c/EPA
           United States
           Environmental Protection
           Agency
           Environmental Sciences Research EPA-600/8-79-015a
           Laboratory •       June 1979
           Research Triangle Park NC 27711
           Research and Development
A Lagrangian
Photochemical Air
Quality Simulation
Model

Adaptation to the
St. Louis—RAPS
Data Base
Volume I.
Model Formulation
                                          OF

-------
                  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
environmental 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  SPECIAL REPORTS series. This series is
reserved for reports which are intended to meet the technical information needs
of specifically targeted user groups. Reports in this series include Problem Orient-
ed Reports, Research Application Reports, and Executive Summary Documents.
Typical of these reports include state-of-the-art analyses, technology assess-
ments, reports on the results of major research and development efforts, design
manuals, and user manuals.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                                     EPA-600/8-79-015a
                                                     June 1979
    A LAGRANGIAN PHOTOCHEMICAL AIR QUALITY SIMULATION MODEL

         Adaptation to the St. Louis - RAPS Data Base

                  Volume I. Model Formulation
                               by

Fred Lurman, Daniel Godden, Alan C. Lloyd, Richard A. Nordsieck
          Environmental Research and Technology, Inc.
                Environmental Analysis Division
                      2625 Townsgate Road
              Westlake Village, California  91361
                    Contract No. 68-02-2765
                        Project Officer

                       Jack H. Shreffler
              Meteorology and Assessment Division
          Environmental Sciences Research Laboratory
               Research Triangle Park, NC  27711
          ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
              OFFICE OF RESEARCH AND DEVELOPMENT
             U.S. ENVIRONMENTAL PROTECTION AGENCY
               RESEARCH TRIANGLE PARK, NC  27711

-------
                                DISCLAIMER
     This report has been reviewed by the Environmental Sciences Research
Laboratory, 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 recommendation for use.
                                AFFILIATION
     Dr. Shreffler, the Project Officer, is on assignment to the Meteorology
and Assessment Division, Environmental Sciences Research Laboratory, from the
National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
                                     ii

-------
                                  ABSTRACT
     A Lagrangian photochemical air quality simulation model has been adapted
to St. Louis Missouri/Illinois metropolitan region and the Regional Air Pol-
lution Study (RAPS) aerometric and emissions data base.  This adaptation was
performed to provide a means for EPA to independently assess the validity of
a state-of-the-art Lagrangian photochemical model.

     Chemical kinetic oxidation mechanisms involving hydrocarbons, nitrogen
oxides and sulfur oxides and a vertical diffusion formulation developed by
Environmental Research and Technology Inc. for modeling reactive pollutants
in the troposphere are described.  Methods for determining model input para-
meters are discussed and model results for ozone, nitrogen dioxide, carbon
dioxide, sulfur dioxide, and sulfate are presented for three summer days in
1976.  In considering so few simulations, no firm conclusions concerning
model reliability are possible, although predicted pollutant concentrations
are of reasonable levels.  Most noteworthy for future users, the results
suggest that the model may predict less ozone than is actually generated in
St. Louis.  Uncertainty in initial conditions of ozone and organic species is
likely responsible for this discrepancy between observed and computed values.
                                     iii

-------
                                CONTENTS
                                                                  m
ABSTRACT
LIST OF ILLUSTRATIONS                                             V1_
                                                                   i x
LIST OF TABLES
ACKNOWLEDGEMENTS                                                    X
1.   INTRODUCTION                                                 1-1
2.   MATHEMATICAL FORMALISM OF THE ATMOSPHERIC DIFFUSION
     MODEL                                                        2-1
     2.1   Governing Equations                                    2-1
     2.2   Validity and Restrictions of the Lagrangian
           Air Parcel Concept                                     2-4
     2.3   Numerical Methods                                      2-5
           2.3.1   Spatial Discretion of the Governing
                   Equations                                      2-5
           2.3.2   Solution of the Ordinary Differential
                   Equations                                      2-9
3.   FORMALISM OF THE EDDY DIFFUSIVITY ALGORITHM                  3-1
     3.1   K  in the Surface Layer                                3-1
            z>
     3.2   Evaluation of Surface Layer Parameters                 3-2
     3.3   K  in the Ekman Layer                                  3-3
     3.4   Application of Meteorological Data                     3-9
4.   DEVELOPMENT OF A HOMOGENEOUS GAS PHASE MODEL FOR
     THE PREDICTION OF OZONE AND SULFATE FORMATION                4-1
     4.1   Formulation of a Chemical Mechanism for the
           HC/NO  System                                          4-1
                j\
           4.1.1   Chemical Mechanism Development                 4-2
           4.1.2   Kinetics and Mechanism for the
                   Photochemical Model                            4-2
           4.1.3   Salient Features of the Inorganic
                   Mechanism                                      4-5
           4.1.4   Salient Features of the Organic
                   Mechanism                                      4-12
           4.1.5   Validation of Photochemical Model              4-18
           4.1.6   Results and Discussion                         4-19
     4.2   Formulation of a Chemical Mechanism for S0?
           Oxidation                                              4-45
                                    v

-------
                         CONTENTS (CONTINUED)

           4.2.1   SO- Oxidation Chemistry                         4-45
           4.2.2   Validation of the HC/NO /SO  System             4-50
           4.2.3   Results and Discussion                          4-54
5.   ADAPTATION OF THE MODEL TO THE ST. LOUIS REGION
     AND RAPS DATA BASE                                            5-1
6.   METHODS AND RESULTS FOR TEST DAY SIMULATIONS                  6-1
     6.1   Air Parcel Trajectories                                 6-1
     6.2   Meteorological Conditions                               6-3
     6.3   Air Quality Conditions                                  6-15
     6.4   Source Emission Strengths                               6-20
     6.5   Initial Pollutant Concentrations                        6-20
     6.6   Simulation Model Results                                6-27
7.   CONCLUSIONS AND RECOMMENDATIONS                               7-1
8.   REFERENCES                                   ,                8-1
                                   VI

-------
                       LIST OF ILLUSTRATIONS


Figure                                                            Page

 2-1     Spatial Mesh Description                                 2-6
         Geometric Parameters of Spatial Mesh                     2-6
         Discretized Air Parcel (not to scale)                     2-6

 3-1     Method of Updating Temperature Sounding                  3-11

 4-1     Scheme For Photooxidation of a Typical Aromatic
         Hydrocarbon Using Toluene as an Example                  4-16

 4-2     Simulation of Smog Chamber Experiment  GC  116             4-20

 4-3     Simulation of Smog Chamber Experiment  GC 119             4-25

 4-4     Simulation of Smog Chamber Experiment  GC 133             4-28

 4-5     Simulation of Smog Chamber Experiment  GC 135             4-31

 4-6     Simulation of Smog Chamber Experiment  GC 138             4-34

 4-7     Simulation of Smog Chamber Experiment  GC 150             4-37

 4-8     Simulation of Smog Chamber Experiment  GC 156             4-40

 4-9     Simulation of Smog Chamber Experiment  B-S-114            4-55

 4-10    Simulation of Smog Chamber Experiment  B-S-110            4-57

 4-11    Simulation of Smog Chamber Experiment  B-S-107            4-60

 5-1     RAPS Measurement Station Locations                       5-2

 5-2     Clear Sky Ultraviolet Radiation                          5-4

 6-1     Air Parcel Trajectory For 6-29-76                        6-2

 6-2     Air Parcel Trajectory For 7-13-76                        6-4

 6-3     Air Parcel Trajectory For 7-14-76                        6-5

 6-4     Vertical Temperature Profiles (6-29-76)                  6-9

 6-5     Vertical Temperature Profiles (7-13-76)                  6-10

 6-6     Vertical Temperature Profiles (7-14-76)                  6-11

 6-7     Initial Pollutant Concentration Vertical
         Profiles 6-29-76                                         6-24

 6-8     Initial Pollutant Concentration Vertical
         Profiles 7-13-76                                         6-25
                                    vii

-------
                 LIST OF ILLUSTRATIONS (CONTINUED)


Figure                                                            Page

 6-9     Initial Pollutant Concentration Vertical
         Profiles 7-14-76                                         6-26

 6-10    Trajectory Model Concentration Predictions
         For 6-29-76                                              6-29

 6-11    Trajectory Model Concentration Predictions
         For 6-29-76                                              6-30

 6-12    Trajectory Model Concentration Predictions
         For 7-13-76                                              6-32

 6-13    Trajectory Model Concentration Predictions
         For 7-13-76                                              6-33

 6-14    Trajectory Model Concentration Predictions
         For 7-14-76                                              6-35

 6-15    Trajectory Model Concentration Predictions
         For 7-14-76                                              6-36
                                    viii

-------
                               LIST OF TABLES


Table


 3-1     Relationship of Pasquill Stability Categories
         To Ranges of the Parameter a (Fulle, 1975)           3-4

 4-1     Detailed Composition of Hydrocarbon Surrogate
         Mixture                                              4-3

 4-2     Generalized Reaction Mechanisms                      4-6

 4-3     Chemical Species Symbol Definitions                  4-8

 4-4     Surrogate Hydrocarbon Smog Chamber Runs Used
         For Model Testing and Development                    4-43

 4-5     Rate Constants Used For Hydrocarbons (ppm~
         rain  )                                               4-43

 4-6     Additional Reactions Used For HC/NO /SO
         Homogeneous Reaction Mechanism     X   X             4-51

 4-7     Initial Conditions For Battelle Smog
         Chamber Simulations                                  4-53

 6-1     Meteorological Conditions For 6/29/76                6-6

 6-2     Meteorological Conditions For 7/13/76                6-7

 6-3     Meteorological Conditions For 7/14/76                6-8

 6-4     Vertical Eddy Diffusivity Coefficients
         For 6/29/76                                          6-12

 6-5     Vertical Eddy Diffusivity Coefficients
         For 7/13/76                                          6-13

 6-6     Vertical Eddy Diffusivity Coefficients
         For 7/14/76                                          6-14

 6-7     Air Quality Along Trajectory For 6/29/76             6-16

 6-8     Air Quality Along Trajectory For 7/13/76             6-17

 6-9     Air Quality Along Trajectory For 7/14/76             6-18

 6-10    Area Source Emissions Entrained Along
         Trajectories                                         6-21

 6-11    Point Source Emissions Entrained Along
         Trajectories                                         6-21
                                    ix

-------
                             ACKNOWLEDGMENTS

     The work reported in this document was performed under contract to
United States Environmental Protection Agency, Environmental Science
Research Laboratory, Meteorological Assessment Division.  The project
monitor for the effort was  Dr. Jack Shreffler.
     This study was carried out by a project team led by Mr. Fred Lurmann
of the Environmental Analysis Division of Environmental Research and
Technology, Inc., at Santa Barbara, California.  The project supervisor
was Dr. Alan Eschenroeder, Manager of that Division.  Other members of
the project team included Dr. Alan Lloyd, Mr. Daniel Godden, and
Mr. Richard Nordsieck.
     We wish to thank the following programs, organizations, and indivi-
duals for their support and technical assistance in the model develop-
ment program which preceded and made possible this study.

     (1)  Coordinating Research Council and members of the
          CRC-CAPA-12 Committee
     (2)  Los Angeles Reactive Pollutant Program (LARPP)
     (3)  Environmental Research S, Technology's ARTSIM Research
          Program
     (4)  National Science  Foundation
     (5)  Electric Power Research  Institute

-------
                            1.    INTRODUCTION

     The objective of this study is to provide a state-of-the-art
Lagrangian photochemical air quality simulation model suitable for pre-
dicting atmospheric pollutant concentrations in the St. Louis, Missouri/
Illinois Metropolitan Area.  The Environmental Research § Technology,
Inc., (ERT) Lagrangian Photochemical Diffusion Code has been adapted for
use with the Regional Air Pollution Study (RAPS) meteorological and emis-
sions data base and implemented on the Environmental Protection Agency's
UNIVAC 1110 computer for this purpose.  In adapting the model to the St.
Louis region, a deliberate effort has been made to fully utilize the
extensive data base provided by the RAPS field program.  The model has
been exercised for simulations of three summer days during 1976.  These
days were selected by the Environmental Protection Agency (EPA) to use
to demonstrate the model's operational capabilities.  A user's manual
has been prepared to provide instructions for proper use of the computer
programs and to document their computational procedures.  Since it is
the intent of EPA to exercise the code extensively, an integral part of
this study has been to maximize the computational efficiency of the
model and its interfaces with the RAPS data base.
     This report is organized into two volumes.  Volume I describes the
mathematical, meteorological, and chemical formalism incorporated in the
air quality simulation model.  It provides an overview of the adaptation
of the model to the St. Louis region and its data base.  The methodology
and results of air quality simulations for the test days are presented
and interpreted.  Exercises for evaluation of the model's predictive
capabilities are suggested.  Volume II consists of the user's manual for
the code, which includes FORTRAN listings of the computer programs and
instructions for their use.
                                   1-1

-------
      2.   MATHEMATICAL FORMALISM OF THE ATMOSPHERIC  DIFFUSION  MODEL

2.1  Governing Equations

     The fundamental approach used to determine pollutant  concentration
in the ambient atmosphere involves solving the differential  equations
governing the conservation of pollutant mass.  The general conservation
equation for a particular pollutant may be written in vector form,  as
follows:

          3c.
          _-!  =  _v - (Vc ) + V • (DVc ) + R  + S                 (2-1)
          dt              1            111

with      c.  =  concentration of species i
          i   =  1, 2, 3, ..., n species
          V   =  the wind velocity with components u, v, and w
                 in the x, y, and z directions
                 3  •*•   3  •*   3  f
          V   =  IE1 + ay ' +li7k
       x,y,z  =  component directions
       i,j,k  =  units vectors in directions x, y, z, respectively
         D    =  molecular diffusivity tensor
         R.   =  rate of generation of i   species by chemical
                 reactions
         S.   =  emission source strength for i   species
                                                                   3c.
Equation (2-1) states that the time rate of concentration  change, -r—,
                                                                   ot
is equal to the net effect of four processes:

     (1)  advection (or transport) of pollutant, V •  (uc.)

     (2)  molecular diffusion, V • (Dc.)> of the pollutant;

     (3)  the change due to chemical reactions, R.; and

     (4)  emissions source strength, S., of the pollutant.
                                   2-1

-------
     The concentration and wind can be expressed in terms of turbulent
deviations from their time-averaged values:
          c.  =  c. + c.
           i      11
          V
thus      u   =  u + u'
          w   =  w + w1

where bars above the quantities denote time-averaged values, and primes
indicate turbulent eddy fluctuations.
     By introducing the above expressions into Equation  (2-1), taking
time averages of each term, expanding, and rearranging terms, the  fol-
lowing equation is obtained for the conservation of mass of  species  i in
a turbulent atmosphere:
3c.   3(uc.)    3(vc.)   3(wc.)    3(u'c.')    3(v'c.')    3(w'c. '
  i     v  ij      ^  i        \.'     ^   i       v   i
                                      3x          3y          3z
                                                                 (2-2)
In order to reduce Equation  (2-2) to  a  form  tractable  for solution,  the
following assumptions are made:
(1)   Molecular diffusion is negligible in comparison to
     turbulent diffusion,  hence D.  =0.
(2)   Atmospheric flow is incompressible, hence

       3}L + lZ+ 3Ji  -  o
       3x   3y   3z  ~  °
       e turbulent
     may be defined as follows:
      (3)  The  turbulent  eddy diffusion coefficients K ,  K ,  K
                                                      x   y   z
                                   2-2

-------
                           3c.
                .'  =  -K
                i        x
            v'c.'  =  -K
               i        y 9y
            w'c.'  =  -K
                           9c.
                            i
               'i        z  9z

Introducing these assumptions into Equation  (2-2)  gives;
9c.   _ 9c.   _ 3c.   _  9c.
    + u+v     + w
                                                                   (2-3)
Further simplification can be achieved by  introducing  the  following
additional assumptions:

     (1)  The vertical velocity component  w  can be  neglected.
     (2)  Horizontal eddy diffusion can be neglected.
     (3)  The horizontal wind field can be considered  uniform
          flow  (i.e., no vertical or horizontal wind shear).

With these assumptions, Equation  (2-3) reduces to:

3c\   _ 9c".   _ 9"c.
—- + u -r-i + v T-^  =  4-   K  ~-  I + R. + S.                    (2-4)
9t      9x      9y      9z  I  z 9z                                 ^   J

This equation can be solved in the fixed x,  y, z  Eulerian  coordinate
system, or it can be transformed to a Lagrangian  coordinate  system,
moving with horizontal velocity V, V = ui  +  vj, which  results  in further
simplification.  The transformation to the Lagrangian  coordinate system
reduces Equation (2-4) to:

9c".
                                   2-3

-------
This is the form of the governing equation assumed in the ERT Lagrangian
Photochemical Diffusion model.  Implicit in the coordinate transforma-
tion is the concept of an air parcel (or column) moving across a region
with conservation of mass within its boundaries.

2.2  Validity and Restrictions of the Lagrangian Air Parcel Concept

     Since the ERT Lagrangian Photochemical Diffusion Model solves the
form of the conservation of mass equation shown in Equation (2-5), the
appropriateness of the assumptions used in its formulation deserves dis-
cussion.
     The assumption to neglect the vertical component of the wind velo-
city is believed to be realistic for regions with relatively smooth
terrain.  In regions with rough terrain, the occurrence of convergent
or divergent flows caused by topographic constraints makes this assump-
tion unrealistic.
     The assumption to neglect horizontal and vertical wind shear is
believed to be realistic under certain meteorological conditions and for
limited modeling time periods.  During daylight hours with moderate or
strong wind velocities, the effects of shear are normally most important
within a thin layer above the surface.  Thus, if the wind speed used in
the model to advect the parcel is chosen to be representative for the
region above this layer, the assumption is realistic.  For meteorological
conditions characterized by light winds, the assumption is somewhat less
reliable (as is the hourly averaged wind data).  However, the light-wind
conditions, for which vertical shear is most pronounced, occur more
frequently at night, when the ground-level concentrations are lowest due
to lower source strengths and slower chemical rates of transformation.
In addition, it is believed that possible errors, introduced by the
assumption, increase with time, so that the recommended duration for
air parcel simulations is realistically limited to 8 to 12 hours.  For
practical purposes, then the  zero-shear assumption is accepted with
recognition of the associated uncertainties under some conditions.
     Horizontal diffusion can realistically be neglected when the geo-
graphic distribution of emission strengths is fairly uniform in the prox-
imity of the air parcel.  Although area source emissions in an urban
airshed generally have this characteristic, the superposition of major
                                   2-4

-------
point source emissions significantly reduces the uniformity of total emis-
sions.  To account for this phenomenon, the model has been formulated to
include the approximate effects of lateral dispersion on emissions from
major point sources in the source term, S. , of Equation (2-5).  Classi-
cal Gaussian plume solutions to the expression for mass conservation are
used to determine the downwind distribution of mass from major point
sources near the air parcel's trajectory.  The model estimates discrete
net amounts of point source emissions that are entrained into the pass-
ing air parcel.  Thus, while the model's formalism does not account for
lateral diffusion on a continuous basis, the effects due to major point
sources are incorporated in an approximate manner.  This approximation
is generally realistic for an urban airshed where the area source
emissions exceed the point source contributions.
     In summary, the assumption of conservation of mass in a Lagrangian
air parcel is realistic in regions with smooth terrain during daylight
hours with significant wind velocities, provided that special treatment
is given to lateral diffusion from major point sources.
                                 •
2.3  Numerical Methods

     2.3.1  Spatial Discretion of the Governing Equation

     To solve Equation (2-5), the discrete spatial analog of the equa-
tion is formulated using finite difference approximations (also known as
the method of lines).  The spatial discretion employed allows variable-
size mesh specification.  Given the spatial mesh description, as shown
in Figure 2-1, the continuous-time-discrete-space approximate is:
dt
                   m  c
                   [ '
          K.
                 [I] cjtl , R.. S.                                C2-6)


where c.  =  vector of n species concentration at j   spatial
             mesh point.
                                   2-5

-------
      Figure 2-1  Spatial Mesh Description
h.
c. . •
J+l



c. •
J

CJ-I •
T
A Z.
J
/ y

L

    Ca)  Geometric Parameters of Spatial Mesh

cs
C4
C3
C2
Cl
7" 7 1 1 1 I 1 1 1 I I I I I I I I i I I I i
t
r
1 1 1 1
    (b)   Discretized Air Parcel  (Not to Scale)
                        2-6

-------
     K. !   =  vertical eddy diffusion coefficient between  (j-1)

      J ^          th
              and j   mesh points


     K. ,   =  vertical eddy diffusion coefficient between  j

                       th
              and (j+1)   mesh points


     Az.   =  distance between j   and  (j+1)   mesh points


     h.    =  cell height, defined by h. =  (Az. - + Az.)/2


     [I]   =  identity matrix


     R.    =  vector of n species chemical rates for the j


              mesh points


     S.    =  vector of n species emission source rates for

      -1       -th    ,     . «.
              j    mesh points



The boundary conditions for the equations are zero flux of pollutant


mass through the ground and top edge of the air parcel.  At the ground,


which is surface mesh point 1, the boundary condition is:
           l          -  0                                         (2-7)




This boundary condition is incorporated into the finite difference equa-


tion for the surface mesh point as follows:




Let:      8cl     C2 " C0        .    A

                                     AZ0  =  AZ1
Thus:     CQ
Eliminating c» and Az_ in Equation  (2-6) results in;



          dc

                                               Ri + si
Similarly, for the top mesh point, node n, the zero flux boundary  condi-
                                                                    •

tion results in CN   = c  , and letting Az  = AzN  . results in:
                                  2-7

-------

In practice, to achieve the simulated reflection at the boundaries,  K%
and KN+^ are chosen equal to K ^ and 1C, ^,  respectively.
     Surface deposition is treated as a sink for pollutant  mass  in the
air parcel.  In the present formulation, surface deposition is repre-
sented as a component,  S , of the surface node source term  S .   Given
the surface deposition velocity V, for a particular species,  this  compo-
nent is:
              =  v cp  i
               •  vdLl  h
i.e., the surface flux is the product of the deposition velocity and the
concentration to the power p.
     The matrix form of Equation (2-6), which results after the incor-
poration of the boundary conditions,  is written as:

          ^  =  QC + R + S                                       (2-10)

where     C  =  [C^ C2,...Cn]T
                              T
             R—  fn   D     D 1
             ~*  Li >  O * * * *  -1
                              T
          s  -  [Sl, s2,...sn]
and
        2K
          1i 1                  1 1
        dr-M          IT-  [I]
        zihi                zini
                                               K.
                      ml    -*-'Z   .   "-2  I FT]      2^2
                     "I A, Vi     AT V.  J 1"1J   *_ U
                                               22

Q  =
 If    \
 2%  \
Il2h2;
                                                                  ^-^  m
                                                   AZN-1HN       AzN-lhN
                                   2-8

-------
It is important to note that this matrix has block-tridiagonal structure.
Certain advantages associated with the use of matrices with this struc-
ture greatly reduce the computational effort for simultaneous solution
of the equations.

     2.3.2  Solution of the Ordinary Differential Equations

     The continuous-time-discrete-space equation for n species at N mesh
points results in an n x N system of ordinary differential equations.
The solution to the system of equations is obtained by numerical inte-
gration, using a method first developed by Gear.  The Gear-type integra-
tion algorithm used in the model is called EPISODE (Experimental Package
of the Integration of Systems of Ordinary Differential Equations) and
was developed by Hindmarsh and Byrne (1975) at the Lawrence Livermore
Laboratory.  A complete description of EPISODE is given by Hindmarsh and
Byrne (1975), and will not be repeated here.  The advantages of EPISODE
are:

     (1)  its ability to integrate stiff equations;
     (2)  its use of a variable-order variable-step method;
     (3)  its ability to handle large systems of equations; and
     (4)  its economical operation due to its ability to take
          large steps by accurate tracking of the solution.

Most importantly, the backward differencing implicit method used by
EPISODE assures numerical stability, regardless of spatial mesh size and
equation stiffness.
     The portion of the EPISODE package that solves the linear matrix
equation QX = B for X by lower-upper (LU) decomposition at each time
step has been modified to take advantage of the block-tridiagonal struc-
ture of Q.  The mathematical formalism of LU decomposition on these type
equations will not be presented here, but it is worth noting that it is
believed to be the most efficient method available.
                                    2-9

-------
            3.   FORMALISM OF THE EDDY DIFFUSIVITY ALGORITHM
     The vertical eddy diffusivity, K  , is the parameter which  controls
                                     Zi
the rate of vertical pollutant transfer in the atmosphere.   For surface-
based emissions of pollutants, large values of K  result in  vigorous
                                                Li
vertical mixing and lower surface concentrations.  Small values of K
                                                                    Li
result in limited mixing and allow for surface build-up of pollutants.
     This section of the report describes the theoretical basis and
practice application of an algorithm to determine K  coefficients  as
functions of height and time.

3.1  K  in the Surface Layer

     If it is assumed that atmospheric pollutants diffuse in the same
manner as momentum, then the vertical  eddy diffusivity in the surface
layer is given by:

                 ku^z
                  m

where     k  =  von Karman constant  =  0.35  (Businger,  et  al,  1971)
         u^  =  friction velocity
          z  =  height above ground
          L  =  Monin-Obukhov length
         (j>   =  the non-dimensional wind shear.

     By definition,   l-pj equals 1 for neutral  conditions  (j- =  OJ.   For
non-neutral conditions in the surface layer,  Businger, et  al  (1971)
developed the following formulae:
*
m
                                                        (3-2a)
              r) =  I1 + 4'7I   for stable conditions (r >  °  )
                    r       i-3*
                 =   1 - 15|-     for unstable conditions  (-f < 0)   (3-2b)
                    L      LJ                             \L    /
                                   3-1

-------
              Substitution of Equations  (3-2a and  3-2b)  into Equation  (3-1)  yields


              formulae for determining K  in the surface layer  for  non-neutral  condi-


              tions:
                        K   =  i 4. A  7L  f°r tJie  stable  case                    (3-3a)
                         2i     X "* T" • / f
                                      L





                            =  ku^z  (1  -  15f p    for  the unstable case.        (3-3b)

                                     \      L/




              In the neutral case, the  equation  for  unstable conditions,  (3-3b),  is


              used with a large negative  value of  the  Monin-Obukhov length (L =  -10 ),


              which effectively reduces the equation to the  form for neutral condi-


              tions.


                   The depth of the  surface layer, and hence,  the extent  to which the


              surface formulae for K are applied, depends upon stability.  Above the
                                     2i

              surface layer, in the  region  of the  planetary  boundary layer (PBL)


              often referred to as the  Ekman layer,  different  formulations for K  must
                                                                                 L*

              be used.  These formulations  are presented  in  Section 3.3.





              3.2  Evaluation of Surface  Layer Parameters




                   The friction velocity  u^ is determined in the model from the equa-
              tion
                                                                                 (3-4)
/
               where    z   =  wind measurement height
                         w


                        z   =  the roughness length.
               Representations of § ( ^\ for stable and unstable conditions are given in


               Equations (3-2a and 3-2b).  For the stable case, then,
                                                                                  (3-5)
                               .  /Zw\   4.7 ,       ,
                               In —  + —r— (z  - z )
                                 \z /    L  *• w    oj
                                 \ O '
                                                  3-2

-------
Benoit's (1977) expression for the integral in Equation  (3-4) is incorpo-
rated for unstable conditions, i.e.,
ku(zw)
z
In — + In
0
/ 2 + 1
U2 + l
) (<0 + i)2l+ , tan_, ?
) (? +i)2! " ^w ta" "°
                                                                       (3-6)
                                         j
where
                1 - 15f
     The Monin-Obukhov length  (L) is determined by first evaluating the
parameter a (Fulle, 1975), defined as
          a  =
                 10 *
                             +  0.0025
(3-7)
The Pasquill stability class corresponding to a given value of a is
determined as shown in Table 3-1, and then, L is determined within each
class as a function of roughness length  (z ) using the method suggested
by Colder (1972).

3.3  K  in the Ekman Layer

     O'Brien (1970) suggested an interpolation method to determine
diffusivities between the top of the surface layer and the top of the
planetary boundary layer  (PEL).  Taking  into account the physical re-
quirement that the first derivative of K^ be continuous with height in
the Ekman layer, O'Brien formulated the  second-order equation,
                                    3-3

-------
                             TABLE 3-1
           RELATIONSHIP OF PASQUILL STABILITY CATEGORIES
            TO RANGES OF THE PARAMETER a  (FULLE,  1975)

          	                          Stability Class
    a. <_7.Q                           A   (extremely unstable)
  7.0 < a _< 8.0                       B   (unstable)
  8.0 < a _< 8.75                      C   (slightly unstable)
 8.75 < a <_ 9.5                       D   (neutral)
  9.5 < a ^ 11.25                     E   (slightly stable)
11.25 < a <_ 13.5                      F   (stable)
 13.5 < a                             G   (extremely stable)
                                3-4

-------
             (ZA - ZB)
J_  L  - K   •  '      ^    /8n      • ~ "B - "A
                        K,, - K
           I H K \
(z -
(-)
\  / -,—
                                        z. - Z-.
                               z=z_      A    B.
                                  D
                                                                      (3-8)
where     z  =  height at which K  is to be determined
         z.  =  height of the PEL
          A
         z_  =  height of the surface layer
         K,  =  value of K  at the top of the PEL
          A               z          r
         KR  =  value of K  at the top of the surface layer.

     The height of the PEL, z., in the neutral case was determined using
                             A
the relationship (Blackadar and Tennekes, 1968)
          ZA  =  c ^                                              (3-9)

where     u^  =  the friction velocity
           f  =  the Coriolis parameter.

The coefficient c has a value of 0.35 (Zilitinkevich, 1972).
     In the non-neutral case the height calculated from Equation  (3-9)
is corrected for stability.  For stable conditions the equation is
          ZA  =

(Zilitinkevich, 1972).  Reported values of c are 0.72  (Businger and Arya
(1974) 0.22 (Wyngaard 1975), 0.40  (Brost and Wyngaard, 1978), and 0.27
(Rao and Snodgrass, 1978).  For the present algorithm, a value of 0.35
was used.  This value was chosen because it is near the center of the
range reported and it is consistent with the value of c used in Equation
(3-9).  The effective mixing height, z., is not allowed to exceed the
                                      A
value calculated for neutral stability  (Equation' 3-9).
     For unstable conditions, the equation used is

                           u **
                                    3-5

-------
after Zilitinkevich (1972).  The coefficient c was arbitrarily assigned
a value of 0.35 to be consistent with the above formulations.   In this
case, z. is not allowed to be less than the neutral case value, and, for
the most unstable conditions, a maximum value of 2500m was assigned.
     The determination of the height of the PBL from Equation (3-11) is
not critical.  An inversion, which limits the extent of vertical mixing,
is generally present at some height well below z..  In this case, the
                                                J\
mixing height (H) determined in the conventional manner is used in place
of z. in Equation (3-8).  This is the height in the most recent tempera-
ture sounding at which the potential temperature equals the potential
temperature at the surface.  That is, H is the height at which a line
of slope -0.0098 °C/m (the dry adiabat) extended up from the observed
surface temperature intersects the temperature profile.  The height
calculated by Equation (3-11) is used in the model as a secondary lid to
define an upper limit to mixing for cases where no inversion is present
in the temperature sounding.
     The height of the surface layer in the neutral and stable cases is
approximated as one-tenth the height of the PBL (Haltiner and Martin,
1957).  That is,

          ZB  =  0.1zA.

     For unstable conditions, ZD is estimated by
                               D
          ZB  =  -5L ,

from Myrup and Ranzieri (1976), not to exceed the value

          ZB  =  0.1zA.

     A very small value of K  is assigned at the top of the boundary
                            Z
layer and above.  That is,
                      2
          K.  =  0.1 m /sec
           f\
                      2
              =  6.0m /min.
                                    3-6

-------
     The value of K  at the top of the surface layer is determined by


evaluating the appropriate surface layer equation for z = z .   In stable


conditions, for example, Equation (3-3a) becomes
          Kr
                 1 * 4.7jl
     It is also necessary when using Equation (3-8) to know the deriva-


tive of K  with respect to z evaluated at height ZR.  This is crucial to
         2                                        O

the determination of the shape of the K  curve in the Ekman layer.  In


the stable case, Equation (3-3a) is differentiated and ZD substituted for
                                                        D

z to yield
3K\

9z/
  'Z=Zr
                                                                  (3-12)
     In the unstable case, the surface layer formula, Equation (3-3b),


applies only within the layer where mechanically-induced turbulence is


dominant in generating vertical mixing.  At some height, or more specif-

                         2
ically, at some value of •=-, turbulence induced by buoyancy effects is
                         Lt

the dominant factor.  Myrup and Ranzieri (1976) formulated an equation


to describe buoyant eddy diffusion:
              =  c
          /  0.4  z\

          \   k   V
                               1/3
(3-13)
They suggested that this equation be used for •=-  < -5.  A value of 0.58
                                              L

for c is used to achieve continuity between Equation  (3-13) and the surface


layer formula (Equation 3-3b) at ~- = -5.  This is incorporated into the


present K  algorithm as follows.


     In the case when




          -5L  <  0.1H,



where H is determined as described above, then K  is determined from


Equation (3-13)  for



          -5L  <  z  <  0.1H,
                                   3-7

-------
and the derivative of Equation (3-13),  evaluated at ZD,  is  used in
                                                     D
Equation (3-8).  That is,
                      2    0.4
                      3    —  ~      U*
On the other hand, when

          -5L  _>  0.1H ,

that is, for slightly unstable conditions,  (Equation (3-13)  is not used
at all and the derivative of Equation (3-3b)  evaluated at zn,
                                                           D
         (H)=
                                                    -3/4
             Z=ZB
is used in Equation (3-8).
     It is commonly observed that in the early morning a surface-based
radiation inversion is surmounted by a deep neutral layer which is,  in
turn, capped by a subsidence inversion.  The mixing which occurs in  the
elevated neutral layer is simulated by calculating K  as though the
layer extended to the surface.  That is, if the top of the surface-based
inversion is at 100m and the base of the elevated subsidence inversion
is at 1,000m, then K  is calculated between 100 and 1,000 in the same
                    ti
manner as if there were no surface-based inversion, except for one minor
modification.  To simulate a realistic transition across the upper
boundary of the surface inversion, a smoothing factor is applied to  the
calculated values of K  in the neutral layer.  This factor is defined as:
                 z - z,
S,  =
                      t
                ZA- ZT
                1/4
                                                                 (3-16)
where     z   =  height above surface
         ZT   =  height of top of surface inversion
         z.   =  neutral boundary layer height (see Equation 3-9).
                                    3-8

-------
3.4  Application of Meteorological Data

     The meteorological data base is used to evaluate the parameters
used in the K  algorithm as follows.
     The information required to evaluate the parameter a (see Equation 3-7)
                                                                3T
and subsequently L, includes the vertical temperature gradient (V-) and
                               n.                                0 Z
the vertical wind speed shear (-^-).   The surface temperature gradient in
                               oZ
the morning, when the temperature increases from its minimum value, is
                                                                    ST
interpolated between an estimate of the maximum midday lapse rate  (-57)
and the neutral lapse rate (-0.98 °C/100m).  The maximum midday temperature
lapse rate is estimated from the nearest midday sounding.  The depth of
                                          ST
the layer of superadiabatic lapse (i.e., -•*— > 0.98) is also estimated
from the available vertical temperature profile data.  During this time,
the temperature profile is updated each hour above the surface superadiabatic
layer by extending a dry adiabat from the temperature at the top of the
surface layer to its intersection with the initial profile.
     In the afternoon, after the maximum surface temperature is reached
and during the period when the surface temperature initially decreases,
the temperature gradient is calculated from the surface temperature and
the temperature at the top of the superadiabatic layer at the time of
the maximum surface temperature.
     Once the surface temperature has decreased such that the lapse rate
in the surface layer is less than the neutral lapse rate, the surface
lapse rate is estimated for each hour in a different manner.  This
estimation is accomplished by determining the slope of a line extending
from the current surface temperature and intersecting the most recently
updated profile at the height of the surface inversion top present in
the morning's profile.  This method is illustrated in Figure 3-1.
     The wind shear, -5—, at the surface was estimated as
          3z    z
                 w
where u(z ) is the wind speed along the trajectory at measurement
height zw.  The estimate of |£ is used along with the temperature
gradient, as estimated above, to determine a value of L from the procedure
described in Section 3.2.
                                   3-9

-------
HEIGHT
   SFC
                             TEMPERATURE
 Figure 3-1  Method for updating sounding following afternoon
             temperature maximum.  Solid line is temperature
             sounding at time of surface temperature maximum  (Tm)
             Dashed line is subsequent update for a lower
             surface temperature (T ).  Z.is depth of the

             morning's surface  stable layer.
m
                         3-10

-------
     The wind shear in  an  elevated layer is approximated by first making
an estimate using the logarithmic wind law for neutral conditions:
~te
                               In
                         -  k(z2 - 2l)

where u? and u1  are the  estimated wind speeds at heights z2 and z
corresponding to the top and  the bottom of an elevated layer.   The
friction velocity,  u^, is determined initially from the logarithmic
law using the interpolated  surface wind:
                 ku(z )
                     w
                                           (r\ \
                                           3—j  ,  determined from
Equation (3-18),  is used with  the temperature gradient from the most
recent profile to determine  an initial value of L for the layer as
described in Section 3.2.  This value of L is then used to determine
                               rv
an initial non-neutral  value of •_— from the equations

                                -1/4
            /u   \   'nx
for unstable conditions,  and
         /£)   .(*)   (1*4.7^)
         \3z's    \3z/n   V       L /
for stable conditions,  where  ZT  is the height of the top of the
layer. This procedure is  repeated until the difference between the non-
neutral -g— on two successive  iterations is sufficiently small to result
in a difference of less than  1%  in the corresponding values of L.   This
final value of L is then  used to determine the stability within the
elevated layer.
     The estimated stability  is  used with the measured wind speeds each
hour to determine the friction velocity, u^, using Equation (3-5)  or
(3-6).  As mentioned in Section  3, hourly-measured surface temperatures
and available vertical  temperature soundings are used to determine the
mixing height, H.

                                  3-11

-------
3.5  Typical Profiles

     Typical profiles of vertical  eddy diffusivity are shown in Figure
3-2 and 3-3.  Figure 3-2 depicts a profile under stable conditions,
as with a surface-based inversion.  Figure 3-3 shows a typical  profile
for an unstable mixed layer capped by an elevated inversion.
                                   3-12

-------
 4001
300-
200-
100  -
                 I      I
                                            T     i      r
                   K, (m2 /sec)
    Figure 3-2.  Vertical  Eddy Diffusivity Profile for Slightly
                Stable Conditions (L=1060m,  and u* = 0.09 m/sec)
                          3-13

-------
      700
     600 J
     500 H
     400 H
o>
I
     300H
      20CH
      rxH
                         Kz (mT/sec.)
           Figure 3-3.  Vertical Eddy Diffusivity Profile for

                        Unstable Conditions (L=-27m,  u*=0.36,
                        and  H=670m).
                             3-14

-------
           4.  DEVELOPMENT OF A HOMOGENEOUS GAS PHASE MODEL FOR
               THE PREDICTION OF OZONE AND SULFATE FORMATION

     This section describes the formulation of a chemical mechanism to
describe the gas phase reactions which produce ozone (0_) and sulfate
(S0.=0 from irradiated mixtures of hydrocarbons (HC), nitrogen oxides
(NO ) and sulfur dioxide (S0~).  The majority of the work for this part
   X                        w
of the model development was carried out under funding from the Coordi-
nating Research Council and the National Science Foundation.  Only
slight modifications were made under this EPA program.
     It is recognized that major advances have been made in our under-
standing of certain segments of the mechanism (e.g., 0, - olefin reac-
                                                      O
tions, aromatic hydrocarbon photooxidation) since the formulation of
this mechanism.  However, due to time constraints of this program, these
latest changes are not reflected in the mechanism.  Updating of the
chemical mechanism incorporating the latest information is currently
being carried out under continued funding from the Coordinating Research
Council.
     The following discussion describes initially the chemistry occur-
ring in the hydrocarbon/nitrogen oxides (HC/NO ) system and subsequent-
                                              J\
ly, the additional reactions necessary to describe the homogeneous gas
phase conversion of S02 to sulfate.  This is followed by a description
of the reactions used to computer model the hydrocarbon/NO /SO  system
                                                          A.   A
and a presentation of the comparison of the predicted results against
smog chamber data.

4.1  Formulation of a Chemical Mechanism for the HC/NO  System
                                                      A
     The chemical kinetic mechanism for the HC/NO  system, including the
                                                 A.
appropriate rate constants, is discussed in detail below.  Briefly, it
is a lumped hydrocarbon model in which the hydrocarbons are partitioned
among five classes, namely, alkanes, alkenes, aromatics, formaldehyde
and the higher aldehydes.  In addition, some of the other oxygenated
hydrocarbons, such as ketones, may be placed into the higher aldehyde
category if desired.
     A detailed discussion is given of the inputs into the chemical
mechanism formulation.  This was felt necessary since the quality of
                                    4-1

-------
the results of model predictions is a strong function of the input data
(kinetic, mechanistic and experimental).

     4.1.1  Chemical Mechanism Development

     The chemical mechanism used was tested against various smog chamber
data in a systematic manner.  The following chamber data (from the
University of California, Riverside chamber facility) (Pitts, et al,
1977a; Pitts, et al, 1976; Pitts, et al, 1975a) have been utilized:

     •    n-butane/NO  mixtures
                     X
     •    propylene/NO  mixtures
                      J\.
     •    n-butane/propylene/NO  mixtures
                               A.
     •    surrogate hydrocarbon mixtures representative of the atmosphere
          and composed of 4 alkanes, 4 olefins, 2 aromatics, 2 aldehydes,
          CO, CH., and acetylene.  The detailed composition is shown in
          Table 4-1 (Pitts, et al, 1975a).

     The calculations, including those involving SO- described later,
were performed on an IBM 360/75 computer and a CDC 6400 computer, using
a chemical kinetics program written in FORTRAN IV.  The list of species,
reactions and rate constants, and stoichiometric coefficients were input
to the program, forming a set of kinetic differential equations.  These
equations were integrated, using an ordinary differential equation
integration package developed by Gear (1971) for the variable-step-
variable-order integration of coupled differential equations.  Calcula-
tions were done in single precision with an error bound of 0.1%.
Integrations ran smoothly with no problems encountered due to stiffness.

     4.1.2  Kinetics and Mechanism for the Photochemical Model

     In developing the chemical module for the atmospheric model,
extensive use has been made of previous and current studies.  These
include those by Niki, et al (1972); Demerjian, et al (1974); Hecht, et
al (1974); Whitten and Hogo (1977); Graedel (1976); and Carter, et al
(1978).  However, in order for the chemical scheme to have practical
applications in terms of airshed models, the number of species has to be
                                    4-2

-------
                             TABLE 4-1
       DETAILED COMPOSITION OF HYDROCARBON SURROGATE MIXTURE
                (modified from Pitts, et al, 1975a)
Compound Class
Detailed Composition and Concentration
 (parts per billion as carbon, ppbC)
 Alkanes


 Alkenes


 Aromatics

 Aldehydes

 Miscellaneous
Ethane (160), Propane (40), n-Butane  (785),
2,3-Dimethyl butane (615)

Ethene (84), Propene (35), cis-2-Butene  (60),
2-Methyl-2-Butene (70)

Toluene (115), m-Xylene  (325)

Formaldehyde (54), Acetaldehyde  (5)

Acetylene (101), Acetone  (6), Methane  (2800),
Carbon Monoxide  (7000)
                                   4-3

-------
carefully limited.  Thus, inherently, the mechanism is an approximation
adapted to the treatment of multicomponent mixtures of great complexity
in contrast to the models developed by the majority of the above workers.
     Our chemical mechanism has been significantly changed from that of
Hecht, et al (HSD) (1974); however, our partitioning of the hydrocarbons
into classes is very similar to that used by HSD with the exception of
the fact that the aldehydes are split between formaldehyde and the
higher aldehydes.
     As indicated above, the formulation of a chemical kinetic scheme
for use in an atmospheric model requires a substantial reduction in the
level of chemical detail to be treated.  For example, the numerous
individual hydrocarbons must be "lumped" into various representative
categories.  The type of lumping employed can vary.  Thus, we have
partitioned the hydrocarbons into five classes, while other workers have
used fewer classes (Hecht, et al, 1974) or have treated the total
reactive hydrocarbons in terms of propylene (Graedel, et al, 1976;
Wayne, et al, 1971) or a mixture of propylene and n-butane  (Dodge,
1977)  -- the latter two compounds representing the highly reactive and
lesser reactive species, respectively, in the atmospheric mixture.
     In view of the differences in reaction mechanisms, we felt at the
outset that the hydrocarbons should be divided among alkanes, alkenes,
aromatics, formaldehyde and the higher aldehydes.  As with any lumping
technique, errors are incurred, for example, an average, mole-weighted
rate constant, which is representative of the class, has to be chosen.
Where there is a wide variation in rate constants within a class, this
assumption leads to an over or underprediction in reactivity during
different stages of the reaction.  A problem inherent with our approach
is the fact that the grouping, represented by the symbol "R" for the
generalized case, does not get degraded as rapidly in the model as would
happen under experimental conditions, e.g., reaction 35  (Table 4-2)
shows that R occurs on both sides of the equation although a carbon atom
has been lost through the production of CO.  We have tried to minimize
the effect of this approximation by including nitrate and other stable
product formation, and by differentiating among the alkyl groups origi-
nating from alkane and alkene  species, e.g., alkanes are denoted by PA
and alkenes by OLEF.
                                    4-4

-------
     The extensive body of smog chamber data obtained by the Statewide
Air Pollution Research Center of the University of California, Riverside,
in its studies for the California Air Resources Board (Pitts, et al,
1975; 1976) has proved invaluable in checking the simulations against
experimental observations under varying HC/NO  ratios and total concen-
                                             A
trations of HC and NO .
                     J\
     The reaction mechanism for the HC/NO  system is presented in
                                         X
Table 4-2.  Table 4-3 lists the definitions of the chemical species
symbols used in Table 4-2.  The additional reactions necessary to
describe the S0_ oxidation are presented later in this report.   The
rate constants shown in Table 4-2 are operative at 305°K and refer to
the hydrocarbon mix used by Pitts, et al, in their smog chamber studies.
In instances where M or 0  occur in the reaction mechanism, their
concentration has been included in the rate constant to give a pseudo
first or second order rate constant.

     4.1.3  Salient Features of the Inorganic Mechanism

     The inorganic mechanism for hydrocarbon/NO /air mixtures has been
                                               A
generally well characterized.  The mechanism has been discussed exten-
sively in previous publications (Niki, et al, 1972; Demerjian, et al,
1974; Hecht, et al, 1974; Whitten and Hogo, 1977; Graedel, et al, 1976;
Carter, et al, 1978), and the major reactions of 0 , NO, N0_ and water
are generally well known.  However, continuing research has uncovered
some important reactions, such as the formation of pernitric acid,
HO NO,,, which have previously been neglected.  In addition, recent work
on OH reactions with NO, N02 and CO has led to the use of higher values
of their rate constants.  These recent studies have shown the importance
of studying reactions under atmospheric conditions of temperature and
pressure.
     Many of the rate constants were taken from the National Bureau of
Standards Technical Note #866, entitled "Chemical Kinetics and Photo-
chemical Data for Modeling Atmospheric Chemistry" (Hampson and Garvin,
1975).  In some cases, particularly for the organic reactions, kinetic
data are generally taken from the work of Carter, et al (1978), while
the remainder were derived in studies carried out for the Coordinating
Research Council (Lloyd and Tashima, 1977) and the National Science
                                   4-5

-------
           TABLE 4-2

GENERALIZED REACTION MECHANISMS
                       Rate Constants (ppm   min~ )
                        at 505°K and 1 Atmosphere
 1

 2

 3

 4

 5

 6

 7

 8

 9

10

11

12

13

14

15

16

17

18

19

20

21

22



23
N02 + hv  =  0 + NO

0 + 0  (+ M)  =  0
   + NO  =  N02 -i-
                       M)
NO + N02 + H20  =  2HONO

HONO + HONO  =  NO + N02

HONO + hv  =  OH + NO

OH + NO + M  =  HONO + M
OH + N02 + M  =  HN03 + M

OH + CO  =  H02 + C02

H02 + NO  =  N02 + OH
N03 + NO  =  2N02
            =  2HN0
OH + OLEF  =  A0
A02 + NO  =  N02 + AO

AO + 0   =  RCHO + HCHO + H0
03 + OLEF = 0.5HCHO + 0.5RCHO + 0.25H02 +

            0.25RC03 + 0.50H + 0.5RCH02

RCH02 + N02  =  N03 + 0.5HCHO + 0.5RCHO
                                 3.20E-01

                                 4.12E+06

                                 2.50E+01

                                 2.20E-09

                                 1.40E-03

                                 8.96E-02

                                 1.50E+04

                                 1.50E+04

                                 4.40E+02

                                 1.20E+04

                                 1.20E+03

                                 4.00E+00

                                 8.40E+03

                                 5.00E-02

                                 1.30E+04

                                 5.60E+03

                                 5.00E-06

                                 2.40E+01

                                 4.00E+04

                                 2.90E+04

                                 4.10E+05



                                 2.00E-01

                                 2.30E+04
                  4-6

-------
TABLE 4-2  (CONTINUED)
                Rate Constants (ppm   min  )
                 at 505°K and 1 Atmosphere
24
25

26
27

28
29
30
31

32
33
34
35

36
37
38
39
40
41
42
43
44
45

46
RCH02 + NO
0 + OLEF =

OH + PA =
PA00 + NO
2
PA02 + NO
PAO = R02
PAO + 02 =
PAO + NO

R02 + NO =
OH + RCHO
RCHO + hv
RCO_ + NO
3
RCO + N09
•3 ^
= N02 + 0.5HCHO + 0.5RCHO
.3EPOX + .3RCHO + .4H02 +
•4R02
H20 + PA02
= N00 + .85PAO + .15R00
2 2
= NTRA
+ .5HCHO + .5RCHO
.5KET + .5RCHO + H02
= .85NTRA + .15RCHO +
.15HONO
NO + PAO
= RC03 + H20
= R02 + H02 + CO
= C00 + N00 + RO,
222
= PAN
PAN = RC03 + N02
HCHO + hv
HCHO + OH
R02 + H02
03 + WALL
AR + OH =
AROH + OH
ARCO + NO
AR + OH =

ARO + NO =
= 2.H02 + CO
= H20 + H02 + CO
= R02H+02
= LOSS OF 03
AROH + H02
= ARCO
= N02 + RCHO
H20 + ARO
i
NO,, + H00 + AR CHO
2.90E+04

2.30E+04
3.80E+03
2.90E+04

2.60E+03
1.40E+05
6.70E+04

2.30E+04
2.90E+04
2.20E+04
1.40E-04
2.90E+04

1.70E+04
6.60E-02
2.30E-04
2.10E+04
4.20E+03
3.21E-04
2.24E+04
5.02E+04
2.90E+04
5.60E+03

2.90E+04
          4-7

-------
Species

  AO

  A02

  AR

  ARCO


  ARO



  AR'CHO

  AROH

  CO

  co2

  D

  EPOX

  HCHO

  HONO

  HN03



  H02

  H20


  H2°2

  HV

  KET

  M

  NO

  NO,,
             TABLE 4-3

CHEMICAL SPECIES SYMBOL DEFINITIONS


                   Symbol Designation

 Alkoxy radical equivalent of A02

 Product of OH addition to olefin in the presence of 0~

 Aromatic hydrocarbons

 Product of addition of OH to a cresol
 in the presence of 0_

 Product of H-abstraction from side chain alkyl
 group on benzene ring followed by addition of
 0? to radical formed

 Aromatic aldehyde

 Cresol

 Carbon monoxide

 Carbon dioxide

 Criegee intermediate  (RCH02)

 Epoxide formed from 0 atom addition to olefin

 Formaldehyde

 Nitrous acid

 Nitric acid

 Pernitric acid

 Hydroperoxyl  radical

 Water

 Hydrogen peroxide

 Photon

 Ketone

 Any  third body,  such  as  N2 or 02

 Nitric oxide

 Nitrogen dioxide
                                 4-8

-------
                    TABLE 4-3 (CONTINUED)
Species

  NO
    J

  N2°5

  NTRA

  0


  °2

  °3

  OH

  OLEF

  PA

  PAN

  PAO

  PA02



  R

  RCHO

  RC03

  RO
  R02H
                  Symbol Designation

Nitrate radical

Dinitrogen pentoxide

Organic nitrate

Oxygen atom (ground state)

Oxygen

Ozone

Hydroxyl radical

Alkenes (olefinic hydrocarbons)

Alkanes (paraffinic hydrocarbons)

Peroxyacetyl Nitrate

Alkoxy radical formed by PA

Alkyl peroxy radical from the 0,, addition
to the radical formed by H-abstraction
from a paraffinic hydrocarbon

Generalized alkyl group (e.g., C2H5' C?H-, etc.)

Aldehydes other than formaldehyde

Acyl peroxy radical

Alkoxyl radical

Alkyl peroxy radical

Product of disproportionation between H0? and R0_
                                 4-9

-------
Foundation (Martinez, et al, 1977).  The majority of the following
discussions has appeared in these two reports but will be repeated here
for completeness.
     There are recent experimental data for several of the inorganic
reactions of interest.  These are discussed in detail below:

     The Chemistry of Nitrous Acid

     The formation of nitrous acid by the reaction

          NO + N02 + H20 i 2HONO

has been investigated recently by several workers, including Chan, et al
(1976); Cox (1975); and Zafonte, et al (1975).  This reaction together
with the reaction

          OH + NO + M -*• HONO + M

are presumed to be the major sources of nitrous acid formation in
chamber studies, and probably in ambient air.  We have used the data
obtained by Chan, et al (1976), for the formation and destruction of
nitrous acid.  In addition, we have used the photolysis rate constant
obtained by Cox and Derwent (1977).
     In order to account for the radical initiation in smog chamber
studies (Whitten and Hogo, 1977; Carter, et al, 1978; Dodge,  1977; Niki
and Weinstock, 1975; Wu, et al, 1976) and to take into account chamber
wall contamination (Bufalini, et al, 1972), we have used a portion of
the equilibrium HONO concentration to provide an initial radical source
to initiate the photochemical smog reactions  (Whitten and Hogo, 1977;
Jeffries, et al, 1975).  As discussed in Section 4.1.6, the values used
are usually one quarter of the equilibrium HONO concentration.  One
problem in using this method to account for the radical initiation is
that it tends to shorten the ozone initiation time, but does not
adequately account for radical production later in the run.  In essence,
it affects the time to ozone maximum but not the value of the maximum.
However, if no significant initial radical source is used in simulating
smog chamber experiments, then the reactivity of the system, in terms of
time to N0? maximum, rate of NO oxidation, etc., is invariably under-
predicted.  As discussed below, it may be unnecessary to use initial
                                  4-10

-------
HONO as a radical initiator under ambient conditions since there may be
sufficient aldehydes present.

     The Reaction of H02 with NO

     There have been several studies (Hampson and Garvin, 1975) of the
reaction

          H02 + NO ->- N02 + OH  .

We have used the value reported by Howard  (1977) , who obtained a rate
constant of 1-2 x 10  ppnf  rain"  based on a laser magnetic resonance
study of HO-.

     The Formation of Pernitric Acid From The Reaction of H02 With N02

     The importance of pernitric acid in photooxidation studies involving
nitrogen oxides has become increasingly evident in the last few years
(Niki, et al, 1976a; Levine, et al, 1977).  While earlier studies
(Simonaitis and Heicklen, 1974; Cox and Derwent, 1975; Simonaitis and
Heicklen, 1976) showed the formation of nitrous acid from this reaction,
there now seems to be reasonable agreement that pernitric acid is the
major, if not the only, product, Niki, et al (1976a; 1977a) .
Pernitric acid decomposes in a manner similar
to peroxyacetyl nitrate with a rate constant which obeys the Arrhenius
expression
           k = 1-3 x 1014 exp  ("2RT°0)   sec"1   (Graham, et al, 1977).

     The value chosen for the HO,NO,, decomposition rate affects the
final 0, peak since H02N02 acts as a reservoir for N02 .  A short life-
time means that the net effect of reaction (11) on 0, production is not
significant, and vice versa.
                                 4-11

-------
0  Decay at the Smog Chamber Walls

     In order to model 03 behavior in a chamber, one has to take into
account the 0  reactions with the walls of the chamber.  This is usually
calculated by measuring the rate of decay of ozone in a nonirradiated
chamber and quoting a half-life for the dark decay.  A half-life of 35
hours for dark decay was used (Darnall, 1977) for the simulations of the
surrogate data from the UCR-SAPRC glass chamber, indicating that the
surface is very inert to ozone reaction.

     4.1.4  Salient Features of The Organic Mechanism

     The standard organic reactions used in modeling studies have been
discussed in detail by many workers (Niki, et al, 1972; Demerjian, et
al, 1974; Hecht, et al, 1974; Whitten and Hogo, 1977; Graedel, et al,
1976; Hendry, et al, 1978; Carter, et al, 1978).  Thus, the present
discussion of individual reactions is limited to a few key aspects in
the treatment of the organic system.

     Olefin Reactions With 07
     ~~ ~"""  '                   O
     The major role in photochemical oxidant formation played by the
reactions of 0_ with olefins has been known for many years and described
              O
in detail elsewhere (Niki, et al, 1972; Demerjian, et al, 1974; Hecht,
et al, 1974; Whitten and Hogo, 1977; Graedel, et al, 1976; Carter, et
al, 1978).  Recently (Niki, et al, 1976; Niki, 1977; Walter, et al,
1977), studies have shown that the reaction produces fewer radicals than
previously thought.  Thus, a major advance was made with the definitive
identification of a stable secondary ozonide by Niki, et al  (1976),
which confirmed the earlier but less definitive work of Hanst, et al
(1958).  Therefore, it appears that Criegee biradicals  (RCHCL • ) are
formed under atmospheric conditions and last sufficiently long to react
bimolecularly.  We have assumed that they react rapidly with NO and N02.
However, there is still significant uncertainty concerning the relative
rates of molecular  (path A) compared to fragmentation routes  (path B)  in
                                    4-12

-------
the overall reaction
     0  + R-CH = CHR^O.SRCHO + 0.5R1CHO + O.SRCHO^  + 0.5R1CH02  (A)

     0_ + R-CH = CHR -^O.SRCHO + O.SR^HO + 0.5RCO + O.SR^O + OH  (B)

where RCHO •  and R CHO™-  represent the Criegee intermediates.  Early mod-
eling studies depicted the 0_-olefin reaction in terms of path B.  The
net reaction and overall rate constant are shown in Table 4-2.  Equal
partitioning between the two paths has been assumed in the current study.
     In a surrogate hydrocarbon model, the uncertainty of the mechanism
of the ozone-olefin reaction is compounded by the uncertainty inherent
in choosing an appropriate olefin representative of the bulk olefin
mixture.  Such a choice has a bearing on both the rate constant and the
mechanism.
     In this study, a molar-weighted rate constant was chosen based on
the composition of the surrogate hydrocarbon mixture used in the smog
chamber experiments.  Thus, the olefins used in the surrogate mixture
are ethylene, propylene, cis-2-butene, and 2-methyl 2-butene, as shown
in Table 4-1.  The value of the molar-weighted rate constant derived
lies between the value for the reaction of 0, with propylene and that
for cis-2-butene, while the mechanism is largely representative of that
for propylene.

     Olefin Reactions With OH

     The consumption of olefins, during the early stages of reaction
before a significant build-up of ozone occurs, is attributed to the
reaction with the hydroxyl radical  (Niki, et al, 1972; Demerjian, et al,
1974).  Olefins react rapidly with OH.  Cvetanovic  (1976) has shown that
the mechanism is largely one of addition, with about 65% of reaction
occurring by addition to the internal carbon atom, while the other 35%
occurs by addition to the terminal carbon atom.  The mechanism used in
this study ignores any hydrogen abstraction from the olefin.
     Following addition of OH and subsequently of 0^ to the radical so
produced (reactions 18-20), the mechanism predicts the conversion of two
molecules of NO to NO , which is consistent with the findings of the
work of Carter, et al (1978).  This is also generally consistent with
                                  4-13

-------
the recent studies of Niki (1977).   This mechanism differs from that
used by Demerjian, et al (1974), since the reaction of CL with the
hydroxy-type radical (RCHOH), shown as reaction 21 in Table 4-2, leads
to the production of H02 and an oxygenated species, rather than addition
of 02 to the RCHOH radical as follows;

          RCHOH + 02 ->• RCH(OJOH

     Olefin Reactions With Oxygen Atoms

     Under atmospheric conditions,  the major portion of the oxygen
atoms generated by NO^ photolysis combines with molecular oxygen to form
ozone.  However, oxygen atoms react rapidly with olefins, particularly
the internally-bonded olefins, and, thus, in the early stages of reac-
tion when the olefin concentration is reasonably high, the overall rate
of reaction of 0 atoms with olefins can be significant and helps to ini-
tiate radical production and to commence the process of ozone formation
(Niki, et al, 1972; Demerjian, et al, 1974; Hecht, et al, 1974; Whitten
and Hogo, 1977; Graedel, et al, 1976; Carter, et al, 1978).  Our mecha-
nism reflects the recent work of Singleton and Cvetanovic (1976), and
the reaction mechanism is shown by reaction 24 in Table 4-2.  It is seen
that some stable products, such as epoxides, are formed part of the
time, while other products result from radical fragmentation processes.
The partitioning among the two will undoubtedly change depending upon
the olefins, but we have chosen to use the results for propylene as
being typical of the mixture under study.  This should also be a reason-
able assumption for olefins encountered in ambient air.

     The Photooxidation of Alkanes
     It is assumed  in this  study that  the only mechanism  for  the  con-
sumption of alkanes under ambient  conditions  is that  due  to reaction
with OH.  Since the C-H bond  strengths in the paraffins are reasonably
high (95-98 kcal mol" ), the  only  species for which hydrogen  abstraction
reactions are  significantly fast is OH.  The  alkane photooxidation
mechanism is probably the best understood of  all  the  hydrocarbon  species
(Whitten and Hogo,  1977; Carter, et al,  1978) and is  represented  by
reactions 26-33 in  Table 4-2.
                                    4-14

-------
     The mechanism used in this study includes the results reported by
Carter, et al (1976) and Darnall, et al (1976) -- namely, the formation
of nitrates from the longer chain (^C.) alkanes in the R0» + NO reaction,
as shown below, and the isomerization of alkoxy radicals formed from
C.  hydrocarbons.  For example, the ratio of rate constants for the
reaction

           (27)  PA02 + NO -*- N02 + PAO

to that of

           (28)  PACL + NO ->- Nitrate

is about 11.2, as determined earlier by Darnall, et al  (1976), in a
chamber study of n-butane/NO /air mixtures.
                            A.
     The mechanism also takes into account nitrate formation from the
normal reaction of alkoxy radicals with NCL  (reaction 33); in this case
the alkoxy radical is denoted by PAO, which is the radical derived from
the paraffins.  Naturally, the choice of 11.2 for the ratio of k0_/k_0
                                                                z /  £o
may vary in going to ambient air when significant proportions of  longer-
chain alkanes are present.  Thus, the choice of this ratio in any model
application has to be examined in light of any detailed hydrocarbon data
available  from urban airshed studies specific to the region under
consideration.  Otherwise, the value of 11.2 may be used.

     Aromatic Hydrocarbon Photooxidation

     At the time when this study was being carried out, very little
information existed on the photooxidation of aromatic compounds under
photochemical smog conditions.  The mechanism shown in Table 4-2, taking
toluene as a representative aromatic hydrocarbon, included the known
results of that time.  Thus, the mechanism reflects the known addition
(Davis, et al, 1975; Hansen, et al, 1975; Doyle, et al, 1975; Lloyd, et
al, 1976)  of OH to the aromatic compound followed by subsequent reac-
tions  shown.  In addition, it is assumed that abstraction of hydrogen
atoms  from the side chain occurs about 20% of the time  (Perry, et al,
1977; Hendry, 1977), with the net result of the formation of an aldehyde,
e.g.,  from toluene, benzaldehyde would be produced, as  shown in Fig-
ure 4-1.
                                   4-15

-------
Addition of OH
     CH,
CH,,
                      0,
             + OH
                                       OH
                               o-cresol
                   HO,
     CH.
 CH
          OH
      OH
               OH
                                  /\
                                 OH H
                                        H
Abstraction by OH
   1  CH.
             + OH
                                                  NO
    CH,
                NO,
 OH/XH


"RCHO"
                                 OH


                                 ^
                                 \
                                  H
                                                  NO
                                                  NO,
                                     CHO
              + 0,
                 + HO,
                             benzaldehyde
       Figure  4-1   Scheme  for Photooxidation of a Typical  Aromatic

                   Hydrocarbon Using Toluene as an Example
                                 4-16

-------
     The mechanism considers the possibility of ring-opening following
addition of OH and 0_ to the cresol formed from the addition of OH to
toluene.  Such ring-opening has been suggested by O'Brien (1975) and
Hendry (1977).  Likely products from ring-opening would involve multi-
functional oxygenated compounds, and we have chosen to represent these
compounds by an aldehyde, RCHO.  This is done to conserve the number of
species involved in the model, but also to take into account the possi-
bility that the products resulting from ring-opening will be photo-
reactive and create further radicals.  This technique has been used
previously (Pitts, 1975) .
     The reaction mechanism is currently being refined as more data on
the aromatic photooxidation steps become available.  The revised mechanism
is being formulated and tested under another program sponsored by the
Coordinating Research Council.

     Aldehyde Photolysis

     It has been known for some time that aldehydes are key radical
initiators in the photooxidation systems present in ambient air and in
smog chambers (Pitts, et al, 1976; Demerjian, et al, 1974; Dodge and
Hecht, 1975; Dodge and Whitten, 1976) .  Quantum yields for radical
production for aldehydes in air are highly uncertain, although it has
been shown that the rate constants for aldehyde photolysis are very
important in terms of photochemical smog modeling  (Dodge and Hecht,
1975).
     Our mechanism differentiates between formaldehyde and the higher
aldehydes.  We feel this is important in view of the different radicals
formed and the slightly different shape of the formaldehyde absorption
curve compared with the higher aldehydes (Calvert and Pitts, 1966) .
This is most manifested by the absorption at longer wavelengths by
formaldehyde compared with the other aldehydes, and this is reflected in
the urban airshed model by a difference in their photolysis rate con-
stants with altitude.  In addition, by differentiating the aldehydes in
this way, more realistic predictions of PAN concentrations are possible
since formaldehyde does not form PAN.
     Our model calculations are sensitive both to the initial concentra-
tions used for aldehydes, as well as to the rate constants used for
                                   4-17

-------
photolysis.  We have chosen rate constants for aldehyde photolysis based
on measurements obtained in the glass chamber at UCR and the quantum
yields of Lee, et al (1976).  We have taken into account the recent work
of Weaver, et al  (1976), by combining the two possible photooxidation
paths, one which is dependent on the oxygen concentration, and the other
which is independent.  In addition, we have assumed that the formyl
radical formed upon photolysis of formaldehyde reacts with 0  as follows:

          HCHO + hv -*- H2 + HCO

          HCO + 02  -*- H02 + CO

This pathway has been shown by Niki, et al  (1977) to be the major one
occurring under atmospheric conditions, rather than the alternative
suggestion:

          HCO + 0  -*- HCO
                 <£       O
      In view of the sensitivity of  the model to both the rate constant
and the initial conditions, it would be highly desirable for better data
to be available both on the photochemistry of the aldehydes under
ambient conditions and on measurements in ambient air under varying
levels of photochemical smog intensity.  Results of measurements of
formaldehyde quantum yields obtained after  completion of this work have
been  recently summarized  (Lloyd,  1978).

      4.1.5  Validation of Photochemical Model

      A number of smog chamber  runs  were modeled to ascertain the appro-
priateness of both  the  inorganic  mechanism  and the mechanism used  to
describe  the photooxidation of the  alkanes  and the alkenes.  Specifi-
cally, runs were  carried out for  mixtures of n-butane/NO   and propy-
                                                        .A.
lene/NO   in air, and  a binary  mixture  of propylene and  n-butane in  air
       •X
containing NO  .  These  chamber data were obtained for the  EPA at the
              .A.
smog  chamber  facility at  the University of  California,  Riverside  (Pitts,
et  al,  1977).   The  evacuable,  teflon-lined  chamber irradiated by  a solar
simulator was used  in these runs  (Beauchene, et al,  1973).
                                    4-18

-------
     4.1.6  Results and Discussion

     Figure 4-2 is included as being typical of the results obtained for
the n-butane/propylene mixtures.  This figure shows that the mechanism
used in this study predicted well the decay of both the paraffin and the
olefin, although in the later stages of the irradiation, the model
slightly overpredicted the decay.  While the predicted ozone initiation
time was slightly shorter than the experimental value, the overall
agreement between the experimental and predicted values is good.  The
overprediction of the PAN concentration is reflected in the underpre-
diction of the later N02 concentrations.  Little effort was expended in
refining the details of the mechanism since the overall goal is to model
a complex hydrocarbon mixture under ambient conditions.
     As indicated earlier, the smog chamber data chosen to represent a
complex hydrocarbon mixture was that obtained at SAPRC, under a Califor-
nia Air Resources Board contract.  In contrast to the above data, these
were obtained in an all-glass chamber operated under conditions of room
temperature and pressure and irradiated externally by banks of black
lights  (Pitts, et al, 1975, 1976).  Table 4-4 shows the specific runs
chosen for model validation purposes.  These span a variety of hydro-
carbon/NO  ratios to provide a reasonably stringent test for the model.
         J\.
     In order to employ appropriate rate constants for this complex
hydrocarbon mix, the mole-weighted rate constants for the reactions of
OH with alkanes, aromatics and olefins, the reactions of 0 atoms with
olefins, and the reaction of ozone with olefins were obtained.  These
values were derived by weighting the appropriate rate constants of the
elementary reactions to reflect the concentration of the individual
components in the mix.  In this way, reasonably representative rate
constants were obtained.  The rate constants used are shown in Table 4-
5, while the mole-weighted values used in our computations and derived
from these values and the individual compound concentrations present in
the surrogate mixture, are included in Table 4-2.
     Figures 4-3 to 4-8 show the comparison of the computed values and
experimental data obtained for the various smog chamber runs selected
for model validation.  The mechanism used for these runs is the same as
that shown in Table 4-2.  The initial concentration of MONO was kept at
one quarter the equilibrium value for four runs but was placed at 1/16
                                   4-19

-------
  741
  556
  .371
(ppm)

  .185
             Figure 4-2   Simulation of Smog Chamber Experiment EC 116
                                           JL
                  60
                          120
    180
TIME (min)
240
300
360
  391 r
  293

[NO]

  196

(ppb)

    98
                                                           Legend
                                                        Experimental values
                                                        Computed values
.1
                                           J_
                  60
                           120
                                           180
                                      TIME (min)

                                         4-20
                                                   240
                             300
                                                                           360

-------
        Figure  4-2   Simulation of Smog Chamber Experiment  EC 116
                               (continued)
     ,406r
     .304-
[NOJ
     .203-
(ppm)
     .101 -
                     60
120         180
  TIME (min)
           300
      824 •-
      618
 [OLEF]
      412
 (ppb)
      206
       .5
                                                      Legend

                                                • Experimental  values
                                               —•Computed values
                                   I
                     60
120          180

   TIME (rain)
240
300
                                        4-21

-------
 4.oor
  3.78


[PA]


  3.56


(ppm)


  3.34
  3.12L
             Figure 4-2   Simulation of Smog Chamber  Experiment   EC  116
                                    (continued)
                                      Legend

                                 • Experimental values

                                 — Computed values
                 60
            120
            180

        TIME (min)
240
300
                                                                              360
  .299r
  .224

 [PAN]
(ppm)


   .074h
      0-
      0
                                            I
60
120
                                           180

                                       TIME (min)
                                     240
                                      300
                          360
                                        4-22

-------
  .463
             Figure  4-2   Simulation of Smog Chamber Experiment  EC 116
                                     (continued)
  .349


[HCHO]


  .234


(ppm)


  .119
                                Legend

                             Experimental values

                            1 Computed vlaues
  ,005
                  60
120
    180

TIME (min)
240
300
360
  ,550,-
                  60
120
    180

TIME  (min)
240
300
360
                                        4-23

-------
 ,274r
 .206

[KET]
 .137

(ppm)

 .068
            Figure 4-2  Simulation of Smog Chamber Experiment  EC 116
                                   (continued)
                                          J_
                                                	I
                                                 360
                 60
120
   180
TIME (min)
240
300
 3.96r
 2.97

[NTRA]

 1.98

(pphm)

   .99
                                 Legend
                            • Experimental values
                            — Computed values
                                            I
                  60
120
             180
         TIME  (min)
                                                       240
                            300
                                                 360
                                         4-24

-------
   .244
   .183-
   .122
(ppra)


   .061
              Figure 4-3  Simulation of Smog Chamber Experiment  GC 119
                                            _L
                  60
120
    180
TIME (min)
240
300
                                                                                 360
   .301,-
                                                          Legend
                                                       Experimental  values
                                                       Computed values
   .0121
                               120
            180
        TIME (min)
                                                        240
                            300
                                                                                360
                                         4-25

-------
  .245 r
              Figure 4-3  Simulation of Smog Chamber Experiment  GC 119
                                     (continued)
  .194 -
[N02]
  .143
(ppm)

  .092
                                    Legend
                               • Experimental values
                               ""•'Computed values
  .041
                  60
120
    180
TIME (min)
 240
300
360
  1.80 i-
  1.35  -

 [PAN]

   .90
 (pphm'

   .45
                                            _L
                  60
 120
    180
TIME  (min)
240
300
360
                                         4-26

-------
             Figure 4-3   Simulation  of  Smog Chamber Experiment   GC  119
                                     (continued)
 .131 r
  .107

[HCHO]

  .083

(ppm)

  .059
  .035
                                            JL
                  60
120
    180
TIME (min)
                                                        240
             300
                                                  360
  .143r
  .113

 [RCHO]
                                   Legend
                              • Experimental values
                              —• Computed values
  .083

 (ppm)

  .053
   .023
                                            I
                   60
 120
     180
 TIME (min)
240
300
                                                                                 360
                                         4-27

-------
        Figure 4-4  Simulation of Smog Chamber Experiment  GC 133
     326r
     .244
     .163
Cppm)
     .082
                    90
                                  I
              1
180          270
  TIME (min)
                                                              •    o   •
360
450
     8.40r
     6.35 -
[NO]
     4.29 -
(pphm)
     2.24 -
                   Legend
               Experimental values
               Computed values
                                 180          270
                                    TIME (min)
                         360
             450
                                      4-28

-------
        Figure 4-4  Simulation of Smog Chamber Experiment  GC 133
                               (continued)
    6.96r-
    1.30
                   90
180         270
  TIME (min)
360
450
    1.5 Or
    1.12
[PAN]
      .75
Cpphm)
      .38
                     90
                                 _L
                   Legend
               Experimental values
               •Computed values
              _L
 180          270
  TIME  (min)
                                                          360
             450
                                    4-29

-------
        Figure 4-4  Simulation of Smog Chamber Experiment GC 133
                               (continued)
   11.30r
    8.60

[HCHO]


    5.90

Cpphm)


    3.20
      .50
                     80
                                  Legend

                             • Experimental values
                            — Computed values
160
                                    TIME
                         240
                                                          320
                                     400
    10.04t-
     7.81 -
[RCHO]
     5.57
(pphm)
     3.34
      1.10
          0
90
             _L
             _L
                                  180         270

                                    TIME (min)
                                     360
                                      450
                                     4-30

-------
       Figure 4-5  Simulation of Smog Chamber Experiment GC 135
    ,297 r-
    .223
    .149

(ppm)

    .074
   3.00r-
    2.25

 [NO]

    1.15

 (pphm)

     .77
                                              I
                   90
                   180         270
                    TIME  (min)
360
450
                                       Legend
                                  • Experimental values
                                  —•Computed values
•  «
    • •  • •
                •   •
                           • • •
                   90
                  180         270
                    TIME  (min)
                                                         360
            450
                                    4-31

-------
       Figure 4-5  Simulation of Smog Chamber Experiment GC  135
                              (continued)
   2.66r-
   2.15

[N02]

   1.63

Cpphm)

   1.12
     .61
                                 _L
                         _L
                    90
            180          270
               TIME (min)
                         320
           450
   12.73r-
    9.54

[PAN]

    6.36

(PPb)

    3.18
        0
                                 _L
90
                          _L
                                      Legend
                                •  Experimental values
                                — Computed values
180          270
   TIME (min)
360
450
                                     4-32

-------
        Figure 4-5  Simulation of Smog Chamber Experiment  GC  135
                                (continued)
     .149r-
     .112

 [HCHO]

     .075

 (ppm)

     .393
     .003
                                   I
                     90
180
  TIME
270
360
450
    9.16
    6.94

[RCHO]

    4.73

(pphm)

    2.51
     .30
        Legend
   • Experimental  values
  — Computed values
                                  _L
JL
                     90
                                 180          270
                                   TIME  (min)
            360
                                     450
                                     4-33

-------
              Figure 4-6  Simulation of Smog Chamber Experiment GC 138
  3.56-
  2.67-
  1.78-

(pphm)

   .89
  .464r
  .368

 [NO]

  .271

 (ppm)

  .174
  ,078L
      0
                  60
            120
             180
        TIME (min)
                240
             300
            360
                                        Legend
                                  •  Experimental  values
                                  —.Computed values
                                            j_
60
120
    180
TIME (min)
240
300
360
                                         4-34

-------
 .305r
 .22S-
[N02]
 .154


(ppm)


 .073
  .004
     0
  2.00-
             Figure 4-6  Simulation of Smog Chamber Experiment GC 138
                                    (continued)
                                         Legend

                                    • Experimental values

                                   —"• Computed values
                                           _L
  1.50-

  [PAN;

  i.oo

  (ppb;

   .50
60
120
   180

TIME (min)
                                                       240
                                                  300
                                                 360
                 60
             120         180

                      TIME (min)
                                                       240
                                      300
                                         360
                                        4-35

-------
 10.48
  9.33
             Figure 4-6  Simulation of Smog Chamber Experiment GC 138
                                     (continued)
[HCHO]

  8.19


(pphm)


  7.04
                               Legend

                            No  experimental  values

                            Computed values
  5.90U
                  60
120
   180

TIME (min)
240
                                      300
            360
  9.46r
                   60
 120
    180
TIME (min)
 240
300
360
                                         4-36

-------
           Figure 4-7  Simulation of Smog Chamber Experiment GC 150
 16.60r
 12.45
   .83

(pphm)

   .42
                                            I
                  90
180         270
   TIME (min)
360
                                                                    450
  .440r
[NO]
                                                      Legend
                                                •  Experimental  values
                                                — Computed values
  .022
                             180          270
                                TIME (min)

                                      4-37
                        360
            450

-------
            Figure 4-7  Simulation of Smog Chamber Experiment GC 150
                                   (continued)
  ,338r
  .272
[N02]
  .206

(ppm)

  .140
                                          Legend
                                   •   Experimental  values
                                  — Computed  values
  .074
                                I
                  90
            180          270
              TIME (rain)
                        360
             450
  5.44r
  4.08

[PAN]

  2.72

(ppb)

  1.36
                                I
                          I
      0
90
180         270
  TIME (min)
360
450
                                        4-38

-------
  .191 r
  .177


[HCHO]

  ,162


(ppm)


  .148
            Figure 4-7  Simulation of Smog Chamber. Experiment  GC  150
                                   (continued)
   .134
                                _L
             J_
                   90
180          270
  TIME (min)
                                                         360
                                      450
  12.68r
   9.61

 [RCHO]


   6.53


 (pphm)


   3.46
                              Legend

                             Experimental  values

                            1 Computed  values
     .38
                                             _L
                   90
 180         270

   TIME (min)

           4-39
360
450

-------
             Figure 4-8   Simulation of Smog Chamber Experiment GC 156
   .419r
   .314
   .209

(ppm)

   .105
                   90
180          270
  TIME (min)
360
450
  17.40r
  13.14

 [NO]

   8.89

 (pphm)

   4.63
                          Legend
                    •  Experimental values
                   — Computed values
     .37
                                    Jt	ft.
                                 _L
              JL
                    90
  180          270
         TIME  (min)

          4-40
 360
 450

-------
           Figure 4-8  Simulation of Smog Chamber Experiment GC 156
                                  (continued)
 15.1ST
 12.16-
[N02]
   9.r
(pphm)

   6.19-
   3.20
      0
                               _L
90
180          270
   TIME (min)
360
                                                  450
  1.60r
  1.20

 [PAN]

    .80

 (pphm)

    .40
                                             I
                                                           Legend
                                                         Experimental values
                                                         •Computed values
                  90
            180          270
               TIME  (min)
                         360
             450
                                       4-41

-------
       Figure 4-8   Simulation of  Smog  Chamber  Experiment  GC  156
  8.41r
  1.00
                   90
                                                      Legend
                                               •  Experimental values
                                               ••"* Computed values
180          270
   TIME (min)
360
450
  11.86
   8.98

[RCHO]

   6.10

Cpphm)

   3.22
    .03
                                             I
                   90
180          270
   TIME (min")
 360
                                                                     450
                                    4-42

-------
                                TABLE 4-4
              SURROGATE HYDROCARBON SMOG CHAMBER RUNS USED
                   FOR MODEL TESTING AND DEVELOPMENT*
Run (UCR#)
119
133
135
138
150
156
NO
.301
.084
.030
.464
.440
.174
N02
.041
.013
.007
.050
.074
.032
NO Tot. HC
X
.340
.097
.037
.523
.523
.206
.499
.318
.305
.482
.501
.471
CO
7.45
4.05
3.84
7.9
8.6
6.8
HCHO
.038
.005
.027
.060
.134
.10
RCHO HC/NO
.A.
.023
.011
.003
.003
.004
.003
1.5
3.3
8.2
0.92
0.96
2.3
0 Max
.24
.33
.22
.03
.17
.42
*A11 concentrations in ppm
                                TABLE 4-5
           RATE CONSTANTS* USED FOR HYDROCARBONS (ppm   min  )
Compound
OH
°3
0(3P)
 Ethane
 Propane
 n-Butane
 2,3,-Dimethylbutane
 Ethene
 Propene
 Cis-2-Butene
 2-Methyl-2-Butene
 Toluene
 m-Xylene
4.2x10
3.2x10-'
4.4x10-
7.6x10-
3.7x10
1.2x10'
8.8x10-
3.5x10^
2.5x10
1.6x10
1.8x10
8.9x10
-3
-2
-1
-1
1.1x10*
5.2x10-
2.4x10^
9.3x10^
*Taken from Hampson and Garvin (1975); Pitts, et al  (1976); Anderson,  (1976)
                                   4-43

-------
and 1/2 for runs 150 and 156, respectively, to account for the reac-
tivity of the mixture in terms of time to NO- maximum etc.  Consequently,
it can be seen that the mechanism does as well as might be expected from
a lumped species mechanism although probably not as well as one with a
larger number of reactions to account for all the different species in-
volved.  In general, the mechanism tends to underpredict the final 0,
                                                                    O
slightly, but overpredicts it in the case of runs 135 and 138.
     It can be seen from Figures 4-4 and 4-5 (Runs 133 and 135, respec-
tively) that the computed NO concentration is significantly lower after
about 90 minutes, than the experimental values.  This is an experimental
artifact whereby, at the low NO  concentrations used for these particu-
lar experiments, there is a slow infusion of NO  from the room air which
                                               A.
maintains a fairly constant minimum NO  concentration.  The computed
results assume that there is no introduction of NO  during the irradia-
                                                  ^v
tion.  Consequently, the results of modeling these runs should be viewed
with the above observation in mind.
     In the majority of runs, the experimental values of N07 at the end
of the irradiation are less than the computed values.  This indicates
that the current mechanism is not fully accounting for all the NO  loss
                                                                 .A.
mechanisms.  This has been rectified in the atmospheric model by increas-
ing kir
     The predicted ozone initiation times are in reasonable agreement
with the experimental values and should not lead to significant problems
when the mechanism is included in the atmospheric model.  The 0_ initia-
                                                               *J
tion times are used as a guide in determining the initial HONO to be
used, and this value is chosen to give the best agreement between the
computed and experimental 0_ initiation times.
     As indicated earlier, the formulation of so-called aldehyde produc-
tion in the aromatic mechanism means that the computed PAN readings will
generally be high compared with the experimental data.  Thus, the scheme
reflects this expectation in that the computed values are uniformly
greater than the experimental measurements.
     The aromatic mechanism shown is subject to large uncertainty and  is
being revised as new experimental data become available.  A test of the
aromatic mechanism could be made if the aromatic portion  in the surro-
gate hydrocarbon data from the SAPRC smog chamber studies was varied
sufficiently to be able to test the importance of aromatic hydrocarbon
mechanisms in the overall hydrocarbon photooxidation and  in ozone

                                   4-44

-------
production.  The proposed reaction mechanism for aromatic photooxidation
will require modification as more data become available.
     Sensitivity of the results from this chemical mechanism to various
input parameters and a discussion of uncertainties effecting model re-
sults are presented elsewhere (Martinez, et al, 1977).

4.2  Formulation of a Chemical Mechanism for SCL Oxidation

     4.2.1  SO  Oxidation Chemistry

     A discussion of the atmospheric chemistry of S02 should include
both homogeneous and heterogeneous oxidation processes.  Homogeneous
processes are generally assumed to be gas phase homogeneous paths,
whereas, heterogeneous processes usually include those involving aero-
sols, surfaces and water vapor, although by definition, both categories
encompass a much broader spectrum of reactions.  A crucial question in
the atmospheric oxidation of SO- is the relative importance of homoge-
neous gas phase chemistry compared to the heterogeneous routes (Hidy and
Burton, 1975; Calvert, 1974; Wilson, 1976).  Heterogeneous reactions are
assumed to be most important in areas of high relative humidity and/or
aerosol loading and also in pollutant plumes, which may contain catalysts
such as iron, vanadium or manganese.  Homogeneous processes are generally
understood to be significant in areas of high photochemical activity
with elevated levels of ozone and other oxidants.
     The following discussion considers homogeneous gas phase processes
only.  For modeling purposes, we have assumed that any reaction forming
SO, will eventually form sulfate, SO..  The reactions are listed accord-
ingly.
     Sulfur dioxide does absorb ultraviolet radiation  (Calvert and Pitts,
1966), but unlike nitrogen dioxide, it does not dissociate to produce an
                            O
oxygen atom until about 2180A -- far shorter wavelengths than found in
the lower atmosphere.  However, the role played by excited SO,, ( EL),
formed in the atmosphere by absorption of the available radiation, is
insignificant, and Calvert, et al  (1978) estimate that the maximum rate
of oxidation of SO- by direct photo-absorption is less that 0.02% per
hour.  In view of the fact that atmospheric oxidation rates ranging
anywhere from .1% to > 10% per hour have been observed, it is apparent
                                   4-45

-------
that the direct photo-absorption reaction will be of very minor signifi-
cance.  Thus, other processes make the major contribution towards the
atmospheric oxidation of SCL.  These processes have been listed pre-
viously by other workers (e.g. Calvert, 1975; Wilson, 1976; Hidy and
Burton, 1975) and involve inorganic radical intermediates, such as
0 atoms, OH, HC^, and NCL radicals, and organic radical intermediates,
such as RO, RCL, and RCO_.   The importance and occurrence of these
species as contributors to S02 oxidation are discussed separately below.
     We have postulated a number of reactions, which on the basis of
current information we feel to be important in the conversion of SCL to
sulfate in the presence of hydrocarbons and NO :   these six reactions,
                                              A
shown in Table 4-6, have been added to the previous list of 46 reactions
used to model the hydrocarbon/NO  system.  The importance of various
                                X
reactions are discussed initially, while subsequent discussions in this
report describe the ability of this reaction scheme to reproduce the
results of selected smog chamber studies.

OH + S02 + M •» HS03 + M •*•* S0y|

     It is generally recognized (Calvert and McQuigg, 1975; Sander and
Seinfeld, 1976; Calvert, et al, 1978) that this is the most important
single gas phase reaction consuming S02 in the atmosphere.  The rate
constant for this reaction has been studied by many workers, but a reli-
able value applicable to atmospheric conditions is less easy to obtain
since the reaction is in the pressure dependent region for pressures
normally studied under laboratory conditions.  In addition, several
workers (Davis and Klauber 1975; Payne, et al, 1973; Cox, 1974, 1975;
Wood, et al, 1975; Castleman, et al, 1975; and Castleman and Tang, 1977)
studied this reaction in competition with the reaction

          OH + CO -* C02 + H

which is shown as reaction (9) in Table 4-2.  This reaction has been
shown to exhibit a pressure dependence at the low pressure often used in
laboratory studies (Sie, et al, 1976; Cox, et al, 1976; Chan, et al,
1977).  Consequently, the rate constants obtained for OH + S0~ from the
competitive rate studies must be increased by approximately a factor of
2 to bring them in line with the new measurements of k_.

                                    4-46

-------
     In addition to these competitive rate studies, several results have
been obtained using the flash photolysis-resonance fluorescence techni-
que.  These results include studies by Davis and Schiff  (1974), Harris
and Wayne (1975), and Atkinson, et al (1976).  The latter workers
obtained a high pressure limit of 8.3 x 10    cm  mol    sec   by extra-
polation of their data to the high pressure limit.  The  study of Gordon
and Mulac (1975) was carried out at one atmosphere total pressure and
435° K, and they obtained a value of 1.8 x 10"   cm  mol"  sec
     We have chosen the value of 1.2 x 10    cm  mol   sec   (using a
corrected value for k ) in our model calculations.  This is equivalent
            3    -1-1
to 1.76 x 10  ppm   min  .  This value is about the same as that obtain-
ed by Calvert, et al, (1978), by taking an average of the results of the
most comprehensive data sets at atmospheric pressure.
     For modeling purposes, we have assumed that this reaction will
eventually produce sulfate, although it is recognized that intermediary
steps are involved.  The specific reactions are shown in Table 4-6.

H00 + S00 -> OH + SOT -»-* SO,
	.£—'	£	J	4
     It is evident from this reaction that OH radicals are produced in
the oxidation of SCL by HCL.  The only reported determination of the
rate constant for this reaction is that by Payne, et al  (1973).  They
obtained a rate constant for this reaction relative to the reaction HCL
+ HO- -»• H20? + 0- and obtained a value of 8.7 x 10   cm  mol   sec
However, if the value for k (H0~ + H0~) recommended by Hampson and
                                          -15   3    -1    -1
Garvin (1975) is used, a value of 1.5 x 10    cm  mol    sec
(2.2 ppm   min  ) is obtained.  However, a recent preliminary result
obtained by Thrush, et al,  (Atkinson, 1978), suggests that the rate
constant is at least two orders of magnitude less than this.  We have
used a value of 0.03 ppm   min   for the rate constant in our model.
This preliminary value is likely to be modified slightly as more work is
carried out.
     It has been suggested that the product of this reaction could be
one of addition, giving HO-SO,.,.  However, there is little evidence for
this.  Thus, Calvert, et al (1978) estimate that this reaction would
make only a very minor contribution to the overall measured rate for H0~
                                  4-47

-------
0 + SCL + M -> SO, + M ->-*• SO,
	2.	3	4
     This reaction of ground state oxygen atoms with SCL is generally
regarded to be of little significance in the atmospheric oxidation  of
S02.  This is because most of the oxygen atoms produced from the photol-
ysis of ambient NCL react with oxygen molecules to produce ozone.
     The rate constant for this reaction has been measured by a number
of people and these determinations have been summarized and/or evaluated
by Schofield (1973), Hampson and Garvin (1975), and Westenberg and  DeHaas
(1975).  The recent experimental work of Westenberg and DeHaas indicates
that the earlier measurements were a factor of 2 too high.  The measure-
ments of Westenberg and DeHaas give a value of 4.9 x 10~   cm  mol"
sec~  for an atmosphere of nitrogen.  Calvert, et al (1978) using data
                                                                     -14
for nitrogen and oxygen as third bodies, obtained a value of 5.7 x  10
cm  mol   sec"  applicable to atmospheric conditions.  We have used the
latter value in our calculations.  Whichever value is used, the contri-
bution of this reaction to atmospheric SCL oxidation is negligible, al-
though in theory its importance in SCL oxidation in stack gas plumes may
be more significant.

N05 + S02 -» SCL + NO,,

     The oxidation of SCL by NCL radicals formed largely by the reaction
of NCL with 0,, has been invoked in some modeling studies, e.g. Durbin,
et al, (1975) as contributing in a major way to SCL oxidation; this
would be particularly true in the later stages of photooxidation of
hydrocarbon/ NO /SO  mixtures where the NO, concentration increases
               A   A                      *J
significantly.  However, studies by Daubendiek and Calvert  (1975) and
Davis and Klauber  (1975) find that the rate constant for this reaction
has a value of around 10    cm  mol   sec   .  Thus, this reaction is
unlikely to be of  any significance during the daytime oxidation of  N02,
although it is possible that if the NO, concentration increases signifi-
cantly during nighttime hours  (as a result of  the reaction of 0, with
                                                               O
NO-), then this reaction could assume more importance in the oxidation
of S0?.  This reaction has not been included in our model.
                                   4-48

-------
Reactions of Organic Radicals with SO,,

     The reactions which may be considered to be of importance for S0_
oxidation by organic radicals include the alkylperoxy radicals, acyl-
peroxy radicals and the Criegee intermediates formed in the reaction of
ozone with olefins.
     Previous modeling studies (Calvert, 1974; Calvert and McQuigg, 1975)
have shown that alkylperoxy radicals could play a significant role in
terms of SCL oxidation.  However, it is only recently that an estimate
has been obtained for the rate constant for the reaction
This has been obtained by Whitbeck, et al  (1976), who observed the kinet-
ics of the decay of methylperoxyl radicals spectroscopically following
its generation in the flash photolysis of a mixture of azomethane,
oxygen and SO-.  A rate constant of (5.3 - 2.5) x 10"   cm  mol   sec
was obtained for a combination of this reaction and the alternative
pathway



However, Calvert, et al (1978), estimate that the latter pathway will
make only a minor contribution to the overall rate constant.  Thus, this
rate constant is sufficiently large to make the reaction of alkylperoxy
radicals contribute about the same amount to SO- oxidation as that con-
tributed by HO- based on these calculations.
     The reaction of acylperoxy radicals with S02 has not been studied
directly and the contribution which it makes to SO- oxidation is subject
to some uncertainty.  However, it now appears that the reaction

          CH_CO_ + SO- -»• CH_CO-
            •J  
-------
mixture has a negligible effect upon the PAN yields  (Fox and Wright,
1977) .   If the reaction
               + S02 -> S03 + RC02

was important, then it would compete with the reaction RCO_ + NO- to
                                                          O     £*
produce PAN.  Thus, a change in PAN concentration should be noted upon
the introduction of SO  if indeed the reaction of RCO_ with SO- were
important.  Since this is not the case, the results give the same con-
clusion as that obtained by Pate, et al, and this reaction has not been
included in our simulations.

Reaction of SO,, with the Products of Ozone/Olefin Reactions

     It is well known that the direct reaction of S0_ with 0_ is slow.
However, Cox and Penkett (1971) found that the addition of an olefin to
a mixture of SO,, and ozone led to a rapid increase in the S0? oxidation
rate.  This has been interpreted as being due to the oxidation of SO,, by
products of the ozone/olefin reaction, such as the Criegee intermediate,
RCHO,,.  However, later studies (Calvert and McQuigg, 1975; Calvert, et
al , 1978) suggest that most of this increased oxidation is due to other
products of the ozone/olefin reaction, such as hydroxyl radicals and
hydroperoxyl radicals.  The contribution of the Criegee intermediate to
S09 oxidation is uncertain.  In this current model we have assigned a
                 3-1-1
value of 2.9 x 10  ppm   min   for the rate constant for SO™ oxidation
by the Criegee intermediate.  This value is a factor of 10 less than the
rate constant for RCHO- reacting with NO.  This chosen value may be
high, but was chosen on the basis of comparison with predicted and
experimental smog chamber data for sulfates.

     4.2.2  Validation of HC/NO /SO  System
                               X   X
     Table 4-6 shows the additional chemical reactions and rate constants
which have been utilized to predict the homogeneous gas phase conversion
of SO  to sulfates in the presence of hydrocarbon/NO  mixtures.  It was
     <-                                              X
assumed  that all the S0_ formed would eventually produce sulfates and
this supposition is apparently a reasonable one to make  (Calvert, et al,
1978) .   It  can be  seen that the  species chosen  for the oxidation process
                                   4-50

-------
                               TABLE  4-6
                ADDITIONAL  REACTIONS  USED  FOR HC/NO  /SO
                                                  X  X

                     HOMOGENEOUS  REACTION  MECHANISM


0 +
OH +
HO,
RCHO
R02
A02
so4
NOTE:



Reaction
S02 + M -> SO. + M
so2 + so4
+ S02 -»• OH + S04
2 + SO -»• S04 +• RCHO
+ SO- -> RO + S04
+ S02 -> AO + S04
+ Wall -»• Loss of S04
SO. is defined as being any ultimate sulfate
product in most cases will be S0_, which will
fate, SO .
Rate Constant
ppm~ min
8.40E+1
1.76E+03
3.0E-02
2.9E+03
8.0E+00
8.0E+00
3.12E-05
species. The initial
then convert to sul-

(1).    AO- is the product of the addition of OH to  an olefin in the  pre-

       sence of 0?.   (See above).
                                  4-51

-------
are 0, OH, HO , RO  (including AO  formed from the addition of OH to an
olefin in the presence of oxygen), as well as the Criegee intermediates.
The basis for the choice of rate constants has been detailed above.
     The data base initially used to test the reaction mechanism was
that obtained from the Battelle chamber simulations and supplied to us
by Dr. B. Dimitriades of the EPA.  These results were obtained in a 17.8
cubic meter smog chamber.  S02 concentrations were measured using a
Beckman 906 analyzer, while sulfate concentrations were obtained from
aerosol size distributions measured with a Thermosystems electrical
aerosol analyzer, using the assumption that equilibrium existed between
sulfuric acid aerosol in the condensed and vapor phases.  Independent
chemical measurements of sulfate apparently showed that this method was
very accurate.  The measured sulfate concentration-time history proved
to be very difficult to reproduce with the model in spite of numerous
changes in the mechanism and rate constants to try to obtain a good fit
with the sulfate experimental data.
     The initial conditions for the specific chamber runs chosen to test
the mechanism are shown in Table 4-7.  Unfortunately, at the time of
this model development exercise a paucity of reliable and comprehensive
smog chamber data for systems including S02 existed.  Thus, the task of
obtaining a reliable chemical mechanism, tested in detail against an
extensive smog chamber data base could only represent a preliminary
study.
     In order to check our chemical mechanism against the Battelle cham-
ber data, we chose Battelle run #114, which contained propylene in a
mixture of NO  in air.  Our complete chemical mechanism was modified by
             .A.
zeroing the rate constants associated with the reactions of the paraffins,
aromatics and SO • in addition, the rate constants for the reactions of
0, OH and 0, with olefins were modified from Table 4-5 to make them
           J
applicable to propylene.  Since the initial values of HCHO and RCHO were
not given in the experimental details, we chose values which appeared
reasonable based on the experimental data from the U.C. Riverside smog
chamber facility.  Of course, such a procedure is far from satisfactory,
but faced with a lack of experimental measurements, there appeared
little alternative.
                                  4-52

-------
                               TABLE 4-7
                    INITIAL CONDITIONS FOR BATTELLE
                       SMOG CHAMBER SIMULATIONS
                           Initial Concentrations of Reactants (ppm)
Battelle Run #             NO        N02        Propylene        S02

   S - 107               0.328     0.113           1.03        0.474
   S - 110               0.392     0.099           1.10        0.480
   S - 114               0.414     0.095           0.95        0.0
                                  4-53

-------
     4.2.3  Results and Discussion

     Reasonable agreement was obtained between the predicted and experi-
mental data as shown in a typical plot, Figure 4-9, although full equi-
librium value of HONO was required to produce sufficient initial radicals
to match the experimentally observed initial rate of NO oxidation.  Even
then, later in the run, the mixture appears more reactive than the com-
puted predictions, indicating that a continuous wall radical or molecule
flux may be a better approach to modeling these results than choosing a
high initial concentration of HONO.
     The effect of adding S09 to the propylene/NO  mixture in air was
                            Z                    X
modeled by including the SO,, reactions in the chemical mechanism and
testing the predictions against Battelle data, such as runs S-107 and
S-110.  Two plots, typical of the many obtained, are shown in Figures
4-10 and 4-11.  The major conclusion from these calculations is the fact
that it proved impossible to get a reasonable fit simultaneously for NO,
N02 and 03 in addition to that for SO  and SO .
     It is evident from the results shown in the figures that much work
remains to be carried out to elucidate the mechanism of homogeneous gas
phase sulfate formation.  Such work should encompass smog chamber
studies, as well as basic kinetic and mechanistic studies involving S0~.

     4.2.4  Conclusions of Current Modeling of Homogeneous
            SO  Oxidation

     While progress is being made in our understanding of the homogeneous
processes controlling the oxidation of S02 under atmospheric conditions,
many areas require further detailed study.  Thus, more information is
required in order to develop and test a reliable chemical mechanism to
describe the homogeneous transformation of S0~ to sulfate under ambient
conditions.  Among the areas for research are the following:

     •    The kinetics of the reaction of SO  with  intermediates, such
          as RCHO«, and A0_  (which is the product of the addition of
          OH to an olefin in the presence of 0?), are not known and
          require study.
                                    4-54

-------
     Figure 4-9  Simulation of Smog Chamber Experiment B-S-114
  ,384r
                  60
120         180
   TIME (rain)
            240
300
  .414 r
  .311

[NO]

  .208

Cppm)

  .104


  .001
                                                    Legend
                                                 Experimental values
                                                 Computed values
J_
_L
                  60
120          180
  TIME (min)
           240
300
                                 4-55

-------
    Figure 4-9  Simulation of Smog Chamber Experiment  B-S-114
                            (continued)
  .400r
  .314
[N02]
  .229


(ppm)


  .143
  ,057
                 60
120         180

   TIME (min)
240
300
  .950T
  .047
                  60
                                                     Legend

                                                 • Experimental values
                                                 — Computed values
 120          180

   TIME  (min)
 240
300
                                  4-56

-------
    Figure 4-10  Simulation of Smog Chamber Experiment B-S-110
  .600r
                 60
120         180
   TIME (min)
240
300
  .600 r
  .450

[NO]

  .300

Cppm)

  .150
                     Legend
                 • Experimental values
                 —• Computed values
                 60
120          180
   TIME (min)
240
300
                                 4-57

-------
     Figure 4-10  Simulation of Smog  Chamber Experiment  B-S-110
                              (continued)
   .750r
   .614-
[so2]
   .478

(ppm)

   .343
   .20'
                   60
120         180
  TIME (min)
240
                                                                    300
  12.00
   9.00
   6.00
 Cpphm)

   3.00
                   Legend
             •  Experimental values
            —-• Computed values
                                             I
                   60
 120          180
   TIME  (min)
 240
                                                                    300
                                   4-58

-------
       Figure 4-10  Simulation of Smog Chamber Experiment B-S-110
                               (continued)
    ,600r
    .453-
[N02]
    .306-
(ppm)
    .160-
    .013
                    60
120         180
  TIME (rain)
240
300
   1.500
                                                       Legend
                                                 •  Experimental  values
                                                 — Computed values
    ,017
                    60
120         180
   TIME (min)
240
                                                                     300
                                    4-59

-------
      Figure 4-11   Simulation of  Smog  Chamber Experiment  B-S-107
   .750r
   .604
[so2]
   .458
(ppm)


   .311
   ,164
    150
   .113
    .075

 (ppm)

    .038
                                                      Legend
                                                   Experimental  values
                                                   Computed values
                   40
80          120
  TIME (min)

                    40
 80           120
  TIME  (min)
160
                                                                     200
 160
200
                                    4-60

-------
product data for systems containing mixtures of hydrocarbons,
nitrogen oxides and SCL in air are practically nonexistent.  A
detailed search should be made for organic sulfur compounds
and organic-sulfur-nitrogen compounds.
the smog chamber data for the HC/NO /SCL system are much less
                                   A.   L,
extensive than those available for the HC/NO  system.  Thus,
                                            .A
more smog chamber studies are required which contain a variety
of mixed hydrocarbons, as well as single hydrocarbons in the
presence of NO  and S0_.  Only when the results of such studies
become available, can the proposed mechanism for the homo-
geneous oxidation of S0_ be rigorously tested.
until the type of data outlined above become available,
studies using the best currently available data, together
with sound estimates of unknown parameters, should continue
to be used to advance our knowledge of the behavior of S0?
and sulfates in HC/NO  mixtures in air.
                          4-61

-------
                 5.   ADAPTATION OF THE MODEL TO THE ST.
                      LOUIS REGION AND RAPS DATA BASE

     The adaptation of the trajectory model to the St. Louis region in-
volved not only development of software to interface the programs with
the RAPS data base, but also development of procedures to account for
the unique characteristics of the region and the data base.
     The RAPS meteorological data base includes surface data from the
Regional Air Monitoring System (RAMS) stations and data for higher ele-
vations from the Upper Air Sounding Network (UASN) stations.  The loca-
tions of these measurement stations are shown in Figure 5-1.  The RAMS
meteorological data used in the modeling includes surface wind speed,
wind direction, temperature, near surface vertical temperature gradient,
and solar radiation data.  Retrieval software was developed to directly
interface the RAMS data archive (on magnetic tape) with the meteoro-
logical module of the code.  The data requirements from the UASN data
archive were vertical temperature profiles (0-3000 meter elevations).
Automated interfaces were not developed for the UASN data retrieval
since the data requirements were quite small.
     In adapting the model to the RAMS wind data, procedures were
employed to account for differences in wind measurement heights above
the surface.  Since most of the instruments were positioned on 30-meter
towers, the wind speed data from stations with 10-meter towers were
adjusted to equivalent 30-meter wind speeds.  The formulation of this
adjustment is based on a stability-dependent power law wind profile
(Sellers, 1965), as shown below:
                <.33 , stable
                .18 , neutral
                .14 , unstable
                                   5-1

-------
                                                         to
                                                         c
                                                         o
                                                         rt
                                                         o
                                                         o
                                                        c
                                                        o
                                                        td
                                                        ^
                                                        to
                                                         C
                                                         0

                                                         0)
                                                        CD
                                                        a,
                                                         i
                                                        LO


                                                         0)
                                                         ^H


                                                         bO
                                                        •H
5-2

-------
     Each hour's stability was classified by computing the hour's re
gional average temperature gradient and using the following criteria:
                        >  -.005 °C/m        stable
          -.005  >      >  -.015 °C/m        neutral
                 —  dz  —

                    ~  <  -.015 °C/m        unstable
                    dz  —

The net result of this correction is the multiplication of 10-meter wind
speed by 1.437, 1.219, and 1.166 for stable, neutral, and unstable
conditions, respectively.  These adjustments were made for the wind data
from RAMS Stations 108, 110, 114, 115, 116, 117, 118, and 121.  No
adjustments were made to the corresponding wind direction data since the
available methods were not believed to be reliable.
     Another feature incorporated by the model accounts for differences
in the surface roughness length for the urban and rural portions of the
modeling region.  Since the urban region is approximately circular, a
surface roughness island is simulated in the model.  The urban region
within five kilometers of the island center (identified by UTM x and y
coordinates of 739.5 and 4280.5 km) is assumed to have a surface rough-
ness length of one meter.  The rural region located beyond ten kilometers
from the center is assumed to have a roughness length of 0.25 meters.
In the suburban region between five and ten kilometers from the center,
the surface roughness is assumed to vary linearly with distance between
1.0 and 0.25 meters.
     In adapting the model to the RAMS data base, procedures were incor-
porated to compute and interpret the ultraviolet radiation data.  The
ultraviolet radiation for each hour was calculated by subtracting the
measured solar radiation for wavelengths greater than 3950 Angstroms
from the total solar radiation data.  The regional average ultraviolet
radiation was computed from data collected at six stations.  This
regional average of observed radiation was compared with the expected
clear-sky ultraviolet radiation function shown in Figure 5-2.  The ratio
of the observed to the expected ultraviolet radiation was computed and
is referred to as the sky clearness ratio in this study.  The clearness
                                    5-3

-------
           Solar

        Ultraviolet
         Radiation*

       (Langleys/min)
                       .10
                       .08
.06
                       .04
                       .02
                          0   10   20   30  40   50   60   70  80    90

                                    Solar Zenith Angle (°)
*For wavelengths less than 3950 Angstroms and near
 sea-level elevations.

             Figure 5-2.  Clear-Sky Ultraviolet Radiation
                                   5-4

-------
ratio is used in the model to adjust the clear-sky ultraviolet-dependent
photodissociation rates to the appropriate levels.  These computational
procedures were fully automated for this study.
     The clear-sky radiation function applies to solar radiation with
                                     o
wavelength less than 3950 Angstroms (A) and for near sea-level elevations.
It was developed for this study based on both observed and computed
clear-sky solar radiation for St. Louis reported by Bergstrom and
Peterson (1977).   Since their data only included radiation below 3800
o
A, an upward adjustment of 31 percent was incorporated into the UV
                                    o
function for wavelengths below 3950 A.  This adjustment was based on the
                                           o
ratio of solar actinic fluxes through 3950 A to actinic fluxes through
     o
3800 A computed by Peterson (1976) for solar zenith angles between 20
and 60 degrees.
     Terrain maps for the St. Louis region were examined, and it was
concluded that the topographic features within the region were too small
to present significant barriers to the wind fields.  Thus, the air
trajectory generation submodule of the code was set up without geographic
barriers normally used to constrain air parcel trajectory paths.
     Program modifications were incorporated (as optional features) to
accommodate the variable size grid square characteristic of the RAPS
emissions grid.  The RAPS area source data are compiled  (by EPA) for
grid squares ranging in size from 1 x 1 to 10 x 10 kilometers, and each
square is referenced by a single identifier.  The program was modified
to determine air trajectory grid square crossing times on this grid and
to retrieve corresponding emissions data by the single identifiers.
     The program was adapted to a 100 x 100 kilometer emissions grid for
the St. Louis region.  The grid is positioned with its southwest corner
(origin) located at UTM x and y coordinates of 680 and 4,230 kilometers,
respectively. The grid, referred to as the ERT grid in this study, is a
subregion of the RAPS emissions grid.  The ERT grid encompasses 1,664 of
the 1,988 RAPS grid squares and all of the RAMS Stations.  Emissions
from area sources outside of the ERT grid are quite small, and are not
considered important for simulation of St. Louis air quality.  Point
source emissions outside of the ERT grid may be more important.  For
this reason, there are no geographic boundaries imposed on the use of
the RAPS point source emissions data by the model.
                                    5-5

-------
     In summary, there were no major problems in adapting  the model  to
the St. Louis region.   The wealth of data provided by the  RAPS program
and the forethought given to the design of the data base made for  a
straightforward adaptation of the model to the region.
                                   5-6

-------
            6.  METHODS AND RESULTS FOR TEST DAY SIMULATIONS

     This section describes the methods and model results for the simu-
lations of air quality in the St. Louis region on June 29, July 13, and
July 14, 1976.  Air parcel trajectories, meteorological conditions, air
quality data, initial species concentrations, and predicted concentra-
tions for these days are described and interpreted.

6.1  Air Parcel Trajectories

     Hourly surface wind measurements from the RAMS Stations were used
to generate receptor-oriented air parcel trajectories for each of the
test day simulations.  Hourly ozone data for these stations were exam-
ined to isolate the hours and stations with high measured ozone concen-
trations.  Backward trajectory space-time histories were generated from
several stations at several hours to isolate trajectories suitable for
modeling.  One trajectory for each day was selected based on the crite-
ria that:  (1) the receptor recorded relatively high ozone concentra-
tions;  (2) the trajectory start  location was within the modeling region;
and (3) the trajectory start time was near sunrise.  In addition, it was
considered desirable for the trajectory to have a rural start location,
where pollutant concentrations are generally low in the early morning
hours.
     Following selection of a single backward trajectory for each day,
forward trajectories were then generated using the near sunrise start
locations.  The reason for running forward trajectories is twofold.
First,  due to inherent nonlinearities in the method of computing back-
ward trajectories, trajectories  generated in the forward direction are
believed to be more reliable representations of the wind field.  Second,
the forward trajectories can be  extended beyond the selected receptor
location in space and time.  Extension of the trajectory duration per-
mits simulation of concentration histories for an entire day, not just
the portion which precedes the receptor.
     The trajectory selected for June 29, 1976, is illustrated in
Figure  6-1.  This trajectory starts at 6:00 A.M. southwest of St. Louis,
and is  oriented to pass RAMS Station 116 at 1:00 P.M.  The parcel is
advected at moderate wind speeds (4-6 meters/second) across southern St.
St. Louis, and passes quite near RAMS Stations 119, 111, and 105 between
                                   6-1

-------
Location of RAMS stations
and UASN network sites
                                                                Bethalto
                                                               144
                                                             Wood River
            •  RAMS Stations (25)
           A  Central Station No. 101
            o  UASN Sites (4)
                              Crystal City**
                Figure 6-1   Air Parcel  Trajectory For 6-29-76
                                          6-2

-------
11:00 and 12:00 o'clock.  After passing RAMS Station 116 at 1:00 P.M.,
the parcel continues moving in a northeasterly direction, which eventu-
ally carries it out of the modeling region.
     The trajectory selected for July 13, 1976, is illustrated in Figure
6-2.  The trajectory has a 6:00 A.M. start east of St. Louis, and moves
in a northwesterly direction until 1:00 P.M., where it turns north-
northeast to arrive at RAMS Station 122 at 4:00 P.M.  The air parcel is
advected by relatively light winds (1 to 3 meters/second) throughout the
day.
     The trajectory selected for July 14, 1976, is illustrated in
Figure 6-3.  The air parcel space-time tract begins west of St. Louis
at 6:00 A.M. and moves in a westerly direction throughout the day.  It
arrives at RAMS Station 117 at 1:00 P.M. and remains in the modeling
region until 4:00 P.M.  The parcel is advected by wind speeds between 3
and 4 meters/second throughout the day.

6.2  Meteorological Conditions

     The meteorological data provided by the RAMS and UASN measurements
were processed to establish meteorological conditions along the three
test-day trajectories.  These measurements are summarized in the follow-
ing tables and figures.  The hourly surface temperatures, wind speeds,
mixing heights, atmospheric stability, ultraviolet radiation, and sky
clearness are presented in Tables 6-1, 6-2 and 6-3, respectively.  Ver-
tical temperature profiles, obtained from radiosondes released at approx-
imately 0400, 1000, and 1600 hours each day, are shown in Figures 6-4,
6-5 and 6-6.  The surface temperature, wind speed, and vertical tempera-
ture data are used to compute the schedules of mixing coefficient pro-
files shown in Tables 6-4, 6-5 and 6-6.
     Overcast skies characterized the morning hours of June 29, 1976.
Atmospheric conditions were relatively unstable with 4 to 6 meters/second
wind speeds.  A sharp decrease in surface temperature between 1100 and
1200 resulted in stable conditions for that hour.  Surface temperature
reached a maximum of 29.7° C at 1600 hours.  Mixing depth increased from
the 200 meter elevation at sunrise to a maximum of 1080 meters at 1600
hours.
                                   6-3

-------
                \~\~"
Location of RAMS stations
and UASN network sites
                  ))
                                                          Wood River
                                                             Edwardsville
                Figure 6-2    Air Parcel  Trajectory  For 7-13-76
                                        6-4

-------
Location of RAMS stations
and UASN network sites
                                            Jerseyville
                             St. Charles
                                                    • 122
                                                          Alton
                                                                     Bethalto
                                                          1P^i^144
                                                            \§vj VWood River
                                                                      Edwardsville j
               • 125
                                          7C
                                          _>
                                          *m
                                        ^ RAPS
                                        «120st.LoUls>^ms^
                                                  107»  \\»103
                                             112*  !P1£  ..m^.109
                                                                   it. Louis
                                             106^
•  RAMS Stations (25)
A  Central Station No. 101
o  UASN Sites (4)
                    Crystal City"
                                                          110
                                                        • 118
                                                      Columbia
                                                         Waterloo
                                                            • 124
 ggf Collinsville

        14
        HT


   ^ Belleville


0143
                                                                       • 123

                                                                           15
                                                                                                16
                     Figure  6-3   Air Parcel Trajectory For 7-14-76
                                                 6-5

-------

































rH
1
vO

PJ

oa

£_l

















































vO
f1**
^s^
o*
s s^

(/) •(— | V«— /
O """I
6 •'"'
ij r^
< cS
P




CO r~\
bo in tn
C -P (H

'>< "Sp
•H -H 0)
S 
oj o^ to ^j ^j ^o r*** oo ^i^ ^^ ^^ ^»^
i-H i— 1




rHtor-.ootof>orHLr>ocT>cn
trjCNrJ-Tj-Tl-rl-vCiLn^vDvOvO







Lo^fTTLnt^LOLovOTt-tot^-o
rHf>aT3-\ot-voooo>oiC7iOicn
CNCNJCNCNCNCNCMCNCNCNCNCN






rHi-HrHi— IrHrHrHi-H





\D
tn rH
c
O tn
•H d)
p in
aj tn
p rt
CO rH C
CJ 0

^ «J f.1
^L H TO
rC 'l3
p S aj
tn tn rH
in I-H o p
O 3 -H 0)
rH T3 P rH
O O CJ O
g P -H
tl> ,0 CO >
0 3 C3
fn in CO rH
l < g X
P fi -H O ,*
O w tn rH tn
g -H 3 s  4J
4->  CD rH W)iH nJ
frt S-i H c* j i
TO rl i-* r-i ^
•P tO ^Cj tO 'CJ
4J 0 Oj d> d)
O 4) r<*^ 0^ bO ^
P rC g P TO 3
p o o rH tn
-a fH X P 0 rt
0 P 14H rO > 0
P cd TJ co g
_ 1 Tf4 fl\ fl\ ^\ J Lt_]
f^ W >W W W ^1 MH
O CD 4^ ^^ p j rt Q
fn 3 g -H 0 O O
0 10 -H "rj ^ .H -H
+J CO P 0 rH bO +J
C 0 tn rH O 0 CO
r-l S W (X G C£ Oi
rH CM CO J- IT) 
-------






























CM

\o

3
CQ
^*
H















































vQ

•^^
to
rH
••N^
Is*
&
O
PH

CO
55
0
rH
H
rH
Q
2
O
u

CJ
!,-_-! j

o
i-3
o
CO
o
r»i
H

§
















10
to
CO rH
rt
0
rH
CJ

PLO
0) C
rH 0
O -H
•H P
> rt
rt -H
£•1 *1O
4-> rt
rH Oi




J-

(A
o rt
•H rH
rH CJ f~*
0 VO
"ft P rH
tfl -H <— '
O rH
B -H

^C CO
4->



CO f—^
IO t/1
00 4-> rH
.5 -as
X -H 0>

2 32 '— -"


CM
TJ <-^
"d O (A
CM Ci *****
co*£




1-1
o
fH
cj>cjioO'joo
rH rH





00 to *O CT^ to \O vO LO *sO LO CO F*"
tOOOOrH'StOlCMOLOLOrH
rHtOLOt^OOOOOOOOt^LOtOrH
oooooooooooo


























oooooooooooo
•>5tOt~-1>5tOt^OLOOOCM\OO
rHCMCMtO'-trtvOt^OOOrHtO
rH rH rH





LO\OrHO^Cft^CTiVOOVrHr^
CMtOtOtOCMrHrHOrHrHCMrH







LOtOOOvOCNOOLOOOTtOOOit^
OCMtOLOt^OOOrHCMCMCMCM
CMCMCMCMCNCMtOtOtOtOtOtO





\O[--OOO»OrHCMtO^t-LOVOt--
rHrHrHrHrHrHrHrH






C
O f)
•H 0>
4-> M
rt 10
4-> Cti
CO rH Pi
O O
•§ 0 +J
^r $»| (d
K O -H
rC T3
4-> S rt
(/) tO rH
0 0 C
W rH O 4->
O 3 -H 0
rH "T3 P rH
O O rt O
S P -H
0 ,£! CO >
rH VI 1 CO rH
_r^ «g^ 2? 4J
P X < rH
•P to cc 3
0 -H 0
rC > W B X
P fi -H to 0 &
O tO rt rH I/I
B -H 3 rH -4H
O P  -H O C O
0 P 2 X-H-: -H
T3 rt co "d -4n B "O
O > s 0 P tO rH 0 B
rH tO •• rH rH -r!
O rH 0 B fH ^ P
P 0 to rt 3 C to
O P O rH O" rt 0
0 0 rH bO 10 1-3
C? fH Oi (S P
rH 0 0 PH -H
P to ,js o "d
4J 0 P 0 0
O0 rC 60 rH
P X E 4-> Tt rt 3
4J O C fH W
•d fn x o 0 rt
0 P ^4H JD PJ > 0
4-> rt to rt B
rH T3 0 0 fH rH M-l
O 0 4-> P fH rt O
PH rH rt OOP!
fH 3 B -H 0 0 0
0 tfl -H "d «H «H
P rt 4J 0 VO 60 P
C 0 to fH 1 0 rt
rH S W G, rH Oi C£
1-1 CM oo j- in ID



6-7

-------

























to

vO
M

3
H












































VO
^
^^
iH

f^

05
o


CO
2
0
i— i
H
t-H
^^
o


53
u
i— i

o

o

o
OJ
r .
C™1
§


















(A
^* C
 rt
CO »iH
tj *pH
H^ ^^
3


in
in
u rt
•H i-l
fH CJ <— v
fl> vO
"ft t? rH
in -H ^

s *^
•P X>
< rt

CO

CO ^^
in in
004J r<
C X 
X -H 
•H a> 6
S X^-



 ^,
S PHV— '
CO



r-t
(D
fH
O J_) C 1
rt rt o

t, A\
^ o.
CO S
o
H


f-i
3
O
33

to«£S££o££££;£
i-H




\O ^1" f^ 'Sf \O t^*" r** 00 rH LO CO ^^
tO 0s! ^D O^ LO 00 tO O^ 0s! 0s! \O O^
rH tO LO \O 00 00 O^ t***» vO LO tO ^H
oooooooooooo






















oooooooooooo
lOLO^OOOO^SocQLoSS
i— li— li-HCN(SC
in rH O -P
O 3 -H O
rH 13 .(-> i— 1
o o rt o
o> j! w >
Q) 3 U- rt
fn in i co 5-1
t(~l ^ 12* 4J
•P X ^ rH
+J in a 3

,C > in e X
4-> C -H in O ^
O in rt ?H in
6 -H 3 rH MH
O 4-> <4-( O f-i
JH rt <+-i O rt
MH C ^ -H T) -P D
O CO 13 fn 3 I-H
in -H o C o
^3 rt ^^ ^3 ^H S n3
O > CO 13 T-t — 0)
C o 3 i) u in 4J
rH 1 >-, rt
X 0> -P W rH  e -H bo 4->
•P (U tn rt 3 C in
O +J O rH CT' rt 0
rt) fli r — 1 (JA (A ^J
•r-> g 0 O rt O
rt ^ D- C +J
JH O 0) P, -H
P to !^ O ""O
+J ^ O O rt
O ] * I4H o o . ^ fls
+j rt in rt g

rH T)  P +-> f-t rt o
& rH rt O O C
r7 3 g -H O O O




6-8

-------
           Figure 6-4  Vertical Temperature Profiles  (6-29-76)
           1400
           1200
           1000
Height Above
    Mean
 Sea Level
    O)
800
            600
            400
            200
                  UASN 142
                                                       /Release\
                                                       \  Time /
                                     UASN 141  UASN  141   /Release\
                                                          (station/
               16   18   20   22    24   26    28   30    32    34
                            Temperature  ( C)
                                    6-9

-------
           Figure 6-5  Vertical Temperature Profiles   (7-13-76)
         1600 r
         1400 -
         1200 '
 u •  u*.
 Height
 Above
Mean Sea
 Level    80°
  (m)

          600
          400
          200
                      UASN  143
 1634.    /Release\
          \ Time  /
         UASN 144 /Release\
UASN 141          (station/
             14    16    18    20   22    24    26   28    30   32    34
                                Temperature (  C)
                                     6-10

-------
            Figure 6-6  Vertical Temperature Profiles  (7-14-76)
Height

Above
Mean
2800



2600



2400



2200



2000



1800



1600



1400
Sea

Level  1200


(m)
       1000



        800



        600



        400



        200
                  (Release  Time)     353.
               (Release Station)    UASN  142
                                                    UASN 141  UASN 144
           12   14  16    18    20  22    24    26    28   30    32    34    36   38


                                  Temperature  (  C)
                                        6-11

-------
                                TABLE 6-4


           VERTICAL EDDY DIFFUSIVITY COEFFICIENTS* FOR 6/29/76



                                Heights (meters)
Hour
6
7
8
9
10
11
12
13
14
15
16
17
0-120 120-300 300-600 600-1200
9 236 327 126
41 166 140 27
487 389 842 507
1,486 550 867 527
1,782 6,202 9,845 1,767
137 738 1,495 363
3,099 11,193 18,901 6,756
2,136 7,654 13,583 10,764
1,701 7,383 13,520 10,833
1,403 5,940 13,019 10,898
104 5,352 7,509 2,520
104 5,352 7,509 2,520
                                                             2
* Average K  coefficient for the height range shown in meters /minute.
                                    6-12

-------
                               TABLE 6-5


          VERTICAL EDDY DIFFUSIVITY COEFFICIENTS* FOR 7/13/76




                                Height (meters)

Hour               0-140      140-350      350-700      700-1400
6
7
8
9
10
11
12
13
14
15
16
17
26
858
1,818
1,905
1,758
1,306
1,209
624
1,114
1,325
1,494
286
6
9
787
2,829
4,387
4,037
3,824
1,988
3,553
4,230
4,772
947
6
6
6
6
453
6,561
6,613
3,491
6,262
7,471
8,434
285
6
75
12
8
6
3,301
7,482
4,233
7,709
9,271
10,492
6
                                                             2
* Average K  coefficient for the height range shown in meters /minute.
                                   6-13

-------
                                TABLE 6-6


          VERTICAL EDDY DIFFUSIVITY COEFFICIENTS* FOR 7/14/76




                                Height  (meters)

Hour               0-250      250-625    625-1250    1250-2500
6
7
8
9
10
11
12
13
14
15
16
17
11
146
781
1,873
5,086
6,096
4,767
3,605
3,962
3,633
3,617
4,939
90
71
104
159
13,837
16,835
13,241
10,015
11,008
10,093
10,060
13,735
6
6
6
7
18,981
24,039
19,337
14,625
16,075
14,740
14,732
20,113
6
6
6
6
6,393
13,342
13,973
10,569
11,617
10,652
10,837
14,794
                                                              2
* Average K   coefficient  for the height  range  shown  in meters /minute.
                                   6-14

-------
     For July 13, 1976, the processed meteorological data indicate a
more typical summer day for St. Louis.  Clear skies and continuously-
increasing surface temperatures until 1600 hours resulted in unstable
atmospheric conditions throughout the day.  The mixing height increased
from a sunrise elevation of 140 meters to a maximum of 1300 meters at
1700 hours.  The relatively light winds resulted in a somewhat smaller
vertical dispersion than those calculated for June 29, 1976.
     Meteorological conditions for July 14, 1976, were also typical of
summer conditions for the St. Louis region.  Clear skies, rapidly
increasing surface temperatures, and moderate wind speeds combined to
create conditions of vigorous vertical mixing.  The atmosphere was
unstable, with the mixing height extending to 2500 meters by 1500 hours.
Surface temperatures and mixing heights are significantly higher than
those on the previous day.  The range of surface temperatures was from
24.8° C at 0600 to 35.1° C at 1500 hours.

6.3  Air Quality Conditions

     The air quality measurements from the RAMS Stations were inter-
polated along the three test trajectories.  Ambient concentrations at
the three closest stations are weighted by the square of their inverse
distance to the trajectory node to estimate concentrations in the
Lagrangian air parcel.  Interpolated air quality estimates along each
trajectory are listed in Tables 6-7, 6-8 and 6-9 for ozone (0_), carbon
monoxide (CO), methane (CH ), total hydrocarbons (THC), nitric oxide
(NO), nitrogen dioxide (N0_), and sulfur dioxide (S0?).  The concentra-
tions listed for a particular time represent the average value of meas-
urements taken during the hour following the time listed.
     The air quality measurements for June 29, 1976, are fairly typical
for a. summer day in St. Louis.  The ozone concentrations ranged from  .06
and .07 parts per million (ppm) between 1300 and 1700 hours.  Nitric
oxide and nitrogen dioxide concentrations were below .01 ppm, except
during the noon hour where they slightly exceed this value.  The carbon
monoxide concentrations were generally low, between 0.10 and 0.40 ppm,
with the exception of a 1.37 ppm peak at 1400 hours.  In addition, all
the sulfur dioxide monitors reported 0.0025 ppm, which is the lower
detectable limit of the measurement instrumentation.  These air quality
                                   6-15

-------
                           TABLE 6-7
           AIR QUALITY ALONG TRAJECTORY FOR 6/29/76
Hour
6
7
8
9
10
11
12
13
14
15
16
(ppm)
.0293
.0220
.0298
.0382
.0436
.0445
.0439
.0655
.0621
.0599
.0693
CO
Cppm)
.133
.240
.108
.086
.216
.362
.278
.197
1.37
.165
.176
CH4
(ppm)
1.52
1.50
1.47
1.47
1.45
1.38
1.59
1.57
1.51
1.49
1.52
THC
(ppmc)
1.48
1.96
1.50
1.44
1.47
1.44
1.66
1.66
1.53
1.49
1.48
NO
(ppm)
.0031
.0052
.0033
.0028
.0033
.0028
.0102
.0025
.0025
.0026
.0025
N02
(ppm)
.0055
.0144
.0082
.0061
.0111
.0069
.0114
.0052
.0057
.0076
.0065
so2
(ppm)
-
.0025
.0025
.0025
.0025
.0025
-
-
-
.0025
_
Interpolated to trajectory nodes from the three closest RAMS Stations,
                                  6-16

-------
                            TABLE 6-8
          AIR QUALITY ALONG THE TRAJECTORY  FOR 7/13/76
our
6
7
8
9
10
11
12
13
14
15
16
17
&
.0335
.0391
.0447
.0475
.0631
.0909
.1305
.1412
.1260
.1288
.1430
.1413
CO
.120
.114
.089
.139
1.25
.577
.492
.334
.149
.118
.082
.108
CH4
(ppm)
1.63
1.63
1.61
1.73
1.76
1.81
1.77
1.69
1.66
1.58
1.58
1.61
THC
(ppmc)
1.71
1.56
1.61
1.63
1.79
1.95
1.93
1.74
1.53
1.54
1.52
1.56
NO
(ppm)
.0025
.0025
.0025
.0036
.0052
.0037
.0025
.0025
.0025
.0025
.0025
.0025
N02
(ppm)
.0025
.0025
.0029
.0031
.0112
.0204
.0222
.0152
.0167
.0165
.0051
.0067
                                                               (ppm)
"Interpolated  to  trajectory nodes  from the three closest RAMS Stations.
                                 6-17

-------
                            TABLE  6-9
            AIR QUALITY ALONG TRAJECTORY FOR 7/14/76J
our
6
7
8
9
10
11
12
13
14
15
16
17
(ppm)
.0213
.0335
.0489
.0599
.0649
.0630
.0945
.0942
.0870
.0808
.0693
.0596
CO
(ppm)
.222
.222
.152
.148
.159
.311
.066
.106
.115
.172
.166
.216
CH4
(ppm)
1.71
1.71
1.64
1.61
1.53
1.46
1.49
1.38
1.46
1.48
1.47
1.47
THC
(ppmc)
1.66
1.67
1.58
1.54
1.50
2.69
1.43
1.54
1.46
1.46
1.44
1.42
NO
(ppm)
.0037
.0025
.0025
.0025
.0025
.0025
.0025
.0025
.0025
.0025
.0025
.0035
N02
(ppm)
.0100
.0069
.0040
.0031
.0083
.0073
.0052
.0040
.0027
.0026
.0029
.0029
                                                              (ppm)
Interpolated to trajectory nodes from the three closest RAMS Stations.
                                 6-18

-------
measurements are consistent with expected values for the observed meteor-
ological conditions of June 29, 1976.  The moderate wind speeds and
unstable atmosphere combined to rapidly advect and vertically disperse
the pollutant mass in the atmosphere.  Reduced ultraviolet radiation,
due to overcast conditions, limited the photochemical production of
ozone to levels below the Federal Standard of 0.08  ppm.
     The air quality measurements for July 13, 1976, show somewhat
higher concentrations than those for June 29.  The ozone concentrations
reached 0.12 to 0.14 ppm between 1200 and 1700 hours.  Although the
nitric oxide reached a maximum of only 0.005 ppm, the nitrogen dioxide
exceeded 0.02 ppm from 1100 to 1300 hours.  Carbon monoxide measurements
ranged from 0.1 ppm, a typical rural background level for St. Louis, to
a maximum of 1.25 ppm at 1000 hours.  These somewhat higher ozone and
nitrogen dioxide concentrations are consistent with the meteorological
conditions of light winds, clear skies, and higher temperatures.
     The interpolation of air quality measurements for July 14, 1976,
shows ozone concentrations which ranged between 0.08 and 0.094 ppm
during the 1200 to 1600 hour time period.  The nitric oxide and nitrogen
dioxide measurements remained quite low, below 0.0037 and 0.0083 ppm,
respectively.  Similarly, the carbon monoxide concentration measurements
ranged between 0.06 and 0.22 ppm, which are also quite low.  These low
precursor concentrations are consistent with the meteorological trends
for this day.  Vigorous vertical mixing of pollutants introduced near
the surface to elevations as high as 2500 meters prevented significant
build-up of pollutant mass at the surface.
     Methane measurements for all three days ranged between 1.38 and
1.81 ppm, which equal or slightly exceed the global background methane
concentration, and are believed to be typical of the St. Louis region.
The total hydrocarbon measurements range from 1.42 to 2.69 ppmc.  It is
important to note that the total hydrocarbon measurements are smaller
than the methane measurements for approximately half the hours examined.
This occurrence indicates that the non-methane hydrocarbons concentra-
tions are generally low since the differences (between methane and total
hydrocarbons are within the accuracy limits of the instrumentation
employed in the measurement program.
                                  6-19

-------
6.4  Source Emission Strengths

     Each of the three test-day trajectories has associated schedules of
area source and point source emission rates based on the RAPS emissions
inventory (provided by EPA) for the corresponding day.  These emissions
schedules include emission rates for nitrogen oxides, four classes of
reactive hydrocarbons (alkenes, alkanes, aromatics, and aldehydes),
carbon monoxide, and sulfur oxides.  Tables 6-10 and 6-11 list the total
number of moles of each species entrained into the air parcels from area
sources and point sources, respectively, for each of the three test-day
trajectories.  The data show that the majority of nitrogen oxide, reac-
tive hydrocarbon, and carbon monoxide emissions are due to sources
included in the area source emissions inventory.  The major contribution
of sulfur oxides to the emission schedules is due to sources included in
the point source emissions inventory.  In comparing the data for the
three different trajectories, it appears that the total emissions for
June 29 and July 14 are quite similar, yet the emissions for the July 13,
1976, trajectory are generally lower.  The July 13 trajectory did not
traverse as much of the emission-intensive metropolitan region as did
the June 29 and July 14 trajectories.

6.5  Initial Pollutant Concentrations

     The simulation model requires specification of initial pollutant
concentrations as inputs.  The interpolated RAMS air quality data are
used for the surface concentration inputs to the model.  Concentrations
at elevations above the surface are estimated by means of the following
assumptions.  Nitric oxide, hydrocarbon, carbon monoxide, sulfur dioxide
and sulfate concentrations above the mixing height  (or inversion base)
are chosen as one-half the measured surface concentration.  The concen-
tration at the mixing height is assumed to be two-thirds of the surface
concentration.  Concentrations at  elevations between the surface and
mixing height are chosen by assuming a  linear rate of change from the
surface.  Although this formulation is  somewhat arbitrary, it is consis-
tent with the physical expectation of precursor pollutant profiles which
normally exhibit highest concentrations near the surface and sources.
                                   6-20

-------
                    TABLE 6-10




AREA SOURCE EMISSIONS ENTRAINED ALONG TRAJECTORIES

Date
6/29/76
7/13/76
7/14/76



Date
6/29/76
7/13/76
7/14/76
Moles
NOX PARF OLEF AROM ALDE CO
554. 201. 136. 58. 39. 10481.
184. 69. 49. 26. 22. 3913.
608. 272. 168. 67. 45. 12445.
TABLE 6-11
POINT SOURCE EMISSIONS ENTRAINED ALONG TRAJECTORIES
Moles
NOX PARF OLEF AROM ALDE CO
103. 30. 9. 39. .2 852
148. .05 .05 .02 .05 45
98. 18. 1.4 32. .12 117

SOX
23.
10.
25.



SOX
245
216
143
                          6-21

-------
     Initial vertical profiles of ozone concentrations are input to the
model, assuming that the lowest ozone concentrations occur at the sur-
face.  At model mesh point elevations below the mixing height, the ozone
concentrations are chosen as one and one-half times the surface measure-
ment.  Above the mixing height and below 1500 meters, the ozone concen-
trations are estimated to be twice the surface values.  Above 1500
meters, the ozone concentrations are estimated to be three times the
surface value.
     These methods for choosing initial pollutant vertical profiles are
based on typical shapes of vertical profiles measured by Meteorology
Research, Inc.*, during aircraft measurements between the hours of 0500
and 0900 in August, 1977.  The location of the measurements was in a
rural region near Rockport, Indiana, which is similar to the rural start
locations outside of St. Louis modeled in this study.  Note that the
methods described here provide approximate initial conditions suitable
for summer morning trajectories with rural start locations.  Uncertain-
ties in these approximations could easily be plus or minus one hundred
percent.
     An additional task in choosing initial concentrations is the parti-
tioning of individual hydrocarbon class concentrations from the reactive
hydrocarbon data.  Given a reactive hydrocarbon measurement [RHC] in
parts per million as carbon (ppmC), the individual hydrocarbon class
concentrations in ppm are estimated as follows:

           [Alkanes]            =   .15*  [RHC]
           [Alkenes]            =   .03*  [RHC]
           [Aromatics]          =   .03*  [RHC]
           [Formaldehydes]      =   .02*  [RHC]
           [Other Aldehydes]    =   .01*  [RHC]

These partitioning formulae are based on the assumption of a somewhat
aged reactive hydrocarbon mix, so that there are relatively large
fractions of the less reactive alkanes and aromatics, and smaller
fractions of the more reactive alkenes and aldehydes, than indicated by
the St. Louis emissions  inventory.  The  assumption of an aged atmos-
pheric hydrocarbon mixture is appropriate since the  formulae are applied
 *Blumenthal,  1978.
                                   6-22

-------
uniformly to calculate initial conditions at all elevations and in rural
locations with low emission densities.
     An additional procedure is incorporated in the determination of
initial conditions for hydrocarbons as a result of uncertainties in the
RAMS measurements.  A minimum reactive hydrocarbon concentration of 0.02
ppmC is assumed when the difference between total hydrocarbon and methane
measurements is below this level.
     The initial pollutant concentration inputs to the model are shown
in Figures 6-7, 6-8 and 6-9 for the three test days, respectively.  The
figures indicate vertical profiles for nitric oxide (NO) , alkanes (PARF) ,
alkenes (OLEF) , aromatics (AROM) , formaldehyde (HCHO) , other aldehydes
(RCHO) , ozone (0 ) , carbon monoxide (CO) , sulfur dioxide  (S0_) , and
sulfate (S04).
     Initial nitrogen dioxide concentrations are calculated by the
simulation model from the photostationary state equation  shown below:

                               kl
           [N02]  =  [NO] [03]  ^
where k1 and k_ are the rate constants for the reactions:
       J.      «J

                      kl
          N02 + hv    -£  NO + 0
          NO
     The concentrations of the radical species OH and H0_ are  input with
                               -8       -6
vertically uniform values of 10   and 10   ppm, respectively.   Initial
H20 concentrations are also input with uniform profiles, and the  values
assumed for the three test days are  shown below.

          Date                   HgO Concentrations
         6-29-76                   1.7 x 104
         7-13-76                   1.7 x 104
         7-14-76                   2.2 x 104
                                   6-23

-------
Figure 6-7  Initial Pollutant Concentration Vertical Profiles
                    6 AM          6-29-76
l^UU
1000
800
Helght 600
O)
400

?nn
£f\J\J
n
-
. [NO]
. [PA]

•



i^UU
1000
800
[SO ]
£
600

400
^v
^yV^
x 2°:
.
[0j]
.

•
>
/
. ~s<







1

[CO]



x. inversion base
^^-x^
.^^^
         0 .001 .002 .003
       Concentration (ppm)
                            0   .02   .04   .06   .08  .10   .12   .14
                                  Concentration  (ppm)
         1200

         1000

          800
Height
 (m)      600

          400

          200
                             frlCHO
                                                inversion base
                   0    .1   .2    .3   .4   .5  .6
                        Concentration (ppb)
                           6-24

-------
                                 Figure  6-8
              Initial Pollutant Concentration Vertical Profiles
                            6 AM           7-13-76
        1500 r
        1250
        1000

Height   75°
 (m)     500

         250
[SO,]
[PA]
1500 r

1250

1000

 750

 500

 250
                .005   .01   .015
             Concentration (ppm)
         1500

         1250

         1000
Height
  (m)      750

          500
          250
  [RCHO]
                   COLEFJ
        [HCHC3
[CO]
                                                              inversion base
                         .025   .05  .075   .10  .125
                           Concentration  (ppm)
                                      inversion base
        0.5      1.0        1.5        2.0
              Concentration (ppb)
                                                             2.5
                                     6-25

-------
      Figure 6-9  Initial  Pollutant  Concentration  Vertical  Profiles
                           6 AM           7-14-76
Height
 (m)
2500 p

2000

1500

1000

 500
[PA]
[NO]
                  .001  ,002  .003  .004
                Concentration  (ppm)
2500
2000
1500
1000
500
n
••
•
-
[°3V
,/,


/


/
/

[so2]
k. inversion
^\^ base
, ;\,
                                    .02   .04   .06   .08   .10
                                    Concentration  (ppm)
2500
2000
1500
Height
(m) 1000
500




_
-




KHQ
\

1 .


EcK
\
i
2 .

^

\
\i i
3 .4 .5
2500
2000
ts°4l 1500
1000
^"^- 500
^^^_^

.6 .7 .8 .9 1.0
-
-
.
-
i
.1


[CO]
\
\ ,
.2 .:
                      Concentration  (ppb)
                                              Concentration  (ppm)
                                      6-26

-------
6.6  Simulation Model Results

     The ERT Lagrangian Photochemical Diffusion Model was exercised to
simulate pollutant concentrations along the three test-day trajectories.
The methodology developed for applying the model in the St. Louis region
was applied consistently on the three days without adjustment to improve
model performance on a particular day.  The results are presented in
Figures 6-10 through 6-15, which show the predicted surface concentra-
tions of ozone, nitrogen dioxide, nitric oxide, carbon monoxide, sulfur
dioxide, and sulfate, and the interpolated RAMS measurements as func-
tions of time.
     In examining the model's performance relative to interpolated
measurements, it is important to recognize that there are situations
where the interpolated measurements may not be truly representative of
the air parcel concentrations.  Situations arise where the interpolated
data represent measurements made at 20 to 60 kilometers away from a
given air parcel location along the trajectory.  Other situations have
been observed where the three closest stations used in the interpolation
have recorded concentrations which differ by more than two orders of
magnitude.  This indicates certain monitors are influenced by local
sources, which may or may not influence the air parcel in question.  It
is believed that these situations are somewhat rare, and, in general, it
is felt that the interpolated measurements represent the best available
means of comparing concentrations computed in a Lagrangian reference
frame to measurements recorded in an Eulerian reference frame.  Neverthe-
less the reliability of the interpolated measurement, as a basis of com-
parison, is believed to decrease significantly when the distance to the
stations exceed the air parcel dimensions.
     The model results for the June 29, 1976, trajectory should be
viewed with the knowledge that approximately 90% of the source emissions
are entrained into the air parcel between 1000 and 1400 hours.  The pre-
dicted ozone concentrations for the trajectory agree well  (+_ .005 ppm)
with the interpolated measurements between sunrise and noontime.  In the
afternoon, the model underpredicts the ozone data.  This discrepancy is
largest, 0.02 ppm, at 1300 hours when the air parcel passes Station 116.
The computed ozone curve shows a reduction between 1100 and 1200 hours
caused by both increased nitric oxide concentrations and a reduction in
                                   6-27

-------
ultraviolet radiation.  The maximum computed ozone concentration of
0.055 ppm occurs around 60 kilometers northeast of downtown St. Louis.
Computed nitrogen dioxide concentrations for this trajectory do not
agree as well with the interpolated data as do the ozone predictions.
The nitrogen dioxide measurements appear somewhat scattered between 5
and 14 ppb, while the computed values show a sharp rise to a midday peak
of 16 ppb and a gradual decline in the afternoon hours.  Computed nitric
oxide concentrations follow trends in the data more closely than nitro-
gen dioxide.  Until 1030, both predicted and measured values for this
species are at or near the lowest detectable limit of the instrumenta-
tion.  The model's NO concentrations increase to a maximum value of
6 ppb at 1130, and the interpolated measurements indicate a peak of
10 ppb at 1200 hours.  Afternoon NO concentrations decline to values
near the lowest detectable limit (2.5 ppb), which is consistent with the
measurement records.  The computed carbon monoxide concentrations agree
fairly well (+_ .2 ppm) with the interpolated data, except for a signif-
icant discrepancy at 1400 hours.  Upon closer examination of this
discrepancy, we find that the three stations, from which the inter-
polated value of 1.37 ppm was determined, were measuring widely varying
CO levels.  The stations, distances, and measurements are listed below.

          Station          Distance (km)           [CO]  (ppm)
            116                15.3                  3.67
            115                15.9                    .15
            123                16.3                    .05

This tabulation shows that the apparently high interpolated value is due
to one high measurement at RAMS 116.  Thus, the interpolated value is
felt to be unrepresentative of the region, and the discrepancy is under-
standable.  Sulfur dioxide concentrations predicted by  the model exhibit
trends that are consistent withe the trajectory's SO  emission schedule.
The peak computed S0_ concentration (7.5 ppb) occurs at 1230 when the
air parcel is over southern St. Louis.  All the available measurement
records indicate an S02 level of 2.5 ppb, which is the  lower detectable
limit of the instruments.  Prior to 1100, the computed  concentrations
are below this level.  After 11 o'clock, they are above this level, but
since there is only one measurement record after this time, it is
                                   6-28

-------
                                 Figure 6-10
           Trajectory Model Concentration Predictions for 6-29-76
[03]
(ppm)
[N02]
(ppb)
[NO]
(ppb)
.07

.06
.05

.04

.03

.02
 20

 15

 10

  5

  0


 15

 10

  5
                 7   8    9   10
                           11   12  13
                            Time
              14   15   16   17
                     Legend
                   • Interpolated RAMS
                       measurements
                 -  " Model prediction
                                                           	LDL"
                                                  j	i	i
                      8   9   10
11   12
 Time
13   14  15   16   17
                      8   9   10
                           11   12
                            Time
          13   14  15   16   17
             * LDL = Lower Detectable Limit of Instrumentation
                                    6-29

-------
[CO]

(ppm)
[so2]
(ppb)
 (ppb)
                                 Figure 6-11
           Trajectory Model Concentration Predictions for 6-29-76
1.4

1.2


1.0

  .8

  .6

  .4
  8

  6


  4

  2

  0




2.0

1.5


1.0


0.5


  0
             678
                       Legend
                     • Interpolated RAMS
                        measurements
                    — Model prediction
                                                                j
                          9   10
                           11   12
                            Time
          13    14    15   16    17
                                                           	LDL*
                             10
11   12

  Time
13   14  15   16   17
                     10   11   12  13

                           Time
              14   15   16   17
               *LDL = Lower Detectable Limit  of Instrumentation
                                    6-30

-------
difficult to draw conclusions about the model's S02 predictive capability
from these limited data.  Similarly, since hourly sulfate measurements
are not routinely archived by the RAPS network, it is not possible to
make an assessment of the accuracy of the sulfate predictions.  Never-
theless, the computed sulfate concentration history for this trajectory
is interesting and unique in this study.  The almost constant afternoon
concentrations show a balance between the effects of SCL conversion to
SO. and SO. surface deposition.
     The model results for the July 13, 1976, trajectory indicate that
the model underpredicts the interpolated data for all pollutants.
Recall that the total emissions used in the simulation of this trajec-
tory are only 1/3 to 1/2 as large as those for the other trajectories,
yet the interpolated RAMS data indicate generally higher atmospheric
concentrations.  The model's computed ozone shows good agreement with
the data until 1100 hours, after which these data are underpredicted by
0.03 to 0.06 ppm.  The model predicts a peak ozone concentration of
0.098 ppm at 1600 hours when it passes Station 122, where the measured
ozone was 0.143 ppm.  Computed nitrogen dioxide concentrations for the
trajectory are well below the observed values, with the largest dis-
crepancies occurring between 1000 and 1500 hours.  The computed nitric
oxide concentrations are generally quite low, which is consistent with
the measurement records, since most measurements are near the lowest
detectable limit for the instrumentations employed.  The time of the
model-predicted peak nitric oxide concentration agrees with the observed
data, but the magnitude of the computed peak is only 40% of the observed
maximum.  The carbon monoxide concentration predictions are also well
below observed levels between 0900 and 1300 hours.  The consistent
underprediction by the model on this trajectory may be caused by a
number of factors.  First, the trajectory path generated from the wind
data has the smallest levels of pollutant emissions of any of the tra-
jectories in this study.  It appears that this trajectory may have
missed some major sources of pollutants which influenced the air mass.
The wind data show light winds for most of the day.  Thus, there is
believed to be more uncertainty in the accuracy of the trajectory path
on this day than the others in the study.  The underprediction may also
be attributable to an overestimate of the strength of vertical dis-
persion for this particular day.

                                   6-31

-------
                                Figure 6-12

           Trajectory Model Concentration Predictions for 7-13-76
(ppm)
.150

.125


.100

.075


.050


.025
         20 -
         15 -
[N°2]     10
       5.00
[NO]

CPPb)
            Legend
          • Interpolated  RAMS
             measurements
         — Model prediction
            67    89    10   11  12   13   14  15   16   17

                                   Time
                                                                 LDL*
            67    89
                       10   11  12

                            Time
13   14  15   16   17
            67    8    9   10   11  12   13   14  15   16   17
                                                    «	•	LDL"
            *LDL = Lower Detectable Limit of Instrumentation
                                    6-32

-------
                                Figure 6-13
           Trajectory Model Concentration Predictions for 7-13-76
 [CO]
.(ppm)
1.50

1.25

1.00

 .75

 .50

 .25
       10.0

         7.5

 [S0]    5.0
 [so4]
 (ppb)
        l.SOp
        1.25
         .75
         .50
                                                         Legend
                                                         Interpolated RAMS
                                                           measurements
                                                         -Model prediction
^

_ .
1


1 1 1 t 1 1
             67     8     9    10   11    12    13    14   15    16    17
                                    Time
             67    8    9   10  11   12  13   14   15   16   17
                                    Time
                 78    9   10   11   12   13  14   15   16   17
                                   Time
                                   6-33

-------
     For the July 14, 1976, trajectory, the model predicts ozone concen-
tration below the observed levels.  The interpolated RAMS data indicate
a maximum ozone level of 0.094 ppm at 1300 when the computed ozone
concentration in the parcel has a value of 0.046 ppm.  The computed
ozone increases to a maximum value of 0.074 ppm at 1700, and the inter-
polated data show a decline to 0.060 ppm by this time.  The model's
nitrogen dioxide predictions show good agreement with the data at 0800
and 0900 hours.  The model's peak nitrogen dioxide concentration is
equal to the observed peak value  (8.3 ppb), yet it is predicted at
1200 hours instead of 1000, when the maximum measurement was recorded.
The computed afternoon nitrogen dioxide concentrations decline at a rate
similar to that indicated by the observed measurements, but they over-
predict the observed values by 2 to 3 ppb.  Predicted nitric oxide
concentrations exceed the interpolated data during the midday hours.
The model predicts a maximum concentration of 4.7 ppb nitric oxide at
1200 hours, but all of the data between 0700 and 1600 indicate 2.5 ppb,
the lowest detectable limit of the instrumentation.  Prior to 1030 and
after 1330, the computed nitric oxide values are below 2.5 ppb, which is
consistent with the measurement records.  The predicted carbon monoxide
concentrations agree well  (+_ ,07 ppm) with the data between 0600 and
1100 hours.  The model predicts a peak carbon monoxide concentration of
0.38 ppm at 1130, and the maximum interpolated value along the tra-
jectory is 0.31 ppm  at 1100 hours.  In the afternoon, the model over-
predicts the interpolated data with the largest discrepancies occurring
between 1200 and 1400, and smaller discrepancies between 1500 and 1700.
Unfortunately, there are no sulfur dioxide measurements archived for
this day.  Hence, no basis exists for  interpretations of the sulfur
dioxide or sulfate predictions.   It should be pointed out that the
initial concentrations of SCL and SO.  are arbitrary  and believed to be
somewhat higher than might have been observed on this day.  They were
purposefully chosen  somewhat high to demonstrate the  effects of the
chemistry at somewhat higher concentrations.
     In summary, over the  three test-day  trajectories,  the model pre-
dictions give  best agreement with the  interpolated RAMS air quality data
for ozone and  carbon monoxide, and poorest agreement  for nitrogen
dioxide.  In examining the results for the different  days, the model
                                   6-34

-------
                                Figure 6-14

          Trajectory Model Concentration Predictions for  7-14-76
67    8   P     10
[N02]

(PPb)
[NO]
5.0

3.75


2.5


1.25
                                   11   12

                                    Time
                                     13   14  15   16   17
                                                          Legend
                                                          Interpolated RAMS
                                                             measurements
                                                          Model  prediction
                                                               -? - LDL*
 67    89    10
                                   11  12

                                    Time
                                     13  14   15   16   17
                                                                 	LDL*
67    8    9   10
                                  11   12  13

                                   Time
                                         14   15  16   17
               *LDL = Lower Detectable Limit of Instrumentation
                                     6-35

-------
                                 Figure 6-15
           Trajectory Model Concentration Predictions for 7-14-76
[CO]

(ppm)
[so2]
(PPb)
 .4

 .3

 .2

 .1
 10

  8

  6

  4


  2
                                        _L
                              10
                           11  12

                            Time
    13   14  15   16   17
                                   _L
J_
                          9   10
                           11  12

                            Time
                                                         Legend
                                                        • Interpolated  RAMS
                                                           measurements
                                                       —Model prediction
     13    14    15   16
              17
 [so4]
 Cppb)
1.4

1.2

1.0

 .8


 .6
                               10
                           11  12

                            Time
     13
14   15  16   17
                                       6-36

-------
performed best on June 29, 1976, and worst on July 13, 1976.  These
simulations are viewed as being preliminary demonstration runs for a
complex simulation model.  Given the wealth of data from the RAPS
program, refinements in the methods for selecting trajectories and
initial conditions can be made when the model has been exercised for a
greater number of days in the St. Louis region.  Such refinements have
proved important in other geographic areas, and may improve the model's
predictive capabilities for the St. Louis application.
                                  6-37

-------
                   7.  CONCLUSIONS AND RECOMMENDATIONS

     The primary objective of this study has been to provide a Lagrangian
photochemical diffusion model to the EPA.  The principle conclusions from
the study are that the model has been successfully adapted, documented,
and demonstrated for use in the St. Louis region, and that its results
show a tendency to underpredict measured ozone concentrations.
     It is our understanding that the primary reason for the delivery of
the model is to provide the means for the EPA to conduct an assessment
of the accuracy and validity of the model.  In light of this plan, a
list of recommendations has been compiled which pertains to the model
evaluation.  The recommendations cover possible refinements in the
modeling of St. Louis, running the model for purposes of evaluation, and
how the evaluation may be carried out.
     It is suggested that prior to the evaluation of the model, it be
exercised more extensively with the RAPS data base.  The purpose of
these exercises would be to develop procedures for trajectory selection
and initial pollutant concentration designation that are more sophisti-
cated than the ones used in this study.  Particular attention should be
given to the choice of initial conditions and their vertical distribu-
tions because there is considerable uncertainty in these inputs and the
model predictions are usually sensitive to them.  The exercises would
involve examining any available elevated pollutant concentration data
for St. Louis, as well as examining the model's performance using
different initial concentration profiles.  These exercises should not be
confused with altering the model formulation, which they are not.
Rather, the exercises would consist of investigations to reduce the
uncertainty in the inputs, using real data and feedback from the model.
     It is suggested that a relatively large number of days (20-50) and
trajectories be simulated with the model for the evaluation.  These
selected test days should include a spectrum of meteorological condi-
tions and have data bases as complete as possible.  Also, since each of
the model's simulations provides information for only one space-time
tract through the region, the model should be exercised for several
trajectories for each test day.  These procedures will enhance the
statistical significance of the model evaluation results.
                                   7-1

-------
     An important aspect of the evaluation, which should be addressed,
is how to evaluate Lagrangian concentration predictions relative to data
collected in an Eulerian reference frame.  In this study, the model
results have been compared to the data interpolated from the three
closest stations, using reciprocal distance squared weighting of the
individual station data.  A relatively large maximum distance criteria
was used, which allowed data to be interpolated from stations as far as
sixty kilometers away from the air parcel location.  Given that the
resolution of the model is approximately a one kilometer wide space-time
tract, such large distances would be unexceptably large for a rigorous
evaluation of the model.  From a practical standpoint, the maximum
distance cannot be chosen too small because this would greatly reduce
the number of points of comparison.  Thus, as a compromise, it is
suggested that the model results be compared to station data measured
within five kilometers of the air parcel locations.  An obvious exception
to this criteria would be when either a station or the air parcel, but
not both, is a short distance downwind of a large emissions source.  In
this case, a maximum distance criteria of less than one kilometer should
be applied.
     Lastly, the model performance evaluation should be accompanied by
some model sensitivity analyses to increase its validity and define
areas for further research.  In particular, model sensitivity to input
parameter variations needs to be established.  The magnitude of these
variations should be comparable to the uncertainty in the input data.
Given the large number of inputs to the model, this task can become a
huge one.  So, from a practical point of view, the matrix of runs with
input perturbations must be limited and carefully chosen.  Nevertheless,
they should include meteorological, emissions, and chemical input data.
The results of such analyses are quite valuable when interpreting the
accuracy of the model.  For it is believed that the results of a per-
formance evaluation of a model reflect not only accuracy of the physical
formulation, but also the accuracy of the inputs.
     Since the accuracy and extent of the RAPS data base exceed any
other known air pollution modeling data base, the model evaluation
studies  should reveal the upper limit of air quality predictability with
current modeling technology.  It is hoped in the course of the evalua-
tion that both strengths and weaknesses of air quality simulation models
                                    7-2

-------
be elucidated which may subsequently further the state-of-the-art of the
technology.
                                   7-3

-------
                          8.  REFERENCES
Anderson, L.G. 1976.  J. Geophys. Res.  14: 151.

Atkinson, R., R.A. Perry and J.N. Pitts, Jr. 1976.  Rate Constants for
     the Reactions of the OH Radical With NCL (M = Ar and N_) and S02
     (M = Ar).  J. Chem. Phys. 65(July 1): No.l: 306-310.

Beauchene, J.H., P.J. Bekowies, J.M. McAfee, A.M. Winer, L. Zafonte and
     J.N. Pitts, Jr. 1973.  A Novel 20 KW Solar Simulator Designed for
     Air Pollution Research.  Proceedings of Seventh Conference on Space
     Simulation (NASA Special Publ. 336).  Paper No. 66(Nov. 12-14):
     811-825.

Bergstrom, R.W. and J.T. Peterson 1977.  Comparison of Predicted and
     Observed Solar Radiation in an Urban Area.  Journal of Applied
     Meteorology, 16: 1107-1116.

Bergstrom, R.W. and J.T. Peterson 1977.  Comparison of Predicted and
     Observed Solar Radiation in an Urban Area.  Journal of Applied
     Meteorology, lf>_ pp 1107-1116.

Blackadar, A.K. and H. Tennekes 1968.  Asymptotic Similarity in
     Neutral Barotropic Planetary Boundary Layers.  J. Atmospheric Sci.
     25;  1015-1020.

Blumenthal, D.L. 1978.  Aircraft Measurements of Pollutant and Meteoro-
     logical Parameters in the SURE Network.  Meteorology Research, Inc.,
     report submitted to the Electric Power Research Institute for
     contracts RP-862-3 and RP-862-4.

Brost, R.A. and J.C. Wyngarrd 1978.  A Model Study of the Stably
     Stratified Planetary Boundary Layer, Manuscript, CIRES, University
     of Colorado/NOAA, Boulder.

Bufalini, J.J., S.L. Kopczynski and M.C. Dodge 1972.  Contaminated Smog
     Chambers in Air Pollution Research.  Environ. Lett. _3_: No.2: 101-
     109.

Businger, J.A., J.C. Wyngaard, Y. Izumi and E.F. Bradley 1971.  Flux-
     Profile Relationships in the Atmospheric Surface Layer.
     J. Atmospheric Sci. 28: 181-189.

Businger, J.A. and S.P.S. Arya 1974.  Height of the Mixed Layer in the
     Stably Stratified Planetary Boundary Layer.  Advances in Geophysics.
     ISA: 73-92.

Calvert, J.G. 1974.  Modes of Formation of the Salts of Sulfur and
     Nitrogen in an NO -SO^-Hydrocarbon Polluted Atmosphere.  Conference
     on Health Effects of Atmospheric Salts and Gases of Sulfur and
     Nitrogen in Association With Photochemical Oxidant.  Volume II -
     Reference Documents, T. Timothy Crocker, ed., ARE Contract No. 3-
     197, University of California (January): VIII-1 - VIII-16.
     (Available from NTIS as PB 251 233.)


                                   8-1

-------
Calvert, J.G. and J.N. Pitts, Jr. 1966.  Photochemistry.   New York:
     John Wiley § Sons, Inc.

Calvert, J.G. and R.D. McQuigg 1975.   The Computer Simulation of the
     Rates and Mechanisms of Photochemical Smog Formation.  Chemical
     Kinetics Data for the Upper and Lower Atmosphere.   Int.  J.  Chem.
     Kinet. Symp. No. 1: 113-154.

Calvert, J.G., Fu Su, Jan W. Bottenheim and Otto P. Strausz 1978.  Mechanism
     of the Homogeneous Oxidation of Sulfur Dioxide in the Troposphere.
     Atmos. Environ.  12: No.1-3: 197-226.

Carter, William P.L., Karen R. Darnall, Alan C. Lloyd,  Arthur M. Winer
     and James N. Pitts, Jr. 1976.  Evidence for Alkoxy Radical
     Isomerization in Photooxidations of C.-C, Alkanes Under Simulated
     Atmospheric Conditions.  Chem. Phys. Lett. £2_(August 15): No.l:
     22-27.

Carter, W.P.L., A.C. Lloyd, J.L. Sprung and J.N. Pitts, Jr. 1978.
     Int. J. Chem. Kin.  Accepted for Publication.

Castleman, A.W., Jr, Richard E. Davis, H.R. Munkelwitz, I.N. Tang and
     William P. Wood 1975.  Kinetics of Association Reactions Pertain-
     ing to H2SO.Aerosol Formation.  Chemical Kinetics of Association
     Reactions Pertaining to H2SO. Aerosol Formation.  Chemical
     Kinetics Data for the Upper and Lower Atmosphere.   Int. J.  Chem.
     Kinet.  Symp. No. 1: 629-640.

Castleman, A.W., Jr. and I.N. Tang 1977.  Kinetics of the Association
     Reaction of Sulfur Dioxide With the Hydroxyl Radical.
     J. Photochem.  6_: No.5: 349-354.

Chan, Walter H., Robert J. Nordstrom, Jack G. Calvert and John H. Shaw
     1976.  Kinetic Study of HONO Formation and Decay Reactions in
     Gaseous Mixtures of HONO, NO, N0?, HO, and N~.  Environ. Sci.
     § Technol. l_0(July) : No.7: 674-682.

Chan, Walter H., Robert J. Nordstrom, Jack G. Calvert and John H. Shaw
     1976.  An IRFTS Spectroscopic Study of the Kinetics and the
     Mechanism of the Reactions in the Gaseous System, HONO, NO, N02,
     H20.  Chem. Phys. Lett. 3J7_(Feb. 1): No.3: 441-446.

Chan, Walter H., William M. Uselman, Jack G. Calvert and John H. Shaw
     1977.  The Pressure Dependence of the Rate Constant for the
     Reaction: HO + CO -> H + S0?.  Chem. Phys. Lett.  45_  (Jan. 15):
     No.2: 240-243.

Cox, R.A.  1974.  Photolysis of Nitrous Acid in the Presence of
     Carbon Monoxide and Sulfur Dioxide.  J. Photochem. 5: No.4:
     291-304.                                           ~

Cox, R.A.  1975.  The Photolysis of Gaseous Nitrous Acid - A Technique
     for Obtaining Kinetic Data on Atmospheric Photooxidation Reactions.
     Chemical Kinetics Data for the Upper and Lower Atmosphere.
     Int. J. Chem. Kinet.  Symp. No.l; 379-398.

                                     8-2

-------
Cox, R.A. and S.A. Penkett 1971.  Photo-Oxidation of Atmospheric S02-
     Nature.  229 (Feb. 12) No.5285: 486-488.

Cox, R.A. and S.A. Penkett 1971a.  Oxidation of Atmospheric SO, by
     Products of the Ozone-Olefin Reaction.  Nature. 23_0(Apr. 2):
     Review Suppl.: 321-322.

Cox, R.A. and R.G. Derwent 1975.  Kinetics of the Reaction of Hydro-
     peroxy With Nitric Oxide and Nitrogen Dioxide.  J. Photochem.
     4_: Nos.1-2;  139-153.

Cox, R.A. and R.G. Derwent 1977.  The Ultraviolet Absorption Spectrum
     of Gaseous Nitrous Acid.  J. Photochem.  6_: No.l; 23.

Cox, Richard A.,  Richard G. Derwent and Pauline M. Holt 1976.  Relative
     Rate Constants for the Reactions of OH Radicals With H^, CO, NO,
     and HONO at Atmospheric Pressure and 296 K.  J. Chem. Soc.,
     Faraday Trans. I.  72^: Pt.9: 2031-2043.

Cvetanovic, R.J.  1976.  Chemical Studies of Free Radicals of Atmospheric
     Interest.  Paper No. 31, 12th International Symposium on Free
     Radicals, Laguna Beach, California.  January 4-9.

Darnall, K.R. 1977.  Private communication.

Darnall, Karen R., William P.L. Carter, Arthur M. Winer, Alan C. Lloyd
     and James N. Pitts, Jr. 1976.  Importance of R02 + NO in Alkyl
     Nitrate Formation From C.-C, Alkane Photooxidations Under Simulated
     Atmospheric  Conditions.  J. Phys. Chem.  SCKAug. 12): No.17:
     1948-1950.

Daubendiek, Richard L. and Jack G. Calvert  1975.  A Study of the N 0 -
     S09-0, Reaction System.  Environ. Lett.  8^: No. 2: 103-116.
       ^  O                   -- — -  -  _ _ ll_..Mji-M_mi_.. 1

Davis, D.D. and Gary Klauber 1975.  Atmospheric Gas Phase Oxidation
     Mechanisms for the Molecule SO-.  Chemical Kinetics Data for the
     Upper  and Lower Atmosphere.  Int. J. Chem. Kinet.  Symp. No. 1:
     543-556.

Davis, D.D. and R. Schiff 1974.  Unpublished work.

Davis, D.D., W. Bellinger and S. Fischer 1975.  A Kinetics Study of the
     Reaction of  the OH Free Radical With Aromatic Compounds.  I.
     Absolute Rate Constants for Reaction With Benzene and Toluene at
     300° K.  J.  Phys. Chem. 79(Jan.): No.3: 293-294.

Demerjian,  Kenneth L., J. Alistair Kerr and Jack G. Calvert  1974.
     The Mechanism of Photochemical Smog Formation.  Advances in
     Environmental Science and Technology.  4_: 1-262.  John Wiley and
     Sons,  c.1974.  James N. Pitts, Jr. and Robert L. Metcalf, eds.
                                    8-3

-------
Dodge, M.C. 1977.  Combined Use of Modeling Techniques and Smog Chamber
     Data To Derive Ozone-Precursor Relationships.   International
     Conference on Photochemical Oxidant Pollution and Its Control -
     Proceedings.  U.S.  Environmental Protection Agency.  EPA-600/3-
     77-OOlb.  I_I_(Jan.): 881-889.  (Available from NTIS as PB 264  233.)

Dodge, M.C. and T.A. Hecht 1975.  Rate Constant Measurements Needed to
     Improve a General Kinetic Mechanism for Photochemical Smog.
     Chemical Kinetics Data for the Upper and Lower Atmosphere.
     Int. J. Chem. Kinet.  Symp. No. 1: 155-163.

Dodge, Marcia C. and Gary Z. Whitten 1976.  Aldehydes and Photochemical
     Smog.  Paper presented before the Division of Environmental
     Chemistry, American Chemical Society Centennial Meeting, New  York,
     New York (April 4-9).

Doyle, George J., Alan C. Lloyd, K.R. Darnall, Arthur M.  Winer and
     James N. Pitts, Jr. 1975.  Gas Phase Kinetic Study of Relative
     Rates of Reaction of Selected Aromatic Compounds With Hydroxyl
     Radicals in an Environmental Chamber.  Environ. Sci. § Technology.
     9_(Mar.) : No.3: 237-241.

Durbin, P.A., T.A. Hecht and G.Z. Whitten 1975.  Mathematical Modeling
     of Simulated Photochemical Smog - Final Report, June 1974-June 1975.
     EPA-650/4-75-026, U. S. Environmental Protection Agency.  (June)
     (Available from NTIS as PB 246 122.)

Egan, B.A. and J.R. Mahoney 1972.  Applications of a Numerical Air
     Pollution Transport Model to Dispersion in the Atmospheric Boundary
     Layer.  J. Appl. Meteorology.  11: 1023-1039.

Fox, D.L. and R.S. Wright 1977.  Photochemical Smog Mechanisms --  HC-
     NO -SCL Systems.  Paper presented at the 4th International Clean
   •  Air Congress.  Tokyo.  (May 16)

Fulle, D.J.  1975.  A Lapse Rate-Wind Shear Classification of Turbulent
     Diffusion.  M.S. Thesis, University of Utah, Department of
     Meteorology.

Gear, C.W. 1971.  Algorithm 407 - DIFSUB for Solution of Ordinary
     Differential Equations.  Commun. ACM. 14_(March): No.3: 185-190.

Colder, Donald  1972.  Relations Among Stability Parameters in the
     Surface Layer.  Boundary-Layer Meteorology.  _3_: 47-58.

Gordon, Sheffield and William A. Mulac 1975.  Reaction of the OH(X2II)
     Radical Produced by the Pulse Radiolysis of Water Vapor.  Chemical
     Kinetics Data for the Upper and Lower Atmosphere.  Int. J. Chem.
     Kinet.  Symp. No.  1: 289-299.

Graedel, T.E. 1976.  Sulfur Dioxide, Sulfate Aerosol, and Urban Ozone.
     Geophys. Res. Lett.  3_(March): No.3: 181-184.
                                    8-4

-------
Graedel, T.E., L.A. Farrow and T.A. Weber 1976.  Kinetic Studies of the
     Photochemistry of the Urban Troposphere.  Atmos. Environ.  10:
     No.12: 1095-1116.

Graham, Richard A., Arthur M. Winer and James N. Pitts, Jr. 1977.
     Temperature Dependence of the Unimolecular Decomposition of
     Pernitric Acid and Its Atmospheric Implications.  Chem. Phys.  Lett.
     SUOct. 15): No.2: 215-220.

Haltiner, G.J. and Frank L. Martin 1957.  Dynamical and Physical
     Meteorology.  New York: McGraw-Hill Book Company: 227.

Hampson, Robert F. Jr. and David Garvin eds. 1975.  Chemical Kinetic
     and Photochemical Data for Modeling Atmospheric Chemistry.  U.S.
     National Bureau of Standards, Technical Note #866, U.S. Department
     of Commer. (June): 1-113.

Hansen, D.A., R. Atkinson and J.N. Pitts, Jr. 1975.  Rate Constants for
     the Reaction of OH Radicals With a Series of Aromatic Hydrocarbons.
     J. Phys. Chem.  79_(Aug. 14): No. 17: 1763-1766.

Hanst, P.L., E.R. Stephens, W.E. Scott and R.C. Doerr 1958.
     Atmospheric Ozone Olefin Reactions.  The Franklin Institute.
     Philadelphia, Pennsylvania.
Harris, G.W. and R.P. Wayne 1975.  Reaction of Hydroxyl Radicals With
     NO, NO,
     610-61-
NO, N02, and SO-.  J. Chem. Soc.  Faraday Trans. I.  71: Pt. 3:
"" '17.
Hecht, Thomas A., John H. Seinfeld and Marcia C. Dodge 1974.  Further
     Development of Generalized Kinetic Mechanism for Photochemical
     Smog.  Environ. Sci. S Technology.  S^(April):  No. 4: 327-339.

Hendry, D.G. 1977.  Private communications from M.C. Dodge, Progress
     Report to M.C. Dodge (May).

Hendry, D*G., A.C. Baldwin, J.R. Barker and D.M. Golden 1978.  Computer
     Modeling of Simulated Photochemical Smog.  EPA-600/3-78-059,
     U.S. Environmental Protection Agency (June).  (Available from NTIS.)

Hidy, G.M. and C.S. Burton 1975.  Atmospheric Aerosol Formation by
     Chemical Reactions.  Chemical Kinetics Data for the Upper and
     Lower Atmosphere.  Int. J. Chem. Kinet.  Symp. No. 1: 509-541.

Hindmarsh, R.D. and G.D. Byrne 1975.  EPISODE, Lawrence Livermore
     Laboratory, UCID-20112.

Howard, Carleton J. 1977.  Kinetics of the Reaction of HO- With NO,.
     J. Chem. Phys.  £7 (Dec. 1): No.11: 5258-5263.

Jeffries, H., D. Fox and R. Kamens 1975.  Outdoor Smog Chamber Studies.
     Effect of Hydrocarbon Reduction on Nitrogen Dioxide.  EPA-650/3-
     75011, U.S. Environmental Protection Agency (June).  (Available
     from NTIS as PB 245 829.)
                                    5-5

-------
Lee, Edward K.C., Roger S. Lewis and Richard G. Miller 1976.   Photo-
     chemistry of Formaldehydes:  Past and Present.   The 12th Informal
     Conference on Photochemistry - Extended Abstracts, National Bur.
     Stand., U.S. Dept. Commer. (June 28-July 1):  C2-1.

Levine, S.Z., W.M. Uselman, W.H. Chan, J.G. Calvert and J.H.  Shaw 1977.
     The Kinetics and Mechanism of the H02-N02 Reactions; the Signi-
     ficance of Peroxynitric Acid Formation in Photochemical  Smog.
     Accepted for publication in Chem. Physics. Lett.

Lloyd, Alan C., Karen R. Darnall, Arthur M. Winer and James N. Pitts,  Jr.
     1976.  Relative Rate Constants for Reaction of the Hydroxyl Radical
     With a Series of Alkanes, Alkenes, and Aromatic Hydrocarbons.
     J. Phys. Chem.  8j3(April 8): No.8: 789-794.

Lloyd, Alan C. and Cindie Tashima 1977.  Chemical Mechanism Development
     for Modeling the Los Angeles Reactive Pollutant Program  (LARPP)
     Data.  Prepared for Coordinating Research Council by Environmental
     Research £ Technology, Inc.  Preliminary Draft (June 16).

Lloyd, Alan C. 1978.  Tropospheric Chemistry of Aldehydes.  Environmental
     Research § Technology, Inc., paper presented at Chemical Kinetic
     Data Needs for Modeling the Lower Troposphere (Reston, Virginia)
     (May).

Martinez, J. Raul, Khan T. Tran, Alan C. Lloyd and George M.  Hidy 1977.
     Development of an Atmospheric Model for Sulfate Formation.
     Environmental Research § Technology, Inc., Document No.  P-1534
     (Sept.).  Prepared for National Science Foundation.

Myrup, L.O.  and A.J. Ranzieri 1976.  A Consistent Scheme for Estimating
     Diffusivities to be Used in Air Quality Models.  California Dept.
     of Transportation Report #CA-DGT-TL-7169-3-76-32.

Niki, H. 1977.  Private communication.

Niki, H., E.E. Daby and B. Weinstock 1972.  Mechanisms of Smog Reactions.
     Adv. Chem. Ser. No. 115: 16-57.

Niki, H. and B. Weinstock 1975.  Environmental Protection Agency
     Seminar.  Washington, D.C. (February).

Niki, H., P. Maker, C. Savage and L. Breitenbach 1976.  IR Fourier-
     Transform Spectroscopic Studies of Atmospheric Reactions.  The 12th
     Informal Conference on Photochemistry - Extended Abstracts, National
     Bur. Stand., U.S. Dept. Commerc.  (June-July): N2-1 - N2-4.

Niki, H., P. Maker, C. Savage and L. Breitenbach 1976a.  Fourier-
     Transform Spectroscopic Studies of Gas-Phase Ozone-Olefin Reactions.
     ACS Centennial Meeting, New York, New York (April 4-9).
                                   8-6

-------
Niki, H., P.O. Maker, C.M. Savage and L.P. Breitenbach 1977a.  Fourier
     Transform IR Spectroscopic Observation of Pernitric Acid Formed
     Via HOO + NCL -> HOONO-.  Chem. Phys. Lett.  4_5(Feb. 1): No.3:
     564-566.

Niki, H. , P.O. Maker, C.M. Savage and L.P. Breitenbach 1977.  Fourier
     Transform IR Spectroscopic Observation of Propylene Ozonide in the
     Gas Phase Reaction of Ozone—Cis-2-Butene—Formaldehyde.
     Chem. Phys. Lett.  4^S(Mar. 1): No.2: 327-330.

O'Brien, J.J. 1970.  A Note on the Vertical Structure of the Eddy
     Exchange Coefficient in the Planetary Boundary Layer.
     J. Atmospheric Sci.  27: 1213-1215.

O'Brien, R.J. 1975.  ACS Meeting, Los Angeles, California, October.

Pate, C.W., R. Atkinson and J.N. Pitts, Jr. 1976.  Rate Constants for
     the Gas Phase Reaction of Peroxyacetyl Nitrate With Selected
     Atmospheric Constituents.  J. Environ. Sci. £, Health.  Part A:
     All; No. 1: 19-31.

Payne, W.A. . L.J. Stief and D.D. Davis 1973.  A Kinetics Study of the
     Reaction of H0~ With S0? and NO.  J. American Chem. Soc.
     95_(Nov. 14): No.23: 7614-7619.

Perry, R.A., R. Atkinson and J.N. Pitts, Jr. 1977.  Kinetics and
     Mechanism of the Gas Phase Reaction of OH Radicals With Aromatic
     Hydrocarbons Over the Temperature Range 296-473 K.  J. Phys. Chem.
     81_(Feb. 24): No.4: 296-304.

Pitts, J.N., Jr. 1975.  Int. Conf. on Env. Monitoring, Las Vegas, Sept.

Pitts, J.N., Jr., A.C. Lloyd and J.L. Sprung, 1975.  Ecology, Energy and
     Economics.  Chem. Britain.  ll_(July~): No.7: 247-256.

Pitts, James N., Jr., Arthur M. Winer, Karen R. Darnall, George J. Doyle
     and John M. McAfee 1975a.  Chemical Consequences of Air Quality
     Standards and of Control Implementation Programs:  Roles of Hydro-
     carbons, Oxides of Nitrogen and Aged Smog in the Production of
     Photochemical Oxidant.   ARB-R-3-017-75-52, Air Resources Board,
     State of California (July).  (Available from NTIS as PB 269 376.)

Pitts, James N., Jr., Arthur M. Winer, Karen R. Darnall, George J. Doyle
     and John M. McAfee 1976.  Chemical Consequences of Air Quality
     Standards and of Control Implementation Programs:  Roles of Hydro-
     carbons, Oxides of Nitrogen and Aged Smog in the Production of
     Photochemical Oxidant.   ARB Contract No. 4-214 (May).  University
     of California.

Pitts, J.N.,  Jr., A.M.  Winer, K.R.  Darnall, A.C. Lloyd and G.J.  Doyle
     1977.   Hydrocarbon Reactivity and the Role of Hydrocarbons, Oxides
     of Nitrogen, and Aged Smog in the Production of Photochemical
     Oxidants.  International Conference on Photochemical Oxidant
     Pollution and Its Control - Proceedings:  Volume II, EPA-600/3-
     77-OOlb, U.S.  Environmental Protection Agency (January): 687-704.
     (Available from NTIS as PB 264 233.)

                                   8-7

-------
Pitts, James N., Jr.,  Karen R.  Darnall, Arthur M.  Winer and John M.  McAfee
     1977a.  Mechanisms of Photochemical Reactions in Urban Air - Volume
     II.   Chamber Studies.  EPA-600/3-77-014b, U.S.  Environmental
     Protection Agency (February).   (Available from NTIS as PB 265 593.)

Peterson, J.T. 1976.  Calculated Actinic Fluxes (290-700 run) For Air
     Pollution Photochemistry Applications, EPA-600/4-76-025.

Rao, K.S. and H.F. Snodgrass 1978.   The Structure of the Nocturnal
     Planetary Boundary Layer,  National Oceanic and Atmospheric
     Administration, ATDL Contribution File No. 78/9.

Sander, Stanley P. and John H.  Seinfeld 1976.  Chemical Kinetics of
     Homogeneous Atmospheric Oxidation of Sulfur Dioxide.  Environ.  Sci.
     5 Technol. 10(Nov.): No.12: 1114-1123.

Schofield, Keith 1973.  Evaluated Chemical Kinetic Rate Constants for
     Various Gas Phase Reactions.  J.  Phys. g Chem.  Ref. Data.  2^No.l;
     25-84.

Sellers,  W.D. 1965.   Physical C_lima_tolqgy_.  Chicago:  University of
     Chicago 152-153.

Shir, C.C. 1973.  A Preliminary Numerical Study of Atmospheric
     Turbulent Flows in the Idealized Planetary Boundary Layer.
     J. Atmospheric Sci.  _30: No.7: 1327-1339.

Sie, B.K.T., R. Simonaitis and J. Heicklen 1976.  The Reaction of OH
     With CO.  Int.  J. Chem. Kinet.  8>: No.l: 85-98.

Simonaitis, R. and Julian Heicklen 1974.  Reaction of H02 With NO and
     N02>  J. Phys.  Chem.  78_(Mar.  28): No.7: 653-657.

Simonaitis, R. and Julian Heicklen 1976.  Reactions of H02 With NO and
     N02 and of OH With NO.  J. Phys.  Chem.  8£(Jan. 1): No.l: 1-7.

Singleton, D.L. and R.G. Cvetanovic 1976.  J. Am. Chem. Soc. 98: 6812.

Thrush, B.A. and co-workers 1978.  Private communication from
     R. Atkinson.

Walter, Theodore A., Joseph J.  Bufalini and Bruce W. Gay, Jr. 1977.
     Mechanism for Olefin - Ozone Reactions.  Environ. Sci. £ Technol.
     l_l_(Apr.) : No.4: 382-386.

Wayne, Lowell, Roy Danchick, Mel Weisburd, Allan Kokin and Arnold Stein
     1971.  Modeling Photochemical Smog on a Computer for Decision-
     Making.  J. Air Pollut. Control Assoc.  21_(June): No.6: 334-340.

Weaver, James, James Meagher and Julian Heicklen 1976.  Photo-Oxidation
     of CH3CHO Vapor at 3130 A.  J. Photochem. 6(Dec.): No.2: 111-126.

Westenberg, A.A. and N. DeHaas 1975.  Rate of the Reaction 0 + S02 + M
     -»• S03 + M.  J.  Chem. Phys.  6^(Dec.  15): No.12: 5411-5415.

-------
Whitbeck, Michael R., Jan W. Bottenheim, Stuart Z.  Levine and
     Jack G. Calvert 1976.  A Kinetic Study of CH_02 and (CH2)_C02
     Radical Reactions by Kinetic Flash Spectroscopy.  The 12tK Informal
     Conference on Photochemistry - Extended Abstracts.  Natl. Bur.
     Stand., U.S. Dept. Commer. (June-July); Kl-1.

Whitten, G.Z. and H. Hogo 1977.  Mathematical Modeling of Simulated
     Photochemical Smog - Final Report.  EPA-600/3-77-011, U.S.
     Environmental Protection Agency, June 1975/June 1976.  (Available
     from NTIS as PB 263 348.)

Wilson, W. et al 1976.  Sulfates in the Atmosphere.  ACPA Paper
     73-30-06.  Presented at 69th Annual Meeting of the APCA, Portland,
     Oregon.

Wood, W.P., A.W. Castleman, Jr. and I.N. Tang 1975.  Mechanisms of
     Aerosol Formation From SCL.  J. Aerosol. Sci.   ^(Sept.): No.5:
     367-374.

Wu, C.H., S.M. Japar and H. Niki 1976.  Relative Reactivities of HO-
     Hydrocarbon Reactions From Smog Reactor Studies.  J. Environ. Sci.
     § Health - Environ. Sci. § Eng.  All: No.2: 191-200.

Wyngaard, J.C. 1975.  Modeling the Planetary Boundary Layer - Extension
     to the Stable Case.  Boundary-Layer Meteorology.  9_: 441-460.

Zafonte, L., P.L. Rieger and J.R. Holmes 1975.  Some Aspects of the
     Atmospheric Chemistry of Nitrous Acid.  Presented at 1975 Pacific
     Conference on Chemistry and Spectroscopy.  Los Angeles, California.
     Oct. 28-30.

Zilitinkevich, S.S. 1972.  On the Determination of the Height of the
     Ekman Boundary Layer.  Boundary-Layer Meteorology.  3: 141-145.
                                    8-9

-------
                                  TECHNICAL REPORT DATA
                           (Please read Instructions on the reverse before completing)
1  REPORT NO.
   FPA-finn/a-79-mfia
                                                          3. RECIPIENT'S ACCESSION-NO.
4. TITLE ANDSUBTITLE
 A LAGRANGIAN PHOTOCHEMICAL AIR QUALITY SIMULATION MODEL
 Adaptation to  the  St.  Louis - RAPS Data Base
 Volume  I.  Model Formulation
                                                          5. REPORT DATE
                                                            June 1979
                                                          6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
 Fred Lurmann, Daniel  Godden Alan C. Lloyd and
 Richard A. Nordsieck
                                                          8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
 Environmental Research and Technology,  Inc,
 2625 Townsgate  Road
 Westlake  Village,  CA  91361
                                                          10. PROGRAM ELEMENT NO.
                                                           1AA603A AA-045 (FY- 79)
                                                          11. CONTRACT/GRANT NO.

                                                           68-02-2765
12. SPONSORING AGENCY NAME AND ADDRESS
 Environmental  Sciences Research Laboratory - RTP,  NC
 Office of  Research  and Development
 U.S. Environmental  Protection Agency
 Research Triangle  Park, NC  27711
                                                          13. TYPE OF REPORT AND PERIOD COVERED
                                                           Final
                                                          14. SPONSORING AGENCY CODE


                                                           EPA/600/09
15. SUPPLEMENTARY NOTES

   Volume II.User's Manual  —  EPA-600/8-79-015b, June  1979
16. ABSTRACT

      A Lagrangian  photochemical air quality  simulation model  has been adapted to
 the St.  Louis,  Missouri/Illinois metropolitan  region  and the Regional Air Pollution
 Study  (RAPS)  aerometric and emissions data base.   This adaptation was performed to
 provide  a means for EPA to independently assess  the  validity of a state-of-the-art
 Lagrangian  photochemical  model.

      Chemical  kinetic oxidation mechanisms involving  hydrocarbons, nitrogen oxides
 and sulfur  oxides  and a vertical diffusion formulation developed by Environmental
 Research and  Technology Inc. for modeling reactive pollutants in the troposphere  are
 described.  Methods for determining model input  parameters are discussed and model
 results  for ozone, nitrogen dioxide, carbon  dioxide,  sulfur dioxide, and sulfate
 are presented for  three summer days in 1976.  In  considering so few simulations,
 no firm  conclusions concerning model reliability are possible, although predicted
 pollutant  concentrations are of reasonable levels.  Most noteworthy for future
 users, the  results suggest that the model may  predict less ozone than is
 actually generated in St. Louis.  Uncertainty  in initial conditions of ozone and
 organic  species is likely responsible for this discrepancy between observed and
 computed values.
17.
                               KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
                                             b.IDENTIFIERS/OPEN ENDED TERMS  C. COSATI Fielc^/GlOUp
 * Air  pollution
 * Hydrocarbons
 * Nitrogen  oxides
 * Sulfur  oxides
 * Ozone
      aptation
                         * Mathematical
                            models
                         * Photochemical
                            reactions
St. Louis, MO
Missouri/Illinois  region
13B
07C
07B
12A
07E
13 DISTRIBUTION STATEMENT

         RELEASE TO PUBLIC
                                             19. SECURITY CLASS (ThisReport)

                                                 IINP.I ASSTFIFD
                                             20. SECURITY CLASS (Thispage)

                                                 llNCi ASSTFTFD
                         21. NO. OF PAGES

                            156.
                                                                        22. PRICE
EPA Form 2220-1 (9-73)

-------