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 months—roughly April to October—the 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 schemes—it 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—
     I—t
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
I—I

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
i—i
>
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
I—r

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.
 1—J-»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                         (D—5)

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
                                                                             A—Measured
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	• A—Measured
               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

-------