO - 22%
liVA £f£f/O
acetaldehyde - 0.43 ppmC
Static side reached an 0_ max of 0.45 at 1500 PST.
160
-------
.500
.450
.400
.350
! .300
0.
« .250
o
j .200
\ .150
o
.100
.050
.000
i ' i ' i ' i ' i
i ' i ' i ' i ' i
OCTOBER 13, 1976, BLUE CHflMBER -
8 9 10 11 12 13 14 15 16 17 18 19
HOURS, CORRECTED TO PST
.500
.450
.400
.350
.300
.250
.200
.150
.100
.050
.000
Figure 69. NO, N0», and 0_ profiles from simulation experiment of October 13, 1976.
-------
.500
Ni
.350
.300
« .250
o
o
N .200
. .150
o
.100
.050
.000
I ' I ' I ' I ' I ' I ' I ' 1 ' I ' I ' I '
OCTOBER 13, 1976^RED CHflMBER _
.500
A5Q
8 9 10 11 12 13 11- 15 16 17 18 19
HOURS, CORRECTED TO PST
.350
.300
.250
.200
.150
.100
.050
.000
Figure 70. NO, N02» and 0_ profiles from static experiment of October 13, 1976.
-------
2.00
1.80 -
7 1.60 h-
I 1.40
0 1.20
"5 1.00
o
-; 0.80
85 en
cc
ce
or
o
CO
0.60
0.10
0.20
0.00
T^
i
I
i i
OCTOBER 13, 1976 -
Solar Radiation
Ultraviolet Radiation
**
89 10 II 12 13 14 15
HOURS, PST
16 17 18
.910
.819
.728
.637
.546 f
^M m**^
3^
.364 ~
.273
.182
.091
.000
Figure 71. Solar radiation, ultraviolet radiation, and N0_ photolysis rate constant from
October 13, 1976. (UV scale is 64.24 mLy/min maximum.)
-------
Run #9 October 15, 1976
Time: EDT
Weather: Clear
The intent of this run was to explore static run conditions
which would produce ozone equivalent to that of the LARPP run,
and to build up a second-day photochemical system for use in
next day's simulation. Slug injections were made at 0530 EDT.
initial conditions:
Blue Red
HC 1 ppmC 1.5 ppmC
NO 0.16 ppm 0.24 ppm
X
N02 20% 20%
HC/NO 6.22 6.22
X
The NO meter was measuring blue chamber only, thus red NO and
N0_ profiles are not known. The blue chamber conditions pro-
duced an ozone concentration about 1 pphm higher than the LARPP
day at 1530 EDT (equivalent to 1400 PST). The red side overshot,
producing 0.26 ppm at 1530 EDT.
164
-------
.500
1 ' ' ' ' I ' I ' I ' I
OCTOBER 15, 1976
.000
8
9 10 11 12 13 If 15 16 17 18
HOURS, CORRECTED TO PST
Figure 72. NO, N0_, and 0_ profiles from static experiments of October 15, 1976. Blue side:
1 ppmC HC mix Xdashed line); Red side: 1.5 ppmC HC mix (solid line); NO and NO
for blue side only.
-------
Run #10 October 16, 1976
Time: Corrected to PST
Weather: Clear
In this run, the contents of the blue chamber from the previous
day had been diluted so as to obtain a residual ozone concentration
of about 0.07 ppm. The same ramp and profiles injections employed
in Run #8 were made on top of the old material in the blue chamber.
Provisions were also made for dilution of the blue chamber using
material from the red chamber. The red chamber contained material
from the previous day also, but no new precursors were introduced.
This system was very fast. Ozone production was quite rapid com-
pared with previous runs. The 0_ max of 0.21 ppm occurred at 1300
PST. J
166
-------
.500
.450
350 -
t .300
o.
o
; .250 -
£ .200
". .150
o
.100
.050 h-
.000
| [ | [ | i | I | [ | I I [ I I i i i I
OCTOBER 16, 1976, BLUE CHflMBER _J 450
8 9 10 11 12 13 14 15 16 17 18 19
HOURS, CORRECTED TO PST
Figure 73. NO, NCL, and 0. profiles for simulation experiment of October 16, 1976. Performed
with same material from previous day in chamber and same dilution with air from
red side of chamber.
-------
00
.500
.450
.400
.350
t .300
Q.
.250
o
N .200
o
2. .150
o
^ .100
.050
.000
I ' I ' I ' I ' I ' I ' I
I ' I ' I ' I '.
OCTOBER 16, 1976, RED CHflMBER -
1 I r I , I t I i I i I t I t i t -I i I i I , I t I
8 9 10 11 12 13 11- 15 16 17 18 19
HOURS, CORRECTED TO PST
.500
.450
.400
.350
.300
.250
.200
.150
.100
.050
000
Figure 74. NO, NO and 0. profiles for October 16, 1976, red side. Materials left from the
previous day only.
-------
2.00
1.80
1.60
vo
E
's
O
CJ
o
a
en
or
QZ
CE
o
co
1.20
1.00
0.80
0.60
0.40
0.20
0.00
I ' I ' I ' 1 ' I ' I ' I ' I ' I ' ! ' I ' I ' I '
OCTOBER 16, 1976
Solar Radiation
Ultraviolet Radiation
I.I.
70.0
60.0 <
50.0
40.0 o
30.0 §
20.0
10.0
8 9 10 11 12 13 14 15 16 17 18 19
HOURS, EOT
0.0
Figure 75. Solar radiation and ultraviolet radiation for October 16, 1976.
-------
exhaust (California driving cycle). Since acetylene and CO are emitted
almost exclusively from automobiles, they concluded that some as yet
unknown process was removing NO from the air, or at least from the gas
X
phase. They were faced, when validating their model, to reduce the
emission inventory NO source strengths by a factor of 4 to obtain
X
reasonable 0- predictions. Reynolds et al. (87) refer to the substan-
tial loss of NO observed in chamber irradiations using auto exhaust as
Jt
opposed to that observed for pure hydrocarbon-NO systems. Both groups
X,
speculated that the loss was probably a conversion to the aerosol phase.
Lee et al. (89), found that the nitrate content of exhaust particulates
was increased by a factor of 175 if the gases were first irradiated
until the oxidation of NO to NO- was well under way. Crittenden's (90)
data on nitrogen aerosol difunctionals show double diurnal maxima which
roughly correspond to the hours of heaviest traffic. Difunctionals not
containing nitrogen displayed diurnal behavior similar to secondary
species such as PAN or 0~ with a single maximum in the afternoon. Thus,
the chemistry of recently emitted auto exhaust probably involves con-
version of NO to nitrate. The first step in this process may be the
rapid oxidation of some NO to NO^ before the exhaust can be substan-
tially diluted.
Chan et al. (91) hypothesize that Reaction 38 may result in the
oxidation of 25% of the NO in auto exhaust
NO + NO + 02 J 2 N02 (38)
to N02 during this dilution process. In addition to this, Reactions 39
and 40
NO + N02 + H20 £ 2HONO (39)
HONO + hv -> HO + NO (40)
may provide a source of hydroxyl radicals to rapidly initiate the typi-
cal free radical NO oxidation pathway described in Chapter I. The
estimates of the effectiveness of these processes in oxidizing NO to
N02 depends upon assumptions concerning the dilution experienced by auto
exhaust puffs on a molecular scale. Very little information is available
170
-------
concerning this phenomenon; however, it is reasonable to speculate that
wind speed is an important factor in dilution of auto exhaust. The
interaction of the wind with roughness elements on the ground is the
major source of small-scale turbulence necessary for mixing of the
exhaust puffs with surrounding air. Thus, the higher the wind speed,
the less important the above reactions should become. The very low wind
speed during operation 33 would tend to emphasize their effects.
The second factor which could result in conversion of NO to N02 in
the morning was the turbulent entrainment of ozone-bearing air from
above the inversion base. The 0 concentration below 1000 feet and
above the inversion base was 6 to 7 pphm by 0920 PST. From 0815 to 0915
PST, the mixing depth rose from 500 to 1000 feet. Thus, the 0 in
this time period could account for the oxidation of 3500 pphm-feet of
NO. At 0915, the total NO below the inversion was about 40,000 pphm-
X
feet. Thus, during this time period, about 8-9 percent of the total
nitrogen oxides could have been oxidized by direct reaction with ozone.
Thus, the injection of NO as 35% N0_ is a reasonable estimate of
X ^
the effects of various localized processes which affect the N09/N0 ratio.
£» X
Acetaldehyde Injection
Another modification of the experimental procedure was the addition
of acetaldehydes to the hydrocarbon mixture for all the runs during
October. Dimitriades and Wesson (92) stated that the total aldehyde
levels in auto exhaust from automobiles at that time was about 5 percent
of the total hydrocarbon on a carbon atom basis. They reported the
typical composition of exhaust aldehydes in mole percent as follows:
formaldehyde 60
aliphatic with more than 1 carbon 32
benzaldehyde 5
aromatic with more than 7 carbons 3
The contributions of exhaust aldehydes to the reactivity of a mixture
containing the typical constituents of auto exhaust is a function of
aldehyde type. Aldehydes react primarily in three ways: (1) photolysis
171
-------
to give radicals; (2) photolysis to give stable species; and (3) re-
action with hydroxyl radicals.
Formaldehyde, when photolyzed to radicals, produces two hydroperoxy
radicals, since k. , is
4.6
CH20 + hv (X < 3700 A) -> H + HCO (41)
H + 02 -> H02 (42)
HCO + 02 -> CO + H02 (33)
HCO + 02 + M -> HC002 + M (43)
expected to be greater than k. 7 (8). On HO attack, formaldehyde yields
H /
one HO- radical (by way of HCO).
HO + CH20 -» H20 + HCO (44)
On photolysis to radicals, acetaldehyde yields one H02 (by way of HCO)
and the CH- radical. CH- may participate in further chain oxidation, to
convert one NO to N0_ or to generate another HO-. On hydrogen abstrac-
tion by an oxidizing agent such as HO, acetaldehyde may produce PAN when
NO concentration is low, or may oxidize two NO molecules to NO- and
regenerate an H0_ by chain degradation. Thus, the principal difference
between the effects of formaldehyde and acetaldehyde from a mechanistic
point of view is that formaldehyde is expected to have a greater effect
on NO to NO- conversion at the beginning of a run, while higher aliph-
atic aldehydes have the potential for slightly greater direct contri-
butions to the conversion of NO to NO- per carbon atom over a longer
time. In the outdoor chamber, it may be expected that the effects of an
early initial production of H0_ would be more important than the longer
term effects since timing of events is quite important under diurnal
light variation. Also, it was thought from the September simulations
that conversion of NO to NO- was slower than the LAKPP runs in the early
morning. Thus, it was decided to add aldehydes in amounts to simulate
the effects of aldehydes in the early portion of the runs.
172
-------
Because formaldehyde is difficult to inject (adsorbs to walls of
manifold and chamber), it was decided to simulate the effects of all
aldehydes by addition of acetaldehyde. The photolysis rate constant,
considering only radical producing photolysis, for formaldehyde is
almost twice that for acetaldehyde in simulated sunlight (93).
Formaldehyde is 60 mole percent of exhaust aldehydes, and it would
require the addition of twice that many moles of acetaldehyde on the
basis of photolysis rate and roughly two times that again on the basis
of H02 yield just after photolysis. Thus, it was believed that about
four times as many moles of acetaldehyde would be required to simulate
the initial effects of formaldehyde on the photochemical system. Since
aldehydes are about 10 mole percent of the total hydrocarbons, it would
require about 22 mole percent acetaldehyde to simulate the initial
effects of the mixed aldehydes present in auto exhaust. This is roughly
10 percent of the total hydrocarbons on a carbon atom basis. Since the
hydrocarbon injection rates based on source reconciliation estimates did
not include an estimate of aldehydes in the air, it was decided to add
acetaldehyde such that they would represent 10 percent of the total
hydrocarbon on a carbon atom basis after addition, with no change in the
amount of hydrocarbon mix added.
Solar Time Correction
The run of October 6 was the first run for which the weather and
experimental technique were both reasonably good, although the afternoon
was partially cloudy. It was noted that the N0-N02 crossover and the
onset of 0» production were occurring at later times than during LARPP
operation 33. It was thought that this may be due to differences
between the solar time on November 5 in Los Angeles and the solar time
on October 6 in Chapel Hill. The TJV radiation measured during the LARPP
operation at the downtown LAAPCD headquarters plotted against pacific
standard time was compared to the UV measured on October 6 versus east-
ern daylight time. When the time scales were initially matched hour for
hour, it was necessary to translate the October 6 UV curve 90 minutes to
the left to make it match the LARPP curve. Thus, the simulation injec-
tion and dilution profiles had been applied to the chamber 1 1/2 hours
173
-------
too early In the solar day to match the events of 1AEPP. In subsequent
runs, a correction was made for this by setting the internal computer
clock back 90 min from eastern daylight time. The time for these runs
is reported as "corrected to PST." Thus, injection and dilution events
were made to occur at approximately the same solar time as the atmos-
pheric events they simulate.
Summary
The best simulations of LARPP operation 33 are runs 7 and 8 (Octo-
ber 12 and 13). The NO, N02, and 0- profiles for these two experiments
are shown with the LARPP zero-gradient region concentration profiles in
Figures 76 and 77. It may be seen from Figure 76 that the N0_ concen-
tration in the simulation was above the observed N0« until about 1300
PST. The match between the observed and simulated NO was excellent.
The 0_ concentration observed along the trajectory was a little less
than twice that observed in the simulation.
On October 13, the experimental N0« values remained higher than the
LARPP values for the entire experiment. The 0_ values, as in run 7,
were below the observed ozone concentration during the morning, but
ozone production after noon was more rapid. Thus, by 1400 PST the ozone
had climbed to within 2 pphm of the observed value.
The hydrocarbon mix concentrations for both October 14 and 15 are
compared with the NMHC regression line from source reconciliation in
Figure 78. The mix concentrations represent the mix concentration in
the chamber if none were removed by reaction since it was calculated
from the concentrations of relatively unreactive n-pentane. The rate of
removal of n-pentane is probably quite low, as may be seen from the
nearly flat mix profiles at the end of these two experiments, after
dilution and injection had terminated. The hydrocarbon concentrations
were slightly below the source reconciliation line.
The differences between the profiles of secondary species in the
simulation and along the LARPP trajectory will be considered in detail
in Section 7.
174
-------
-J
Ui
0.5
0.4
E
Q.
a.
. 0.3
ro
1 O
0,2
i1
0,1
---2s.
***** ^
il 13 15
TIME, HOURS PST
17
19
Figure 76. Comparison of NO, N02, and 0- profiles for October 12 simulation (dashed line) and
LARPP zero-gradient region (solid line).
-------
0,5
0.4
E
Q.
CL
- 0.3
rO
O
< 0,2
-------
5
4
o
£
a.
o
3 L
2 i-
0
SOURCE RECONCILIATION
OCTOBER 13
OCTOBER 12
II 13 15
TIME, HOURS PST
17
Figure 78. Comparison of NMHC source reconciliation regression line with NMHC mix concentrations
estimated from n-pentane peak for simulations of October 12 and 13.
-------
The static experiments performed during the experimental period
were not carefully watched and checked to be sure the proper injections
had been made; the simulation experiments consumed nearly all the time
of those involved in the experimental work. Also, the concentrations
employed were calculated from an earlier version of the INJECT program
which did not take the mixing height into account. These concentrations
are higher than would be used from the total injection over the total
volume employed in the simulation experiments. The hydrocarbon con-
centration should have been 2 ppmC with an NO concentration of 0.32
X
ppm. The experimental conditions and the ozone concentrations obtained
are summarized in Table 13.
TABLE 13. SUMMARY OF CONDITIONS AND RESULTS OF STATIC RUNS
Run
No.
Date
Initial Conditions
HC,
ppmC
N0x,
ppm
N09,
sr
HC/NO
Weather
Ozone,
@1400
PST
ppm
Maxi.
1
2
3
4
5
6
7
8
9
Sept.
Sept.
Sept.
Oct.
Oct.
Oct.
Oct.
Oct.
Oct.
1
6
25
4
6
11
12
13
15
2.96
3.90
4.85
3.44
3.27
4.33
4.38
3.84
1
0.56
0.53
0.55
0.50
0.50
0.50
0.50
0.50
0.16
20
28
40
20
19
29
21
22
20
5.
7.
8.
6.
6.
8.
8.
7.
6.
29
36
82
88
54
66
76
68
25
overcast
overcast morn.
foggy until
10:30
cloudy after
09:45
partly cloudy
after noon
partly cloudy
after 11:30
clear
clear, warm
clear
0.
0.
0.
0.
0.
0.
0.
0.
0.
03
03
10
09
06
23
30
40
12
0.19
0.33
0.59
0.18
0.28
0.26
0.33
0.45
0.17
blue
9 Oct. 15
red
1.5 0.24 20
6.25 clear
0.17
0.27
178
-------
Based on the results from the three clear days, as shown in Fig-
ure 79, the increase of 03 produced with initial hydrocarbon is noted
for experiments with a small range of HC/NO ratios. It may be con-
2S
eluded that a static experiment with 2 ppmC hydrocarbon mix and a
HC/NO ratio similar to these runs would produce more ozone than was
X
observed along the LARPP trajectory (0.15 ppm at 1400 PST) or in the
simulation experiments.
In experiment number 10, October 16, the effect of the presence of
material remaining from the previous day was tested. The chamber con-
tents from the 1 ppmC static run the day before had been diluted until
the ozone reached about 8 pphm. This was to allow for dark reactions of
ozone during the night, and was estimated to leave 5 to 6 pphm by morn-
ing. In addition, some of the diluent air was drawn from the red side
of the chamber which contained an active second-day photochemical sys-
tem. While this experiment was not a direct simulation of LARPP condi-
tions, it may in an exaggerated way, demonstrate the effect of convec-
tive entrainment of the photochemical system above the inversion base
observed in the morning of operation 33. The reactions were greatly
accelerated and the experiment produced more 0, than was observed along
the LARPP trajectory.
179
-------
0.5
0.4
0.3
E
Q.
°-0.2
uT
o
M
O
O.I
--+ OZONE 1400 PST
o OZONE MAXIMUM
1234
NMHC CONCENTRATION, ppmC
Figure 79. CL concentrations at 1400 hours and maximum produced
versus initial NMHC concentration during static
experiments on clear days.
180
-------
SECTION 7
DISCUSSION
The results of experimental simulations were presented in the
previous chapter. The simulations of October 12 and 13, 1976, most
closely simulate the precursor emissions and dilution which occurred
along the trajectory in LAEPP operation 33. Several important vari-
ables, however, were not under the control of the experimenter, namely,
solar radiation, temperature and reactant inhomogeneity. The effects of
these factors on the simulation must be evaluated in order to explain
differences between the simulation results and the LARPP data. Only the
experiments of October 12 and 13 (runs 7 and 8) will be considered in
detail.
SOLAR RADIATION
Solar radiation influences the formation of photochemical smog in
many ways. The most direct effect is that the photolysis rate of NO- and
other species which photolyze is dependent upon the intensity and spec-
tral distribution of solar radiation. Solar radiation also influences
air temperature, mixing depth, and wind field characteristics. For the
purpose of comparing the simulation to the LARPP operation, only the
effects on photolysis rate need be considered in this section. The
effects of temperature will be considered elsewhere, and the other
effects were accounted for in specifying the conditions for simulation.
The photolysis of NO is the most important photochemical reaction
in the smog system. As discussed in Chapter I, only photons with wave-
lengths less than about 410 nm are likely to cause photodissociation of
N0_ after absorption. The intensity of actinic radiation in the zero-
gradient region at any time of day is a function of zenith angle of the
181
-------
sun, the concentration of absorbing species and scattering particles
(particularly along the path of the direct solar beam), and the albedo
of the earth's surface.
Calvert (26) computed k.. for LARPP operation 33 based on solar
constant data of Johnson (94) and Ackerman (95) which is roughly 10 per-
cent higher than the values of some other investigators. For estimating
the attenuation due to 0., absorption, aerosol, and molecular scattering,
Calvert used the "standard relations adjusted to the conditions that
approximate Operation #33" from Leighton (3). An CL column of 2.8 mm
was employed but no reference was made to the estimates of the water and
dust factors used to calculate the transmissity for particulates.
Calvertfs estimates of the ultraviolet (UV) (290-390 nm) flux on a
horizontal surface above the boundary layer agrees incredibly well with
actual UV radiometer measurements from Mt. Disappointment. When an
attempt was made to duplicate these calculations using Johnson's solar
constant data, the same 0, column and Leighton's tabulated attenuation
coefficients for molecular scattering, it was found that some adjustment
of the water and dust coefficients was necessary to obtain agreement
with Mt. Disappointment radiometer values from operation 33. Figure 80
shows radiometer readings and the estimated UV flux from theory for a
water vapor factor of 2 and a dustiness factor of 2. These water and
dust values seem to be somewhat high compared with those cited by Leighton;
however, it was more important to obtain a good fit with the radiometer.
It may be that the solar constant data employed is too high, thus forc-
ing the use of higher water and dust attenuation to fit the observations.
Calvert (26) calculated k-/k3 taking into account absorption within
the boundary later by N0_, reflection at the earth's surface and attenua-
tion of the reflected beam by N0~. He compared the k./k_ values for
albedoes of 0, 0.25 and 0.5 with [0 ][NO]/[NO] from the LARPP data.
Under homogeneous conditions, k^k- should be approximately equal to the
concentration ratio. The best estimate of albedo for wavelengths im-
portant in NO- photolysis is about 0.05 according to Peterson (96).
Interpolating to this albedo in Calvert's figure, it was noted that the
182
-------
10.0
30.0 -
4-
O
- 20.0
Q
s £
C*i r*"*i
cr
10.0
0.0
TT
40.0
30.0
20.0
10.0
6.0 8.0
10.0 12.0 H.O
-HOURS. PST
16.0 18.0
0.0
Figure 80. Comparison of Mt. Disappointment UV radiometer measurements with theoretical
calculation of UV radiation from 290 to 390 nm above the boundary layer in
Los Angeles on November 5, 1973.
-------
vast majority of ratio points fall well above the k-/k- line. Calvert
noted that this may be due to inhomogeneity.
Computations were performed to estimate k- and k-/k^ as described
in Appendix D. The resulting values are shown in Figures 81 and 82.
ATMOSPHERIC INHOMOGENEITY
General Characterization
Before the effects of atmospheric inhomogeneity can be estimated,
the general nature of inhomogeneity must be considered. The mixing of
one fluid into another takes place by two distinct processes when the
fluids are miscible (97). First, the fluid is broken up into clumps, or
packets. The shape of the clumps will depend upon the mechanism of
mixing. For instance, the shape of clumps of auto exhaust of a moving
auto will probably be somewhat elongated in the direction of motion.
The mean size of these clumps decreases with continued mixing so long as
the size of some of the clumps is greater than the smallest scale size
of turbulent eddies. It should also be noted that if the clump sizes
are much greater than the scale size of eddies, then breakup of the
clumps occurs only around the edges of the clumps, and thus is rather
slow.
Concurrently, molecular diffusion occurs across the boundaries of
the clumps. This has the effect of diluting the material within the
clumps and the boundary between clumps and diluent fluid becomes pro-
gressively blurred. This process, given sufficient time, will reduce a
mixture of miscible fluids to uniformity even in the absence of turbu-
lence although it is rather slow unless turbulence has reduced the size
of the clumps. If diffusion does not occur (as with immiscible fluids),
then the fluids cannot be made homogeneous regardless of turbulence.
The Scales of Inhomogeneity
In considering the effects of inhomogeneity on the interpretation
of the LARPP simulation, it is useful to distinguish between two dis-
tinctly different effects.
184
-------
00
O.45
0.40
0,35
0.30
0.25
C
'I
0,20
0,15
0,10
0.05
10
II
12
13
o 200 FT
A 400 FT
14
TIME, HOURS PST
Figure 81. Theoretical estimates of k.., the rate constant of N0» photolysis at four altitudes
above ground level in Los Angeles on November 5, 1973.
-------
2,0
oo
200 FT
400 FT
600 FT
800 FT
14
15
13
Figure 82.
(0 ti 12 13
TIME, WOURS PST
Theoretical estimates of k../k3, the ratio of the N02 photolysis rate constant to the
rate constant of the NO-0- reaction at four altitudes above ground level in Los
Angeles on November 5, 1973.
-------
Small Scale Inhomogeneity
Small scale inhomogeneity affects individual measurements. Since
each instrument has a finite averaging time and the helicopters were in
motion, each measurement is a spatial composite, a weighted average
concentration over some distance. Because many of the species are
reactive, their spatial average concentrations may not be identical to
their concentrations under well-mixed conditions. To demonstrate this
effect, consider two parcels of air which have arrived at a given alti-
tude level. Parcel 1 has just arrived from ground level and contains
0.005 ppm of 03, 0.09525 ppm of N02, and 0.30476 ppm of NO. Parcel 2,
just entrained from above the inversion, contains 0.05 ppm of 0_,
0.03788 ppm of N02, and 0.01212 ppm of NO, as shown in Figure 83. Both
parcels are in photostationary state at a k- /k_ of 0.016 ppm. If meas-
urements of the NO, N02, and 03 are taken within either air parcel, the
photostationary state equation for ozone,
kx [NO] [03]
k [N02]
should be very nearly correct. The only processes causing deviations
from this equality are a few reactions which may compete for 0- with the
NO reaction (for example, reaction with NO- or alkenes) . On the other
hand, consider an instrumented aircraft measuring NO, NO,, and 0« con-
<& J
centrations which crosses the air parcel boundary in such a way that
the instrument readings represent the average of the concentrations
within the two air parcels: 0.02750 ppm of 03> 0.06657 ppm of N02, and
0.15844 ppm of NO. If the k,/k3 is calculated from Equation 17 using
measured concentrations, a value of 0.065 ppm is founded: about 4 times
greater than the true k./k.. This is quite similar to Calvert's (26)
results. Inhomogeneity on a sub-measurement scale was probably respon-
sible for a large portion of the disagreement of the theoretical k^/kg
and the estimates from Equation 17 in Calvert's results.
When such small scale inhomogeneity exists, the average rate of
reactions of order greater than one is not necessarily equal to the rate
based on mean concentrations. In the example given above, the reason
187
-------
P2
[03] =0.05 ppm
0.03788 ppm
[NO] = 0.01212 ppm
[NO ] =0.05 ppm
X
Figure 83. Two hypothetical parcels of air at photostationary
state at k-,/k3 = 0.016 ppm.
188
-------
for the large difference between the true ^/k.j and [NO][03]/[N02] is that
the rate of the N0-03 reaction is much slower than would be calculated
from the average NO and 0^ concentrations. Let the instrument readings
for NO and 03 be represented by [NO] and [0,] (spatial averages from a
single measurement) and the deviations of the true concentration from
the instrument reading at any point in space be represented by [NO]' and
[03]'. Then the actual concentrations at a point are given by
[NO] = [NO] + [NO]1 (18)
[03] = [03]
[03]f
(19)
Substituting these concentrations into the rate equation for the NO-0,.
reactions yields:
d[03]
~~dt
k3(([NO]
[NO]') ([03]
(20)
the rate of reaction at any point. The average rate of reaction over
/\ /\
the spatial region from which the [NO] and [0 ] were measured requires
averaging both sides of the above equation.
d[03]
~dt
(21)
[N0]'[03]')
[N0][03] = [N0][03]
[N0][03]' = [N0][03]' = 0
[N0]'[03] = [N0]'[03] = 0
(22)
Thus,
d[03]
- -~-- k3([NO][03] + [N0]'[03]')
(23)
The fluctuation term on the right of Equation 23, [NO]' [03]', is equal
to the covariance of the act
the equation may be written
to the covariance of the actual NO and 03 concentrations (71). Thus,
189
-------
d[0 ]
- -~- = k3([NO][03] + cov(NO,03»
Under photostationary state conditions the rate of ozone destruction by
NO is approximately equal to the rate of N0_ photolysis. Thus, the
ratio of k1 to k« under inhomogeneous conditions is:
k, [N0][0_] + cov(NO,0,)
3, 3_ 1_ (25)
k3 [N02]
where the hat concentrations represent single instrument readings from
composite sampling such as those obtained by aircraft cutting across
many air parcels during a single measurement.
In the above example concerning the two air parcels, the covariance
2
of NO and 0 is -0.00329 ppm ; this yields the correct value of k-/k3 in
Equation 25.
Equation 25 is very useful. If k-/k_ may be estimated by some
method and if the NO, N0_, and 0., concentrations are measured, the
covariance of NO and 0« may be estimated in the atmosphere for each
measurement. It should be repeated that in Equation 25, the hat con-
centrations represent single measurement values. In a well-mixed smog
chamber the covariance is zero; thus, Equation 25 reduces to the more
familiar form under homogeneous conditions.
Large Scale Inhomogeneity
Large scale inhomogeneity results in variation of helicopter meas-
urements about the pattern means, rather than a variation of the com-
position of the material comprising a single sample. Based on the
helicopter speed, the scale of such inhomogeneity is on the order of
2
5x10 feet, the distance between 6-second measurements. This scale of
inhomogeneity is best shown by the projection plots in Figures 11 through
17. This variability results in an alteration of the photostationary
state equation for the pattern average ozone concentration.
190
-------
Solving for the photostationary state ozone in Equation 25 yields:
/* \ / \
k /[N07]\ /cov(NO,0_)\
[03] = irKH-
J k3\fNO] ] \ [NO]
Thus, the mean photostationary state 0 concentration around a heli-
copter pattern is
cov(NO,0)
[03] = ^ -^- i- (27)
3 [NO] [NO]
Since the covariance term is, for most measurements, negative, the
atmosphere is capable of maintaining a greater mean photostationary
state 0- concentration than is the smog chamber ait the same mean N00 to NO
j 2
ratio and the same k.. /k,.
_L 3
In the above discussion, it was not the intention to imply that the
N0-0_ reaction is the only one affected by inhomogeneity in the atmos-
phere. Because the NO-0- reaction is the most rapid second order reac-
tion in the atmosphere (in ppm/min), it is the best candidate for dif-
fusion limitation. Other reactions whose rates approach those of N0-0_
may be diffusion limited, for instance some of the reactions of HO or
organic radicals (71). In general, the criterion for judging the
importance of inhomogeneity for a particular reaction is that the rate
of chemical reaction (in concentration/time units) must be at least as
rapid as the rate at which turbulent mixing tends to bring the reactants
together. Donaldson and Hilst (72) state this principle and character-
ize the potential for inhomogeneous effects by the ratio
N = -^__ (28)
A Ka Ca
where D is the molecular diffusivity of the reactants (assumed to be
equal for two different reactants, X is the dissipative scale length of
turbulence in the atmosphere (taken to be 10 cm), KM is the reaction
rate constant and ~C is the reactant concentration. In evaluating N,
a _
Donaldson and Hilst let C equal one ppm for all reactions, which leads
to exceedingly high values for the rates of some reactions involving
191
-------
species which have concentrations much lower than one ppm in the atmos-
phere. Thus, their conclusions concerning the importance of inhomo-
geneity for particular reactions must be looked on with skepticism.
The influence of inhomogeneities, while seeming to have a strong
influence on the 0- along the trajectory did not seem to greatly effect
the NO or N02 profiles except as pointed out in Section 6 (rapid oxida-
tion of NO in relatively undiluted, irradiated auto exhaust). Thus, in
this work the primary concern is with the increased photostationary
state 0- concentration relative to the average N02 and NO.
In exploring the theoretical aspects of the effects of inhomo-
geneity, an effect of light on the ozone concentration was also shown,
as in Equations 26 and 27.
The average NO-/NO ratio was also computed for each helicopter
pattern so that the pattern mean cov(NO,0_)/[NO] could be estimated.
This is the portion of the photostationary state 0_ concentration which
resulted from atmospheric inhomogeneity. The average product of the NO
and 0_ concentrations was also determined so that the pattern average
covariance could be calculated as follows:
1
cov(NO,03) = :£- [N02] - ([N0][03]) (29)
Figure 84 shows the pattern average covariances of NO and 0,. It is
shown here that the covariance reached large negative values above the
inversion base in the early morning. Within the zero-gradient region,
there is no temporal trend evident; values fluctuated about a mean of
-3 2
about -1 x 10 ppm .
It should be pointed out that these are simply estimates of the
true atmospheric covariances. There are several possible sources of
error in these estimates. (1) Errors in estimating k- are possible but
are not likely to be large since excellent agreement was obtained between
theory and the radiometer on Mt. Disappoint. (2) The instrumental
error, including interference, is another possibility. The values for
NO probably include PAN and possibly nitric acid. This would tend to
192
-------
U>
6.0
8.0
10.0
HOURS PST
12.0
Figure 84. Estimated helicopter pattern average covariance of NO and 0» from Operation 33.
-------
lead to some overestimation. (3) In general, the instrument responses
are weighted averages of concentrations in air from more than one air
parcel. The accuracy of estimation of the average concentration is
affected by the shape of instruments' weighting functions, the heli-
copter air speed, the size of the air parcels and the true variability
of pollutant concentrations from one parcel to the next.
The pattern average ratios of the covariance to the NO concentra-
tion are shown in Figure 85. It is interesting to note that the most
negative values of this ratio occurred in the morning above the inver-
sion base. Inhomogeneity seems to account for 3 to 5 pphm of 0. in this
region. Thus, the absence of vigorous turbulent mixing in this region
y>
allowed segregation of 0 and NO. The high value of cov/[NO] at 1257
PST resulted from a small increase in the average N0_ concentration.
The trajectory seems to have passed over a source of N02 at this time.
Emission of NO- would tend to produce air parcels with higher co-existent
concentrations of both NO and 0_. For the purpose of comparing smog
-* *.
chamber results with the LARPP trajectory, the cov/[NO] values from
outside the zero-gradient region need not be considered. Also, the high
positive value at 1257 PST seems to have affected boundary-layer 0, for a
very short time and the brief emission of N0« was not simulated since it
seemed to have little affect on atmospheric 0_. Thus, excluding these
points, the mean cov/[NO] was found to be -1.90 pphm with a standard
deviation of 2.24 pphm.
An interesting feature of the data is that the variability of
cov/[NO] exceeds the variability of the pattern mean ozone. A hand-
fitted smooth curve was passed through the pattern mean 0» points in
Figure 27 and the deviations of pattern means from this line were
measured. The root mean square deviation from the line was 0.58 pphm;
XV
this is only one-fourth the deviation exhibited by cov/[NO]. Thus, it
must be concluded that, while Equation 27 is applicable at any time, it
does not fully describe the dynamic behavior of the relationships among
03, N02> NO, and the covariance of NO and 0 . This topic is discussed
in Appendix E. The covariance of NO and 0_ is the result of processes
which tend to reduce inhomogeneity (such as fresh emission and "
194
-------
O.L
6
o.
a.
O
ZL
10.0
HOURS PST
Figure 85. Estimated helicopter pattern average ratio of covariance of NO and 0 to the NO
concentration. 3
-------
photochemical oxidation of NO). The average 0. concentration is, in
general, not as greatly affected as is cov/[NO]. For instance, if
turbulent mixing brings a portion of the atmosphere to homogeneity in
which the covariance had originally been negative, the increase in the
N07 to NO ratio partially compensates for the loss of 0» due to the
mixing of parcels containing low 0_-high NO. (See Appendix E.) Thus,
A, «5
the use of the cov/[NO] estimated from the atmosphere to show the effects
of inhomogeneity in comparing the LARPP data with the chamber simulation
is correct; however, it must not be assumed that, if the atmosphere were
well-mixed, it would maintain an 0_ concentration equal to the mean
^
observed ozone minus cov/[NO]. Clearly such mixing would change the NOj
to NO ratio.
TEMPERATURE
The effect of temperature on the photochemical reaction complex is
just beginning to be understood from a mechanistic point of view.
Experiments in the UNC outdoor smog chamber have shown possible temp-
erature effects. Figure 86 shows three propylene experiments at nearly
the same initial NMHC, NO, and N0? concentrations. There was very
little difference in the solar radiation on these days; however, the
temperature profiles did differ. October 15 was the coolest of these
days for most of the day. At 0800 EDT, the temperature was about 47° F
rising to 77° F by 1300 EDT. October 12 showed nearly the same temper-
ature profile between 0800 and 1040 EDT. After 1100 EDT, the tempera-
ture remained 3 to 4 F above the temperature on October 15. October 16
was hotter than the other two days for the entire run. From 0800 to
1115 EDT, the temperature profile was about 5 to 6° F above the profile
of October 12; this difference decreased to about 3° F by 1300 EDT. In
this series of runs, the maximum ozone produced increased with the
temperature.
Jeffries et al. (52) developed a mechanistic explanation for the
effect of temperature. A typical treatment of PAN formation and decom-
position is the following:
196
-------
CH3C(0)02- + NO -» N0 + CHC(0)0- (45)
3
k, 800 ppm min"
CH3C(0)02- + N02 * PAN (46)
k46 = 100 ppm"1 min"1
PAN -> N03 + CH3- + C02 (47)
-3 -1
k^7 = 3x10 min
Recent work by Hendry et al. (98), indicates that reaction 47 is not
significant. Instead PAN decomposition is the reverse of reaction 46:
PAN -> N02 + CH3C(0)02 (48)
k4g = 1.507xl017 exp(-12931/T)min~1
3 1 1
k,g = 2.45x10 ppm min
2 -1 -1
k,, = 7.36x10 ppm min
5he sequence of reactions 45, 46, and 48 is then analogous to the reac-
tions of N03 and N-0-, reactions 6, 9, and 10.
Another similar reaction sequence is:
H02 + NO * N02 + OH (49)
H02 + N02 * HONO + 02 (50)
HONO ^-> NO -I- OH (51)
Hendry (1976) suggested that reaction 50 is not an elementary reaction
but goes by way of an intermediate.
HO + N02 + HOON02 (52)
HOON02 -» H02 + N02 (53)
HOON02 -» HONO + 02 (54)
197
-------
vo
00
ex
D.
1.00
0.90
0.80
0.70
_ 0.60
°" 0.50
«
§ 0.^0
g 0.30
0.20
0.10
0.00
OCTOBER 12, 15,
8
9 10 11
HOURS, EOT
12
Figure 86. Comparison of NO, N02> and 03 profiles from static propylene experiments in UNC outdoor
chamber. Oct. 16 ( ). NMHC = 3.88 ppmC; Oct. 12 ( ), NMHC = 3.91 ppmC:
Oct. 15 ( ), NMHC = 3.88 ppmC.
-------
The similarity of reactions 45, 46, and 48 with 49, 52, and 53 suggest
that reaction 53 may have a large temperature dependence as does 48.
Under conditions of a typical diurnal temperature profile, reac-
tions 45, 46, and 48 should have the effect of storing some quantity of
N02 early in the day as PAN. As temperature increases, the equilibrium
concentration of PAN decreases due to the increased rate of the decom-
position reaction, causing a release of additional NO This is the
probable cause of most of the difference in the NO profiles in Figure
87. This contributes to the 03 concentration to the extent that it
alters the N02/N0 ratio. At this time of day, however, this ratio is
much more sensitive to changes in the NO concentration than changes in
the NO- concentration. At lower temperatures, reactions 51 through 54
have the effect of converting NO- to NO, since reaction 54 can compete
with 53 (at 77° F, k5,/k5_ is about 1/12). As the temperature rises,
the specific rate of 53 increases much more rapidly than the specific
rate of 54 (at 93° F, k54/k53 is about 1/25, and at 105° F, it is
about 1/60). Thus, the rate of conversion of N0_ to NO by this pathway
decreases with increasing temperature. This can greatly alter the
N02/N0 ratio, which in turn changes the photostationary state 0-.
Jeffries et al. (52) presented three model runs for hydrocarbon mix
using the same initial concentrations but different temperature pro-
files. The results are shown in Figure 87. It may be seen that morning
temperature seems to have little effect but the afternoon temperature
shows considerable effect on the 0 concentrations produced. Note that
the NO- and NO profiles are changed very little, but, since 03 is a
function of the ratio of NO- to NO, the 0- profiles show a substantial
temperature effect.
DISCUSSION OF COMPARISONS
The direct comparison of the LARPP zero-gradient region concen-
trations with the simulation experiments of October 12 and 13 were pre-
sented in Figures 76 and 77. To evaluate how closely the chemistry in
the chamber matched the chemistry in the atmosphere, the effects of
199
-------
1.00
co
o
o
0.00
0. 60. 120. 180. 2*0. 300. 360. *20. *80. 5*0. 600. 660. 720.
MINUTES
Figure 87. PKSS model of 0.22 ppm N0x> 2.00 ppmC EC mix for three temperature profiles; ( )
73° F to 105° F; ( ) 73° F to 93° F; (....) constant 77° F.
-------
variables not under experimental control must be quantified, and the
magnitude of possible errors discussed.
It should first be pointed out that the lines representing the
zero-gradient region profile in Figures 76 and 77 are smoothed repre-
sentations of the zero-gradient region pattern averages as shown in
Figure 88. The vertical deviations from these smooth lines is often 1
pphm and can be as high as 2 or 3 pphm. Thus, there is some small
uncertainty as to the placement of the LARPP concentration profile
lines. The NO profiles seem to have matched the NO profiles of the
LARPP data extremely well. In the latter portion of the runs NO con-
centrations go quite low; small errors in measurements take on great
importance. At this time, small changes in the NO concentration dras-
tically affect the N02/N0 ratio and, therefore, the 0~ concentration.
Thus, it is difficult to ascertain the effect of small disagreements
between the LARPP and chamber NO on the 0_ concentration comparison.
The N0_ concentration in the simulations initially remained above
the LARPP N0_- In the October 12 experiment, the NO concentration is
quite close to the LARPP N0_ after 1230 PST. This is probably fortui-
tous since no attempt had been made to simulate the small rise in NO
observed at 1230 PST in the LARPP data. On October 13, the NO remained
above the LARPP values because simulation dilution was discontinued
after 1200 PST. Also, the NO- measurements in LARPP and the simulations
are performed with chemiluminescent instruments. These instruments are
known to have interferences in the NO mode such as PAN and nitric acid.
A
It is not known how much the interfering species contributed to the N02
measurements in either LARPP or the simulations. In Los Angeles, PAN is
thought to be roughly 10 percent of the total oxidant (24). This is not
greatly different from the PAN to oxidant ratio from numerical models of
the UNC hydrocarbon mix (99).
The assumption that hydrocarbon and NO emissions remain in the
X
same ratio for the entire operation is indeed a simplification, but was
expedient in lieu of a reliable means of estimating N02 and NO emissions
along the trajectory. This could be refined in further simulations,
perhaps by use of emissions inventory information, at least to weight
201
-------
o
N>
. TUW
.300
2:
a.
OL.
m
f»
0 .200
a
*ZL
or
M
Sf
III If
.100
«
o
"Z.
«bM>
.000
1 I ' 1 ' I l I ' 1 ' 1 ' 1 '
* NO
* N02
O Q^
*
4-
a a
«H +H M
x a
a * a
a
+ MM
MM MM
- B M ««a BM O O-
+ + + MM OMB O C
O
M O H
ooo us a
OO -"""!
ooo
+ 0
o + ooo o
O 4- +
OO * + -I- 4- -
O ++» + +
OOO
r ! r 1 ° r 1° ° r 1 r 1 r 1 V *V* I***"
6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 1*
.400
.300
- .200
.100
000
PflCIFIC STflNDRRD TIME
Figure 88. NO, N0», and 0_ helicopter pattern average concentrations from LARPP Operation 33.
-------
the NMHC/NOx injection rate ratio differently at various times during
the simulation.
One further consideration is that the measured concentrations of
03, NO, and N02 are affected to some extent by dark reactions in the
sampling manifold. A method was devised for correcting the data for the
03-NO reaction as described in Appendix F. The correction is greatest
around 03~NO crossover since this is when the product of the 0_ and NO
concentrations is greatest. Applying this correction to the data of
October 12, it was found that the correction had a maximum value of 1
pphm, and decreases before and after crossover. This increases the
agreement between the simulation and the atmosphere. (While the cor-
rection is small in considering concentration alone, it is quite im-
portant in comparing [03][NO]/[N02] in the chamber with measured k-/k3.)
The difference in light intensity is another factor to be consid-
ered. Figure 89 shows the range of k-/k» values calculated for the
LARPP operation, as well as k1/k3 values from October 12 and 13. The ^
chamber values are based on measurements taken inside the smog chamber
(Figures 68 and 71). Since the k.. sensor was sometimes shielded by
structural components of the chamber it was necessary to fair a smooth
line through the k- points to obtain the approximate average chamber k-.
k» was calculated as a function of chamber temperature as:
k3 - 1330 exp (-1200/T) ppnf1 min"1 .
k-/k_ was higher for both simulations than for the LARPP operations;
however, it should be pointed out that there are discrepancies between
k../k_ determined from the k., sensor and temperature data and k1/k3
determined from Equation 17 using corrected concentrations. For Octo-
ber 12, the sensor yields k-^kg values of 0.022 and 0.013 ppm at 1030
PST and 1430 PST versus corresponding concentration ratio values of
0.0144 and 0.00583 ppm. Thus, the disagreement was about 7 to 8 ppb for
these two times; this is more important in a relative sense at the
latter time since the absolute values of k-/k3 are smaller. Possible
errors are present in both means of quantitating k-^k^ The following
items may cause error in the evaluation of Equation 17.
203
-------
K)
O
IS
8
10 II 12 13
TIME, HOURS PST
Figure 89. Comparison of ratio of NO- photolysis rate constant to NO-O, reaction rate constant,
k-/k_, inside chamber on October 12 and 13, 1976, to the range of estimated k../k,
values for altitudes from 200 feet to 300 feet in LARPP Operation 33.
-------
(1) The concentration values were only corrected for reaction of 0
and NO, when in reality other dark reactions are taking place, for
instance, C>3 and N02, and 03 and olefin. These reactions would result
in a negative bias in estimating k^/k...
(2) The interferences present in the N0_ measurement would also
produce lower k./k3 estimates.
(3) The ratio is extremely sensitive to small errors in the NO
concentration late in the day.
(4) Equation 17 itself is a simplification which does not account
for reaction of 0 with species other than NO. This, just as items (1)
and (2) above, produces a negative bias in the k../k_ estimate.
For the k^ sensor, Sickles and Jeffries (85) considered the effect
of normally distributed errors in all input variables. The 95 percent
confidence interval at a measured k.. of 0.5 min~ was + 10%; at a
measured k of 0.1, it was + 8.7%. Thus, it is likely that the kj/k-
estimate from the k.. sensor is closer to the true value.
If the chamber k-i/k, had equaled the LARPP values, the 0 simula-
tion profiles would have been lower, thus increasing disagreement. At
1400 PST, the 0_ concentration of October 12 would be about 4 ppb lower
and, on October 13, it would be about 2 pphm lower. This accounts only
for the effect of light intensity on photostationary state, given the
same NO and N0» concentration. It is not possible to take the effect of
light intensity into account regarding the rate of conversion of NO to
N07 in such a complex system short of modeling, however, it is not
expected to be large compared with the effect on photostationary state
(based on comparison of October 12 with October 13).
Atmospheric inhomogeneity, as discussed earlier, adds an average of
2 pphm to the LAKPP 0_ profile. This accounts for about 13 percent of
the LARPP 0- concentration at 1400 PST and increasingly larger per-
centages earlier in the day- Recognition of this effect tends to in-
crease agreement of the LARPP 03 with the simulation 03 profiles.
205
-------
The differences in temperature between LAKPP and the simulations of
October 12 and 13 are shown in Figure 90. The small temperature dif-
ference between LAKPP and October 12 is not expected to greatly in-
fluence 0- concentration. The temperature difference in the afternoon
of October 13 ranges from 8 to about 18 degrees. Judging from the
Figure 87 the 0 produced by this simulation may have been reduced by 20
to 30 percent if run under the temperature conditions of the LAKPP
operation.
In addition to the effect of variables just discussed, instrument
errors add to the uncertainty of the comparison. Also, differences in
calibration method may cause the comparison to be biased although this
is very difficult to evaluate through retrospection.
To summarize, it may be stated that there are two classes of fac-
tors affecting the comparison: random and systematic. The random
factors are the following:
(1) Errors in concentration measurements which are due to offsets
in calibration. This could easily amount to + 10% considering the sum
of the errors in LAKPP and chamber data for higher concentrations and
much greater percentages for NO late in the day.
(2) Random errors in estimates of k^/k, in the chamber have an
approximate 95 percent interval of + 10% and the estimates in the atmos-
phere may be in error by this amount also.
The systematic factors are the following:
(1) Differences in light intensity, accepting the estimates of
k-^/k , add an additional 20 to 30 percent to the 0_ values for October 13,
and somewhat less for October 12, thus decreasing the agreement.
(2) Dark reactions may add up to 16 to 20 percent to the 0 con-
centration observed in the simulation at 0_-NO crossover, thus increas-
ing agreement.
(3) Temperature differences probably plays little part in the com-
parison of October 12 with LAKPP, but may be responsible for as much as
30% of ozone production of October 13, thus decreasing agreement.
206
-------
to
o
100
90
80
70
- 60
LJ
50
£40
CL
UJ
20
10
LARPP
OCT 12
OCT 13
8 9 10 II
TIME, HOURS PST
12
13
14
Figure 90. Comparison of the chamber temperatures on October 12 and 13 with range of helicopter
pattern average temperatures from LARPP Operation 33.
-------
(4) Atmospheric inhomogeneity accounted for about 2 pphtn of the
observed LARPP 0_, thus increasing agreement.
Considering the possible magnitudes of errors and the effects of
light, temperature, and inhomogeneity, it may be concluded that agree-
ment between the LARPP data for the zero-gradient region and the simu-
lation results is quite good. The agreement is especially good if one
considers some of the simplifying assumptions which were made in re-
lating atmospheric processes to chamber injection and dilution, for
example, (a) the use of constant NMHC/NO injections; (b) the use of a
2v
constant composition hydrocarbon mix; (c) the straight-line approxima-
tion of source reconciliation results; and (d) the use of numerical
computation to calculate NMHC flux values in the atmosphere.
The higher values of k-./k_ would have been expected to produce
higher 0_ values in the simulation than in the atmosphere if the NO./NO
ratio were the same as the average ratio in the atmosphere, even after
correction for inhomogeneity to the 0, concentration. The conclusion to
be drawn is that the N02/N0 ratio was in fact lower in the chamber than
in the atmosphere. For reasons discussed earlier, it is very difficult
to obtain good estimates of this ratio directly using typical chemilum-
inescent monitoring instruments. Since the values 0- and k../k., are
probably subject to less error than the nitrogen oxide concentrations,
it is possible to estimate NCL/NO ratio as the ratio of 0» to k.,/k,.
Note that in the atmosphere, it is not the true average ratio, but the
ratio if the atmosphere were homogeneous. This may be termed the effec-
tive NO-/NO ratio. The ratios were compared for three times of day in
Table 14. It is shown that the effective ratio in the atmosphere was
greater than the simulation ratio, even when corrected for inhomogeneity
by subtracting 0.02 ppm from the LARPP 0. values. Differences of this
magnitude could not result from differences in N0«, but could more
easily be caused by small differences in the NO concentration. For
instance, if the N02 values are approximately correct, the difference in
N02/N0 ratio between LARPP and October 12 (corrected for dark reactions)
can be explained by a difference in NO concentrations of 1 pphm. This
208
-------
TABLE 14. ESTIMATES OF N02/N0 RATIO FOR SIMULATIONS
OF OCTOBER 12 AND 13 AND THE EFFECTIVE NO /NO RATIO FOR THE
LARPP DATA CORRECTED AND UNCORRECTED FOR INHOMOGENEITY
Time, PST
N02/N0
LARPP
Uncorrected
Corrected
October 12
October 13
1000
1200
1400
2.33
5.7
10.9
1.2
4.5
9.4
1.0
2.8
4.6
1.0
2.4
6.3
indicates that the simulations had either a lower radical flux or more
NO to be oxidized than the LARPP trajectory.
Recent experiments in the UNC outdoor smog chamber suggest a pos-
sible cause for lower chamber NO oxidation rates, and thus lower 0.^. A
static experiment was performed at a hydrocarbon mix concentration of 1
ppmC and an NO concentration of 0.35 ppm with 20 percent N02 on both
2C
sides of the chamber. In addition, one side contained formaldehyde at a
concentration equal to 10 percent of the total nonmethane carbon while
the other side contained acetaldehyde at a concentration equal to 10 per-
cent of the total nonmethane carbon. The formaldehyde side oxidized NO
to N0? more rapidly and produced a maximum 03 concentration of 0.24 ppm
versus 0.08 ppm maximum 03 in the acetaldehyde side. This indicates
that the presence of formaldehyde could play a much greater role in the
0 formation potential than was thought at the start of this exper-
imental program. Thus, the rationale for the use of acetaldehyde only
(presented in Section 6) was probably incorrect. Further simulations
should use a mix of formaldehyde and acetaldehyde representative of the
atmosphere.
209
-------
Another factor was the convective entrainment of what appeared to
be second-day material from above the inversion base. As results from
October 16 indicate, the presence of second-day material can enhance the
rate and extent of NO oxidation.
Another simplification is the assumption of no horizontal diffusion
and the subsequent evaluation of NMHC vertical flux. After this work
was completed, Crane, Panofsky, and Zeman (100) estimated vertical flux
of CO after 9:15 PST by assuming values of the emission rates at the
ground. The disagreement of their flux estimates with the estimates in
this report lead Feigley and Jeffries (101) to conclude that some cross-
wind diffusion of CO away from the trajectory had occurred. It is
probable that some NMHC and NO also diffused away from the trajectory,
X
which would imply that both injection and dilution rates were somewhat
underestimated. While the concentrations of nonreactive species in-
jected into the chamber would not be changed, the concentration of
reactive precursors would be increased by increased injection and di-
lution. It is difficult to determine how this would affect the ozone
concentration since it would increase both the NO to be oxidized and the
reactive hydrocarbon available for chain reactions which result in NO
oxidation.
210
-------
REFERENCES
1. Haagen-Smith, A. J., C. E. Bradley, and M. M. Fox. Ozone Formation
in Photochemical Oxidation of Organic Substances. Indus. Eng.
Chem., 45(9):2080-2089, 1953.
2. Stephens, E. R., P. L. Hanst, R. C. Doerr, and W. E. Scott. Reac-
tions of Nitrogen Dioxide and Organic Compounds in Air. Indus.
Eng. Chem., 48:1498-1504, 1956.
3. Leighton, P. A. Photochemistry of Air Pollution. Academic Press,
New York, 1961.
4. Wayne, L. G. The Chemistry of Urban Atmospheres. Los Angeles
County Air Pollution Control District, 1962.
5. Haagen-Smit, A. J., and L. G. Wayne. Chapter 6. In: A. C. Stern,
ed., Air Pollution, Vol. I, Academic Press, New York, 1968. p. 149.
6. Pitts, J. N., Jr. Adv. in Environ. Sci. Technol., 1:289, 1969.
7. Altshuller, A. P., and J. J. Bufalini. Environ. Sci. Technol.,
5:39, 1976.
8. Demerjian, K. L., J. A. Kerr, and J. G. Calvert. Adv. Environ.
Sci. Technol., 4:1, 1974.
9. Environmental Protection Agency. National Primary and Secondary
Air Quality Standards and Air Pollution Control. Fed. Reg., 36(21):
Part II, 1971.
10. Stephens, E. R. Chemistry of Atmospheric Oxidants. JAPCA, 19(3):
183, 1969.
11. Seinfeld, J. H. Air Pollution: Physical and Chemical Fundamentals.
McGraw-Hill, New York, 1975. p. 151.
12. Seinfeld, J. H. Air Pollution: Physical and Chemical Fundamentals.
McGraw-Hill, New York, 1975. p. 165.
13. Hampson, R. F., Jr., and D. Garvin. Chemical Kinetic and Photo-
chemical Data for Modeling Atmospheric Chemistry. National Bureau
of Standards Technical Note 866. U.S. Government Printing Office,
Washington, DC, 1975. p. 59.
211
-------
14. Jeffries, H. E. 1975. personal communication.
15. Gartrell, G., Jr., and S. K. Friedlander. Relating Particulate
Formation to Sources: The 1972 California Aerosol Characterization
Study. Atmos. Environ., 9:279, 1975.
16. McNeils, D. N. Aerosol Formation from Gas Phase Reactions of Ozone
and Olefin in the Presence of Sulfur Dioxide. PhD Thesis, Univer-
sity of North Carolina, Chapel Hill, North Carolina, 1974.
pp. 148-152.
17. Hasar, R. B., K. T. Whitley, and B. Y. H. Liu. Physical Mechanisms
Governing the Los Angeles Smog Aerosol. J. Coll. Inter. Sci.,
39:211-224, 1972.
18. Heisler, S. L. Gas-to-Particle Conversion in Photochemical Smog:
Growth Laws and Mechanisms for Organics. PhD Thesis. California
Institute of Technology, Pasadena, California, 1976.
19- Zeman, 0., and H. Tennekes. Parameterization of the Turbulent
Energy Budget at the Top of the Atmosheric Boundary Layer. J.
Atmos. Sci., 1976. In press.
20. Edinger, J. G. Changes in the Depth of the Marine Layer Over the
Los Angeles Basin. J. Appl. Meteorol., 16(3):219, 1959.
21. Edinger, J. G. Modification of the Marine Layer over Coastal
Southern California. J. Appl. Meteorol., 2:706, 1963.
22. Roth, P. M., P. J. W. Roberts, M. -K Liu, S. D. Reynolds, and J. H.
Seinfeld. Mathematical Modeling of Photochemical Air Pollution
II. A Model and Inventory of Pollutant Emissions. Atmos. Environ.,
8:97-130, 1974.
23. Eschenroeder, A. Q., and J. R. Martinez. Concepts and Applications
of Photochemical Smog Models. Photochem. Smog Ozone React. R. F.
Gould, ed., American Chemical Society, Washington, DC, 1972. pp.
101-168.
24. Air Quality Criteria for Photochemical Oxidants. National Air Pol-
lution Control Administration, Pub. No. AP-63, Washington, DC,
1970.
25. Panofsky, H. A. 1976. personal communication.
26. Calvert, J. G. Test of the Theory of Ozone Generation in Los
Angeles Atmosphere. Environ. Sci. Technol., 10(3):248-256, 1976.
27. Hecht, T. A., and J. H. Seinfeld. Development and Validation of a
Generalized Mechanism for Photochemical Smog. Environ. Sci. Technol.,
6:47, 1972.
212
-------
28. Wayne, L. G., R. Danchick, M. Weisburd, A. Kokin, and A. Stein.
Modeling Photochemical Smog and Computer for Decision Making.
Pacific Conference on Chemistry and Spectroscopy, 6th Western
Region Meeting, ACS, San Francisco, California, October 6-9, 1970.
29. Air Quality Criteria for Hydrocarbons. Ma, NAPCA Publication No.
AP-64. U.S. Department of HEW, PHS, NAPCA, 1970. p. 5-4.
30. Huess, J. M., G. J. Nebel, and J. M. Cblucci. National Air Quality
Standards for Automotive Pollutants A Critical Review. J. Air
Poll. Cont. Assoc., 21(9):535-544, 1970.
31. Altshuller, A. P. Relationships Among Pollutants Undergoing Atmos-
pheric Reactions in CAMP Cities in the U.S. Presented at the 63rd
Annual Meeting of the APCA, APCA Paper No. 70-113, June 14-18, St.
Louis, Missouri, 1970.
32. Angell, J. K., C. R. Dickson, and W. H. Hoecker, Jr. Relative Dis-
persion Within Los Angeles Basin as Estimated from Tetroon Triads.
Air Resources Laboratories, National Oceanic and Atmospheric Admin-
istration, Silver Spring, Maryland, November 1974.
33. National Academy of Sciences. Panel on Emission Standards and
Atmospheric Chemistry. A Critique of the 1975-76 Federal Auto-
mobile Emission Standards for Hydrocarbons and Oxides of Nitrogen,
May 22, 1973.
34. Dimitriades, B. Effects of Hydrocarbon and Nitrogen Oxides on
Photochemical Smog Formation. Environ. Sci. Techno1., 6(3):253-260,
1972.
35. Los Angeles Air Pollution Control District. Los Angeles County Air
Pollution Control District Views on the Emission Control Strategy
to Achieve the Oxidant Air Quality Standard. Los Angeles, California,
April 5, 1974.
36. Los Angeles Air Pollution Control District. Answers to EPA Ques-
tions Posed in April 22, 1974 Letter (Covington to Lunche). Los
Angeles, California, June 21, 1974.
37. Fox, D. L., R. M. Kamens, and H. E. Jeffries. Photochemical Smog
Systems: Effect of Dilution on Ozone Formation. Sci., 188:1113,
1975.
38. Shuck, E. A., J. N. Pitts, and J. K. S. Wan. Relationship Between
Certain Meteorological Factors and Photochemical Smog. Inter. J.
Air Water Poll., 10:689-711, 1966.
39. Chock, D. P. and S. B. Levitt. A Space-Time Correlation Study of
Oxidant and Carbon Monoxide in the Los Angeles Basin. Atmos.
Environ., 10:107-113, 1976.
213
-------
40. Paskind, J., and J. R. Kinosian. Hydrocarbon, Oxides of Nitrogen
and Oxidant Pollutant Relationships in the Atmosphere Over California
Cities. APCA Paper No. 74-31. Presented at the 67th Annual Meet-
ing of the APCA, Denver, Colorado, June 9-13, 1974.
41. Edinger, J. G., M. H. McCutchan, P. R. Miller, B. C. Ryan, M. J.
Schroeder, and J. V. Behar. Penetration and Duration of Oxidant
Air Pollution in the South Coast Air Basin of California. J. Air
Poll. Cont. Assoc., 22(11):882-886, 1972.
42. Earth, D. S. Federal Motor Vehicle Emission Goals for CO, HC, and
NO Based on Desired Air Quality Levels. Air Poll. Cont. Assoc.,
20*519-523, 1970.
43. Larsen, R. I. A Mathematical Model of Air Pollution Concentration
Averaging Time and Frequency. J. Air Poll. Cont. Assoc., 19:29, 1969.
44. Larsen, R. I. A Mathematical Model for Relating Air Quality Meas-
urements to Air Quality Standards. Preliminary Publication No. AP-
89. Environmental Protection Agency, Office of Air Programs,
Research Triangle Park, North Carolina, 1971.
45. Hamming, W. J., R. L. Chass, J. E. Dickinson, and W. G. MacBeth.
Motor Vehicle Control and Air Quality: The Path to Clean Air for
Los Angeles. APCA Paper No. 73-73. Presented at the 66th Annual
Meeting of the APCA., Chicago, Illinois, June 24-28, 1973.
46. Korth, M. W., A. H. Rose, Jr., and R. C. Stahman. Effects of
Hydrocarbon to Oxides of Nitrogen Ratios on Irradiated Auto Exhaust.
J. Air Poll. Cont. Assoc., 14:168, 1964.
47. Kamens, R. M., H. E. Jeffries, D. L. Fox, and L. Alexander. A Smog
Chamber Study of the Potential Effects of Hydrocarbon Reductions on
Nighttime NO- Concentrations. Atmos. Environ., 11:225-229, 1977.
48. Trijonis, J. C. Economic Air Pollution Control Model for Los
Angeles County in 1975. Environ. Sci. Technol., 8(9):811-826,
1974.
49. Hamming, W. J., and J. E. Dickinson. Control of Photochemical Smog
by Alteration of Initial Reactant Ratios. Presented at the 57th
Annual Meeting of the APCA, APCA Paper No. 64-110, Houston, Texas,
June 1964.
50. Myrabo, L. N., P. Schleifer, and K. Wilson. Oxidant Prediction
Model for Land Use and Transportation Planning. Calif. Air Environ.,
1974. In press.
51. Trijonis, J. C. Economic Air Pollution Control Model for Los
Angeles County 1975. PhD Thesis, California Institute of Tech-
nology, Pasadena, California, 1972.
214
-------
52. Jeffries, H. E., R. M. Kamens, and C. E. Feigley. Outdoor Simula-
tions of Air Pollution Control Strategies. Progress Report EPA
Grant 800916. Department of Environmental Science and Engineering
School of Public Health, Univ. of North Carolina, Chapel Hill, North
Carolina, 1977.
53. Stephens, E. R., and F. R. Burleson. Distribution of Light Hydro-
carbons in Ambient Air. J. Air Poll. Cont. Assoc., 19(12):929-
936, 1969.
54. Bailey, B. S., and K. H. Ludlum. Factors in Achieving the Oxidant
Standard. Presented at 26th Southeastern Regional Meeting of
American Chemical Society, Norfolk, Virginia, October 24, 1974.
55. Merz, P. H., L. J. Painter, and P. R. Ryason. Aerometric Data
Analysis Time Series Analysis and Forecast and an Atmosphere
Smog Diagram. Chevron Corporation, Los Angeles, California,
1970.
56. Dimitriades, B. An Alternative to the Appendix-J Method for Cal-
culating Oxidant- and N02-Related Control Requirements. Inter-
national Conference on Photochemical Oxidant Pollution and its
Control Proceedings. Vol. II. EPA-600/3-77-001b. Raleigh, North
Carolina, 1976. p. 871.
57. Dimitriades, B. On the Function of Hydrocarbon and Nitrogen Oxides
in Photochemical Smog Formation. U.S. Bureau of Mines. Report of
Investigations RI-7433, Bartlesville, Oklahoma, 1970.
58. Dodge, M. C. Combined Use of Modeling Techniques and Smog Chamber
Data to Derive Ozone-Precursor Relationships. International Con-
ference on Photochemical Smog Pollution and its Control Proceed-
ings. Vol. II. EPA-600/3-77-001b. Raleigh, North Carolina, 1976.
p. 88.
59. OAQPS, 1977.
60. Parker, R. 0., and J. R. Martinez. Los Angeles Reactive Pollutant
Program (LARPP) Data Archiving and Retrieval. ERT Document No. P-
1464-W, Environmental Research and Technology, Inc., Concord,
Massachusetts.
61. Shepard, D. A Two-Dimensional Interpolation Function for Irregu-
larly-Spaced Data. Proceeding - 1968 ACM National Meeting, 1968.
62. Lea, D. A. Vertical Ozone Distribution in the Lower Troposphere
near an Urban Pollution Complex. J. Appl. Meteorol., 1(2):252-
267, 1968.
63. Edinger, J. G. Vertical Distribution of Photochemical Smog in Los
Angeles Basin, Environ. Sci. Technol., 7(3):247-252, 1973.
215
-------
64. Johnson, W. B., and H. B. Singh. Oxidant Layers Aloft: Their
Origin and Significance. International Conference on Photochemical
Oxidant Pollution and its Control. Raleigh, North Carolina,
September 12-17, 1976.
65. Edinger, J. C. Preliminary Analysis of LARPP Data. Unpublished
Report to the Coordinating Research Council, New York, New York,
1975.
66. Jeffries, H. E., J. E. Sickles, II, and L. A. Ripperton. Ozone
Transport Phenomena: Observed and Simulated. APCA Paper No. 76-
14.3. Presented at APCA Annual Meeting, Portland, Oregon, 1976.
67. Lumley, J. L., and H. A. Panofsky. The Structure of Atmospheric
Turbulence. Interscience, New York, 1964.
68. Trijonis, J. C., and K. W. Arledge. Utility of Reactivity Criteria
in Organic Emission Control Strategies for Los Angeles. Report for
Contract No. 68-02-1735. Environmental Protection Agency, Research
Triangle Park, North Carolina, 1975.
69. Panofsky, H. A. A Model for Vertical Diffusion Coefficients in a
Growing Urban Boundary Layer. Boundary-Layer Meteorology, 9:235-
244, 1975.
70. Deardorff, J. W., and G. E. Willis. Advances in Geophysics. 18b.
Landsberg and Van Miegham, eds. Academic Press, New York, 1974,
pp. 187-200.
71. Martinez, J. R. Comments on Dynamics and Kinetics in Environmental
Problems Are They Separable? Proceedings of the Symposium on
the Chemical Aspects of Air Quality Modeling. L. D. Kornreich,
ed., University of North Carolina, Chapel Hill, North Carolina,
April 17-19, 1974.
72. Donaldson, C. duP., and G. R. Hilst. Environ. Sci. and Technol.,
6(9):312, 1972.
73. Hilst, G. R. Dynamics and Kinetics in Environmental Problems
Are they Separable? Proceedings of the Symposium on Chemical
Aspects of Air Quality Modeling. L. D. Kornreich, ed., University
of North Carolina, Chapel Hill, North Carolina, 1974. p. 43.
74. Lamb, R. G. Continued Research In Mesoscale Air Pollution Simula-
tion Modeling: Vol. Ill - Modeling of Microscale Phenomena. EPA-
600/4-76-016c. U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina, 1976.
75. Mayrsohn, H., and J. H. Crabtree. Source Reconciliation of Atmos-
pheric Hydrocarbons. Atmos. Environ., 10:137-143, 1976.
216
-------
76. Mayrsohn, H., and J. H. Crabtree. Source Reconciliation of Atmos-
pherzc Hydrocarbons, personal communication to Basil Dimitriades.
U.S. Environmental Protection Agency, Research Triangle Park, North
Carolina, 1975.
77. Mayrsohn, H, personal communication to Basil Dimitriades, U.S.
Environmental Protection Agency, Research Triangle Park, North
Carolina, 1975.
78. Hurn, R. W. Mobile Combustion Sources. Chapter 33, Air Pollution,
Vol. Ill, A. C. Stern, ed., 2nd edition, Academic Press, New York,
New York, pp. 55-95, 1968.
79. Mayrsohn, H., J. H. Crabtree, M. Kuramoto, R. D. Southern, and
S. H. Mano. Source Reconciliations of Atmospheric Hydrocarbons
1974. Atmos. Environ., 11:189-192, 1977.
80. Keith, R. 1976. personal communication. Los Angeles Air Pollu-
tion Control District.
81. Calvert, J. G. A Preliminary Analysis of the Data from LARPP Oper-
ation 33 of November 5, 1973. Report to Coordinating Research
Council. New York, New York, 1973.
82. Geiger, R. The Climate Near the Ground. Howard University Press,
Cambridge, Massachusetts, 1965.
83. Jeffries, H., D. Fox, and R. Kamens. Outdoor Smog Chamber Studies:
Effect of Hydrocarbon Reduction on Nitrogen Dioxide. Publication
No. EPA-650/3-75-011, U.S. Environmental Protection Agency, Raleigh,
North Carolina, 1975.
84. Jeffries, H., D. Fox, and R. Kamens. Outdoor Smog Chamber Studies:
Light Effects Relative to Indoor Chambers. Environ. Sci. Technol.,
10:1006, 1976.
85. Sickles, J. E., and H. E. Jeffries. Development and Operation of a
Device for the Continuous Measurement of Oka for Nitrogen Dioxide.
Publication No. 396. Department of Environmental Sciences and
Engineering, School of Public Health, University of North Carolina,
1975.
86. Kamens, R. M., H. E. Jeffries, and T. Johns. An Automatic Chromato-
graphic Method to Measure C.,-Cg Ambient Hydrocarbons. Presented at
ACS Meeting, New Orleans, Louisiana, March 17, 1977.
87. Methods of Determination of Velocity, Volume, Dust and Mist Content
of Gases. Bulletin WP-50. Western Precipitator Division, Joy
Manufacturing Company, Los Angeles, California, 1968.
217
-------
88. Reynolds, S. D., M. -K. Liu, T. A. Hecht, P. M. Roth, and J. H.
Seinfeld. Mathematical Modeling of Photochemical Air Pollution -
III. Evaluation of the Model. Atmos. Environ., 8:563-596, 1974.
89. Lee, R. E., Jr., R. K. Patterson, W. L. Crider, and J. Wagman.
Concentration and Particle Size Distribution of Particulate Emis-
sions in Automobile Exhaust. Atmos. Environ., (5):225, 1971.
90. Crittenden, A. L. Analysis of Air Pollutants by Mass Spectroscopy.
EPA-600/3-76-093. U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina, 1976.
91. Chan, W. H., R. J. Nordstrom, J. G. Calvert, and J. H. Shaw.
Kinetic Study of HONO Formation and Decay Reactions in Gaseous
Mixtures of HONO, NO, N0», H»0 and N7. Environ. Sci. Technol.,
10(17):674-682, 1976.
92. Dimitriades, B., and T. C. Wesson. Reactivities of Exhaust Alde-
hydes. J. Air Poll. Cont. Assoc., 22(33):33-38, 1972.
93. Whitten, G. Z., and H. Hogo. Mathematical Modeling of Simulated
Photochemical Smog. EPA-600/3-77-011. U.S. Environmental Pro-
tection Agency, Research Triangle Park, North Carolina, 1977.
94. Johnson, F. S. The Solar Constant. J. Meteorol., 11:431-439,
1954.
95. Ackerman, M. Ultraviolet Solar Radiation Related to Mesopheric
Processes. Inst. Aeron. Spatials Belg. Brussels, A-77:149, 1970.
96. Peterson, J. T. Calculated Actinic Fluxes (290-700 nm) for Air
Pollution Photochemistry Applications. EPA-600/4-76-025. U.S.
Environmental Protection Agency, Research Triangle Park, North
Carolina, 1976.
97. Danckwerts, P. V. The Definition and Measurement of Some Char-
acteristics of Mixtures. Appl. Sci. Res., A3:279-296, 1952.
98. Hendry, D. G., J. E. Davenport, and R. A. Kenley. Quarterly
Progress Report for Grant entitled "Reactions of Oxy Radicals in
the Atmosphere." Stanford Research Institute, Menlo Park, California,
1976.
99. Jeffries, H. E. personal communication, 1976.
100. Crane, G., H. A. Panofsky, and 0. Zeman. A Model for Dispersion
from Area Sources in Convective Turbulance. Atmos. Environ., 11:893-
900, 1977.
101. Feigley, C. E., and H. E. Jeffries. Analysis of Processes Affect-
ing Oxidant and Precursors in the Los Angeles Reactive Pollutant
Program (LARPP) Operation 33. submitted for publication, 1978.
218
-------
102. Ostle, B. Statistics in Research. 2nd ed. Iowa State University
Press, Ames, Iowa, 1963.
103. Milne, W. E. Numerical Solution of Differential Equations. Dover
Publications, New York, 1970.
104. Ketter, R. L., and D. W. Prawel. Modern Methods of Engineering
Computation. McGraw-Hill, New York, 1969.
105. Carnahan, B., H. A. Luther, and J. 0. Wilkes. Applied Numerical
Methods. John Wiley and Sons, Inc., New York, 1969-
219
-------
APPENDIX A
FLAG FIELD MESSAGES FOR OPERATION 33
The flag field contained in the helicopter records for operation 33
was translated from EBCDIC to 7-track BCD and the messages were printed
out using a program written for this purpose called FLAG. The messages
for each helicopter were handled separately and are presented here in
Tables A-l and A-2.
220
-------
TABLE A-l. SMOG 2 DATA FLAG MESSAGES FOR OPERATION 33
ro
Approximate altitude correction applied
from 72112 through 143620
Pots changed during flight. Data suspect (zero drift)
from 72112 through 143620
No higher than NO
from 72112 through 143620
Poor cal data
from 72504 through 74803
from 84425 through 85159
from 92400 through 93051
from 100642 through 101135
from 110451 through 111307
from 115147 through 115822
from 130322 through 131033
from 142650 through 143253
Clogged plumbing
from 72112 through 143620
NO
instrument
NO
instrument
NO
instrument
-------
TABLE A-2. SMOG 3 DATA FLAG MESSAGES FOR OPERATION 33
K)
NJ
KJ
Instrument outage
from 124750 through 125006
Instrument outage
from 124750 through 125006
Approximate altitude correction applied
from 84736 through 144549
Pots changed during flight. Data suspect (zero drift)
from 84736 through 144549
No higher than NO
from 84736 through 144549
Poor cal data
from 85331 through 90155
from 100930 through 101722
from 114831 through 115738
from 123817 through 124723
Scale change due to span pot no able to adjust inst
to cal source
from 84736 through 144549
Clogged plumbing
from 84736 through 144549
CH, instrument
4
instrument
NO
NO
CH,
NO
instrument
instrument
instrument
instrument
-------
APPENDIX B
SOURCE RECONCILIATION AND ASSOCIATED CALCULATIONS
The source reconciliation method itself has been described by
Mayrsohn and Crabtree (75) and requires only brief discussion here. If
£ is a matrix of source weight fractions of m tracers from n sources
(m>n), A is a column vector containing the n tracer concentrations
observed in the atmosphere, and IJ is a matrix of m weighting coeffic-
ients, it is desired to find the best values of ]J for the overdetermined
set of linear equations:
A = C B_ .
If the least square criterion is applied then the solution for the
coefficients follows:
B. = (Cl'C)"1 (£'A) .
After the coefficients were calculated, the differences between the
expected and observed values of tracer concentrations, D_, was calcu-
lated.
D = £ B_ - A
Mayrsohn and Crabtree (75) calculate the standard error of estimate S:
1/2
S = (D'D/(n-l)r/Z .
This form is incorrect; the sum of squares of the residuals D'D should
be divided by the number of degrees of freedom of the residuals which is
the number of tracer species minus the number of coefficients estimated.
Thus, the standard error was calculated as follows:
S - D'D/(m-n))1/2 .
223
-------
Mayrsohn and Crabtree (76) found that the fit could be improved by
deleting certain sources from the jC_ matrix. Thus, the calculations
above were performed with all sources represented in £ and then various
smaller configurations were explored. The source combination with the
lowest standard error was selected as the final result. Thirteen con-
figurations were tried for each observation. Following the suggestions
of Mayrsohn and Crabtree (76), auto exhaust was never deleted; either
gasoline vaporization or gasoline evaporation were deleted but never
both; either natural gas or LPG was always included.
For the best fitting configuration, the R-squared value for the
regression was calculated by the following approach (102) :
MSE = D'D/(n-m)
_ 2
(A.)
MSS = (A 'A -- )/(n-l)
-- n
2 , MSE
R =
_ 2
The term A. is the mean value of observed concentrations. R provides a
measure of how well the model fits.
Mayrsohn and Crabtree 's (75) use of the chi-square goodness-of-fit
test is not valid for this purpose. The chi-square test assumes an
underlying multinomial distribution, i.e., frequency data.
The total concentration (weight/volume) of hydrocarbons emitted to
produce the observed tracer concentrations may be calculated as follows:
a. = E C.. b.
J Ji i
or
tracer j _ z [ V
tn3 " 1 V
yg tracer j _ y fyg tracer j source i
lug tracer norm, source i
yg tracer norm, source
3
m
224
-------
where "tracer norm" is the particular tracer species by which column i
of £ was normalized for computation of the transpose-product inverse.
a is the value of a tracer species j predicted to be present in the air
by the regression model. Because of the normalization of C, the b.'s have
3
units of (iag tracer norm, source i/m ). Since the source compositions
are known for tracers as well as other species, the total expected
concentrations of any of the emitted species in the absence of reaction
may be calculated. The coefficients b 's are corrected for the effect
of normalization by multiplying by the ratio of the total normalized
tracer concentration over the "tracer norm" concentration.
b,
i i C .
norm i
ye total tracer source i yg tracer norm, source i
or _
m m
yg total tracer source i
yg tracer norm, source i
From the total source compositions, the ratio, R., of tracer species to
total hydrocarbons is calculated. Thus, the total hydrocarbon concen-
tration which would be present in the absence of reaction may then be
determined.
THC = I 3./R.
i X X
or
yg THC in air /yg total tracer source i\ / /yg total tracer source i
3I 3 I / I yg THC source i
m i \ m JI *
Also, the total nonmethane hydrocarbons contributed from each source may
be calculated.
NMHC = THC B± (1 - FPMETH±)
where FRMETH is the weight fraction of methane in source i.
us total MMHC in air = l
3 ,
m
225
-------
APPENDIX C
NUMERICAL COMPUTATION OF VERTICAL
FLUX ALONG A LARPP TRAJECTORY
Because the concentrations of pollutants was not measured up to the
inversion base for much of the day, it was necessary to solve the
equation
dc
3
-
(C-l)
numerically in an altitude-time space of variable dimensions. Such a
solution estimates the concentration at each point in an altitude-time
grid within the region of altitude from h to h/2 where h is the altitude
of the inversion base at each time. Since h increases, the region in
which solution takes place is the shaded region in Figure C-l.
3000
\-
u.
f+
LJ
Q
600
PST
1400
Figure C-l. Region of solution.
226
-------
Equation C-l may be written as
(C-2)
where K'(z,t) is the partial derivative of K(z,t) with respect to alti-
tude. This is a parabolic partial differential equation according to
the criteria of Milne (103) .
To avoid stability problems, an implicit solution was derived
analogous to the formulation of an implicit parabolic solution presented
by Ketter and Prawel (104) . The first and second spatial derivatives
were replaced by first central-difference expressions. The first
forward-difference expression was substituted for the time derivative
yielding the following recurrence formula:
2Az l*i+l, v^i+l,j+l "1+1,J-
The term multiplied by (1-0) is the contribution of concentrations on
the 1th time step and the term multiplied by 0 is the contribution of
concentrations on the i+lth time step. C , ., the concentration on the
, t-h 1"t"-L>J
1+1 time step and the j altitude level, is dependent upon the con-
centrations above and below it on the i+1 time step and the concen-
trations above, below and on the same altitude level on the i time
step. This dependence is shown in Figure C-2.
227
-------
^*L,
TIME INDEX,
Figure C-2. Solution-space indexing.
The values of K and K' are average values across the altitude range to
which they are applied. For instance, K± .+1/2 is the integral of K
from z. to z. _ divided by Az. The expressions for K and K1 used here
K - -A' In ( - 1) (C-4)
h
K' " (C~5)
are:
and
2 -1
where z is altitude, A1 = 84,237 ft hr and h is the altitude of the
inversion base. Special consideration must be made in evaluating K and
K' near h/2. If z is nearly equal to h/2, absurdly high values of K and
K' can result. Thus, if the difference between some z and h/2 was less
than one foot the values of K and K' were computed based on a corrected
z, z , such that
z - h/2 + 1 . (C-6)
Rearranging Equation C-3 so that the i+1*1 concentration terms are on
the left side of the equation and the i terms on the right gives:
J-l/2 +
228
-------
(rKi,M/2
where
H * Cl-9) r
r = ~ and s =
CAz)2 2AZ
Let
CC-8C)
and let b. equal the entire right-hand side of Equation C-7. Equa-
tion C-7 becomes
a. C. . . , + 3. C. . . + Y- C. . . n = b. . (C-9)
j i+l, j+1 pj i+l, j T3 i+l, J-l 3
This equation actually represents a set of equations from j equals 1 to
n-2 where n is the number of altitude grid points inside the range
h _>_ z > h/2 plus 2 for the points just above and below the range. Since
there are n concentrations and n-2 equations, the concentrations just
above and below the region of solution must be specified. The system
of linear equations is then:
an-3 6n-3 Yn-3
a
n-2
Ji+l,n-2
bl * al
un-3
bn-2 - Yn-2
229
-------
or
A X = B (C-10)
Matrix A is a tridiagonal matrix. A solution for X was obtained using a
subroutine described by Carnahan, Luther and Wilks (105). The concen-
tration just below h/2 is known from the helicopter data given the
assumption that there is zero gradient of ground-source pollutants from
about 50 m up to h/2; thus, the lower boundary condition is well estab-
lished. The upper boundary condition, while not known from direct
measurement, may be assumed to be quite low since this is characteristic
of conservative pollutants emitted from the ground in the Los Angeles
basin (65).
In order to initiate computation it is necessary to specify con-
centrations at all altitudes within the region of solution. These
values were determined from helicopter data.
Since mixing depth, initial and boundary conditions were needed at
the space-time grid points, it was necessary to interpolate the data
between actual measurement points. A cubic spline interpolation sub-
routine from the International Mathematics and Statistics Library was
used for this purpose. The computed vertical concentration profiles
were saved in a buffer such that on any time step the profile from the
previous step was available. For each time step, the temporal deriva-
tive at all altitude points was approximated by the first backward
difference expression.
These derivatives were then integrated between h and 150 feet. Both
trapezoid rule integration and Simpsons rule were tested; there was very
little difference between their results. Simpsons rule was used for the
final results. The integral is the flux at j and was listed at time
step i-1. Thus, the fluxes at time i-1 are those required to change the
C to the C . during the one time step for all j.
1J-»J ijj
The initial grid spacing chosen was 0.10 hour and 25 feet. The
effect of reducing both temporal and spatial steps was investigated by
230
-------
comparing the original estimated values of CO flux at various times
during the run with those computed using one-half the altitude spacing
and those using one-fourth the time steps. These comparisons in Table
C-l show that the differences are quite small and that the original
choice of grid spacing was acceptable. Note that for the run with the
shortened time step, the flux value obtained at a given time had to be
averaged with the following three flux estimates before the comparison
was made, since the flux at a particular time represents the average
flux between that time and the next time.
231
-------
TABLE C-l. COMPARISON OF FLUX CALCULATIONS USING
DIFFERENT TIME AND ALTITUDE GRID SPACING
Time,
hours PST
Flux at h/2
2 -1
kg * m hr
At =
Ah -
0.1 hr
25 ft
At - 0.1 hr
Ah = 12.5 ft
At =
Ah =
0.025 hr
25 ft
Flux at 150 ft
i -2 . -1
kg m hr
At = 0.1 hr At = 0.1 hr At = 0.025 hr
Ah = 25 ft Ah = 12.5 ft Ah = 25 ft
OJ
to
8.25- 8.35
10.05-10.15
12.05-12.15
13.85-13.95
6.02 x 10
8
5.78 x 10
-8
5.91 x 10
-8
5.69 x 10
-8
5.46 x 10
-8
5.76 x 10
-8
6.29 x 10
-8
6.27 x 10
-8
6.33 x 10
-8
3.60 x 10
-8
3.60 x 10
-8
3.60 x 10
-8
6.09 x 10
-9
6.07 x 10
-9
,-9
-8
-1.33 x 10
-8
-1.36 x 10
8
6.40 x 10 -4.60 x 10
-1.33 x 10~8 -7.18 x 10"8
-4.60 x 10
,-8
-4.60 x 10
-8
-7.18 x 10
-8
-7.18 x 10
-8
-------
APPENDIX D
CALCULATION OF k FROM THEORY
The calculation employed in this work is similar to that of Calvert
(26) except that the actual mixing depth and N0? concentrations in the
boundary layer were used instead of an empirical fit to ground station
data. Also, k.^ was calculated directly taking into account reflection
at the earth's surface rather than calculating k- without reflection and
correcting afterward. The method generally uses the relations from
Leighton (3) expanded to account for reflection.
TRANSMISSION THROUGH THE ATMOSPHERE
Consider a point within the boundary layer at an altitude above the
ground of %; at this time and location, the mixing depth is H and the
solar zenith angle is z. The downward directed radiant flux on a hori-
zontal surface for wavelength X just above the polluted layer has two
components: I' , the direct beam irradiance, and I' , the sky irradi-
uA SA
ance which is some fraction of the light scattered from the direct beam.
Leighton (3) gives the following relations:
XdA = TsA TaA XoA COS Z ^
^A ' 8 <1-TsA) TaA XoA COS Z ' ^
T is the transmissivity resulting from molecular and particulate
sA
scattering. T is the transmissivity resulting from 03 absorption, IQX
is the solar constant, and g is the fraction of light scattered downward
(taken as 1/2 here). The irradiances are summed over the wavelength
interval AA with a center wavelength A, and the transmissivities are
evaluated at A. This analysis used 10 nm band widths. The solar
constant data of Johnson (94) were used.
233
-------
The transmissivities were determined as follows:
10 aX X 3
where a was interpolated from a table in Leighton (3), p. 19. (0«) is
A
the ozone column and was taken to be 2.8 mm to agree with Calvert.
m - sec z
T . - T . T . (D-4)
sX mX pX
where T . is the molecular component; T .. is the particulate component
mX pA
log T (S ) m (D5)
where (S ,) the theoretical attenuation coefficient is taken from
mX o
Leighton (3).
Iog10 T x - -(3.75 x 10~3 X~2 w (D-6)
+ 3.5 x 10"2 X~°'75 d) m
where w is the water factor; d is the dust factor. It was found that
it was necessary to set w and d equal to 2 to fit UV radiometer values
from Mt. Disappointment for LAKPP operation 33. This fit is shown in
Figure 82. This adjustment must be regarded as an empirical evaluation
of w and d, bearing little relation to the actual water and dust content
of the air measured by independent means.
ATTENUATION OF DOWNWARD DIRECTED RADIATION
The downward directed radiation reaching an altitude, H, within the
boundary layer is given by
I., = I' io-axCsec z (D-7)
dX aX
IsX = rx10-°AClCH-£) (D-8)
which accounts for attenuation by NO within the boundary layer, a
" A
is the extinction coefficient for N©* at X, C is the N02 concentration
and i is a factor relating the average path length of diffuse sky
234
-------
radiation to the vertical thickness of the absorbing layer through which
the beam must pass. Following Leighton's example, i was set equal to 2.
The molar absorption coefficient of N02 used by Calvert (26) were
employed here. The concentration of N02 was determined from helicopter
pattern averages constant up to the inversion base. It is believed that
this method of calculating attenuation is more valid than Calvert's use
of stationary, ground-level UV radiometers to estimate the appropriate
concentration and path length for calculations. These stations may bear
little relation to what happens along the trajectory, and Calvert's
method has very poor temporal resolution in calculating attenuation.
REFLECTED RADIATION
First the radiation reaching the ground was calculated by the
following equations:
I. , = I' HfaXCH S6C Z (D-9)
dtA dA
I , - I' 10~aACiH (D-10)
stA dA
where I, and I . are the portions of the direct and sky irradiance
dtA stA
reaching the ground. Assuming that the radiation reflection at the
ground is diffuse, the reflected irradiance at altitude H, 1^, after
attenuation may be calculated (treating the path length problem similar
to sky radiation path length).
I . = (I,,. + I .) a. l
-------
The quantum yield was taken from Jones and Bayes as presented by Calver
(26).
Calvert's method of calculating k.. differs somewhat from that pre-
sented here. He first calculated k1 for the case of no attenuation in
the zero gradient region and an albedo of zero. Then, the estimate was
improved to give k' by the following relationship:
I . + I + I
ki _ k dt st _ I fD_l
k k ' tD L
The absence of subscript X indicates summation on X. This correction i
highly questionable for several reasons: (1) I, and I are the down-
Qt S t
ward irradiances at the earth's surface, not at altitude £; (2) the
dependence of k.. on radiation direction has been removed in Equation
D-12 by multiplying I, by sec z and I ^ + I . by i, but this is ignored
in D-13; (3) since $ and a are functions of X, the correction of k-
using the irradiances summed on X is not correct the correction must
be performed prior to summing on X.
236
-------
APPENDIX E
THE INTERACTIVE EFFECTS OF INHOMOGENEITY AND
ATMOSPHERIC PROCESSES ON PHOTOSTATIONARY STATE OZONE
As shown by the example in Section 7, the inhomogeneous nature of
the atmosphere can lead to average 0- concentrations either greater or
larger than one would expect based on k, and the average NO- /NO ratio.
If the covariance of NO and 0- is negative, the average 0 will be
higher than expected, and if the covariance is positive, the average 0-
will be less than expected, as may be seen from the equation:
k. [N00] cov(NO,0.)
[0.] = ~ -^ --- 7 - . (E-l)
J 3 [NO] [NO]
The hats on the concentrations represent spatial averaging, either by
composite sampling, or numerical averaging of point concentrations. The
effects of various processes will be illustrated by means of simplified
examples .
Example E-l. Averaging measurements of NO- and NO to calculate
photostationary state 0~.
Consider the two air parcels from Section 7 as shown in Figure E-l.
Each parcel is well mixed within itself to the molecular level. The
0- concentration within each parcel obeys photostationary state at k-^kg
= 0.016 ppm.
k, [NOJ
(E-2)
Instruments aboard a helicopter take a single measurement of NO, N02
and 0- as the helicopter flys through the two parcels in such a way that
each parcel contributes equally to the measurement. The measured quan-
tities are then:
237
-------
[N02] = 0.09525 ppm
[NO] = 0.30476 ppm
Meas . Parcel
[N02]/[NO]
s\
[03],ppm
x\
[N02],ppm
[N0],ppm
[0 ] [NO]
~ , ppm
[N02]
2
cov,ppm
/\
cov/[NO],ppm
NJ
U>
00
1 0.3125
A 0.02750 0.06657 0.15844 0.06545 -0.00329 -0.0208
2 3.1254
3
B
4
Average of A & B
6.249
0.065 0.16773 0.05728
1.8749
0.04625 0.11715 0.10786
0.02220 -0.00104 -0.0182
-0.00217 -0.01946
Figure E-l. Data for example E-l and summary.
-------
Measurement A;
/s
[03] * 0.02750 ppm
x\
[N02] - 0.06657 ppm
A
[NO] = 0.15844 ppm
as shown in Figure E-l.
The k-^/k estimated from simple photostationary state is much too
high
k [0,] [NO]
_ = 2 -
3 [N02]
= 0.06545 ppm . (E-3)
Thus, the photostationary state expressions such as E-2 and E-3 were
rederived to take into account inhomogeneity (Section 7), yielding
Equation E-l and the following:
k [N0][0,] + cov(NO, 0,)
3 [N02]
The covariance of the two parcels is calculated as follows:
cov(NO,03)
2
I (NO.-NOXO-.-O.)
2
-0.00329 ppm
This yields the correct value for k./k,, in Equation E-4.
Now consider parcels 3 and 4 in Figure E-l, which were sampled in a
manner similar to the previous two. They are subject to the same temp-
erature and solar radiation as parcels 1 and 2.
Measurement B;
A
[03] = 0.065 ppm
^
[N02] « 0.16773 ppm
239
-------
[NOT - 0.05728 ppm
cov(NO,03) = -0.00104 ppm2
If measurements A and B are averaged together mathematically,
A.
[0_] = 0.04625 ppm
[N02] = 0.11715 ppm
s\
[NO] = 0.10786 ppm
cov(NO,03) = -0.00217 ppm
Putting these values in Equation E-3 gives too great a value,
k
^ = 0.04258 ppm
and in Equation E-2
r 1 _ (0.016)(0.11715) _ ft
[03] (0.10786) 0.01738 ppm
the ozone concentration is underestimated.
If one attempts to calculate the average photostationary state
ozone by entering the average of A and B in Equation E-l, the result
is a little closer to the true average ozone.
rn 1 (0.016)(0.11715) (-0.00217)
[U3J ~ 0.10786 " (0.10786)
= 0.03750 ppm .
Considering that the mean of a ratio is not necessarily equal to the
ratio of the means, the following exact equation must be applied to
estimate the average photostationary state 0_ from individual meas-
urement values of N0» and NO.
_ , \
k / [N09] \ I cov(NO,0,) \
r« ., _ II __L_ _ _ J_ (E_5)
[NO] / \ [NO] /
240
-------
[03] - (0.016)(1.67423) - (-0.01946)
= 0.04625 ppm
As may be seen from this Table E-l if the photostationary state
03 or k1/k3 are to be estimated from either composite sample measure-
ments, or from averaged measurements, the effect of inhomogeneity must
be considered unless the system is well mixed.
Example E-2. The Effect of Mixing.
If two air parcels containing 03 in photostationary state are mixed
together completely, the resulting concentrations of 03, NO, and NO, may
be determined from the constraint that the stationary state is main-
tained. Thus, letting X equal the amount of 0_ and NO that must be
consumed to maintain stationary state:
kx ([03] - X)([NO] - X)
j = . (E-6)
3 ([N02] + X)
This is a quadratic equation in X whose solution is:
/ 2 (E~7)
k, / k- k,
[N0]+[0 ] + rr- -/ ([N0]+[0_] + £± ) - 4([NO][OJ - ^ [NO.])
J K3 V 3 J K, 2
X = - *
2
(The root formed by addition of the radical is extraneous.) From
Equation E-4, it may be determined that
k..
[N0][0_] - ~ [NO,] = -cov(NO,0_) .
j ICrt t~ -J
Thus,
[N0]+[03] + jp / ([N0]+[03] + ^ )
X = - / + cov(NO,0») . (E-8)
241
-------
TABLE E-l. RESULTS OF VARIOUS APPROACHES TO
ESTIMATE PHOTOSTATIONARY STATE 0- AND k-i/ko
(knowing
and NO)
True average
0.04625 ppm
Est. by
[N02]
[NO]
0.01738
Est. by r±
[NOJ
[NO]
cov
[NO]
0.03750
Est. by r-
[N02]
[NO]
cov
[NO]
0.04625
(knowing
, N02>
True value
0.016 ppm
Est. by
[03] [NO]
[N02]
0.04258
242
-------
The 03 after mixing, [0^ is given by:
[o3]n - [63] - x (E_9a)
and similarly,
[N0]m = [NO] - X and [NO = [W + X . (E-9b,c)
An alternate expression which emphasizes the effect on 0 by alteration
of the N02/N0 ratio follows:
/v
kt [NO ] + X
[°3]m = k -- ~ - ' <
J m k3 [NO] - X
The effect of the value of the original NO-0- covariance before mixing
«3
follows from Equations E-8 and E-9a.
If cov(NO,0,) < 0 then [0,] < [0,]
j j m j
If cov(NO,0_) = 0 then [0,1 - [0_]
j J' m J
If cov(NO,0_) > 0 then [0,] > [00]
5 .3 m 6
Consider the result of mixing parcels 1 and 2 from Example E-l
(Figure E-l).
X = 0.10097 - fw> , J^' - 0.00329
N 4
X - 0.01787 ppm
Thus, the photostationary state 0_ has dropped from 0.02750 ppm to
0.00963 ppm. The ratio of [N02]/[NO] has gone from 0.42016 to
[NO,] ([NO ] + X) 0 08,,,
' ^ 0 14057 = °-6007° '
[NO] ([NO] - X) °>14°57
m
Thus, while the covariance/NO in Equation E-l was raised to zero from
-0.02076 ppm by mixing, the remaining term on the right-hand side in-
creased by 0.00289 ppm to partially compensate for the loss of ozone
by direct reaction with NO.
243
-------
Example E-3. The Effect of Mixing on the Average of Measurements.
Consider the four air parcels from Example E-l. Measurement A
spans parcels 1 and 2, and Measurement B spans parcels 3 and 4.
[0 ] = 0.02750 ppm
[NO-] = 0.06657 ppm
[NO] = 0.15844 ppm
B
[0 ] = 0.06500 ppm
[N02] = 0.16773 ppm
[NO] = 0.05728 ppm
cov(NO,03) = -0.00329 ppm cov(NO,03> = -0.00104 ppm
When the two measurements were averaged, the following were obtained:
IQ3]
0.04625 ppm, [N00] = 0.11715 ppm
[NO] = 0.10786 ppm, cov(NO,03> = -0.00217 ppm
[N02]
[NO]
1.67423,
cov(NO,03)
[NO]
-0.01946 ppm
If parcels 1 and 2 become well-mixed as in Example E-2
and the measurements were performed again, B would be the same but
now A would be
244
-------
[03] = 0.00963 ppm
A
[N02] = 0.08444 ppm
[NO] - 0.14057 ppm
cov(NO,03) = 0
The averaging of measurements A and B now yields:
[03] = 0.03732 ppm, [N02] = 0.12609 ppm
[NO] = 0.09893 ppm, cov(NO,03> = -0.00052 ppm2
[N0_] \ /cov(NO,0
x = 1.76447, ^ | - -0.00908 ppm
[NO] / ^ [NO]
Prior to mixing Equation E-5 evaluates as follows:
kl I [N02]
[NO] / \ [NO]
[03] - 0.02679 + 0.01946 = 0.04625 ppm .
After mixing, it evaluates as:
[03] = 0.02823 + 0.00908 - 0.03731 ppm .
Thus, as was the case with the mixed parcel alone, the contribution of
the cov/NO term to average ozone was reduced and the average NO^/NO
ratio term partially compensates for the increased homogeneity.
Example E-4. The Effect of Photochemical Oxidation of NO to N02 on
the Covariance of NO and 03-
The effect of oxidation of NO to NO, is more difficult to handle
^ A
mathematically. Such oxidation increases [03]. Solving Equation E-l
for the covariance
245
-------
^ A
Also,
cov(NO,0,) = rr [NO,] - [0,][NO] .
3 k3 L J
kj_ [N02]
t03] = kT "[NOT
Thus,
cov(NO,03) = £- ([N02] -
[N02]
[NO]
[NO])
(E-ll)
It is not readily apparent whether NO conversion increases or decreases
covariance from this form since it depends upon the joint probability
distribution of NO, N0~, and some variable representing the amount of NO
oxidized within each parcel. To get an idea of the possibilities, a
condition close to those in the atmosphere will be considered.
Three parcels of air maintain photostationary state 0» at a k./k_
of 0.016.
A measurement across the three parcels yields
[0»] = 0.08 ppm
st,
[N02] - 0.08033 ppm
s±
[NO] = 0.01634 ppm .
Also,
cov(NO,0 ) = -2.2000 x 10~5 ppm2
[N02]
[NO]
5.00000,
[N02]
[NO]
= 4.91616
246
-------
cov(NO,03)
[NO]
-1.3464 x 10~3 ppm .
If 10 percent of the NO in each parcel is oxidized the concentrations
below result.
[NO ] - 0.08326
[03]
[N023
0.09067 ppm
0.08196 ppm
[NO] - 0.01470 ppm
cov(NO,03)
cov(NO,03)
[NO]
-2.1433 x 10~5 ppm2
-1.4581 x 10~3 ppm
[N02]
[NO]
5.66670,
[N02]
[NO]
= 5.57551
Thus, in this example, oxidation of NO to N0» has increased the measured
0» by about 1 pphm. The covariance of NO and 0- was raised slightly,
but the cov/[NO] was slightly lowered indicating increased effect of
inhomogeneity on 0_. Other numerical examples have shown that when the
concentrations in the parcels are greatly different, a great change in
the covariance can result from oxidation with very little change in 0,,.
247
-------
Example E-5. Emission of NO--
Consider air parcels 1 and 2 from Example E-l (Figure E-l). The
NO, is injected into both parcels. The resulting concentrations in each
£ sv
parcel may be determined by Equation E-7 where [NO-] is replaced by the
initial NO concentration in a parcel plus the increase in NO concen-
tration caused by the emission in the absence of reaction.
If emission produced an additional 0.01 ppm of N02 in each parcel,
then
parcel 1
0.32576 -V(0.32576)2 -4(1.5238 - 1.68 x 10~3)
_ 0.32576 - 0.32674 n An_._
X = » -0.00049 ppm
parcel 2
X =
0.07812 -«J(0.07812)2 -4(6.0600 x 10~4 - 7.6603 x 10"4)
v 0.07812 - 0.08212 n __...
X = 5 = -0.00200 ppm
Thus, the new concentrations are:
PI
[03] = 0.00549 ppm
[N02] = 0.10476 ppm
[NO] - 0.30525 ppm
P2
[0,] = 0.052 ppm
[N02] = 0.04588 ppm
[NO] = 0.01412 ppm
measurement
The measured concentration, as shown above would be:
s^
[03] = 0.02875 ppm
248
-------
[N02] - 0.07532 ppm
[NO] - 0.15969 ppm .
Also,
cov(NO,03) = -0.00339 ppm2
cov(NO,0~)
- -0.02120 ppm
[NO]
[NOJ
[NO]
0.47166
Thus, emission of N0_ has had the effect of increasing the measured 0,
while decreasing the covariance of NO and 0_. This may be seen by
comparing the above values with those from Example E-l. The terms in
Equation E-l have been changed by emission of N02 as follows:
cov(NO,03)
[NO]
before
after
IU3J
0.02750 ppm
0.02875 ppm
k3 [NO]
0.00672 ppm
0.00755 ppm
-0.02076 ppm
-0.02120 ppm
It may be seen that the effect of addition of N02 is to increase the
0- concentration by raising the N02/N0 ratio as well as the increased
effect of inhomogeneity (i.e., the cov/NO term).
249
-------
APPENDIX F
CORRECTION OF DATA FOR THE DARK
REACTION OF NO AND 00 DURING SAMPLING
For most purposes correction of smog chamber data for the loss of
NO and 0- and gain of NO- is not worthwhile. Generally, the corrections
are not much more than 1 pphm, however, in the simulation experiments
the maximum ozone concentration only reached about 10 to 15 pphm. Thus,
a method was developed for correcting NO, 0», and N0» data for the
NO - 0- reaction.
Given the measured concentration [NO], [0_], and [N0_], it is
desired to determine the original concentrations at some time, t, in
the past. The rate expression for the reaction may be written in terms
of the decrease of reactant concentrations, x.
|f - k3 [NO][03] (F-l)
= k3 ([N0]-x)([03]-x)
dx , ,.
= kdt
([N0]-x)([03]-x)
The left-hand side of the equation may be integrated by use of partial
fractions.
-x
f ' dx
[N0]-[03] \ Q I [N0]-x
Note that the limits of integration on t have been reversed so that the
reaction will be run in reverse (x is negative).
1
[NO]_[0 j (ln([NO]-x) - ln([03]-x))
x
f
(F-2)
0
250
-------
ln
f([NO]-xf)[03]
[N0]-[03r xn l([03]-xf)[NO]
\ i
Rearranging Equation F-2 yields an expression for the initial reactant
ratio.
([N0]-xf) [NQ]
(lOo]~x.p) [Oql 3 3 f
Let -M- exp (-k ([N0]-[0_])t,) = R . (F-4)
L"-' o J -^ J I
J
Solving for xf gives
[N0]-R[0 ]
J /- r* \
xf =
Then the concentrations prior to reaction are given by
[NO]Q [NO] - xf (F-6)
[03]0 - [03] - xf
[N02lo = [N02] + xf .
Selected sets of measurements from the October 12 simulation were
corrected. The original data and corrected data for 03> NO, and N02
were plotted versus time as shown in Figures F-l, F-2, and F-3. The
correction is greatest between 0900 and 1000 (corrected to PST) reaching
values just above 1 pphm.
251
-------
___C1_J_
2,l.n_
_0.0.8_
-A~--Measured
NJ
Ul
N)
3.36_>
.OZ_
B A
I
Ji /_
A A
-1Z
Figure F-l. Measured and corrected ozone for October 12 simulation.
-------
NT I
0.15
AMeasured
ro
Ui
CO
0.10
0.2S
0.20
1.15
3.10
0.05
-A S-
A . B
0.03 «
A B
16
TI.VE
Figure F-2. Measured and corrected NO for simulation of October 12.
-------
I
i
I
i AMeasured
I
-ja_ , . R=CorrectecU
A
__
Ni
Ui
i
i
I
_a.io.>
i
i
I,
I
1
I
.05 »
I f w~ ^-*»»«« t«*»«*.»*vvv~*>B «.WM*f .v.Bvw«.*vw*.«HW_VH.v»f __. * H_aM«»»*b«. *««_.., vv w V «H»W * «
-_^_______, ft in ^^ ^ ^ ],; .', . i 'i l^V . -. ^.. - 16
TIME
Figure F-3. Measured and corrected NO for October 12 simulation.
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
REPORT NO.
EPA-600/3-79-05Q
2.
3. RECIPIENT'S ACCESSIOI*NO.
TITLE AND SUBTITLE
SMOG CHAMBER VALIDATION USING LAGRANGIAN
ATMOSPHERIC DATA
5. REPORT DATE
May 1979
6. PERFORMING ORGANIZATION CODE
AUTHOR(S)
Charles Eugene Feigley, Harvey E. Jeffries, and
Myra A. Carpenter
8. PERFORMING ORGANIZATION REPORT NO.
PERFORMING ORGANIZATION NAME AND ADDRESS
Department of Environmental Sciences and Engineering
University of North Carolina
Chapel Hill, North Carolina 27514
10. PROGRAM ELEMENT NO
1AA603
11. CONTRACT/GRANT NO.
Grant No. 800916
2. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory-RTF, NC
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
13. TYPE OF REPORT AND PERIOD COVERED
Interim
14. SPONSORING AGENCY CODE
EPA/600/09
5. SUPPLEMENTARY NOTES
6. ABSTRACT
A method was developed for validating outdoor smog chamber experiments as a
means of determining the relationships between oxidant concentrations and its
precursors - hydrocarbons and nitrogen oxides. When chamber experiments were
performed in a manner that simulated relevant meteorological processes and
precursor concentrations, the validation method showed that photochemical smog
reactions observed in the smog chamber generally agreed with data from the Los
Angeles Reactive Pollutant Program (LARPP). The LARPP data consist of detailed
airborne and ground level pollutant and meteorological measurements.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
.lDENTIFIERS/OPEN ENDED TERMS
;. COS AT I Field/Group
* Air pollution
* Ozone
* Hydrocarbons
* Nitrogen oxides
* Chemical reactions
Test chambers
* Mathematical models
* Computer
simulation
13B
07B
076
07D
14B
12A
09B
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)
21. NO. 01
273
20. SECURITY CLASS (Thispage)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
255
------- |