600383035C
A REGIONAL-SCALE  (1000  KM) MODEL OF PHOTOCHEMICAL AIR POLLUTION
           Part  3.   Tests of the Numerical Algorithms
                        Robert G. Lamb
              Meteorology and Assessment Division
            Atmospheric  Sciences Research Laboratory
         Research  Triangle  Park, North Carolina  277!
                       Gerard F. Laniak
                    Program Resources, Inc.
                   Annapolis, Maryland  21401
            ATMOSPHERIC  SCIENCES RESEARCH LABORATORY
               OFFICE OF RESEARCH AND DEVELOPMENT
              U.S.  ENVIRONMENTAL PROTECTION ARENCY
               RESEARCH  TRIANGLE PARK, NC  27711

-------
                                 DISCLAIMER
     This report has been reviewed by the Atmospheric Sciences Research
Laboratory, U.S. Environmental Protection Agency, and approved for publication.
Mention, of trade names or commercial  products does not constitute endorsement
or recommendation for use.

-------
                                  PREFACE






     This is the last in a series of reports describing the development of



the Environmental Protection Agency's Regional  Oxidant Model (ROM).  The



first report described the theoretical  bases of the model, the second



developed a system design for the network of processors that drive the



model, and the present report describes a series of technical  evaluations



of the model's governing equations.  Our objective here is to demonstrate



that the numerical algorithms that constitute the model's predictive



equations are accurate analogies of the differential  equations that describe



the physical and chemical processes that the model is intended to simulate.



We consider this to be a necessary condition for model validity.   A sufficient



condition is that all components of the model jointly — the numerical



algorithms are but a single part — compose a basis for predicting given



features of the species concentrations  that are consistently within given



error limits of the values one would actually observe under the meteorological



and emissions conditions simulated.  Demonstrating that a model satisfies



the sufficient conditions for validity  generally requires comparison of



predictions with observations.  At the  present time preparations are underway



to subject the ROM to tests of this sort.  In this study we make no



comparisons of model predictions with observations.  Rather, our standard



for judging the model's performance are known, exact solutions of the



equations that describe the hypothetical situations that 'we treat.





     This study is a part of the quality assurance program that we have



implemented to achieve and maintain the highest degree of accuracy and



credibility possible.  We have found that in a modeling system as complex





                                    iii

-------
as the ROM, the sources of error are so numerous that meaningful  model

applications are impossible unless  stringent,  comprehensive measures are

taken to erradicate error in every  part of the system.  This is  not to  say

that we believe that errors can be  eliminated  entirely.   We expect that

residual errors will always remain  in much the same way  that sources of

error exist even in instruments of  the highest quality.   In this  context we

view our quality assurance procedures as an effort  to increase the signal-

to-noise ratio of the ROM.  We are  convinced that without such efforts, the

ROM would never achieve the level of reliability necessary to qualify it

for a role in assisting the development of emissions control policies and

air quality management.


                                                 R. G.  Lamb
                                                 April  1985
                                     IV

-------
                                  ABSTRACT






     The regional oxidant model is applied to a series of test problems



whose exact solutions are known.  The predicted concentrations are compared



with the true values to obtain a measure of the accuracy of the numerical



algorithms that comprise the model's governing equations.  Some of the



problems test only the model's chemical kinetics algorithm, others test the



kinetics and transport/diffusion algorithms jointly, and one tests all



three of the models basic algorithms together -- kinetics, transport/diffusion,



and vertical fluxes.





     It is found that the kinetics algorithm produces exact solutions of



the chemical rate equations over the full range of species concentrations



that are likely to be encountered in applications.  A modified version of



the algorithm yields concentrations that are within ± 5% of the correct



values in 1/2 to 1/3 the computer time needed for exact solutions.






     In simulations of the advection of clouds of chemically reactive



compounds, the kinetics and transport algorithms jointly reproduce the correct



shapes and motions of clouds and they predict the peak peak concentration



in the cloud to within 10% of the true value over 48-hour simulation times.





     In applications to continuous finite line sources in steady, spatially



variable flows, the combined algorithms, produced plumes with negligible



pseudo diffusion.  In the case of ozone, the predicted plume centerline



concentration was within 5% of the true value in a plume five grid cells



wide and within 15% of the correct value in a plume two grid cells in width.



Corresponding errors in the CO concentrations were about 50% larger.   In

-------
genera], it was  found  that  ozone  is  among the  species  simulated  best  while




compounds such as nitrous t ;  nitric acid, alkyl  nitrate  and  related  nitrogen



containing species are simulated  poorest.  The predicted  concentrations  of



free radical species are of intermediate  accuracy.   Evidence  was also found



that errors in plume concentration  can  he amplified  when  a  plume crosses  a



second source.  The zone of enhanced error tends  to  be confined  to  the



vicinity of the  second source.  The  accuracies of the  simulated  concentrations



have added significance in  that  the  model employs a  numerical  transport-diffu-



sion scheme that does  not maintain  positive definite concentration.   Negative



concentrations are simply clamped.

-------
                                  CONTENTS
Preface	   iii
Abstract  	     v
Figures	viii
Tables  	xxiv
Acknowledgment	   xxv


     1.  Introduction and Summary 	     1
     2.  Case 1A:  Chemistry Without Transport or Sources   	    13
     3.  Case 2A:  Chemistry with Transport	    57
           Case 2B: Chemistry with transport and vertical mixing   ...    71
     4.  Case 3A:   Chemistry With Transport and Continous Sources  .  .  .   127

References  	264
                                    vn

-------
                                  FIGURES

Number                                                                  Page

 1-1     Schematic illustration of the regional  model  and  the
         network of processors that supply it  information  	     5
 2-1     Temporal  variations in the magnitudes  of the photolytic
         rate constants k\,  k/ and k23 used in  both reactor
         simulations 1A.L and 1A.R (and in all  other experiments
         presented in this report)  	    17

2-2(a)   Results of NO concentration in batch reactor simulations
         1A.L (top) and 1A.R (bottom)	    22

2-2(b)   Same as 2-2(a) but  for N02-  Case 1A.1  (top), 1A.R (bottom).  .    23

2-2(c)   Results for ozone in batch reactor simulations 1A.L (top)
         and 1A.R (bottom)	    24

2-2(d)   Results for olefin  in batch reactor simulations 1A.L (top)
         and 1A.R (bottom)	    25

2-2(e)   Results for paraffin in batch reactor  simulations 1A.L (top)
         and 1A.R (bottom)	    26

2-2(f)   Results for aldehyde in batch reactor  simulations 1A.L (top)
         and 1A.R (bottom)	    27

2-2(g)   Results for aromatic in batch reactor  simulations 1A.L (top)
         and 1A.R (bottom)	    28

2-2(h)   Results for CO in batch reactor simulations 1A.L (top)
         and 1A.R (bottom)	    29

2-2(1)   Results for nitrous acid in batch reactor simulations
         1A.L (top) and 1A.R (bottom)	    30

2-2(j)   Results for nitric  acid in batch reactor simulations
         1A.L (top) and 1A.R (bottom)	    31

2-2(k)   Results for PAN in  batch reactor simulations 1A.L  (top)
         and 1A.R (bottom)	    32

2-2(1)   Results for alkyl nitrate in batch reactor simulations
         1A.L (top) and 1A.R (bottom)	   33

2-2(m)   Results for hydrogen peroxide in batch reactor simulations
         1A.L (top) and 1A.R (bottom)	    34

2-2(n)   Results for atomic oxygen in batch reactor simulations
         1A.L (top) and 1A.R (bottom)	   35

                                    viii

-------
Number                                                                  Page

2-2(o)   Results for nitrate in batch reactor simulations
         1A.L (top) and 1A.R (bottom)	   36

2-2(p)   Results for hydroxyl radical in batch reactor sinulations
         1A.L (top) and 1A.R (bottom)	   37

2-2(q)   Results for pernitric acid in batch reactor sinulations
         1A.L (top) and 1A.R (bottom)	   38

2-2(r)   Results for hydroperoxyl radical in batch reactor sinulations
         1A.L (top) and 1A.R (bottom)	   39

2-2(s)   Results for alkoxyl radical in batch reactor simulations
         1A.L (top) and 1A.R (bottom)	   40

2-2(t)   Results for alkylperoxyl radical in batch reactor simulations
         1A.L (top) and 1A.R (bottom)	   41

2-2(u)   Results for alkoxy radical in batch reactor simulations
         1A.L (top) and 1A.R (bottom)	   42

2-2(v)   Results for peroxyacyl radical  in batch reactor simulations
         1A.L (top) and 1A.R (bottom)	   43

2-2(w)   Results for peroxy radical in batch reactor simulations
         1A.L (top) and 1A.R (bottom)	   44

2-3(a)   Results of NO concentration in batch reactor simulations
         1A.L and 1A.R using a value of .01 for the control
         parameter x in the numerical algorithm.  (Results shown
         in Fig. 2-2(a) use X =.001.)	   45

2-3(b)   Results of N02 concentration in batch reactor simulations
         1A.L and 1A.R using a value of .01 for the control
         parameter x in the numerical algorithm.  (Results shown
         in Fig. 2-2(a) use X =.001.)	   46

2-3(c)   Results of 03 concentration in batch reactor simulations
         1A.L and 1A.R using a value of .01 for the control
         parameter X in the numerical algorithm.  (Results shown
         in Fig. 2-2(a) use X =.001.)	   47

2-3(d)   Results of aldehyde concentration in batch reactor
         simulations 1A.L and 1A.R using a value of .01 for the
         control parameter x in the numerical algorithm.   (Results
         shown in Fig. 2-2(a) use X =.001.)	   48

2-3(e)   Results of PAN  concentration in batch reactor simulations
         1A.L and 1A.R using a value of .01 for the control
         parameter X in the numerical algorithm.  (Results shown
         in Fig. 2-2(a) use X =.001.)	   49

                                     ix

-------
Number                                                                  Page

2-3(f)   Results of alkylperoxyl  concentration  in  batch  reactor
         simulations 1A.L and 1A.R using a  value of .01  for  the
         control parameter X in the numerical  algorithm.   (Results
         shown in Fig. 2-2(a) use X =.001.) .  .  I-	'	    50

2-4(a)'   Results of NO concentration in the hatch  reactor simulations
         1A.L and 1A.R obtained with the modified  numerical  algorithm
         that varies the parameter \ temporally  to effect maximum
         speed and minimum error	    51

2-4(b)   Results of NOg concentration in the batch reactor simulations
         1A.L and 1A.R obtained with the modified  numerical  algorithm
         that varies the parameter \ temporally  to effect maximum
         speed and minimum error	    52

2-4(c)   Results of 03 concentration in the batch  reactor simulations
         1A.L and 1A.R obtained with the modified  numerical  algorithm
         that varies the parameter \ temporally  to effect maximum
         speed and minimum error	    53

2-4(d)   Results of aldehyde concentration  in  the  batch  reactor
         simulations 1A.L and 1A.R obtained with the modified
         numerical algorithm that varies the parameter \ temporally
         to effect maximum speed and minimum error  	
                                                                          54
2-4(e)   Results of PAN concentration in the batch reactor simula-
         tions 1A.L and 1A.R obtained with  the  modified  numerical
         algorithm that varies the parameter \  temporally to effect
         maximum speed and minimum error  	    55

2-4(f)   Results of alkylperoxyl  concentration  in  the batch  reactor
         simulations 1A.L and 1A.R obtained with the modified  numerical
         algorithm that varies the parameter x.  temporally to effect
         maximum speed and minimum error  	    56

3-1      Initial concentration distribution ca(I,J,t0) in the  cloud
         simulated in Case 2A for species a = carbon monoxide.  Also
         shown is ca(I,J,t0 + 48 hr)	       61

3-2      Comparison of simulations by 3 differencing schemes of the
         advection of an ellipsoidal cloud  in a rotating flow.
         Panels a-d display different cross-sections of  the cloud
         (indicated by the upper right corner of each panel) after
         one complete rotation of the cloud, 100 time steps  in the
         case of schemes Q and S, 150 steps in the case  of Z.
         Notation: 0 (circles) = transport  algorithm used in the
         ROM; S (x) = transport scheme of Mahrer and Pielke
         (1978); Z = transport scheme of Zalesak (1979)   	    63

3-2      Continued	    64

3-2      Continued	    65

-------
Number                                                                  Page

 3-2      Concluded	      66

 3-3(a)   Initial  concentration  of  CO in  cross-sections  of.the cloud
          simulated in  experiment  2A.  Diagrams  i"  the 'upper right
          corner of each panel  indicate the  location  of  tne cross-
          section within the cloud.   The  curves  labeled  "chemistry"
          represent the true solution	      74

 3-3(a)   Continued.   Travel time  =  4 hours  	      75

 3-3(a)   Continued.   Travel time  =  8 hours  	      76

 3-3(a)   Continued.   Travel time  =  16 hours   	      77

 3-3(a)   Continued.   Travel time  =  36 hours   	      78

 3-3(a)   Concluded.   Travel time  =  48 hours  	     79

 3-3(b)   Initial  concentration  of  NO in  cross-sections  of  the cloud
          simulated in  experiment  2A.  Diagrams  in  the upper right
          corner of each panel  indicate the  location  of  the  cross-
          section within the cloud.   The  curves  labeled  "chemistry"
          represent the true solution	    80

 3-3(b)   Continued.   Travel time  =  4 hours  	    81

 3-3(b)   Continued.   Travel time  =  8 hours  	    82

 3-3(b)   Continued.   Travel time  =  12 hours   	    83

 3-3(b)   Concluded.   Travel time  =  16 hours   	    84

 3-3(c)   Initial  concentration  of ozone  in  cross-sections  of the cloud
          simulated in  experiment  2A.  Diagrams  in  the upper right
          corner of each panel  indicate the  location  of  the  cros's-
          section  within the cloud.   The  curves  labeled  "chemistry"
          represent the true solution	    85

 3-3(c)   Continued.   Travel time  =  2 hours  	    86

 3-3(c)    Continued.  Travel time  =  12 hours   	    87

 3-3(c)    Continued.   Travel time  =  16 hours   	    88

 3-3(c)    Continued.  Travel time  =  24 hours   	    89

 3-3(c)    Continued.   Travel time  =  36 hours   	    90

 3-3(c)    Concluded.  Travel time  =  48 hours   	    91
                                     XI

-------
Number                                                                    Page

3-3(d)   Initial  concentration  of N02 in  cross-sections  of  the  cloud
         simulated in experiment  2A.   Diagrams  in  the  upper right
         corner of each panel  indicate the  location  of the  cross-.
         section  within the cloud.   The curves  labeled "chemistry"
         represent the true solution	      92

3-3(d)   Continued.  Travel time  = 2  hours	      93

3-3(d)   Continued.  Travel time  = 4  hours	      94

3-3(d)   Continued.  Travel time  = 12 hours  	      95

3-3(d)   Concluded.  Travel time  = 24 hours  	      96

3-3(e)   Initial  concentration  of olefin  in  cross-sections  of the
         cloud simulated in experiment 2A.   Diagrams in  the upper
         right corner of each  panel  indicate the  location of the
         cross-section within  the cloud.   The curves labeled
         "chemistry" represent  the true solution   	      97

3-3(e)   Continued.  Travel time  = 4  hours   	      98

3-3(e)   Continued.  Travel time  = 8  hours.   Insert  in upper
         panel is magnified plot  of major axis  cross-section	      99

3-3(e)   Concluded.  Travel time  = 12 hours.  Insert in  lower panel
         is magnified plot of  minor axis  cross-section	  .      100

 3-4     Initial  CO concentration in  clouds  2B.L  and 2B.R.   Arcing  lines
         labeled  E, M, and C are  48-hour  trajectories  of points originating
         at the edge, midpoint, and center,  respectively, of each cloud  .  . 101

 3-5     Initial  cross-section  of ozone concentration  in cloud  2B.R.
         Diagrams in the upper  right  corner  of each  panel show  the
         location of the cross-section in the cloud.   Curves labeled
         "chemistry" represent  the true solution	102

 3-5     Continued.  Travel time  = 4  hours  (Case  2B.R)	103

 3-5      Continued.  Travel time = 12 hours.  Vertical  mixing  between
          layers  1 and 2 begins at this instant.   (Case  2B.R)	104

 3-5      Continued.  Travel time = 16 hours, 4 hours  after mixing
          (Case 2B.R)	105

 3-5      Continued.  Travel time = 24 hours (Case 2B.R)	106

 3-5      Continued.  Travel time = 36 hours	107

 3-5      Concluded.  Travel time = 48 hours, 36 hours after mixing
          (Case 2B.R)	108

                                    xii

-------
Number                                                                  Page

3-6(a)   Time histories of CO concentration following the center of
         cloud 2B.L, top, and cloud 28.R, bottom.  Curve labeled
         "chemistry" represents the true solution 	  109

3-6(b)   Time histories-of NO concentration following the center of
         cloud 2B.L, top, and cloud 2B.R, bottom.  Curve labeled
         "chemistry" represents the true solution 	  110

3-6(c)   Time histories of ozone concentration following the center
         of cloud 2B.L, top, and cloud 2B.R, bottom.  Curve labeled
         "chemistry" represents the true solution 	 ....  Ill

3-6(d)   Time histories of N0£ concentration following the center
         of cloud 2B.L, top, and cloud 2B.R, bottom.  Curve labeled
         "chemistry" represents the true solution 	  112

3-6(e)   Time histories of olefin concentration following the center
         of cloud 2B.L, top, and cloud 2B.R, bottom.  Curve labeled
         "chemistry" represents the true solution 	  113

3-6(f)   Time histories of peroxy acetyl nitrate concentration
         following the center of cloud 2B.L, top, and cloud 2B.R,
         bottom.  Curve labeled "chemistry" represents the true
         solution	114

3-7(a)   Time histories of CD following the midpoint of cloud 2B.L,
         top, and 2B.R, bottom.  Curves labeled "chemistry" represents
         the true solution	115

3-7(b)   Time histories of NO following the midpoint of cloud 2B.L,
         top, and 2B.R, bottom.  Curves labeled "chemistry" represents
         the true solution	  116

3-7(c)   Time histories of ozone following the midpoint of cloud 2B.L,
         top, and 2B.R, bottom.  Curves labeled "chemistry" represents
         the true solution	117

3-7(d)   Time histories of N02 following the midpoint of cloud 2B.L,
         top, and 2B.R, bottom.  Curves labeled "chemistry" represents
         the true solution	118

3-7(e)   Time histories of olefin following the midpoint of cloud 2B.L,
         top, and 2B.R, bottom.  Curves labeled "chemistry" represents
         the true solution	119

3-7(f)   Time histories of PAN following the midpoint of cloud 2B.L,
         top, and 2B.R, bottom.  Curves labeled "chemistry" represents
         the true solution	120

3-8(a)   Time histories of CO following the edge point of cloud 2B.L,
         top, and 2B.R, bottom.  Curves labeled "chemistry" represents
         the true solution	121

                                    xiii

-------
Number                                                                  Page

3-8(b)  Time histories of NO following the edge  point  of  cloud  2B.L,
        top, and 2B.R, bottom.   Curves labeled  "chemistry"  represents
        the true solution	122

3-8(c).  Time histories of ozone'following th-e edge  point  of. cloud  2B.L,
        top, and 2B.R, bottom.   Curves labeled  "chemistry"  represents
        the true solution	123

3-8(d)  Time histories of N02 following the edge point of cloud 2B.L,
        top, and 2B.R, bottom.   Curves labeled  "chemistry"  represents
        the true solution	124

3-8(e)  Time histories of olefin  following the  edge point of cloud  28.L,
        top, and 2B.R, bottom.   Curves labeled  "chemistry"  represents
        the true solution	   125

3-8(f)  Time histories of PAN following the edge point of cloud 2B.L,
        top, and 2B.R, bottom.   Curves labeled  "chemistry"  represents
        the true solution	126

4-1     Locations and relative  strengths of 4  line  sources  (b,  c,  e,
        and f) simulated in experiment 3A.  Flow speed u  =  .02  radian
        per time step	130

4-2     Isopleths of CO concentration (units =  ppm) at the  end  of  the
        58-hour period simulated in experiment  3A.   Letters b,  c,  e
        and f refer to the sources shown in Figure  4-1	134

4-3     Schematic representation of the continuous  plumes generated
        by sources b, c, e and  f in experiment  3A.   Examples are shown
        of a cross-section and  a Lagrarigian trajectory	136

4-4     Isopleths of ozone concentration at the last hour,  0930 day 3,
        of the line source simulation experiment 3A	137

4-5(a)  Comparison of predicted (solid curve)  and true (dashed)
        CO concentrations in experiment 3A along the cross-section
        Indicated in the insert.  (Travel time  = 7 hrs from sources
        b, c and e)	147

4-5(b)  Same as 4-5(a) except travel time = 34 hrs	148

4-5(c)  Same as 4-5(a) except travel time = 44 hrs	149

4-5(d)  Same as 4-5(a) except travel time = 52 hrs	150

4-6(a)  Comparison of predicted (solid curve)  and true (dashed)
        ozone concentrations in experiment 3A along the cross-section
        indicated in the insert.  (Travel time = 7 hrs from sources
        b, c and e)	,	151
                                    xiv

-------
Number                                                                  Page

4-6(b)  Same as 4-6(a) except travel time = 13 hours	152

4-6(c)  Same as 4-6(a) except travel time = 25 hours	153

4-6(d)i. Same as 4-6(a) except travel time = 34 hours.  .  .	154

4-6(e)  Same as 4-6(a) except travel time = 44 hours	155

4-6(f)  Same as 4-6(a) except travel time = 52 hours	156

4-7(a)  Comparison of predicted (solid curved and true (dashed)
        NO2 concentrations in experiment 3A along the  cross-section
        indicated in the insert.  (Travel time = 7 hrs from sources
        b, c and e)	157

4-7(b)  Same as 4-7(a) except travel time = 13 hours	158

4-7(c)  Same as 4-7(a) except travel time = 25 hours	159

4-7(d)  Same as 4-7(a) except travel time = 44 hours	160

4-7(e)  Same as 4-7(a) except travel time = 52 hours	161

4-8(a)  Comparison of predicted (solid curve) and true (dashed)
        olefin concentrations in experiment 3A along the cross-section
        indicated in the insert.  (Travel time = 7 hrs from sources
        b, c and e)	162

4-8(b)  Same as Figure 4-8(a) except travel time = 13  hours 	  163

4-8(c)  Same as Figure 4-8(a) except travel time = 44  hours 	  164

4-8(d)  Same as Figure 4-8(a) except travel time = 52  hours	165

4-9(a)  Comparison of predicted (solid curve) and true (dashed)
        PAN concentrations in experiment 3A along the  cross-section
        indicated in the insert.  (Travel time = 7 hrs from sources
        b, c and e)	166

4-9(b)  Same as 4-9(a) except travel time = 13 hours	167

4-9(c)  Same as Figure 4-9(a) except travel time = 25  hours.
        Insert shows magnified plot of the predicted and true PAN
        concentration distributions 	 . 	  168

4-9(d)  Same as Figure 4-9(a) except travel time = 34  hours.
        Insert shows magnified plot of the predicted and true PAN
        concentration distributions 	  169

4-9(e)  Same as Figure 4-9(a) except travel time = 44 hours.
        Insert shows magnified plot of the predicted and true PAN
        concentration distributions 	  170

                                     xv

-------
Number                                                                  Page

4-9(f)   Same as Figure 4-9(a)  except  travel  time  = 52 hours	171

4-10(a)  Comparison of predicted (dash-dot)  and  true NO concentration
         (solid curve) along a  Lagrangian  trajectory that  passes
         through the center of  source  e, experiment 3A. .  .•	172

4-10(b)  Comparison of predicted (dash-dot)  and  true NO? concentration
         (solid curve) along a  Lagrangian  trajectory that  passes
         through the center of  source  e, experiment 3A	173

4-10(c)  Comparison of predicted (dash-dot)  and  true ozone concen-
         tration (solid curve)  along a Lagrangian  trajectory  that
         passes through the center of  source e,  experiment 3A 	   174

4-10(d)  Comparison of predicted (dash-dot)  and  true olefin concen-
         tration (solid curve)  along a Lagrangian  trajectory  that
         passes through the center of  source e,  experiment 3A 	   175

4-10(e)  Comparison of predicted (dash-dot)  and  true paraffin
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source  e, experiment 3A.  .  .   176

4-10(f)  Comparison of predicted (dash-dot)  and  true aldehyde
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source  e, experiment 3A.  .  .   177

4-10(g)  Comparison of predicted (dash-dot)  and  true aromatic
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source  e, experiment 3A.  .  .   178

4-10(h)  Comparison of predicted (dash-dot)  and  true carbon monoxide
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source  e, experiment 3A.  .  .   179

4-10(i)  Comparison of predicted (dash-dot)  and  true nitrous  acid
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source  e, experiment 3A.  .  .   180

4-10(j)  Comparison of predicted (dash-dot)  and  true nitric acid
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source  e, experiment 3A.  .  .   181

4-10(k)  Comparison of predicted (dash-dot)  and  true PAN
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source  e, experiment 3A.  .  .   182

4-10(1)  Comparison of predicted (dash-dot)  and  true alky! nitrate
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source  e, experiment 3A.  .  .   183

4-10(m)  Comparison of predicted (dash-dot)  and  true hydrogen
         peroxide concentration (solid curve) along a Lagrangian
         trajectory that passes through the center of source  e,
         experiment 3A	184

                                    xvi

-------
 Number                                                                  Page

 4-10(n)  Comparison of predicted (dash-dot) and true atomic
         oxygen concentration  (solid curve) along a Lagrangian
         trajectory that passes through the center of source e,
         experiment 3A	,	185

 4-10(o)  Comparison of predicted (dash-dot) and true nitrate concen-
         tration  (solid curve) along a Lagrangian trajectory that
         passes through the center of source e, experiment 3A .....  186

 4-10(p)  Comparison of predicted (dash-dot) and true hydroxyl concen-
         tration  (solid curve) along a Lagrangian trajectory that
         passes through the center of source e, experiment 3A 	  187

 4-10(q)  Comparison of predicted (dash-dot) and true hydroperoxyl
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source e, experiment 3A. . .  188

 4-10(r)  Comparison of predicted (dash-dot) and true pernitric acid
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source e, experiment 3A. . .  189

 4-10(s)  Comparison of predicted (dash-dot) and true alkoxyl  radical
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source e, experiment 3A. . .  190

 4-10(t)  Comparison of predicted (dash-dot) and true alkylperoxy
         radical concentration (solid curve) along a Lagrangian
         trajectory that passes through the center of source e,
         experiment 3A	191

4-10(u)  Comparison of predicted (dash-dot) and true alkoxy
         radical concentration (solid curve) along a Lagrangian
         trajectory that passes through the center of source e,
         experiment 3A	  192

4-10(v)  Comparison of predicted (dash-dot) and true peroxyacyl
         radical concentration (solid curve) along a Lagrangian
         trajectory that passes through the center of source e,
         experiment 3A	  193

4-10(w)  Comparison of predicted (dash-dot) and true peroxy
         radical concentration (solid curve) along a Lagrangian
         trajectory that passes through the center of source e,
         experiment 3A	194

4-ll(a)  Comparison of predicted (dash-dot) and true NO concentration
         (solid curve) along a Lagrangian trajectory that passes through
         the outer most  grid cell  of source e, experiment 3A	195

4-ll(b)  Comparison of predicted (dash-dot) and true NO? concentration
         (solid curve) along a Lagrangian trajectory that passes through
         the outer most  grid cell  of source e, experiment 3A	196

                                    xvii

-------
Number                                                                    Page

4-ll(c)  Comparison of predicted (dash-do")  and  true  ozone  concentration
         (solid curve) along a  Lagrangian  trajectory  that passes  through
         the outer most grid cell  of source  e, experiment 3A.  ......   197

4-ll(d)  Comparison of predicted (dash-dot)  and  true  olefin  concentra-
         tion (solid curve) along  a Lagrangian trajectory that  passes
         through the outer most grid cell  of source e,  experiment 3A.  .  .   198

4-ll(e)  Comparison of predicted (dash-dot)  and  true  paraffin  concen-
         tration (solid curve)  along a  Lagranqian  trajectory that passes
         through the outer most grid cell  of source e,  experiment 3A.  .  .   199

4-ll(f)  Comparison of predicted (dash-dot)  and  true  aldehyde  concen-
         tration (solid curve)  along a  Lagrangian  trajectory that passes
         through the outer most grid cell  of source e,  experiment 3A.  .  .   200

4-ll(g)  Comparison of predicted (dash-dot)  and  true  aromatic  concen-
         tration (solid curve)  along a  Lagrangian  trajectory that passes
         through the outer most grid cell  of source e,  experiment 3A.  .  .   201

4-ll(h)  Comparison of predicted (dash-dot)  and  true  carbon  monoxide
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the outer  most  grid cell  of source e,
         experiment 3A	202

4-11(1)  Comparison of predicted (dash-dot)  and  true  nitrous acid
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the outer  most  grid cell  of source e,
         experiment 3A	203

4-ll(j)  Comparison of predicted (dash-dot)  and  true  nitric  acid
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the outer  most  grid cell  of source e,
         experiment 3A	204

4-ll(k)  Comparison of predicted (dash-dot)  and  true  PAN concen-
         tration (solid curve)  along a  Lagrangian  trajectory that
         passes through the outer most  grid  cell of  source  e,
         experiment 3A	205

4-11(1)  Comparison of predicted (dash-dot)  and  true  alky!  nitrate
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the outer  most  grid cell  of source e,
         experiment 3A. .	206

4-ll(m)  Comparison of predicted (dash-dot)  and  true  hydrogen  peroxide
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the outer  most  grid cell  of source e,
         experiment 3A	207
                                   xvm

-------
 Number                                                                     Page

 4-ll(n)   Comparison of predicted  (dash-dot) and true oxygen atom
          concentration (solid curve) along a Lagrangian trajectory
          that passes through the  outer most grid cell of source e,
          experiment 3A	208

 4-ll(o)   Comparison of predicted  (dash-dot) and true nitrate
          concentration (solid curve) along a Lagrangian trajectory
          that passes through the  outer most grid cell of source e,
          experiment 3A	209

 4-ll(p)   Comparison of predicted  (dash-dot) and true hydroxyl
          concentration (solid curve) along a Lagrangian trajectory
          that passes through the  outer most grid cell of source e,
          experiment 3A	210

 4-ll(q)   Comparison of predicted  (dash-dot) and true hydroperoxyl
          radical concentration  (solid curve) along a Lagrangian
          trajectory that passes through the outer most grid cell of
          source e, experiment 3A	  211

 4-ll(r)   Comparison of predicted  (dash-dot) and true pernitric
          acid concentration (solid curve) along a Lagrangian
          trajectory that passes through the outer most grid cell of
          source e, experiment 3A	  212

 4-ll(s)   Comparison of predicted  (dash-dot) and true alkoxyl
          radical concentration  (solid curve) along a Lagrangian
          trajectory that passes through the outer most grid cell of
          source e, experiment 3A	  213

 4-ll(t)   Comparison of predicted  (dash-dot) and true alkylperoxy
          radical concentration  (solid curve) along a Lagrangian
         trajectory that passes through the outer most grid cell of
          source e, experiment 3A	 .....  214

 4-ll(u)  Comparison of predicted  (dash-dot) and true alkoxy radical
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the outer most grid cell of source e,
         experiment 3A	215

4-ll(v)  Comparison of predicted  (dash-dot) and true peroxyacyl
          radical concentration  (solid curve) along a Lagrangian
         trajectory that passes through the outer most grid cell
         of source e,  experiment 3A	216

4-ll(w)  Comparison of predicted  (dash-dot) and true peroxy
         radical concentration  (solid curve) along a Lagrangian
         trajectory that passes through the outer most grid cell
         of source e,  experiment 3A	217

4-12(a)  Comparison of predicted  (dash-dot) and true NO concentration
          (solid curve) along a Lagrangian trajectory that passes
         through the center of source c, experiment 3A	  218

                                    xix

-------
 Number                                                                    Page

 4-12(b)  Comparison of predicted (dash-dot) and true N02 concentration
         (solid curve) along a Lagrangian trajectory that passes
         through the center of source c, experiment 3A. .... 	  219

 4-12(c)  Comparison of predicted (dash-dot) and true ozone concen-
         tration (solid curve) along a Lagrangian trajectory that
         passes through the center of source c, experiment 3A 	  220

 4-12(d)  Comparison of predicted (dash-dot) and true olefin concen-
         tration (solid curve) along a Lagrangian trajectory that
         passes through the center of source c, experiment 3A 	  221

 4-12(e)  Comparison of predicted (dash-dot) and true paraffin concen-
         tration (solid curve) along a Lagrangian trajectory that
         passes through the center of source c, experiment 3A 	  222

 4-12(f)  Comparison of predicted (dash-dot) and true aldehyde concen-
         tration (solid curve) along a Lagrangian trajectory that
         passes through the center of source c, experiment 3A 	  223

 4-12(g)  Comparison of predicted (dash-dot) and true aromatic concen-
         tration (solid curve) along a Lagrangian trajectory that
         passes through the center of source c, experiment 3A 	  224

 4-12(h)  Comparison of predicted (dash-dot) and true carbon monoxide
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source c, experiment 3A. . .  .  225

•4-12(1)  Comparison of predicted (dash-dot) and true nitrous acid
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source c, experiment 3A. . .  .  226

 4-12(j)  Comparison of predicted (dash-dot) and true nitric acid
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source c, experiment 3A. . .  .  227

 4-12(k)  Comparison of predicted (dash-dot) and true PAN concen-
         tration (solid curve) along a Lagrangian trajectory
         that passes through the center of source c, experiment 3A. . .  .  228

 4-12(1)  Comparison of predicted (dash-dot) and true alky! nitrate
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source c, experiment 3A.  .  .  .  229

 4-12(m)  Comparison of predicted (dash-dot) and true hydrogen peroxide
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source c, experiment 3A.  .  .  .  230

 4-12(n)  Comparison of predicted (dash-dot) and true atomic oxygen
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source c, experiment 3A.  .  .  .  231

                                     xx

-------
Number                                                                    Page

4-12(o)  Comparison of predicted (dash-dot) and true nitrate
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source c, experiment 3A. . . .  232

4-12(p)  Comparison of predicted (dash-dot) and true hydroxyl radical
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the center of source c, experiment 3A. . . .  233

4-12(q)  Comparison of predicted (dash-dot) and true hydroperoxyl
         radical concentration (solid curve) along a Lagrangian
         trajectory that passes through the center of source c,
         experiment 3A	234

4-12(r)  Comparison of predicted (dash-dot) and true pernitric
         acid concentration (solid curve) along a Lagrangian
         trajectory that passes through the center of source c,
         experiment 3A	235

4-12(s)  Comparison of predicted (dash-dot) and true alkoxyl
         radical concentration (solid curve) along a Lagrangian
         trajectory that passes through the center of source c,
         experiment 3A	236

4-12(t)  Comparison of predicted (dash-dot) and true alkylperoxy
         radical concentration (solid curve) along a Lagrangian
         trajectory that passes through the center of source c,
         experiment 3A	237

4-12(u)  Comparison of predicted (dash-dot) and true alkoxy
         radical concentration (solid curve) along a Lagrangian
         trajectory that passes through the center of source c,
         experiment 3A	238

4-12(v)  Comparison of predicted (dash-dot) and peroxyacyl
         radical concentration (solid curve) along a Lagrangian
         trajectory that passes through the center of source c,
         experiment 3A. .	239

4-12(w)  Comparison of predicted (dash-dot) and peroxy radical
         concentration (solid curve) along a Lagrangian
         trajectory that passes through the center of source c,
         experiment 3A	240

4-13(a)  Comparison of predicted (dash-dot) and true NO concentration
         (solid curve) along a Lagrangian trajectory that passes
         through the inner edge of source b, experiment 3A	241

4-13(b)  Comparison of predicted (dash-dot) and true N02 concentration
         (solid curve) along a Lagrangian trajectory that passes
         through the inner edge of source b, experiment 3A	242
                                    xxi

-------
Number
4-13(c)  Comparison of predicted (dash-dot)  and  true  ozone  concen-
         tration (solid curve)  along a  Lagrangian  trajectory  that
         passes through the inner edge  of source b, experiment  3A  .  .  .  .   243

4-13(d)  Comparison of predicted (dash-dot)  and  true  olefin concen-
         tration (solid curve)  along a  Lagrangian  trajectory  that
         passes through the inner edge  of source b, experiment  3A  .  .  .  .   244

4-13(e)  Comparison of predicted (dash-dot)  and  true  paraffin concen-
         tration (solid curve)  along a  Lagrangian  trajectory  that
         passes through the inner edge  of source b, experiment  3A  .  .  .  .   245

4-13(f)  Comparison of predicted (dash-dot)  and  true  aldehyde concen-
         tration (solid curve)  along a  Lagrangian  trajectory  that
         passes through the inner edge  of source b, experiment  3A  .  .  .  .   246

4-13(g)  Comparison of predicted (dash-dot)  and  true  aromatic concen-
         tration (solid curve)  along a  Lagrangian  trajectory  that
         passes through the inner edge  of source b, experiment  3A  .  .  .  .   247

4-13(h)  Comparison of predicted (dash-dot)  and  true  carbon monoxide
         concentration (solid curve) along a Lagrangian  trajectory
         that passes through the inner  edge  of source b,  experiment  3A.  .   248

4-13(i)  Comparison of predicted (dash-dot)  and  true  nitrous  acid
         concentration (solid curve) along a Lagrangian  trajectory
         that passes through the inner  edge  of source b,  experiment  3A.  .   249

4-13(j)  Comparison of predicted (dash-dot)  and  true  nitric acid
         concentration (solid curve) along a Lagrangian  trajectory
         that passes through the inner  edge  of source b,  experiment  3A.  .   250

4-13(k)  Comparison of predicted (dash-dot)  and  true  PAN concen-
         tration (solid curve)  along a  Lagrangian  trajectory  that
         passes through the inner edge  of source b, experiment  3A  .  .  .  .   251

4-13(1)  Comparison of predicted (dash-dot)  and  true  alkyl  nitrate
         concentration (solid curve) along a Lagrangian  trajectory
         that passes through the inner  edge  of source b,  experiment  3A.  .   252

4-13(m)  Comparison of predicted (dash-dot)  and  true  hydrogen peroxide
         concentration (solid curve) along a Lagrangian  trajectory
         that passes through the inner  edge  of source b, experiment  3A.  .   253

4-13(n)  Comparison of predicted (dash-dot)  and true  atomic oxygen
         concentration (solid curve) along a Lagrangian  trajectory
         that passes through the inner  edge  of source b, experiment  3A.  .   254

4-13(o)  Comparison of predicted (dash-dot)  and  true  nitrate  concen-
         tration (solid curve)  along a  Lagrangian  trajectory  that
         passes through the inner edge  of source b,  experiment  3A  .  .  .  .   255

                                    xxii

-------
Number                                                                    Page

4-13(p)  Comparison of predicted (dash-dot) and true hydroxyl  radical
         concentration (solid curve) along a Lagrangian trajectory that
         passes through the inner edge of source b, experiment 3A .  . . .   256

4-13(q)  Comparison of predicted (dash-dot) and trjje hydroperoxyl
         radical concentration (solid curve) along a Lagrangian
         trajectory that passes through the inner edge of source b,
         experiment 3A	257

4-13(r)  Comparison of predicted (dash-dot) and true pernitric acid
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the inner edge of source b, experiment  3A. .   258

4-13(s)  Comparison of predicted (dash-dot) and true alkoxyl  radical
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the inner edge of source b, experiment  3A. .   259

4-13(t)  Comparison of predicted (dash-dot) and true alkylperoxyl
         radical concentration (solid curve) along a Lagrangian
         trajectory that passes through the inner edge of source b,
         experiment 3A	260

4-13(u)  Comparison of predicted (dash-dot) and true alkoxy radical
         concentration (solid curve) along a Lagrangian trajectory
         that passes through the inner edge of source b, experiment  3A. .   261

4-13(v)  Comparison of predicted (dash-dot) and true peroxyacyl
         radical concentration (solid curve) along a Lagrangian
         trajectory that passes through the inner edge of source b,
         experiment 3A	262

4-13(w)  Comparison of predicted (dash-dot) and true peroxy
         radical concentration (solid curve) along a Lagrangian
         trajectory that passes through the inner edge of source b,
         experiment 3A	263
                                   xxi

-------
                                   TABLES
Number                                                                  Page

1-1       Chemical reactions included in the'De^erjian/Sche're
            mechanism and the rate constants assumed for each ....      '8

1-2       Summary of the conditions simulated in each of the five
            groups of experiments performed in this report to test
            the accuracies of the regional  oxidant model's numerical
            algorithm.  In all  experiments,•the photolytic rate
            constants undergo the temporal  variations shown in
            Figure 2-1	     10

1-3       Summary of the results of each of the model tests
            described in Table 1-2	     11

2-1       Initial concentrations of each chemical species in
            the batch reactor simulations 1A.L and 1A.R	     16

2-2       Summary of the computer CPU times (VAX 11/780) required
            by the ROM and Gear algorithms  to perform the 48-hour
            simulations for studies 1A.L and 1A.R	     18

2-3       Comparison of computer times (VAX 11/780) required
            by the variable FRAX and Gear algorithms to perform
            the 48-hour batch reactor simulations for studies
            1A.L and 1A.R	    21

4-1       Base emission rates of species used for line sources
            in experiment 3A.  The emission rates of individual
            source cells are fractions (1/3, 1/2, 2/3 or 1) or
            the values shown here (see Figure 4-1)	   132
                                    xxi v

-------
                              ACKNOWLEDGMENTS






     The authors are indebted  to  Ms.  Barbara Hinton  for  her  incomparable



workmanship and patience in  typing  the  manuscript.
                                    XXV

-------
                                 SECTION 1





                          Introduction and Summary








     Air pollution simulation models are not considered to be ready for formal



applications until after they have been "validated".  Although the term



"validation" is not well defined, the validation process is generally taken



to mean the establishment of a quantitative measure of the inherent error



in a model.  Here the distinction is drawn between the component  of error



that is attributable to errors in the input data and the component that is



due to deficiencies in the "physics" and mathematical  algorithms  that



constitute the model itself.  Ascertaining the extent  to which a  model's



poor performance is attributable to its own internal weaknesses is a task



that is plagued by several formidable problems.





     One of these problems is that the measured concentrations that are used



as the standard for judging the model's accuracy contain errors whose



magnitudes are known only approximately.  A second problem is that grid



models predict averages of concentration over large volumes of space --



volumes of the order of 100 km^ — whereas the measured concentrations



represent samples taken virtually at single points in space.  A third, and



perhaps the most significant, problem is the limitation on predictability.



As we discussed in Sections 6 and 7 of Part 1 and in Section 10 of Part 2,



not even a perfect model working with error-free data could predict the



concentration that one would measure at a given site at a given time.



Models can predict probabilities of given concentration values and expected

-------
concentration levels but not the concentration itself.  This limitation



arises from the character of atmospheric motion,  and its magnitude is



determined in part by the type and density of meteorological data that are



used to prescribe the flow field in the model.





     At the present time there does not exist a rational procedure for



model validation that takes all these sources of  uncertainty into account.



Consequently, we will focus our immediate efforts on demonstrating that the



Regional Oxidant Model (ROM) satisfies certain necessary (but not sufficient)



conditions for validity.  To understand what these necessary conditions



are, think of the model as being composed of three parts: physics, numerical



algorithms, and hypotheses.





     The physics describe the chemical reactions, deposition, transport by



the wind, and all other relevant physical processes.  The physics are



described mathematically by a set of differential equations whose solutions



constitute the model's predictions.  Since closed form solutions of the



differential equations are not known, discrete analogues of these equations



must be constructed that are amenable to computer solution.





     Solutions of the discrete equations are produced by the model's



numerical algorithms.  If these algorithms are not properly chosen or are



ill-conditioned, the solutions they yield can differ significantly from the



corresponding solution of the differential equations that they are supposed



to represent.






    Finally, the model hypotheses include the mathematical  descriptions of



physical processes whose spatial and temporal scales are smaller than the

-------
 resolvable scales of the discrete analogue of the corresponding differential



 equations, such as turbulent transport and concentration fluctuations.



 They also include the hypotheses concerning the probaDilities of the individual



 members of the ensemble of flow fields (see part 1, Chapters 6 and 7; and



 Part 2, Section 10).  Let us digress for a moment on tnis last item since



 it may be unfamiliar to the reader.





     In all long range transport models developed before now, i.e., models



 that treat the fate of species beyond distances of the order of 100 km from



 the source, the "ensemble" of flow fields contained only a single member,



 namely the wind field derived from a given set of data using some pre-selected



 objective analysis or interpolation routine.  We showed in Parts 1 and 2 of



 this report that a given set of discrete meteorological data do not uniquely



 specify the wind field.  Rather, they define a set of fields each of which



 is a possible description of the flow that existed during the time the



 observations were made.  We have adopted the position in our regional model



 that the proper way to approach modeling under these circumstances is to



 assign probabilities to each member of the set of possible flows — the



 probability values reflecting additional empirical, historical or other



 information available about the winds in the given area — and to compute



 the concentrations that the given set of sources would produce in each of



the flow fields that comprise the set of most probable flows.  Assigning



quantitative probabilities to each of the possible flows requires a hypothesis



since no theoretical principle is available for this purpose.  From this



 viewpoint, we see that the conventional modeling approach has adopted the



tacit hypothesis that all  of the possible flows have zero probability



except  one, namely that given by the chosen objective analysis routine.

-------
     Let us say that a model  is "valid"  if it produces  concentration



predictions that are consistently within some given error limits  of the



actual concentration that one would observe under the conditions  simulated.



Under this definition necessary conditions for model  validity are that  each



of its three components individually satisfy specific accuracy criteria.



(Strictly speaking, arbitrarily large errors in one component could be



tolerated if sufficiently large errors of a compensating form existed  in



another part.  However, since the three  model components that we  have



defined are inherently distinct, this situation will  not occur in general.



Thus, for all practical purposes we can  assume that each component must



meet certain accuracy standards as a necessary condition for overall model



validity.)





     The purpose of this report is to demonstrate the accuracy of only  the



numerical alogorithms in the regional oxidant model.   In a future study we



plan to present a rational procedure for model verification that  will  allow



us to assess the performance of the model overall.





     To test the numerical algorithms we will apply the model to a series



of rather elementary problems whose exact solutions are known, and compare



the model's predictions in each case with the true values.  Figure  1-1,



which is taken from Part 2, Section 1, will help give a clearer picture of



the specific part of the regional model  that we will be examining.  The box



labeled CORE represents the set of numerical algorithms that  approximate



the differential equations on which the  regional model  is based.   In Part



1, Section 9, we split the governing different!  1 equations  into  three



distinct parts and we developed numerical algorithms for handling  each part

-------
         a.
         Z
II II
>.
I
a.
^ 4
ec t-

o.
O
t-

z
o <


s°
UJ


E
^ K- ^


3
T T

u
"„ <
< J; »-

in
£E
S5
-. ° <


3 = a
in O
o

UJ
D <

O 1
2°

cc o
< ?
UJ
U

: <
O 1-

ir
» s
Figure 1-1.
Schematic illustration of the regional model  and the
network of processors that supply it information.

-------
separately.  One algorithm treats  the  advection  and  horizontal  diffusion



processes, one algorithm handles the chemistry,  and  the  third  approximates



deposition and all  other physical  processes  that affect  vertical  material



fluxes.  In this report we will  assess the  accuracies  o* these  three



algorithms both separately and  jointly.  The objective is  to show that  the



numerical portion of the regional  model satisfies  the  necessary condition



for overall model validity.





     In order to perform the desired tests,  it  is  necessary first to  adopt



a specific chemical mechanism for  the  model. This is  actually  a part of



the physics which,  for flexibility purposes, we  relegated  to the external



module labeled CHEM in Figure 1-1.  Any mechanism can  be used  as long as it



is structured in a  way that is  compatible with  the interface that links the



module CHEM with the basic module  CORE (see Part 2,  Section  1).  For  testing



purposes we will employ the 23 species/36 reaction mechanism developed  by



Demerjian and Schere (1979).  Details  of this scheme are given  in Table 1-1.



One of our interests is to determine whether the accuracy of the algorithm



that handles the chemical kinetics portion  of the model  equation varies



greatly from species to species.





     The test simulations are performed by  assigning to each  of the parameters



in the model input  file  (MIF) (see Figure 1-1)  values characteristic of the



particular situation that we want  to analyze.  For example,  to perform



tests of the transport and chemistry algorithms jointly, we assign values



to the members of the MIF that will prevent vertical fluxes of material



and that describe the flow fields  and   source emissions in each layer as we



want them.  All together, five case studies are conducted to test various

-------
aspects of the model's numerical  algorithms.   The  conditions  simulated  in



each of the studies is summarized in  Table  1-2,  and  a  brief  summary  of  the



findings is presented in Table 1-3 (detailed  discussions  are  provided  in



the remainder of this report).

-------
Table 1-1.  Chemical  reactions  included in  the Demerjian/Schere
            mechanism and the rate constants  assumed  for each.
Reaction
hv
1.
2.
3.
4.
5.
6.

7.

8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.

23.

0 + 02
o3
°3 +
N03
N03 + N02 +



HO
H02 +
H02
H02 + N02
N02
+ M
+ NO
• N02
+ NO
• H20

HONO

+ NO
NO 2
+ NO
-t- M
HOON02
HO +
HO + N02
HO + NO
HO 2
HO
H02 +
OLEF
OLEF
HONO
+ M
+ M
+ 03
+ o3
H02
+ 0
+ o3
OLEF + HO
PARAF + HO



ALD
•> NO +
* °3 +
-> N02 +
* N03 +
-»• 2N02
-»• 2HONO
hv
•* HO +
(02)
H02 +
•»• HONO
* HO +
* HOONO
-y H02 +
* N02 +
* HON02
* HONO
> HO +
-> H02 +
-»• H202
-»• R02 +
-»• R02 +
> R02 +
->• R0?
hv
-> 0.5RO
0
M
°2
°2



NO

co2
+ 02
N02
2 + M
N02
H20
+ M
+ M
202
°2
+ o2
ALD + H02
ALD + H02
ALD


? + 1.5HO? + l.OCO
Rate Constant*
(units3) •
vari
2.
2.
4.
3.
3.

3
7
8
0
4

X
X
X
X
X

van'

4.
4.
1.
1.
3.
9.
1.
7.
3.
1.
3.
5.
1.
3.
5.


1
4
2
5
3
8
5
4
0
0
7
1
4
1
0


X

X
X

X
X
X

X
X
X
X
X
X

vari
abled
10
10
10
10
10

-5 c
-1
-2
4
-3 c

abled

10

10
10

10
10

2

4
-3 c

3
-2 c
10-3 c

10
10
10
10
10
10


2
3
3
-2
4
3

ab!ed

-------
 Table  1-1, continued
Reaction

24.
25.
26.
27.
28.
29.

30.
31.

32.
33.
34.
35.
36.

ALD
R02
RO
R102 •
RO -
R02

R102


AROM
R202
R20
R202
R102

+ HO
+ NO
+ o2
f N02
(- N02
+ 03

+ NO
PAN

+ HO
+ NO
+ o2
+ 03
+ 03


-» 0.3R102 + 0.7H02 + 0.7CO
-> RO +
-> ALD +
•> PAN
> RON02
* RO +
(02)
+ R02 +
-»• RlOo
(02)
> R202
-> R20 +
•» ALD +
-> R20 +
* R02 +
N0?
H02


202

N02
+ N02

+ 2ALD + CO
N02
H02 + 2CO
202
20 2
Rate Constant
(units3 )
1.4 x 104
1.1 x 104
9.0 x 10-1
8.9 x 103
1.0 x 102
2.0
o
4.0 x 10J
1.4 x 10'1 b
n
2.3 x 104
1.1 x 104
8.9 x 10'1
2.0
2.0
*  Values of rate constants that vary by temperature are evaluated here for
   298°K and 1 atm pressure.

a  Rate constant units are ppm'l min'l unless otherwise noted.

b  Units of rate constant are min'l.

c  Units of rate constant are ppnr2 min'1.

d  Photolysis rate constants are based on data compiled by Demerjian, Schere
   and Peterson (1980) and vary as a function of solar zenith angle.  (See text)

Species definitions:

     RN03       Alky! Nitrate
     H02        Hydroperoxyl Radical
     H04N       Pernitric Acid
     RO         Alkoxyl  Radical
     R02        Alkylperoxy Radical
     R20        Alkoxy Radical
     R102       Peroxyacyl Radical
     R202       Peroxy Radical

-------
      Table 1-2.  Summary of the conditions  simulated  in  each  of  the  five
                  groups of experiments  performed  in this report  to test
                  the accuracies of the  regional oxidant  model's  numerical
                  algorithm.  In all  experiments,  the  photolytic  rate constants
                  undergo the temporal  variations  shown  in  Figure 2-1.
Case
1A.L
1A.R
IB
2A
2B.L
2B.R
Horiz.
transport
No
No
No
Yes
Yes
Yes
Horiz.
di f fusion
No
No
No
No
No
No
Vert.
di f fusion
No
No
Yes**
No
Yes
Yes
Initial Sources
concentrations*
Lean
Rich
Rich
Lean, elliptical
cloud
Lean, el liptical
cloud
Rich, elliptical
No
No
No
No
No
No
Remaps



Initial cloi
shown in Fi
Flow field
initial cloi
shown in Fi
3A
Yes
No
No
                                               cloud
Clean
Yes     Sources and
        flow field are
        shown in Fig. 4-1
*   "lean" and "rich" concentrations are defined in Table 2-1.   "Clean" indicates
    negligible concentrations of all species.

**  In this experiment clean fluid is mixed with the contents of the simulated
    batch reactor to approximate the vertical  diffusion process.
                                       10

-------
Table 1-3.  Summary of the results of each of the model  tests described
            in Table 1-2.
Case
Algorithms Tested
              Results
1A
IB
2A
2B
Chemical kinetics
Chemical kinetics
and vertical flux
jointly

Transport and chemical
kinetics jointly
Transport, chemical
kinetics and vertical
flux jointly
Model predictions identical to true solutions
for all 23 species over a 48-hour simulation.
A modified version of the chemical  kinetics
algorithm, called the variable FRAX algorithm,
which is designed for increased execution speed
yields concentrations that are within 5% of-
the true values for all species over the
48-hour simulation.

Error levels same as in 1A.
Percentage errors in simulated peak concentra-
tion in advected clouds:
                                  Species
                                              Travel  time (hrs)

CO
NO
03
N02
Olefin
4
-10
nil
-7
nil
-8
8
-10
+2
-7
nil
-8
16
-12
nil
-8
nil
-18
24
-13
_
-8
nil
-
36
-15
_
-7
-
-
48
-15
_
-6
-
-
                                  Notes:  Negative value indicates model  under-
                                  estimates true value; - indicates species
                                  concentration negligible.
Percentage errors in simulated peak concentra-
tion in advected clouds:
                                  Species
                                              Travel time (hrs)

CO
NO
03
N02*
01 ef i n
4
-8
nil
-
nil
-7
8
-9
+8
-9
nil
-7
16
-8
+2
-8
nil
-
24
-8
_
-9
nil
-
36
-8
nil
-4
nil
-
48
-8
_
nil
nil
-
                                  * No2 underestimated by 10% at hour 8 when
                                    extreme NOX and HC c
                                    [see Figure 3-6(d)).
                           extreme NOX and HC concentrations simulated
                                  Notes: Negative error indicates underprediction;
                                         - indicates concentration negligible.
                                     11

-------
Table 1-3.  Continued.

Case     Alaorithms  Tested
            Results
3A Chemical kinetics and Percentage
inhomogeneous transport concentrati
errors in predicted centerline
on of
plumes from
sources
of
jointly (equations with various widths.
continuous sources)
Species T


CO


03


N02


Olefin


PAN



ravel
(hrs

7
34
52
13
25
34
7
12
18
3
7
12
7
13
20

time
) units
5
-7
-12
-14
-4
-3
-4
-5
nil
nil
-4
nil
nil
-5
-10
-4

Source vli
=grid eel
3
-17
-18

dth
1 dimensio
2
-22
-25
not available
-13
-9
-6
-15
-14
nil
-12
-10
nil
-8
-23
-10
-14
-11
-12
-15
-20
nil
-12
-15
-20
-6
-21
-12
         Same as above but with
         compound sources
Note:  Negative error signifies underpre-
diction.

Percentage errors in predicted peak con-
centration following passage of a plume
from a source over a second, source.
(First plume is three grid cells wide and
crosses second source after a travel time
of 40 hrs.)

Species   Travel time from     error (%)
          second source (hrs)
CO
4
8
-28
-26
                                                     4
                                                     8
                                 -11
                                 -11
                                    See Section 4 for further details.
                                     12

-------
                                 SECTION 2



Case  1A: Chemistry Without Transport or Sources



      In this test we simulate the concentrations of 23 chemical species in


a batch reactor over a 48 hour period.  The objective is to evaluate the


performance of the algorithm that we developed in Part 1, Section 9, for


solving the chemical kinetics portion of the regional model's governing

equations.



      The chemical kinetics are described by a system of nonlinear, ordinary

differential equations of the form


          5Ya     I    I
          —  =  Z    I  kaij Y1 Y                          (2-1)
          at     1=1  j=i  alj     J


where ya denotes the concentration of species a ; I is the total  number of

species present; and ka-jj is the rate constant of the reaction that

involves production of species a from species i and j, or destruction of

species a through its decomposition or its interaction with another species.

Eq. 2-1 describes concentrations in a chemical reactor where transport and


diffusion processes are insignificant.  Hence, it is identical in form to

the chemical kinetics portion of the regional model equations  (cf Part 1,

Eq. 9-24).  Our interest here is in determining how well the solutions of


the numerical  analogue of (2-1) that we formulated in Part 1, Section 9 for

use in the regional  model compare with the exact solutions of  (2-1).


Although the analytic forms of these solutions are not known, we can obtain


virtually exact approximations of them for any values of t using the numerical

technique developed originally by Gear (1971).  In the test problems that


                                     13

-------
we consider here, we will regard the the approximate solutions  of (2-1)



derived from the Gear routine to be the exact solutions,  and  it is  against



these results that we will assess the accuracy of the.solution  algorithm



that we use in the regional  model.





     At this point one might wonder why we bothered to  develop  a new technique



for solving (2-1) when an accurate method already exists.   The  answer is



that the Gear technique requires too much computer time and memory  to make



it practicable in the regional model, or any model of multicell  dimensions.



We have found in preliminary tests of the regional model  that the computer



time required to solve the chemical kinetics portion of the governing



equations is 20 to 50 times larger than that required to  solve  the  transport,



diffusion and vertical mixing portions of the equations combined.  This  was



a surprising finding, especially since the transport terms are  represented



by a 5-th order differencing scheme which requires considerably more computer



time than conventional lower order approximations.  Thus, the overall



efficiency of the regional model code is determined nearly exclusively by



the efficiency of the chemical kinetics solver, rather than the numerical



schemes used for the transport and diffusion processes.





     If the Gear method were used to handle the chemical  kinetics portion



of the regional model's equations, which encompass some 7500  grid cells,



a 24 hour simulation would require about 1 week of CPU time on  a VAX 11/780



computer.  Using the numerical algorithm that we developed for  handling the



kinetics (i.e., Eq. 2-1), the same simulation would require 10-50 hours of



CPU time depending on the accuracy one requires.  Our scheme  was developed



under the constraint of achieving maximum efficiency.  Our task now  is to



determine how much accuracy we have sacrificed for computational speed.



                                      14

-------
     As we noted earlier, out tests consist of two, 48-hour batch reactor



 simulations.   In one test which we shall call case 1A.L, the reactor is



 initialized with a "lean" mix of NOX and hydrocarbons that produces ozone



 concentrations near the current national air quality standard, namely 120



 ppb, after one simulated day.  This particular test will give an indication



 of how well the numerical algorithm can be expected to perform in "typical"



 simulations.





     The second test, Case 1A.R, begins with a "rich" mixture of NOX and



 hydrocarbons that produces ozone levels of the order of 550 ppb -- a value



 more than double the highest hourly ozone concentrations normally observed



 in the Northeastern United States.  The performance of the algorithm in



 this extreme situation will give an indication of whether the accuracy of



 the algorithm is sensitive to variations in the species concentrations.





     The initial concentration values used for each of the 23 species in



 cases 1A.1 and 1A.R are listed in Table 2-1.  In both cases the simulated



 reactor is irradiated with sunlight.  The amplitude of the radiation varies



 in a diurnal manner such that the three photolytic rate coefficients kj, ky



and k23 acquire the magnitudes shown graphically in Figure 2-1.  Note that



the initial instant t = 0 in the reactor corresponds to midnight in the time



frame of the sunlight variations.





     Results of the two test simulations 1A.L and 1A.R are plotted in Figure



2-2a,...w on pages 22 through 44.  The curves labeled "chemistry" represent



the solutions generated by our numerical algorithm and those labeled "Gear"



are the solutions produced by the Gear routine, which we regard as the



exact solutions.  The results show that the solutions of chemical kinetics



                                     15

-------
Table 2-1.  Initial  concentrations of each chemical species in the
            batch reactor simulations 1A.L and 1A.R.
Species.
NO
N02
°3
OLE
PAR
ALD
ARO
CO
HN02
HN03
PAN
RN03
H202
0
N03
HO
H02
H04N
RO
R02
R20
R102
R202

(Rich Mixture)
Case 1A.R
' (PPM)
0.119
3.91 x 10'2
l.C x lO'14
7.80 x 10-2
9.39 x ID'2
9.71 x 10-2
3.01 x 10-2
1.62
1.0 x ID'14
1.0 x 10'14
1.0 x 10-1^
9.42 x 10'13
1.0 x lO'6
1.0 x 10-12
1.0 x ID'14
8.09 x 10-13
1.0 x lO'14
1.52 x 10-13
1.0 x 10-12
1.0 x 10-14
1.0 x 10-12
2.21 x 10-13
4.05 x 10-13
16
(Lean Mixture)
Case 1A.L
'(PPM)
2.66 x ID'5
7.69 x 10'3
7.64 x 13'2
1.56 x ID"3
9.37 x ID"3
1.06 x 10-2
7.80 x lO-4
3.33 x 10-1
6.60 x ID'5
3.62 x 10'2
3.01 x 10-4
4.40 x 10'5
5.53 x ID"5
1.57 x 10-13
8.84 x 10'5
7.24 x 10-9
9.27 x ID'5
1.76 x 10-4
3.42 x 10-9
1.74 x 10-5
5.16 x 10-H
2.23 x 10-6
5.97 x ID'7


-------
       2     4     6     8     10    12    14     IS    18    20    22   24
I  -
  Figure  2-1.
Temporal  variations in the magnitudes  of the  photolytic
rate constants kj, k; and k23 used in  both  reactor
simulations 1A.L and 1A.R (and in all  other experiments
presented in this report).
                                       17

-------
equations (2-1) given by the ROM (regional  oxidant  model)  algorithm are

virtually identical  to the exact solutions  for all  23 species,  over the

entire 48-hour duration of the simulation,  in  both  test  cases  1A.L  and  1A.R.


   .  We conducted a  third test,  IB,  in  which the  contents  of  the  simulated

reactor in case 1A.R were instantaneously mixed with  an  equal  volume of

clean air at hour 12.  Shocking  the  system  in  this  way would  reveal  whether

the accuracy of the  chemistry algorithm is  sensitive  to  the action  of

external  agents, such as turbulent mixing;  and it would  drive  the species

concentration into a third regime, intermediate between  that  of cases 1A.L

and 1A.R, which would reveal further information  on the  sensitivity of  the

algorithm's accuracy to species  concentrations.  The  results  of this test

were also identical  to the corresponding solutions  derived from the Gear

routine.   (For brevity we will not display  the results of  this  test.)


     We conclude from the three  batch reactor  tests that the  algorithm  that

we developed to handle the chemical  kinetics portion  of  the regional oxidant

model is  highly accurate over the entire range of pollutant concentrations

of concern to us in  applied studies.  Moreover, its accuracy  is unaffected

by external  agents such as turbulent mixing,  source emissions or  other

processes that alter species concentrations.   Our tests  also  showed that

computationally the  algorithm is quite  efficient.   The computer times

required  for each of the tests are summarized  in  Table 2-2.


    Table 2-2.  Summary of the computer CPU times (VAX 11/780) required
                by the ROM and Gear  algorithms to perform  the 48-hour
                simulations for  studies 1A.L and  1A.R.

                                            Case
              Algorithm      |        1A.L          1A.R
ROM
Gear
59.6 (sec)
355.6
129.1
409.2
                                     18

-------
     One reason that the Gear times are so large is that at the beginning



of each simulated period this routine computes initial estimates of time



derivatives of various orders for each species.  In the regional model  the



period of each chemistry simulation is only 5 minutes between reinitializa-



tions, because vertical exchange processes must be allowed to operate on tne



concentrations in each of the model's three layers at least this often.



Therefore, in the ROM environment, the Gear routine's initialization require-



ments create a large computational overhead.





     One might argue that the level of precision exhibited by the ROM chemistry



algorithm is unnecessarily high because errors exist in both the physics



and hypotheses portions of the model and in all the input data.  In view of



this it would be practicable to sacrifice some of the algorithm's accuracy



for a further increase in computation speed.  This trade-off can be achieved



easily by increasing the parameter \ that controls the algorithm's integration



time step size (see page 192 of Part 1).  In the batch reactor simulations



1A.L and 1A.R performed above, \ had the value 0.001.  If we increase it by



a factor of 10 to x =0.01, the execution time requirements drop from 59.6 sec



to 35.4 sec for case 1A.L, and from 129.1 to 43 sec for case 1A.R.  The



accuracy penalty that is paid for this increase in speed can be seen in



Figure 2.3a-f, pages 45-50, where we have plotted a few of the best and a few



of the worst results obtained for both cases 1A.L and 1A.R using x = .01.  In



the case of 1A.L, which represents concentrations typical of those that we



would encounter in actual applications, the errors in the predicted concentra-



tions are no larger than 10% for any of the 23 species.  In fact for most of



the species, including all those not shown, the largest error is only a few






                                     19

-------
percent over the entire 48-hour simulations.   However,  as  the  bottom panels



of Figure 2-3 reveal, performance in Case 1A.R is  significantly  poorer.   In



the case of ozone, the predicted concentration is  ove'"  50% too high  on  the



second day of the simulation and for some of  the  other  species,  such as



shown in Figure 2-3f, the errors are still  larger.   Although Case  1A.R



represents conditions much more severe than any that we are likely to



encounter in applications, the magnitude of the errors  revealed  in this  case



show an enhanced sensitivity of the numerical  algorithm's  accuracy to



species concentration when the control parameter  x has  the value .01.



Therefore, in order to realize a high speed algorithm that would not



systematically generate larger errrors in regions  where concentrations  are



high, one of our colleagues, Kenneth Schere,  developed  a modified  version



of the kinetics algorithm in which the parameter  x has  a nominal value  of



.01 but switches to the smaller value .001 wherever the magnitude  of the



local time rate of change of NO concentration exceeds a given  value: 0.5%



sec~l.  It turns out that the temporal behavior of NO is a good  indicator



of conditions in which the accuracy of the algorithm is critical.   We call



this modified algorithm the variable FRAX or variable x algorithm.  Figures



2-4(a)-(f), pages 51-56, show results of new model  runs for cases 1A.L and



1A.R for the same species plotted in Figure 2-3(a)-(f).  The results are



greatly improved and are considered by us to be of sufficiently high quality



to justify use of the variable FRAX algorithm in all applications of the



regional oxidant model (ROM).   (All results presented in this  report utilize



the X =.001 version.)  Table 2-3 compares the computer  time requirements of



the Gear and the variable FRAX algorithms for the two,  48-hour simulations,



Case 1A.L and 1A.R.





                                     20

-------
     Table 2-3.  Comparison of computer times  (VAX 11/780)  required by the
                 variable FRAX and Gear algorithms to perform the  48-hour
                 batch reactor simulations for studies 1A.L and  1A.R.
Algorithm LA.
ROM
Gear
'variable FRAX)
40.
355.
L
1
6
Case
(sec)
1A.R
98.
409.

0
2
     In conclusion, the tests that  we have presented  here  show that  the

algorithm that we use in the regional model  to solve  the chemical  kinetics

portion of the governing equations  produces  solutions with negligible  errors

in a computation time only 1/3 to 1/6 that required  by the highly  accurate

Gear method.  And it provides solutions  with accuracies commensurate with

the error levels in other parts of  the model  and  in  the input  data in  a

computation time 1/10 that required by the Gear routine.
                                     21

-------
            HEM : 001
                                     IA.L
            8   12   16   20   24
            B   12   16   20   2+    4    S    12   16   20   24
               DAY 1
                        DAY 2
                  TIME (HOUR  OF DAY)
Figure 2-2(a).
Results of NO concentration  in batch reactor
simulations IA.L 'top) and 1A.R (bottom).
                               22

-------
                                    IA.L
        4   R   12  16   20   24
                 TIME (HOUR OF DAY)
Figure  2-2(b).
Same  as  2-2(a) but for  N02.
1A.R  (bottom).
                                          Case IA.L (top),
                             23

-------
      O


      X
      Q-  J-
      D_
      O
CHEW : 001


GEAR
                           IA.L
                ,,,,,,!,,,!,,,,,,,,, I .,,,,,,,,,,,,,,
               4    B   12   16   20   24    4   B   12    16   20   24
                   B    12   16   20   24    4    8   1216
                     DAY 1
                                                DAY 2
                                                          20   24
                         TIME (HOUR  OF DAY)
Figure 2-2(c).   Results  for  ozone in batch reactor simulations  IA.L (top)

                 and  1A.R (bottom).
                                      24

-------
                                           IA.L
                    DAY 1
                                                      20   24
                                             DAY 2
                       TIME (HOUR OF DAY)
Figure  2-2(d).
Results  for olefin in batch  reactor simulations  IA.L  (top)
and 1A.R (bottom).
                                   25

-------
                  CHEW : 001
       o
       ~ZL

       O

       0
                                           IA.L
       <  ip-
       Q-  t
               4    8   12   16   20   24   4   8   12   16
                                                           24
                  CHEM : 001
                                            IA.R
                        TIME (HOUR OF DAY)
Figure  2-2(e).
Results for  paraffin  in batch reactor simulations  IA.L  (top)

and IA.R (bottom).
                                    26

-------
                                            IA.L
               4    8    12   16   20  24   4S   12   16   20   24
                                           IA.R
                                          8   12   -6   2024
                        IME (HOUR OF DAY)
Figure  2-2(f).  Results  for aldehyde in batch  reactor simulations  IA.L (top)
               and IA.R (bottom).
                                   27

-------
                 CHEW : 001
                 GEAR
                                            IA.L
              4   8   12   16   20   24
                 CHEM : 001
                 GEAR
                                            S   12   16   20   24
                                            IA.R
                              20
                     DAY 1
                                12
                              DAY 2
                                                    16
                                                        20
                                                            24
                        TIME (HOUR OF DAY)
Figure 2-2(g).
Results for aromatic in batch reactor simulations IA.L (top)
and IA.R (bottom).
                                     28

-------
      o

      ><
  CHEM : 001


  GEAR
IA.L
      £ I
      s  f
      2  L
      o  t
      rr *r
      I—  r
      z  C
      UJ  H-

      £  t

      8'r
      O

      O
           *
      Q_

      CL
      O
      UJ

      O
      2

      O
      o
      o
      o
                   S    12   16   20   2+
                                             8   12    16   20   24
                  CHEM


                  GEAR,
                             IA.R
         Ql I 1 I 1 I 1 I I 1 I I I 1  l.i I I i I I I i , I
                            i ... i ... i  • , • i.
                    DAY 1
                            8    12   16   20


                              DAY 2
                        TIME (HOUR  OF DAY)
Figure 2-2(h).
Results for CO in batch reactor simulations  IA.L  (top)

and IA.R (bottom).
                                     29

-------
                                          IA.L
                      TIME (HOUR OF DAY)
Figure  2-2(i).  Results  for nitrous  acid in batch reactor simulations
               IA.L (top) and 1A.R  (bottom).
                                   30

-------
                                           IA.L
      Ld
      O
      Z
      O
      o


      o
          I-
      O


      X
     D_
     Q_
     S
     LJ
     O
     z:
     o
     CJ

     ro
     O
 CHEW : 001

 GEAR
IA.R
                                               I i i  i I
              4   8   12   16   20   2+   4   S   12    16   20   24
  8    12   18   20   24

    DAY 1
8   12   16   20

  DAY 2
                                                            24


                                                            I
                       TIME  (HOUR OF  DAY)
Figure 2-2(j).
Results for nitric acid in batch  reactor simulations
IA.L (top) and IA.R (bottom).
                                     31

-------
                  CHtM : 001
       CM

       O



       X
       Q.

       CL




       §3
       ce
       o
       z
       o
       o
       <  (r
       CL
                                            IA.L
r • •
~
4 8 12
	 GEAR / >
16 20 24
                                            8    12
                                            IA.R
                       12   16   20   24-


                     DAY 1          ,
 12    16   20   24


DAY 2
                        TIME  (HOUR OF DAY)
Figure  2-2(k).   Results for PAN in batch reactor  simulations IA.L (top)

                and  IA.R (bottom).
                                    32

-------
      in
      O
      D_
      Q_
      O

      <
      cr
      LJ
      O
      •z.
      O
      O

      ro
      O
      -z.
      cr
          F-
         CHEW : 001

        -SeW	
                                   IA.L
3r-
 p
                                             IA.R
                     DAY 1
                                               DAY 2
                        TiME (HOUR OF DAY)
               4    8   12   16   20   24   4    8   12   16   20   24
              4    8    12   16   20   24   4    a   12   16   20   24
Figure 2-2(1).  Results for alkyl nitrate in batch reactor simulations
                IA.L (top) and 1A.R (bottom).
                                     33

-------
                                    4    4    8   12   16   23   24
              4   8   12   16   20   24
                     DAY 1
                            8    12   '.6   20   24

                              DAY 2
                        TIME (HOUR  OF DAY)
Figure 2-2(m).
Results for hydrogen peroxide in batch reactor simulations
1A.L (top) and 1A.R (bottom).
                                     34

-------
                   CHEW : 001

                   GEAR
        O
                                             IA.L
                            16   20   24    4    B   12    16   20
                   CHEM : 001
                                            1A.R
                   8   12    16   20   24
                         IME (HOUR  OF DAY)
Figure 2-2(n).
Results  for  atomic oxygen in batch reactor simulations
IA.L (top) and  1A.R  (bottom).
                                    35

-------
                 &   12   16   20   24
                    DAY 1
                           B   12   16   20   24

                             DAY 2
                       TIME (HOUR OF DAY)
Figure 2-2(o).
Results for nitrate  in  batch  reactor  simulations
IA.L (top) and IA.R  (bottom).
                                     36

-------
                  CHEW : 001
                     DAY 1
                                            IA.L
                                              DAY 2
                        TIME (HOUR OF  DAY)
Figure  2-2(p).
Results  for  hydroxyl radical in batch reactor  simulations
IA.L (top) and  IA.R  (bottom).
                                    37

-------
                  CHEM : 001

                  GEAR
                                           IA.L
                    DAY 1
                                             DAY 2
                        TIME (HOUR OF DAY)
Figure  2-2(q).
Results  for  pernitric acid in batch reactor  simulations
IA.L (top) and  1A.R  (bottom).
                                    38

-------
in
O

X

D_
r\

•ZL
g

£c
       LJ
       O
       -z.
       O
       O

       CM
       O
       o
                                           IA.L

4
S
12
16
20
24
4 8
52
16 20
24
                  CHEM : 001

                  GEAR
                                           1A.R
                        TIME (HOUR OF DAY)
Figure 2-2(r).
         Results for hydroperoxyl radical in batch reactor simulations
         IA.L (top)  and 1A.R  (bottom).
                                    39

-------
         4r-
o
X
                 CHEM : 001

                 G£AR
                                           IA.L
                                               12   16   20   24
                    DAY 1
                                             DAY 2
                       TIME  (HOUR OF DAY)
Figure 2-2(s).
          Results for alkoxyl  radical  in batch reactor simulations
          IA.L (top)  and IA.R  (bottom).
                                     40

-------
       lO
       O

       X
           u
CHEM : 001

GEAR
IA.L
                        TIME (HOUR OF DAY}
Figure 2-2(t).  Results for alkylperoxyl  radical in batch reactor simulations
               IA.L  (top) and 1A.R (bottom).
                                    41

-------
          6r-
                  CHEM : 001

                  GEAR
                                            IA.L
                                                         20
                                            IA.R
                                        4    8   12   16   2:    24

                                              DA.Y 2
        TIM
                                  NP
                                        DAY)
Figure 2-2(u).
Results for alkoxy  radical  in  batch  reactor  simulations
IA.L (top) and IA.R (bottom).
                                     42

-------
                                            IA.L
                   8   12   16   20   24    4    5   ;?   16
                        TIME (HOUR OF DAY)
Figure  2-2(v).
Results  for  peroxyacyl radical  in  batch  reactor simulations
IA.L (top) and  1A.R (bottom).
                                    43

-------
                 CHEM : 001

                 GEAR
                                           JA.L
                       TIME (HOUR OF  DAY)
Figure  2-2(w).  Results for peroxy  radical in batch reactor simulations
               1A.L (top) and 1A.R (bottom).
                                    44

-------
            CHEU : 01

            GEAR
      IA.L
(RELAXED
             8   12   16"  20   24    4   a
                                            16   2D   24
  o

  X
 D_
 D_
 O
 OH
 LJ
 O
 ~z.
 O
 O
     IA.R

(RELAXED  A)
    11-
               DAY 1
        12   IF

      DAY 2
                                               20
                  TIME (HOUR OF DAY)
Figure 2-3(a).  Results of NO concentration  in batch reactor simulations
               IA.L and IA.R using a value  of .01 for the control
               parameter \ in the numerical algorithm.  (Results shown
               in  Fig. 2-2(a) use X =.001.)
                               45

-------
                                       IA.L
                                 (RELAXED
                      16 ~ 20   24 ~  "4   8   12   16
   -* 2|
   O
   Q_
   Q.
   o:

   •z.
   Ld
   O
   Z
   O
   O

   (N
   O
              CHEM : 01

              GEAR
                       IA.R

                  (RELAXED
                  12

                 DAY 1
                      16
                           12

                          DAY 2
                               16
                                       24
                    TIME (HOUR OF DAY)
Figure  2-3(b).
Results of N02 concentration in  batch reactor simulations
IA.L and IA.R using  a value of .01 for the control
parameter X in the numerical algorithm.  (Results shown
in Fig. 2-2(a) use X =.001.)
                               46

-------
    o

    X
    D_
    CL
CHEM : 01

GEAR
     IA.L
(RELAXED  A)
       Ql I I I I i I i I i i i 1 i i i I i i t I i
           4   8   12   16  20   24
                                   4   a   12   IS   20   24
                                           IA.R
                                      (RELAXED A)
                    TIME (HOUR OF DAY)
Figure 2-3(c).   Results of  03 concentration in batch reactor simulations
               IA.L and IA.R using a  value of .01 for the  control
               parameter \ in the numerical algorithm.  (Results shown
               in Fig. 2-2(a) use \ =.001.)
                              47

-------
   o
   X
   Q_
   Q_
CHEW : Qt

GEAR
      IA.L
CRELAXED  A)
                                  *    8   12   16   2D   24
                                      IA.R
                                 (RELAXED
                                                 20   24
                DAY 1
                                        DAY 2
                   TIME (HOUR OF DAY)
Figure 2-3(d).
 Results of aldehyde concentration in batch reactor
 simulations IA.L and IA.R using a value of .01 for the
 control parameter \ in the numerical algorithm.  (Results
 shown in Fig.  2-2(a) use \ =.001.)
                              48

-------
   s  r
   X
   Q_
   Q_
   o


   I
   I—
   IZ
   LU
   O
   z:
   o
   o
   <
   Q_
CHEM : Dl


G£AR
     1A.L

(RELAXED
           4    8   12   16   20   24   4   8   12   16
                                       IA.R

                                 (RELAXED A)
           4    B   12   16   20   24    4    8   \2   16
                 DAY 1
                                         DAY 2
                                                      24
                    TIME (HOUR OF DAY)
Figure 2-3(e).
 Results of PAN   concentration in batch reactor simulations

 IA.L and IA.R using a value  of  .01 for the control

 parameter X in  the numerical algorithm.  (Results shown

 in  Fig. 2-2(a)  use x =.001.)
                               49

-------
   trt

   O


   X
   D_
   Q.
   g

   %
   LJ
   o
   z:
   o
   o

   
-------
     O
                                         IA.L
                                   (VARIABLE  A)
     ^  2,
     O
     Q_
     D_
            4   S   12   16   20
                CHEMISTRY

                GEAR
24    4   3   12   16   .13



        IA.R

   (VARIABLE  PO
                                                        24
                        16
                   DAY 1
                                            i2

                                           DAY 2
                      TIME (HOUR OF DAY)
Figure  2-4(a).  Results of NO concentration in the batch reactor simulations
               IA.L  and  IA.R obtained with the modified numerical  algorithm
               that  varies the parameter X. temporally to effect maximum
               speed and minimum error.
                                 51

-------
     D_
     D_
CHEVISTOY

GEAR
                                        IA.L
                                  (VARIABLE  A)
                                         IA.R

                                   (VARIABLE A)
                  DAY 1
                            12

                           DAY 2
                                                       24
                     TIME  (HOUR OF DAY)
Figure 2-4(b).  Results of N02 concentration  in the batch  reactor simulations
               IA.L and IA.R obtained with the modified numerical algorithm
               that varies the parameter x temporally to  effect maximum
               speed and minimum error.
                                52

-------
      ,-  2r-

      £   t
      X   i
      S:   r
  CHEMiSTOY

  GEAR
      IA.L
(VARIABLE  A)
                              0   24-   4   S   12   16   20   24
                                          IA.R
                              V     (VARIABLE
             4    S   12    16  20  2*
                                     4   8   12   16   2Q   24
                      TIME (HOUR OP DAY)
Figure  2-4(c).
Results  of  03 concentration  in the batch reactor  simulations
IA.L and IA.R obtained with  the modified numerical algorithm
that varies the parameter  x  temporally to effect  maximum
speed and minimum error.

                  53

-------
     x  I
               CHEMISTRY

               GEAR
                         IA.L
                   (VARIABLE
            4   8   12   16   20   2+   4   3   12   16   20
                                        IA.R
                                   (VARIABLE
                                                         24
                   DAY 1
                            DAY 2
                      TIME (HOUR OF DAY)
Figure  2-4(d).
Results  of aldehyde concentration in the batch  reactor
simulations  IA.L and IA.R  obtained with the  modified
numerical algorithm that varies the parameter \ temporally
to effect maximum speed and minimum error.
                                 54

-------
    CM
    o
    Q.
    Q_
o
*
               CHEMlSTTTf

               GEAR
                                   IA.L
                              (VARIABLE  /U
            *    8   12   16   20   24    4    8   -,2   16
       3F
               GEAR
                     \
                      \
                   \
      IA.R
(VARIABLE
                    \
                         \\
LU
O
z
0
o
•z.
<
Q_

<-r
t
£-
p
l
\
i—
P y
oFi..i.../
\
\
\
/ x
' \
i ,1 i i i i i 1 i i • 1 i i i i i ! r>-fc ^ , j i i i , i , i , . , , ; , i , i , ,
                                                   20   24
                     TIME (HOUR OF DAY
Figure 2-4(e).  Results of PAN concentration in the batch reactor simula-
               tions  IA.L and IA.R obtained with the modified numerical
               algorithm that varies the parameter x temporally to effect
               maximum speed and minimum error.
                                 55

-------
irt  °i

x  t	
               CHEUISTW

               GEAR
                         IA.L
                   (VARIABLE
                                    .   IA.R
                                  (VARIABLE
                     TIME (HOUR OF DAY)
Figure 2-4(f).
Results of alkylperoxyl concentration in the batch reactor
simulations IA.L and IA.R obtained with the modified numerical
algorithm that varies the parameter \ temporally to effect
maximum speed and minimum error.

                  56

-------
                                 SECTION 3


Case 2A:  Chemistry with transport

     In the previous experiments, 1A.L and 1A.R, we examined the performance

of the algorithm that solves only the chemical  kinetics portion of the

regional model equations.  In experiment 2A we  will advance one step in

complexity and look at how the algorithm that handles the kinetics and the

algorithm that handles the transport perform when they are coupled together.

In particular, we will consider a combined transport/chemistry problem

characterized by

           dca      5ca      5ca      I   I
           ...  + u ...  + v ...   =  z   z  kaijc-jCj           (3-1)
           at       ax        ay     i=l j=l

where (u, v) is the horizontal wind described by given functions u(x,y,t)

and v(x,y,t) of space and time.  Eq. 3-1 is the form that the regional

model's equations acquire when the horizontal eddy diffusivity K^ and  the

flux parameters that link the three layers of the model vertically are set

to zero.

     The exact solutions of (3-1) can be expressed in terms of the solutions

of Eq.  2-1,  which we evaluated numerically in the previous experiment, by

performing the following transformations of the space coordinates (x,y):

                          t
           I = x - x0 -  / u(x',yI,t')dt' = x - x0(t)               (3-2)
                         to
                          t
           TI = y • y0 -  J v(x',yl,t')dtl = y - y0(t)               (3-3)
                         to
                                     57

-------
where  (x0,y0) is an arbitrary point and  (x',y') is the the point  (x,y) where


 (£,n)  =  (0,0) at time t'.  That is, x1 = x0(t'), y' = y0(t'); and


 (*o> yo)  1S the origin  in  (x,y) space of the  (E,,ri) coordinate system.
     Making use of the chain rule of differentiation, namely
          a   _ at  5     d£  5     an  5

          at   at  at    at  a?    at  an
                                                              (3-4)
          8  _ aC  a     9r|  9

          9x   5x  ac    ax  an
                                                              (3-5)
          a   _ ac  a     a-n  a

          ay   ay  as,    ay  an
                                                              (3-6)
We can express  (3-1) in the form
         at
                 M     a?      as
                 (._ + u __  + v _.)
                 at     ax      ay
                                     ac
k._   + u  —   +  v  —)  —   =
 at       ax       ay   an
                                           I    I
                                           S    Z  kaijc-jc.j      (3-7)
                                          1=1  j=l
After evaluating the derivatives of


(using 3-1 and 3-3), we get
                                       and n  that appear in this equation
          ac
          --
          at
                  - u(x0,y0,t) + u(s + x0,
                 '.- v(x0,y0,t)
                                                       ac
                                                      ac
                                                        a
                                       x0, n + y0»t)] —  =
                                                       ay
                                                                   (3-8)
                                 I    i
where the concentrations are evaluated at (£ + x0, n + ?o) in  (x,y) space
                                     58

-------
We see at once from  (3-8) that at the origin of the ( £, TI) coordinate
system, which is the moving point (x0 (t), y0(t)) in (x,y) space, the
solution of  (3-1) is just the solution of the batch reactor equation (2-1)
that we considered earlier.  This equivalence would not exist were the
horizontal diffusivity K^ nonzero.

     Thus, along any fluid particle trajectory, i.e., along any space-time
path [x0(t), y0(t),t] described by (3-2,3) for given initial  point (x0,y0),
we can obtain the solution of (3-1) with the same precision that we
found solutions of (2-1) in the previous experiments.  And we can compare
these solutions with those given by the transport/chemistry portion of the
regional model along the same paths to assess the joint accuracy of the
algorithms used in the model to describe these two processes.

     The problem that we will consider in this experiment is  that of an
ellipsoidal -shaped cloud of chemical  species transported by a stationary
flow field whose velocity components  (u,v) at any point (x,y) are given by
          u = (y - y0)«                                      (3-9a)
          v = -(x - x0)u>                                     (3-9b)

where to = .02 radian/At and At = 30 min is the time step used in the
transport algorithm.  These expressions describe a fluid in solid body
rotation of angular speed u about the point (x0,y0).  We have chosen this
particular flow field not because it provides a definitive test of the
transport algorithm, but rather because it is a popular test of transport
algorithms with which many modeling investigators are familiar.  Our choice
of an elliptical rather than a circular cloud is motivated by our finding
in Part 1, Section 9 that most finite difference algorithms contain
                                     59

-------
significant sources of error that are not activated unless the transported



field, in this case the species concentrations,  deviates  from forms  that



possess axial symmetry.  We want to excite, all  possible sources of error in



the transport algorithm so that we can see whether the disturbances  that



these sources create are amplified by the kinetics.  Since we found  in  the



first experiment that the errors generated by the kinetics algorithm are



negligible when x = 0.001, we will use this value in the  present experiments.



It follows that any errors that arise in the joint simulation of transport



and chemistry have their origin in the transport algorithm.   One of  our



main interests here is to see whether the chemical kinetics  amplify  errors



generated by the transport algorithm and if so whether the coupling  between



the kinetics and transport processes provides enough positive feedback  for



errors to grow unboundedly.





     The initial concentration of species a (= 1....23) at grid point (I,J)



in the test cloud will be taken to be





          ca(I,J,t0) = f(I,J)Ca(t0)                      (3-10)





where Ca(t0) is the concentration of species a at the center of the cloud



at the initial instant t0, and f is a fraction such that  0 < f < 1.  The cloud



center concentrations Ca are the "lean mixture" values listed in Table 2.1,



page 16.  Figure 3-1 shows the initial concentration distribution of CO in



the cloud at the initial instant t0.  In this experiment the flow field remains



steady and the cloud is transported for a simulated period of 48 hours.



During this time there is no vertical mixing and the photolytic rate constants



vary in the diurnal manner shown in Figure 2-1, page 17.   Figure 3-1 also



shows the simulated distribution of CO in the cloud at the end of the



                                     60

-------
                               SIMULATED CO
                               - IN CLOUD 2A
                                AT T=48  HOURS \
                       CO IN  CLOUD
                        2A AT T= O
Figure 3-1.  Initial concentration distribution ca(I,J,t0) in the cloud
         simulated in Case 2A for species a = carbon monoxide.  Also
         shown is ca(I,J,
48 hr).
                          61

-------
48-hour period.  The relative positions and orientations of the cloud
at the beginning and end of the simulated period  give an indication  of the
speed and vorticity in the flow.

     In the limited space of this report it is not  practical  to describe
the complete spatial and temporal structure of the  simulated  concentrations
of all 23 species.  Therefore, in the present  experiment we will  focus on
the spatial variations in the concentration error field and in  the next
experiments, 2B, where we add vertical  mixing  to  the list of  processes that
we simulate, we will focus on the temporal  variations.

     Spatial features are seen clearest in  plots  of the concentrations
taken at points along cross-sections of the cloud.   Figure 3-2  is an example
taken from part 1, Section 9.  Shown there  are numerical solutions of Eq.
3-1 for the case of a single, chemically inert species  (kal-j  =  0) in a
rotating flow field of the form (3-9).   The circles in  Fig. 3-2 represent
the solutions obtained along the cross-section indicated in the upper
righthand corner of the figure at time t = t0  + 100 At  using the biquintic
(Q) transport algorithm that we use in the regional model.  The triangles
and crosses in the figure represent the corresponding solutions given by the
schemes of Zalesak (1979) (Z) and Mahrer and Pielke (1978) (S), respectively,
The exact solution is represented by the straight,  solid lines.

     One reason for presenting this figure is  to illustrate the two types
of errors in the transport algorithm that are  of primary concern to us in
applications to chemically reactive species.  The first is the distortion
error that is most pronounced in the solutions derived from the Z and  S
                                     62

-------
-CX
       Figure 3-2.
Comparison of simulations by 3 differencing schemes of the
advection of an ellipsoidal  cloud in a rotating flow.  Panels
display different cross-sections of the cloud (indicated by
the upper right corner of each panel) after one complete
rotation of the cloud, 100 time steps in the case of schemes
Q and S, 150 steps in the case of Z.  Notation: Q (circles)
= transport algorithm used in the ROM; S (x) = transport
scheme of Mahrer and Pielke  (1978); Z(A) = transport
scheme of Zalesak (1979).

                      63
                                                                                  a-d

-------
    c.
                          r
                               1.04-

Figure 3-2.   Continued.
                                 64

-------
      d.
Figure 3-2.  Continued.
                                65

-------
     e.
                               1.0--
                           r
                                .8--
                                .6--
                            •X.
                               '..4--
                                          EXACT
                                              Q
                                                                  .-•*•••  i     x...
Figure 3-2.   Concluded.
                                    66

-------
schemes.  Since the chemical reactions are nonlinear, errors in the amplitudes



and phases of the concentration distributions can result in large errors



in the simulated chemical reaction rates.  The second type of error is



negative concentrations.  Both the Q and S algorithms generate errors of



this type, but the Z scheme was specifically designed to eliminate them.



Obviously, negative concentrations are inadmissable in the kinetics algorithm



because they would transform decay processes into mechanisms of chemical



production, and vice-versa.  In the regional model  we avoid this problem



simply by setting any negative concentrations produced by the transport



scheme to zero before they enter the kinetics algorithm.  Negative



concentrations are not generated when the background concentration is larger



than the amplitude of the "undershoot" created by the transport scheme at



the edges of plumes where gradients are large.  For this reason, negative



concentrations are a problem primarily with the radicals and other species



whose background levels are normally very small.  The test simulations that



we are about to present will show whether our simple procedure for handling



negative concentrations causes adverse effects.





     Figure 3-3(a)-(e), pages 74-100, shows the simulated concentrations of



five principal species along the major and minor axes of the ellipsoidal



cloud at various instants during the 48-hour period.  As Figure 3-1 indicates,



the cloud is transported in a direction that is about 30 degrees askew of



the minor axis.  This orientation remains constant throughout the simulation



because the cloud rotates about its center at the same angular speed that  it



moves around the center of the flow vortex.  (The distribution of vorticity



in the flow field defined by 3-9 causes this.)
                                     67

-------
     Recall from the analyses presented at  the beginning of  this  section


that under the conditions simulated in this problem,  the true  concentration


at any point in the cloud at any time t can be derived  from  the batch


reactor equations (2-1) treated in the previous section.   For  example,  the  •


concentrations ca(xj,t;[) of any species a at a given  point K\  in  the cloud
                  ow                                       s*~>

at time t = t^ is the solution of (2-1) initialized with concentrations


C0(x0,t0).  Here t0 represents the initial  instant  in the cloud simulation


and x0 is the point on the back trajectory  through  xj that designates the


position at time t0 of the fluid parcel that is found at x^, at time t. Thus


cross-sections of the true concentration in a cloud can be constructed  by


solving (2-1) for each point in the cross-section.  This was the  procedure


used to derive the profiles of true concentration,  labeled "chemistry"  in


Figures 3-3 through 3-5.




     Looking first at the series of CO concentration  cross-sections  shown


in Figure 3-3(a), pages 74-79, we see that  the transport algorithm preserves


the symmetry of the cloud with a very high  degree of  fidelity.  The  only


distortions are smoothings of the cloud's sharp peak  and edges.   Within the


first four hours of travel, the peak concentration in the cloud  drops


quickly to a value 10% lower than the true value.  However,  during the


remainder of the 48-hour travel period, the error in  the predicted peak


concentration grows at an average rate of only 2.5% per day.  At  the end  of


the two day simulation, the peak concentration in the cloud  is about 15%


low, which is well within the level of accuracy that  we expect of the data


that are used as inputs during model applications.
                                     68

-------
     At the edges of the cloud the transport algorithm causes the



 simulated concentrations to undershoot the background values by an amount



 that is proportional to the concentration gradient at the cloud edge.  This



 is evident in the fact that the undershoot along the cloud's minor axis is



 somewhat larger than that along the major axis.  The worse values are only



 a few percent of the cloud's center concentration.  An important aspect of



 the cloud edge error is that neither its amplitude nor its spatial extent



 increases with time.  It is also noteworthy that the error is symmetrically



 distributed about the cloud.



     The corresponding cross-sections of NO concentration are displayed in



 Figure 3-3(b), pages 80-84.  To facilitate comparison of the relative



 errors from one travel time to another, we have used the same scale for the



 ordinate of each of the NO concentration plots.  The sequence of NO concen-



 tration profiles shown in Figure 3-3(b) illustrates some of the unusual



 phenomena created by the nonlinear chemical processes that are possible



 sources of serious errors in the transport simulation.  We see first that



 following the initial hour 0000, day 215, the NO concentrations decrease



 until  at hour 0400, day 215, the peak concentration has fallen to a value



 only one quarter its original  value.  By 0800, day 215, which is 2 hours



 after sunrise, the NO concentrations have increased abruptly to levels ten



times  the initial  ones; and the distribution of concentration within the



 cloud  has changed from its initial pyramid form to a tooth-shaped pattern



with a concentration deficit at the center of the cloud and a ridge of high



concentrations surrounding the center.  The Figure shows that the transport



 algorithm captures the true shape of the cloud quite well.  The largest



 error  is at the cloud center where the model overpredicts the true concen-



tration by about 10%.



                                     69

-------
     From the standpoint of the transport  algorithm, the  most  significant
aspects of the change in the cloud's  shape is  the  intensification  of
concentration gradients around  the  cloud's edge.   We saw  earlier that  the
magnitude of the concentration  undershoot  just  outside  the  cloud is
proportional to the concentration  gradient at  the  cloud's edge.  It is  not
surprising then that the errors in  the  simulated NO concentrations just
outside the cloud are larger than  those produced in the simulation of  CO.
Figure 3-3(b), page 82, shows that  at hour 0800, day 215, the  NO undershoot
attains a maximum amplitude, coinciding with the time of  peak  NO concentrations
inside the cloud.  At this point the  undershoot is about  15% the peak  value
in the cloud.  An interesting aspect  of the error  field surrounding the
cloud is the apparent absence of chemical  change.  The  plots shown in
Figure 3-3(b) for hours 1200 and 1600,  day 215 (pages 83  and 84),  indicate
that the magnitude of the NO undershoot remains virtually unchanged for 8
hours or longer following its generation even  though NO levels within  the
cloud are declining during this entire period.  By hour 1600,  the  undershoot
is as large as the amplitude of the cloud itself.  However, in absolute
terms the magnitude of the undershoot is only  of the order  of  10~5 ppm NO,
a value much too low to have significant effects on  the chemistry  overall.

     This conclusion is supported by  the results of  the ozone  simulation
shown in Figure 3-3(c), pages 85-91.   Throughout the  simulated 48-hour
period, the model reproduces the ozone concentration with a precision
greater than about 95% over the body  of the cloud.  Subsequent to  hour
0800, day 215, when the simulated NO achieves  the  largest undershoot  at the
cloud's edges, Figure 3-3(c) shows that the ozone  cloud base begins  to
                                     70

-------
 broaden until by the end of the 48-hour period it is about 1 grid cell



 wider than  it started out.  This is apparently a direct consequence of the



 underestimated NO concentrations around the cloud's base; because the



 chemical  reaction NO + 03 -»• N02 + 02 causes an inverse relationship between



 the concentrations of NO and 03.  In any event, the fractional error in the



 simulated 03 levels at the cloud base are much smaller than those that we



 found in  NO.





     Figures 3-3(d), pages 92-96, and 3-3(e), pages 97-100, show the simulated



 N02 and olefin concentration distributions, respectively.  Both of these



 species are nearly completely consumed by chemical reactions well before



 the end of the 48-hour simulation.  During the time they are present, the



 model reproduces their concentration distributions with the same level  of



 precision that it handled the other three species that we just discussed.



 In the last two pages of Figure 3-3(e), which show the olefin concentration



 results, we have added inserts that bring out details in the cloud cross-



 sections when concentration has fallen to very low values.  The results



 show that the model's accuracy remains high throughout the period of declining



 concentration.





 Case 2B:  Chemistry with transport and vertical mixing





     In this experiment we extend the range of testing by adding vertical



 turbulent mixing to transport and chemistry.  The conditions here are the



 same as in experiment 2A, except that rather than being confined to layer 1



 for the duration of the 2-day simulation, the pollutant cloud is allowed to



mix virtually instantaneously with clean air in layer 2 above.  This is
                                     71

-------
done at hour 12 of the first day by  abruptly  changing  the  value  of  the  rms



vertical turbulent velocity on  the  interface  between  layers  1  and 2 from



zero :,o a large value.  Since layer  2 is  1000 m  deep  and  layer 1 is only



300 m thick, mixing causes a reduction of the concentrations of  all  species



by about three-fourths.





     Two elliptical clouds of the form treated in  experiment 2A  are considered



here.   Their initial  locations and  subsequent trajectories  are  illustrated



in Figure 3-4, page 101.  The initial concentration distribution in each cloud



has the form (3-10) (see page 57),  with one cloud, which we  shall refer to



as 2B.L, having the "lean" mix of Ca(t0)  values  at its center  (see



Table 2.1, page 14); and the other  cloud, 28.R,  having the "rich" mix of



Ca(t0) values listed in Table 2.1.





     Cross-sections of the predicted ozone concentration  in  cloud 2B.R  are



shown at selected travel times in Figure 3-5, pages  102-108, in  the same



format that we used in Figure 3-3.   Simulated concentrations of  CO, NO, 03,



N02, olefin, and PAN in clouds 2B.L  and 2B.R  are shown in  Figure 3-6,  7 and



8 (pages 109-126) in the form of time histories  following  three  different



points in each cloud.   One point is  the cloud center,  one  is midway between



the center and the edge, and the third point  an  outermost  grid point.



These points and their resultant trajectories are illustrated  in Figure



3-4 (page 101).  The concentrations  plotted in Figure 3-6, 7 and 8 were



obtained by interpolating the model  output at points  along each  trajectory.



And the true solutions, labeled "chemistry" in the Figures, are  the solutions



of the batch reactor equations (2-1) initialized with the cloud  species



concentra-   ns at the starting point of each  trajectory.





                                     72

-------
     The quality of the model's performance in experiment 2B is not



significantly different from that found in experiment 2A.





     We. conclude from these combined tests of the transport, • chemistry and



vertical mixing algorithms that the solutions of the combined transport and



chemistry equation (3-1) produced by the model are good facsimiles of the



true solutions over the range of species concentration values that we are



likely to encounter in actual problems.  Among the species CO, 03, NO, NO;?



and olefin, the largest error in the simulated concentration within the



cloud was found for CO.  In this case the peak concentration was underestimated



by about 15%.  For the other species the largest errors were between 5 and



10%.  No evidence was found of adverse effects arising from the undershooting



of concentration outside the edges of clouds, which is a characteristic of



the algorithm used to treat the advection terms in the governing equations.



The principal effect was a slight broadening of the simulated ozone cloud.





     Although our conclusions apply strictly to the rather limited conditions



that we have considered here, these tests nevertheless constitute essential



necessary conditions for model validity.  Considering the quality of the



model  performance shown and the invariance of this quality over the range



of conditions that we considered, we are confident that the model can handle



generalized flow fields, diffusion, vertical mixing and species concentrations



with comparable accuracy.  One aspect that we have not yet treated is the



ability of the model  to simulate continuous, discrete sources of pollutants.



We consider this in the next section.
                                     73

-------
     TEST  : CLOUD ADVECT10N TEST
     DATE  : 79215
     HOUR : 000000
                              CO   CONCENTRATION
                                   (PPM_X101  ).
                  CHEMISTTTr'

                  NEROS
                                 2A
                             TOD t30 tOO 1030 1120 1130 1&20 1*20 1KB 1830 17OO 1*30 1*30 2O3D
         10:12 10:13 10:1+ 10:15 10:16 10:17 10:18 10:19 10:20 10:21 10:22 10:23 10:24 10:25 10:26 10:27 10:28

                    CLOUD CELL  LOCATION  (ROW.-COLUMN)
Figure  3-3(a).
Initial  concentration  of CO in cross-sections of  the cloud
simulated  in experiment  2A.  Diagrams in the upper right
corner  of  each panel  indicate the  location of the cross-
section within the cloud.  The curves labeled "chemistry"
represent  the true solution.
                                        74

-------
     TEST   : CLOUD ADVECT10N TEST
     DATE   : 79215
     HOUR  : 040000
                   CHEWiSTRY

                   NEROS
                                CO   CONCENTRATION
                                     (PRM X101 ) .
2A
                                       10tZ> 1139 1130 1MO 14dO 1KB 1RX 17120 1*X 1fcJO
          i    I    T   i   I    '    '    •    •    :   i    ,        '    !    ;    I
         10:12 10:13 10:14 10:15 10:16 10:17 10:18 10:18 10:20 10:21 10:22 10:23 10:2+ 10:25 10:26 10:27 1C:2B

                     CLOUD  CELL LOCATION (ROWrCOLUMN)
Figure  3-3(a).   Continued.   Travel  time  = 4 hours,
                                           75

-------
       TEST  : CLOUD ADVECTION TEST
       DATE  : 79215
       HOUR  : 080000
                                   CO  CONCENTRATION
                                       (PPM X101 )
                     CHEMISTRY

                     NEROS
                                                                     L
                       too taD  «3D  no  ido  too icen nao itx 1*20
          10=12 10:13 10:14 10:15 10:16 10:17 10:18 10:19 10:20 10:21 10*2 10:23 10:2* 1oi25 10-26 10^27 10-28

                      CLOUD  CELL  LOCATION  (ROWrCOLUMN)
Figure 3-3(a).   Continued.   Travel time =  8 hours.
                                           76

-------
      TEST  : CLOUD ADVECTION TEST
      DATE  : 79215
      HOUR : 160000
                   CHEWISTRY

                   NERCS
                                CO  CONCENTRATION
r i i i : i i r — •

	 CHEMISTKV
NEROS


30 KB 1033 TI3J f
-
-1


2A


o inao iaao 1*00 2aa
,rv
I !
         10:12 10:13 10:14 1ollS 1o!t6 10M7
                                   10:19 10:20 10:21 10:22 10*3
                                                         1o25 10:26 1027
                     CLOUD CELL LOCATION  (ROW:COLUMN)
Figure  3-3(a).   Continued.   Travel time  = 16 hours.
                                         77

-------
       TEST
       DATE
       HOUR
CLOUD ADVECT10N TEST
79216
120000
                                  CC   CONCENTRATION
                                      (P<=M X10; )
                     CHEMISTRY

                     NEROS
                                          2A
                                         no nao ixao 1300 1430 IKK 1*30 irao
         10:12 10:13 10:14 10:.3 10:16 10:17 10:18 10:19 10:20 10J21 10^22 1ft23 1ol24 10:2= lo

                     CLOUD  CELL LOCATION  (ROWrCOLUMN)
Figure  3-3(a).   Continued.   Travel time =  36  hours,
                                          78

-------
     TEST  : CLOUD ADVECTION TEST
     DATE  : 79217
     HOUR : 000000
                    CHEMISTRY
                                 CO   CONCENTRATION
                                                       2A
                                                                   L
          one too too x*s «x can toe  TOO  too  *T i~~ '"*> ino ixa ixao 1100 iuo irao IKZI ino
         10:12 10:13 10:14 10:13 10:16 10:17 10:18 10:18 10:20 10:21 10:22 10:23 10:24 10:23 10:26 10:27 10:28

                      CLOUD  CELL LOCATION (ROWrCOLUMN)
Figure 3-3(a).   Concluded.   Travel  time  =  48 hours,
                                           79

-------
     TEST  : CLOUD ADVECTiON TEST
     DATE  : 79215
     HOUR : 000000
                              NO   CONCENTRATION
                                  (P£M X104 )
                  CHEMISTRY

                  NEROS
                                     1
                                     -j
                                     1
                                 2A
«,c~xr i~4i~~««~iL>~~~r»™£~£ ,*
	 CHEMISTRY
	 	 NEROS

0 1(30 1230 1X30 14
-------
TEST
DATE
HOUR
   CLOUD ADVECTION TEST
   79215
   040000
              CHEMISTRY

              NEROS
                           NO   CONCENTRATION
                                (PEM_X1C4 }
                                                            L
     OOO  t-20  220 300  430  &30  130  «0  »30  WO
              CHEMISTRY

              NEROS
                                  1-h
                                        1230 1UO 1OO 1UO 1K20 1731 1UO
                                           2A   t
10:12 10:13 lolu 10:13 lolie io!i7 idiB 1W9 ioao ioai
                                             10:22 10=2+ ioas loizs 10-^7
                CLOUD  CELL LOCATION  (ROWrCOLUMN)
Figure 3-3(b).   Continued.   Travel  time =  4  hours,
                                    81

-------
TEST
DATE
HOUR
CLOUD ADVECTION TEST
79215
080000
              CHEMISTRY

              NE.ROS
                            NO  CONCENTRATION
                                (?5V_X104  }
     I   I   i   i    ;   I   i   '
     COO  13) 230 3Oe  I00 tOO 730
                                IOO 1MO 1130 1230 TJOO 1UD
                                                        1730 1UO 1WO
               CHEMISTRY

               NEROS
                                        2A    ,
    10:12 10:13 10:!+ 10:15 10:16 10:17 10:18 10:19 10^0 1O21 10:22 10:23 10^4 10:23 10:26 10:27

                 CLOUD CELL  LOCATION  (ROWrCOLUMN)
 Figure  3-3(b).   Continued.   Travel  time  = 8 hours,
                                      82

-------
TEST  : CLOUD ADVECTiON TEST
DATE  : 79215
HOUR  : 12QGGO
              CHEM.STRf

              NEROS
                           NO   CONCENTRATION
              CHEMISTRY

              NEROS
                                   1
                                                     1B20 1730 1B21 1J--21
2A   t
     I    I    i   I    i    i    i    i    i   :    i    i    ;    i           ,
    10:12 10:13 10:54 10:13 10:16 10:17 10:18 10:18 10:20 10:21 10:22 10:23 10:2+ 10:23 10^6 1027 1028

                CLOUD CELL LOCATION (ROW:COLUMN)
Figure 3-3(b).  Continued.   Travel  time  = 12 hours,

                                     S3

-------
TEST  : CLOUD ADVECT10N TEST
DATE  : 79215
HOUR  : 160000
               CHEMISTRY

               NEROS
                            NO   CONCENTRATION
                                 (P5MJ<1C4 )
2A
                                                              L
     aaa too ±20 ao too too too  «o  tao  131 ittz; 1130 1230 1*20 i«20 IJUD
               NEROS
                                                 2A   t
                                   1	
                                                    "1	T
    10:12 10:13 10:14 10:13 10:18 10:17 10:18 10:18 10:20 10:21 10:22 10:23 10:2+ 10:23 10:26 10-27 10:28

                 CLOUD  CELL  LOCATION  (ROW:COLUMN)
Figure  3-3(b).  Concluded.  Travel time  =  16 hours.

                                      84

-------
     TEST  : CLOUD ADVECTION TEST
     DATE  : 79215
     HOUR  : 000000
                               03  CONCENTRATION
                   CHEW.'STKY

                   NEROS
                                  2A
             1-20 2:20 MO «3C fcZI  ft3D  730
                   CHEW15TK1T'

                   NEROS
                                KB MO 1030 ItOO 1Z20
                                 2A
         10:12 10:13 10:U 10:15 10:16 10:17 10:18 10:19 10:20 1021 10:22 10:23 10:24 10:25 10;26 10:27 '^
                    CLOUD CELL LOCATION (ROWrCOLUMN)
Figure 3-3(c).
Initial concentration of  ozone in cross-sections of the cloud
simulated  in  experiment 2A.   Diagrams in the  upper right
corner of  each panel indicate the location  of the cross-
section within the cloud.   The curves labeled "chemistry"
represent  the true solution.

                      85

-------
TEST  : CLOUD ADVECTiON TEST
DATE  : 79215
HOUR  : 020000
               CHEMISTRY

               NEROS
                            03   CONCENTRATION
                                 (?£M. XIQ! .)
,
2A

        too toe 300 too loo  us  raa  too uo 1030 1120 1x20 1x20 too iuo 
-------
TEST  : CLOUD ADVECTiON TEST
DATE  : 79215
HOUR  : 12QOOO
               CHEMiSTTTT
               NEROS
                             03   CONCENTRATION
                                 (P£M X10!- )
2A
              r
     030 too £20 3OD «ao  ftao  eao 7:aa too too 1030 1120 1230 1220 1430 iBcao I&SD 17^0
               GHEMISTRT
               NEROS
2A   t
          .                                       	
    10:12 10:13 10:H 10:15 10:16 10:17 10:18 10:19 10:20 10:21 10:22 10:23 10:24 10:25 10:26 10:27 10:28
                 CLOUD  CELL  LOCATION (ROWrCOLUMN)
 Figure  3-3(c).   Continued.   Travel  time =  12 hours.

                                       87

-------
TEST  : CLOUD ADVECT10N TEST
DATE  : 79215
HOUR  : 160000
               CHEMISTRY
               NEROS
                            03   CONCENTRATION
                                 (PgM XI 0:  )
 2A
     OOD iao 220 jao  tao  ut>  too no toe KB 1000 1130 1231 tuo i«» IMO iuo 1721 1100 tno
               CHEMISTRY
               NEROS
2A   ,
      I    i    i    ;    i    i   I    ;    I    i    I    :    I    i    ;    :    i
    10:12 10:13 10:14 10:15 10:16 10:17 10:18 10:13 10:20 10:21 10:22 10:23 10:24 10:25 10:26 10:27 10:28
                 CLOUD  CELL  LOCATION  (ROWrCOLUMN)
Figure 3-3(c).   Continued.   Travel time =  16 hours.

-------
TEST  : CLOUD ADVECTION TEST
DATE   : 79216
HOUR : 000000
               CHEMISTO"

               NERC5
03  CONCENTRATION

    (P=M XI0; }
        I

        1
i2D 
-------
TEST  : CLOUD ADVECTION TEST
DATE  : 79216
HOUR  : 120000
              CHEMISTRY

              NEROS
                           02   CONCENTS,-'
2A   I
i
H
i
1 i ; ! ' : I , ! '
no 120 220 XX tOO &20 tan 730 830 no lft2Q 1120 1230 1X2Q 1«£20 1&20 1&20 1723 1&23 1.93& 2X23
	 CHEMISTRY
.. NEROS

?A . M .
v ;
j ^
    10:12 10:13 10:14 10:15 10:16 10:17 10:18 10:19 10:20 10:21 10:22 10:23 10:24 10:25 10:26 1C:27 10:28

                CLOUD CELL  LOCATION (ROW:COLUMN)
Figure 3-3(c).   Continued.  Travel  time  = 36 hours.

                                    90

-------
TEST
DATE
HOUR
CLOUD ADVECTION TEST
79217
000000
                            03   CONCENTRATION
                                 (P=M XI0!  )
               CHEMISTRY

               NEROS
                                           2A
        130 £20 JC20 4<20 &2D  C20  730  OC20 te20 10OO 1120 1220 1220 14CO 1BC20 1K2D 17ZZO
               CHEMISTRY

               NEROS
                                          2A   t
     I	1	1	;	i	;	1	;	;	•	1	•	,	—,
    10:12 10:13 10:14 10:15 10:16 10:17 10:18 10:19 10:20 10:21 10:22 10:23 10:24 10:25 10-26 10-27 10-28

                 CLOUD CELL LOCATION  (ROWrCOLUMN)
Figure  3-3(c).   Concluded.  Travel  time  = 48 hours,

                                      91

-------
     TEST  : CLOUD ADVECTION TEST
     DATE  : 79215
     HOUR  : 000000
                               N02  CONCENTRATION
                                   (P2M
                   CHEMISTRY

                   NEROS
                                      nan itao 1230 1330 100 iaao 1*20 1/20 IKK IKS
                                                  2A  L
          I    i   i    i    i   i    i   !       i    i    i   i    ;   i    i   i
         10:12 10:13 10:U 10:15 10:16 10:17 1C:1B 10:19 10:20 10:21 10:22 10:23 10:2* 10:25 10:25 10:27 10:28
                    CLOUD CELL  LOCATION  (ROW:COLUMN)
Figure 3-3(d).
Initial concentration of  NC>2 in cross-sections of the  cloud
simulated  in experiment 2A.   Diagrams  in  the upper  right
corner  of  each panel indicate the location of the cross-
section within the cloud.   The curves  labeled "chemistry"
represent  the true solution.
                                        92

-------
TEST  : CLOUD ADVECT10N TEST

DATE  : 79215

HOUR  : 020000
              CHEMISTRT


              NEROS
                           N02  CONCENTRATION

                               (??V XIG" )
2A
                                                           L
     I	1	;   i    I    !    !   i    !	i	:	;	i	T	:	'
    10:12 10:13 10:1+ 10:15 10:16 10:17 10:18 10:18 10:20 10:21 10:22 10:23 10:2+ 10:25 10:26 10:27 10:28

                CLOUD CELL  LOCATION  (ROWrCOLUMN)
Figure 3-3(d).   Continued.   Travel time  = 2 hours.


                                    93

-------
TEST  : CLOUD ADVECT10N TEST
DATE  : 79215
HOUR  : 040COO
               CHEMISTRY

               NEROS
                            N02 CO.NCEN'TRATION
                                  6—
 2A
                                                              L
                                   A—
     OO3 120 130 SX 4OO tOB OB 730  K»  KB 1330 1130 1UO Ii30 1»JO 15JO 10JO 1730 1KX IWO
               CHEMISTRY

               NEROS
2A
    10:12 10:13 10:14 10:13 10:16 10:17 10:18 10:18 10:20 10:21 10:22 10:23 10:24 10:23 10:26 10:27 10:23

                 CLOUD  CELL LOCATION (ROWrCOLUMN)
Figure 3-3(d).   Continued.   Travel  time  = 4 hours.

                                      94

-------
TEST  : CLOUD ADVECTION TEST

DATE  : 79215

HOUR  : 120000
              NEROS
                          N02  CCNCENTRATID"

                               (P=V XI 0" )
2A
            ,
    10:12 10:13 10:14 10:15 10:16 10:17 10:18 10:19 1C:20 10:21 10:22 10^3 10:2+ 10:23 10:26 10:27 11:23

               CLOUD  CELL LOCATION  (ROWrCOLUMN)
Figure  3-3(d).   Continued.   Travel time = 12  hours.


                                   95

-------
TEST  : CLOUD ADVECTiON TEST
DATE  : 79216
HOUR  : OOOOGQ
               CHEMiST"

               NERCS
                            N02 CONCEN'RATiCN
                2A
                                                           \   :   I
                                                              L
        tao 220 Joo 130  KZJ  KB  rao  tx  KB 1330 nao it» tuo iuo luo IBJO 1733 itx mo
               CHEM1STW

               NEROS
,4-
                                   S-T-
                                    1
2A  L
    10:12 10:!3 10:14 10:15 10:16 10:17 10:18 10:18 10:20 10:21 10:22 10:23 10:24 10:25 10:26 10:27 10:28

                 CLOUD CELL LOCATION (ROWrCOLUMN)
Figure  3-3(d).   Concluded.   Travel  time = 24 hours.

                                      96

-------
     TEST
     DATE
     HOUR
CLOUD ADVECTiON TEST
79215
000000
                              OLE CONCENTRATION
                  CHEMISTRY

                  NEROS
                                      2A
             1:20 220 320 *30  fiiZC t2D TrSJ ft20 Sc20 I£t20 JfJO 1220 liJO
                   CHEWISTKY

                   NEROS
                                     2A   ,
            1oil3 1oll+ 10J15 10:1B 10:17 10:1B 10:19
                                        1021
                                                       10:23 1026
                                                                 los
                    CLOUD CELL LOCATION (ROW:COLUMN)
Figure 3-3(e).
     Initial concentration of olefin  in  cross-sections of  the
     cloud simulated in experiment  2A.   Diagrams in the  upper
     right corner  of each panel indicate the location of the
     cross-section within the cloud.   The curves labeled
     "chemistry"  represent the true solution.
                                       97

-------
TEST  : CLOUD ADVECTION TEST
DATE  : 79215
HOUR : 04QGOO
OLE
                                 CONCENTRATION
                                 (°=M XI GJ }
               CHEMISTRY1

               NEROS
                      2A
        120 £2i 300  too  too  tea  «o sao too MOO nao 1221 ii» 100 iuo tuo 1721 isjo IB-JO
               CHEMISTRY

               NEROS
                     2A  i
     10:T2 10:13 10:14 10:15 10:16 10:17 10:18 10:18 10;20 10:21 10O2 10:23 10:2+ 10:25 1CC.2B 10:27 10:28

                 CLOUD  CELL LOCATION (ROWrCOLUMN)
Figure 3-3(e).   Continued.   Travel  time  =  4 hours.

                                     98

-------
TEST  : CLOUD ADVECT10N TEST
DATE  : 79215
HOUR : 080000
                           OLE  CONCENTRATION
                 _•£,-,- .
                                                  2A
                                  i—
        tao  too  £20  «a>  020  120  TOD  KB  tsa icxzo 1120 1220 1120 1100 iuo 1100 1729 1x20
               CHEwisrrrr
               NEF!OS
                                  I-1-
                              2A   L
    10:12 10:13 10:1+ 10:13 10:18 10:17 10:18 10:1S 10:20 10^1 10:22 10:23 103+ 10:23 1Ci26 10:27 10:28
                CLOUD CELL  LOCATION  (ROWiCOLUMN)
Figure  3-3(e).
Continued.   Travel time = 8  hours.   Insert in  upper
panel  is magnified plot of major axis  cross-section.
                  99

-------
      TEST   : CLOUD ADVECTION TEST
      DATE   : 79215
      HOUR  : 120000
                                OLE CONCENTRATION
                    CHEM;STRY

                    NEHOS
                                   2A
                                       i—
             120 £30 w Jj^"r> fcjp **?p Tao  p^?n


                 OLE COHCENTTW10S
                    (P|W X10J


                      3
          CLOUD CELL LOCATICfJ (RO'.Y:COLL''.!M)
                                                1UO 1«JQ 1UQ 1fijd 17JQ
                                       ,030  1B51
                                                          1033 1021  TKZ7
                     CLOUD  CELL LOCATIOM (ROW:COLUMN)
Figure  3-3(e).
Concluded.  Travel  time  =  12 hours.   Insert in  lower panel
is magnified plot  of minor axis cross-section.

                       100

-------
                2B.L
                                       2B.R
Figure 3-4.  Initial CO concentration in clouds  2B.L and 2B.R.  Arcing lines
           labeled E, M, and C are 48-hour trajectories of points originating
           at the edge, midpoint, and center,  respectively, of each cloud.

                               101

-------
     TEST  : DILUTION SIMULATION TEST
     DATE   : 79215
     HOUR : OQOQCO
                              03  CONCENTRATION
                                  P5',< XI0;
                  CHEMISTRY

                  NEROS
                                    ]
                                    j
                                    1
                                    j
                                     !
                                   i4-
                                    j
                                    1
 2B.R
                  CHEMISTRY

                  NEROS
                                   &-T-
2B.R  ,
         8:22 8:23 8:24 8:25  8:26 9:27 9:28 8:28 9-JO  9:J1 8-J2 9:33 8:34 9:35 9J3B gb? 9:38

                   CLOUD CELL LOCATION  (ROW.-COLUMNJ
Figure 3-5.   Initial cross-section of ozone concentration in cloud  2B.R.
              Diagrams  in  the  upper right corner of each panel  show  the
              location  of  the  cross-section in the cloud.  Curves  labeled
              "chemistry"  represent the true solution.

                                     102

-------
TEST
DATE
HOUR
DILUTION SIMULATION TEST
79215
040000
              CHEMISTKY

              NEROS
                           03  CONCENTRATION
                                                            r
                                     2B.R
    -2SJO -1:X  1JO  2130
                       fcJO  ftJO  7-JO ftJO fcJO IftJD 11JO 1£JO 1XB
                                                       1LJQ 1TJC 1&X IB-JO
              CHEMISTRY

              NEROS
                                    2B.R
                                  1—
    9:22 9:23  9:24 9:25 9:26 9:27 9:28  9:29 9:30 9:31  9:32  9:33 9:34 9:35 9:36  9:37 9:38

               CLOUD CELL  LOCATION (ROWiCOLUMN)
Figure 3-5.   Continued.  Travel time  = 4 hours (Case  2B.R).

                                   103

-------
        TEST  :  DILUTION SIMULATION TEST
        DATE  :  79215

        HOUR :  1200GO
                                  03   CONCENTRATION

                                      (F°1.' X10:
                     CHE.MISTrTr


                     NEROS

•JO -1JO tlJO 2JO AX *i3C 6JQ fc» T*-X

	 CHEMISTRY
NEROS
SJO KX 1CJO I1JO 1UO 1UO 14JO 15J3 10JD !

1 PR.R . f
7_» 1SJQ 1R-JO

' \
                      	        i   i    ;    i   ;    ;    ;   :
          8:22  9:23 9:24  9:25 8:26 9:27  9:28 9:29 9:10  9:31 9:32 9:33  9:34 9:35 9:36  9:37 9:38


                     CLOUD CELL LOCATION  (ROWrCOLUMN)
Figure  3-5.  Continued.  Travel  time  =  12 hours.   Vertical mixing between

              layers  1 and  2  begins at this instant.   (Case 2B.R).



                                       104

-------
     TEST
     DATE
     HOUR
DILUTION SIMULATION TEST
79215
160000
                   CHEMiSTFY

                   SEROS
                                03  CONCENTRATION
                                    (P£V XI0! )
                                    2B.R
                                                                L
                    CHEMISTRY

                    NEROS
                                    &JQ fJO 1CJQ 11JO
                                                     1OQ 1SJO 1BJO irJQ 1&JO
                                    2B.R
          9:22 9:23 9:24  9:25  9:26 9:27 8:28  9:29 9:JO gji  9:J2 9:J3 9:^4  9:J5 9:36 9:J7  9:33

                     CLOUD CELL  LOCATION (RO\V:COLUMN)
Figure  3-5.  Continued.   Travel time  = 16 hours,  4 hours  after mixing
              (Case 2B.R).

                                        105

-------
TEST  : DILUTION SIMULATION TEST
DATE  : 79216
HOUR  : OOOQOO
              CHEMISTRY

              NERCS
                          03   CONCENTRATION
                               (P=V XIC: )
2B.R
                                 5—
                                                           L
    -2JD -tJQ  t-JQ £30 A30 *i3D 6JD
              CHEMISTRY

              NEROS
                               SJO WO 10 JO 11JO 12JO 1UO 14*JD 1SJQ 16JO 17 JO 1BJO 1WQ
2B.R  ,
     9:22 9:23 9:24  9:25 8:26 9:27  9:28 9:29 9:30  9;31 9:32 9:33  9:34  9:35 9:36 9:37  9:38

                CLOUD CELL LOCATION  (ROW-.COLUMN)
Figure  3-5.  Continued.   Travel time  = 24 hours (Case  2B.R).

                                  106

-------
TEST   : DILUTION SIMULATION TEST
DATE   : 79216
HOUR  : 12COGO
              CHEMISTRY

              NERCS
                          03  CONCENTRATION
2B.R   !
^ J
--------- ^
UD -1JQ 1JO 2JO £30 AJD UC fcM 7JO ttJO frJO 1CJJO 11 JO
CHrMS5TF7Y _|
NF.ROS 1


£JO 1X30 1+JO 1&JO 1CLJO 17JQ

2B.R t_L_


•KJD IB-JO
N
I ,
                                4—
                                2—
    9:22  9:23 9:24 9:25  3:25 9:27 9:28 9:29 9:10  9:31 5:32  9:33 9:34 9:35  9:36 9:37 9:38

               CLOUD CELL LOCATION  (ROWrCOLUMN)
Figure  3-5.   Continued.   Travel time  = 36 hours.

                                  107

-------
       TEST  : DILUTION SIMULATION TEST
       DATE  : 79217
       HOUR  : OOOGOO
                              03  CONCENTRATION
                   CHEMISTRY


                   NEKOS


•JO .IsJO tao ZJO JJO WO UO 1


T
-J

f '
uo rao EJO u
6-r
1
                                 2B.R
                                                          L
                                              1JJQ 14JO liJO 1BJD trjD 1BJO 1ft
                                            2B.R  L
         9:22
                         9:27
                   CLOUD CELL LOCATION (ROWrCOLUMN)

Figure 3-5.
Concluded.   Travel time = 48 hours, 36 hours after mixinq
(Case 2B.R).                                            3


                      108

-------
O

X
                     CHEMISTRY
          c
          O
                                                   T- O
                                 2B.L
                                                ^  T=48Kr
          O  r
          O  C
             T
                                      24
          Q_
          Q_
          O
         UJ
         O
         •z.
         O
         O
         O
         O
           CHEMISTRY

           ROM,
                                  2B.R
                                               3    12   16   20   24
                                                T= 48 Kr
                 4   8   12    16   20   24    4    8    12   16   20   24
                        DAY 1
                                                 DAY 2
                                                                i  DAY 3
                           TIME (HOUR  OF DAY)
Figure 3-6(a).
      Time histories  of  CO  concentration following the center of
      cloud 2B.L,  top, and  cloud  2B.R, bottom.  Curve labeled
      "chemistry"  represents  the  true solution.
                                    109

-------
             3r
          x
                     CHEMISTRY

                     RDM.
                  2B.L
                                      - O
          O
                  4    S    12   16   20   24    4
                              '6   20   24    4    8   T2   16   20    24
                        DAY 1	,	   DAY 2
                                                                   DAY 3
                           TIME (HOUR OF  DAY)
Figure 3-6(b).
Time histories of NO concentration  following the center of
cloud 2B.L, top,  and cloud  2B.R,  bottom.   Curve labeled
"chemistry" represents  the  true  solution.
                                    110

-------
       o

       X
CHEM1STO-

ROM
2B.L
                       12   16   20   24
                                            8    12
          gi
       D_
       D_
       O

       < st
            -
       Ld
       O
       2
       O
       O
       O
                              2B.R
                                             T=48Kr
                   8   12   16   20   24

                      DAY 1
                                     i i i i i i i i i
                          B    12   16   20   24

                            DAY 2
                         TIME (HOUR OF DAY)
                                                               DAY 2
Figure  3-6(c).  Time histories of ozone concentration following the center
               of cloud 2B.L, top, and cloud  2B.R, bottom.  Curve labeled
               "chemistry" represents the true  solution.
                                   Ill

-------
                                                     = o
                    CHEMISTRY

                    ROM
                 2B.L
                                                              24
                       DAY 1
                                                DAY 2
                                                  DAY 3
                           TIME (HOUR OF DAY)
Figure  3-6(d).
Time histories  of  N02 concentration following the  center
of cloud 2B.L,  top, and cloud 2B.R, bottom.   Curve labeled
"chemistry"  represents the true solution.
                                   112

-------
                                 2B.L
            0" "
                                                  2   16   20   24
                       DAY 1
                                                DAY 2
                                                                DAY 2
                          TIME  (HOUR  OF DAY)
Figure  3-6(e).
Time histories  of olefin concentration  following the center
of cloud  2B.L,  top, and cloud 2B.R,  bottom.  Curve labeled
"chemistry"  represents the true solution.
                                   113

-------
             2
          x

          ^  -
          D-  i_
                     CHEMISTRY

                     ROM
          CM

          O



          X
          Q.

          Q_
          O 3j


          1
          o
          2:
          o
          o
         <
         Q_
f
                     2B.L
                    2B.R
                                                    T=0
                                                  12   16   23   24
T= 48Kr
                 4   8   12   16


                        DAY 1
                                  a    12   16   20   24


                                    DAY 2
                                                                 DAY 3
                           TIME (HOUR OF DAY)
Figure  3-6(f).
   Time histories of  PAN concentration following the

   center of cloud  2B.L, top, and cloud 2B.R, bottom.

   Curve labeled  "chemistry" represents the true solution
                                   114

-------
          _  2

          O

          X
r
          D_  _
          O
          o:
          UJ
          o
          o  _
          o
          o
          o
         ,-10
         o

         X
r
         LU
         O
         2
         O
         o
         o
         o
            4H
                                  = o
                    2B.L
                  4    8
                              16   23
                                                           2Q   24-
CHEM.ST^Sf

ROM/
2B.R
                             T= 46'nr
2
CL
0^
z:
o
§
O 	
F
t
I
4
                     S   12

                        DAY
                              ;s   20
                                                         r=o
Figure 3-7(a).
   Time histories of CO following the midpoint of cloud  2B.L,
   top, and 2B.R, bottom.  Curves labeled "chemistry"  represents
   the true solution.
                                   115

-------
o
T—
X
                   CHEMISTRY

                   ROM
                                                   T=0
2B.L
                              2B.R
         O
                                                T=48Kr
                                                       T=0
                                             a   :2   is   23
                                               ~, AY 2
                                -ni-o
                                 . —' O', ,
Figure  3-7(b).
       Time histories of NO following the midpoint of cloud 28.L,
       top, and 28.R, bottom.  Curves labeled  "chemistry" represents
       the true solution.
                                   116

-------
                                                      T= O
                     CHEMISTRY



                  	ROM  A
                     2B.L

          o  :


          o  ->-
          o  -
             1-
          O
                  4    8   12   16   2Q   24
o   E
-—   i-
X
              _
                     CHEMISTRY



                     ROM
         £-
         o  f


         H
         Z  C
         UJ  -
         o  :
         o
         O
1-
         o  c
                   2B.R
         8    52   16   23   24



           DAY 1-
                                                  12   16   2G   24
                                      T=4SVu-
                                                  '2 "
                                                          20
                           TIME  HGU^ Cr ^-Y
Figure  3-7(c).  Time histories of ozone  following the midpoint of cloud 2B.L,

               top, and 2B.R, bottom.   Curves  labeled "chemistry" represents

               the true solution.
                                   117

-------
                                                     = o
         O
                              2B.R
                                                 T=48Kr
                                                  -.2   -6   iO

                                                DAY 2
                                   \ - ' P rN r - ^
                                   / >_/ r\ v ;,   , —
Figure 3-7(d).
Time histories  of N02  following the midpoint of cloud 2B.L,
top, and 2B.R,  bottom.   Curves labeled "chemistry" represents
the true solution.
                                   118

-------

            O1 ; ' ' I '
                                                     T=0
         CM
         o
         X
         Q_
         D_
         o
         cr
         i—
         2:
         LJ
         O
         O
         o
         Ld
         _J
         O
    CHEMISTRY
    ROM
                2B.R
                                                       16   2D   24
                                               3   12    16   20   24
                                                  DAY 2
                                        r-~
                              V^
                                /
Figure 3-7(e).
Time histories of olefin following the  midpoint  of  cloud  2B.L,
top, and 2B.R, bottom.  Curves labeled  "chemistry"  represents
the true solution.
                                    119

-------
Figure 3-7(f).
Time histories of PAN following the midpoint of cloud 2B.L,
top, and 2B.R, bottom.  Curves labeled "chemistry" represents
the true solution.
                                    120

-------
                                                 '= O
         x  /-
         C  ;

         <  4-
         o  ;

         C  i—
         O  ~-
 2B.L
                                              e
         o
                    8   '2
                                            3   12   16   20
o  ^
X  r-
2  ?

i3?-
    I-
2  h
9  c

14
            F
         LoJ  i-
         z  E
         8'r
                   ROM
2B.R
         o
         o
Figure 3-8(a).   Time histories of CO following the  edge ooint  of cloud 2B.L,
               top, and 2B.R, bottom.  Curves labeled "chemistry" represents
               the true solution.
                                 121

-------
                                                    T= O
                                2B.L
                                                              T=4Shr
0
O
O
2;



t
L
1
(•
r
f-




u 	 / \
'J "+ " 3 •?••<:, i3 24
I DAY 1 n,iy Q
Figure  3-8(b).
Time histories  of NO following the edge  point of cloud 2B.L
top, and 2B.R,  bottom.  Curves labeled  "chemistry" represents
the true solution.
                                   122

-------
            5:-
         o
         X
                2B.L
         <
         UJ
         CJ

         o

Figure  3-8(c).
Time histories  of  ozone following the edge point  of  cloud 2B.L,
top, and 2B.R,  bottom.  Curves labeled "chemistry" represents
the true solution.

                    123

-------
            2.-
          O

          X
     CHEMISTRY

     ROM
T=G
                                • 2B.L
                                                          T*0
                                  20
                        DAY 1
                                3    '2   :6   20   24

                               ._. ^y 2             DAY 3
Figure 3-8(d).
Time histories of N02 following the edge point of cloud 2B.L,
top, and 2B.R, bottom.  Curves labeled "chemistry" represents
the true solution.
                                    124

-------
         c
         X
CHEW:STT?Y

ROM
         C

         <
                        2B.L
                                                   T= O
         O

         O
         O

         i ' '
         _!
         O
        CN
        O

        x
         a.
         o_
        o
g  C
        O
                      2B.R
                                                       T=0
o
o
h 1
0


L
1-
I"
Oi i i i i i i
U 4
1
\
\
\
, r , , V .
8 ,2 o „ ^- . S-2 '6 J2 2i
n.Y 1 DiY 2
Figure 3-8(e).   Time histories of olefin following the edge point  of  cloud  2B.L,
                top, and  2B.R, bottom.  Curves labeled "chemistry" represents
                the true  solution.
                                   125

-------
            2r
         O  r
                     CHEMISTRY

                     nOM
                                                  T=0
                 2B.L
Figure 3-8(f).
Time histories  of  PAN  following the edge point of cloud 2B.L,
top, and 2B.R,  bottom.   Curves labeled "chemistry" represents
the true solution.

                    126

-------
                                 SECTION 4








Case 3A:  Chemistry with transport and continuous sources





     The earlier experiments 1A, 2A and 25 investigated the ability of the



transport and chemistry algorithms to handle the homogeneous forms of their



differential equation counterparts.  That is, situations in which pollutant



species concentrations change only as a consequence of chemical  reaction,



horizontal transport or vertical mixing, but not source emissions.  Since



the primary role of the regional model is to assess the changes  in air



quality that would accompany given changes in the strengths of anthropogenic



sources, it is essential that the model possess the ability to simulate



accurately the fate of species released at arbitrary sites and times within



the model domain.  This is the feature we will examine in experiment 3A.





     One might assume that since our model can simulate isolated clouds



well, it could automatically handle the sequences of puffs that  compose the



plumes produced by continuous sources.  But this is not necessarily the



case.  In independent studies Schere (1984) and Yarmartino (1984) found



that when applied to a continuous source in a uniform flow, the Zalasak



(1979) scheme discussed earlier produced a sequence of large clouds rather



than a continuous plume.  The cause of this error is not certain but it is



likely due to the mechanism built into the scheme that prevents concentrations



from becoming negative.





     As we noted earlier, our transport algorithm can generate negative



concentration; but when this happens we merely reset the values to zero.





                                    127

-------
It has been argued that this procedure  is  unacceptable  because  it  leads  to



a violation of mass conservation  in the simulated  species.   In  our case  the



the deviations in total mass are  typically no  larger  than  a  few percent.



Indeed, .we found no evidence in experiments 2A and 2B that the  total  mass



error is large enough to create significant errors in predicted 03,  NO or



any of the other principal  pollutants.   In experiment 3A,  we will  consider



this matter further.





     The generation of negative concentrations is  associated with  truncation



error in the transport algorithm.  We saw  in experiments  2A  and 2B that  the



magnitude of this error is  proportional to the spatial  gradients in



concentrations.  Since the  grid size of the ROM is about  18  km, there are



a number of sites in the grid network where one cell  contains an entire



small city or source complex while surrounding cells  contain few if  any



sources.  These situations  create maximal  spatial  concentration gradients



and hence maximal truncation error.  Of particular concern to us are



situations in which neighboring,  isolated  sources  produce  parallel plumes.



It is conceivable that the  truncations error in these instances could cause



enough lateral exchange of  species between plumes  that  the chemistry



simulation would be adversely affected.  We will investigate this  matter



in experiment 3A.





     Another possible source of error that we  want to examine is the method



used to handle the outflow  boundaries of the model domain.  Although the



differential equation that  describes advection does not permit  the imposition



of restrictions at outflow  boundaries, the discrete equation that  we use to



model the differential equation requires such  values.  Therefore,  unless the



outflow conditions on the discrete equation, i.e., the  transport algorithm,





                                    128

-------
are properly chosen, the solutions produced by the model  will  not be accurate



facsimiles of the advection process.  In experiment 3A we will  examine



situations where plumes pass through the lateral  boundaries of  the domain.





   '  A final question of interest is the rate at  which errors  grow in the



simulated plume with travel time -- vis-a-vis model applications to multiday-



long distance transport — and the compounding of error that might result



as a plume encounters a new source after traveling for a  day or more.





     To summarize, our objectives in experiment 3A are:



(1)  To examine the ability of the transport algorithm to simulate simple



     continuous source plumes over multi-day travel periods;



(2)  To examine the effects of truncation error on the simulation of parallel



     plumes from isolated sources;



(3)  To determine the rate of error growth in plumes with travel  time and any



     compounding of error upon interaction of a plume with distant sources;



(4)  To investigate error levels at outflow boundaries;



(5)  To assess the impact of errors caused by the clamping of  negative



     concentrations in the transport algorithm.





     The experiment that we have devised to attain all these objectives is



as follows.  A collection of four "line" sources  of various widths,



illustrated in Figure 4-1, emit hydrocarbons and  NOX steadily  into a flow



field identical  to that employed in experiments 2A and 2B, namely a fluid



in solid body rotation of angular speed 
-------
                 c
                          i/a
                      \/z
                                SOURCE
                                 STRENGTHS
                                 (SEE TABLE 4-1 )
                                 DOMAIN
                                  BOUNDARY
Figure 4-1.
Locations and relative strengths of 4 line sources (b, c, e,
and f) simulated in experiment 3A.  Flow speed i = .02 radian
per time step.
                         130

-------
 source b.   The emission rate of each grid cell that composes a source is



 the product of the fraction shown in Figure 4-1 beside that cell and the



 base emission rates given in Table A-l, page 132.  The latter represent the



 highest actual emissions observed in the geographical area covered by the



 regional model.  Specifically, they were taken from the emissions inventory



 at a cell  in the vicinity of New York City.





     The experiment simulates a 58-hour period during which solar radiation



 varies in  the diurnal fashion implicit in the rate constants plotted in



 Figure 2-1, page 17.  During the entire simulation, pollutants are confined



 to layer 1 which has a constant, uniform depth of 300 meters in this experiment,





     The emissions and meteorological conditions simulated in this experiment



 have deliberately been made extreme so that error sources in the model  will



 be stimulated as strongly as is ever likely during actual applications.



 Consequently, the error levels exhibited in this test should provide a good



 measure of the upper bound that we could expect in actual applications.





     The rationale for the sizes and locations of the sources shown in



 Figure 4-1 that are used in experiment 3A is as follows.  First, sources b,



c, and e are positioned so that in the rotating flow field their plumes



will  move parallel  to each other with a distance of about 3 grid cells



 separating each plume.  The widths of these sources have been made different



 (2, 3,  and 5 grid cells widths, respectively) so that we can measure the



sensitivity of the model's accuracy to the widths of the sources simulated.



 Plumes  b, c, and e will  also allow us to infer the extent of lateral



exchange of species among plumes due to truncation error effects.





                                    131

-------
Table 4-1.  Base emission rates of species used for line sources
            in experiment 3A.  The emission rates of individual
            source cells are fractions (1/3, 1/2, 2/3, or 1) of
            the values shown here (see Figure 4-1).
        Species

         NO

         N02

         Olefin

         Paraffin

         Aldehyde

         Aromatic

         CO
Emission rate (pprr «  m
                  sec'1)
       4.91 •  10-3
       3.55
                -4
       10

1.53 .  10-3

3.21 •  lO-3

2.52 •  10-4

5.65 •  10~4

5.55 •  ID'2
                              132

-------
     Source f has been positioned so that the edges of the plumes generated



 by sources c and e will pass over the edges of source f after a travel  time



 of about 40 hours.   With this configuration of sources we will be able to



 measure the extent to which errors in the simulation of plume edges are



 compounded when plumes impact new sources after prolonged periods of travel.



 Finally, source b, the narrowest of the four sources, is positioned so  that



 its plume will encounter no additional  sources and will pass through the



 southern boundary of the model domain after a travel time of about 48 hours.



 The behavior of this plume will provide information not only on the magnitude



 of errors at outflow boundaries but also on the rate of growth of total



 error under the most severe conditions of lateral  concentration gradients



 that we are likely to encounter in actual applied  studies.





     The actual plumes produced by the four sources are shown in Figure 4-2,



 which is a plot of the contours of CO concentration at the end of the 58-hour



 simulation.  Although CO is not chemically inert,  variations in its



 concentration due to chemical processes'are small  enough that CO can be



 regarded as a conservative chemical tracer for our present purposes.  From



this viewpoint we note several qualitative aspects of the model performance



evident in Figure 4-2.





     The first is that lateral spread of the plumes due to truncation error



in the transport algorithm is nil inasmuch as the  concentration contours of



plumes b, c, and e form nearly perfect, concentric circles, as one would



expect in the source/flow configuration (Figure 4-1) simulated here.  (Keep



in mind that in experiment 3A the horizontal diffusivity K^ is zero.)  A



second point is that the peak value of concentration in each plume is well





                                    133

-------
                                      2   .2
Figure 4-2.
Isopleths of CO concentration (units = ppm)  at the end of the
58-hour period simulated in experiment 3A.   Letters b, c, e
and f refer to the sources shown in Figure  4-1.

                       134

-------
preserved, as Is evident in plume b.  Third, at the lower boundary where



plume b exits the model domain and where plume c touches the boundary,



there are no aberrations in the concentration isooletns that would signal



errors generated by the outflow boundary conditions in the transport



algorithm.  Finally, the isopleths show that the simulated plumes  are



continuous rather than disjointed, as Schere and Yamartino found in



applications of the Zalesak transport scheme to continuous sources.





      In order to obtain quantitative estimates of the model 's performance,



we will examine the predicted concentrations of each of the  23 species



along both cross-sections of the plumes normal to the air flow and along



Lagrangian trajectories within each plume.  The cross-sections reveal



spatial variations in accuracy at a given time while the plots of  concen-



tration following a Lagrangian trajectory show error behavior as a function



of travel time from a source.  Figure 4-3 is a schematic representation of



the simulated plumes (shown in Figure 4-2) that illustrates  some of the



cross-sections and trajectories that we will consider.  We will  use Figure



4-3 as an insert in all subsequent concentration plots to identify the



cross-section or trajectory to which the concentrations apply.





     One point to keep in mind in interpreting the concentration fields is



that due to photochemical reactions, many of the species undergo marked



temporal  variations that are synchronized with time-of-day.   One consequence



of this is that concentration isopleths of most reactive species do not



exhibit orderly plume patterns like the CO distribution shown in Figure 4-2.



An example of a temporally varying species is ozone.  Figure 4-4,  page 137,



shows isopleths of ozone at the same hour as the CO isopleths shown in





                                    135

-------
              LAGRANGIAN
               TRAJECTORY
          CROSS-SECTION
                                                        DOMAIN
                                                      BOUNDARY
Figure 4-3.   Schematic representation  of the continuous  plumes generated
            by sources b, c, e and f  in experiment 3A.  Examples are shown
            of a cross-section and a  Lagrangian trajectory.
                                136

-------
                                                     c
                                                                              -"1
Figure 4-4.
Isopleths of ozone concentration at the last hour,  0930 day  3,
of the line source simulation experiment 3A.

                       137

-------
Figure 4-2.  Since this particular plot  is  for  hour  0930,  which  is  only  3



1/2 hours after sunrise, ozone has not yet  been generated  from the  fresh



precursor  emissions of sources.b, c,  e  and f.   As a result  the  ozone



contours give the illusion that all  these sources  have  been  displaced  about.



45 degrees clockwise, i.e., downwind,  from  their actual  locations.   Because



of this complexity in the behavior of  the species  concentrations, a variety



of spatial and temporal cross-sections  are  required  to  form  a comprehensive



picture of the model's performance.   In  the remainder of this section  we



will present and discuss an assortment  of results  obtained in experiment 3A



that will allow us to formulate conclusive  statements regarding  each of  the



modeling questions raised earlier.





     We begin with a sequence of four  cross-section  plots  of CO  shown  in



Figure 4-5(a)-(d), pages 147-150.  The  profiles labeled chemistry represent



the true concentrations.  They were obtained in the  manner described in



Section 3 to generate the cross-sections shown  in  Figures  3-3 through  3-5.



The only difference is that in the present  instance, equation  (2-1) contains



an inhomogeneous term that represents  source emissions.  Sections a and  b of



Figure 4-5 show cuts through plumes b,  c and e  at  7  hours  and  34 hours



travel time, respectively.  Comparing  these two plots we see a  slight



increase in error in the predicted peak  concentration in each  plume with



travel time.  The figure also indicates  that the error  is  inversely propor-



tional to the plume width, as we saw earlier.  According to  Figure  4-5(b),



at the 34-hour point, the model underestimates  the peak concentration  in



plume e (five cells wide) by 10%; by 18% in the case of plume  c  (three



cells wide); and it underestimates the peak in  plume b   (two  cells wide)  by



about 23%.  The cross-sections shown in Figures 4-5(a), (b)  confirm the





                                    138

-------
observation that we made earlier in connection with the CO isopleths in



Figure 4-2 that there is not an appreciable widening of any of the plumes



beyond a few hours travel distance from a source.   At 52 hours distance,



Figure /-5(d) shows that the plumes from sources  c and e have merged with



that from source f to form a single, double-peaked plume.  As one can see



from the symbols along the abscissa of Figure 4-5 that designate the source



locations, the left-hand peak in the plume shown  in Figure 4-5(d), i.e.,



the peak closest to the flow vortex center, is composed only of material



from source e.  The model underestimates this peak by about 13%, which is



consistent with the error growth rate that we found earlier in plume e.



However, the right-hand peak in the plume in Figure 4-5(d) is made up of



material from both sources c and f; and it is underestimated by about 25%,



which is larger than either a 3-cell (source c) or 6-cell (source f) source



plume would produce at this point.  The cause of  this enhanced error is not



clear.  As we shall see shortly, it is not apparent in the concentration



profiles of other species.





     Figures 4-6(a)-(f), pages 151-156, show corresponding cross-sections of



ozone at a number of travel distances.  The first four sections of this



figure, (a)(d), show that up until the point where the plumes first encounter



source f, the model actually simulates the ozone  plumes more accurately



than the CO.  The maximum error is an underestimate of about 10% in plume



b.  The error in the predicted widths of the plumes is comparable to that



found with CO.  Figures 4-6(e) and (f) show that  after the plumes pass



source f, the ozone concentrations at the edges of both plumes c and e drop



markedly, due to reaction with NO emissions from source f.   In fact, the



background ozone that fills the space between plumes c and e before they





                                    139

-------
impact source f is completely erradicated at  the  44-hour travel  point



(Figure 4-6(e)), which is only a  short  distance downwind of  source  f.   The



error level in the simulated ozone at this point  is  an  underestimate of



peak values by about 10".  At the 52-hour travel  point,  Figure  4-6(f),  the



spatial variation of ozone has acquired a rather  complex shape  which is



simulated by the model to within  about  15% of the true  values.   However,



the model significantly overestimates the rate at which  ozone  is  replenished



along the centerline of plume f.   This  error  is undoubtedly  due to  errors



in the simulated NO and N02 concentrations in this region,  rather than  to



erronous lateral diffusion of ozone.





     The error level in the predicted NOg concentrations can be seen in



Figures 4-7(a)-(e), pages 157-161.  The first 3 sections of  the figure,



(a)-(c), show that the peak N02 concentration is  underestimated by  only a



few percent in plume e and by about 15% in the narrowest plume, b.   The



fractional error appears to grow  as the N0£ concentrations  decrease toward



zero.  Figure 4-7(d) indicates that just downwind of source f,  the  predicted



N02 concentration is too large by nearly a factor of two.  Since this  error



is grossly different than that apparent in Figure 4-7(a) at  a  comparable



distance downwind of sources b, c, and e, the large error in the f  plume



must be due primarily to errors in species that  are coupled to N02  chemically,



According to Figure 4-7(e), the error in the predicted N02 concentration



decreases rather rapidly with distance from source f.






     Cross-sections of the predicted and true olefin concentrations are



displayed in Figures 4-8(a)-(d).   The first two  of these indicate an error



level comparable to or lower than that we have seen in any of the species



considered thus far.  Downwind of source f, Figure 4-8(c) and  (d) show an



                                     140

-------
enhanced, asymmetrical error distribution that varies  from a  slight



overprediction on the left side of the plume to and underprediction on the



other side.  The latter error has the larger magnitude,  varying  from about



15% to 40%.  This is the second species in which we have seen a  significant



deterioration of model accuracy following the merger of  two plumes,





     The final species that we will  examine in cross-section  is  the hignly



reactive compound PAN.  Figures 4-9(a)-(f), pages 166-171, reveal  an error



pattern in this species similar to that seen in CO: 10%  to 15% underestimate



of peak concentration in plume b, c  and e prior to intercepting  source f,



and a somewhat larger error downwind of source f.  In  Figures 4-9(c)-(e) we



have added magnified plots of the concentration cross-sections to  show the



fractional error in the predicted PAN levels when the  concentration is very



low.  Except for the left hand side  of plume c, where  concentration is



underpredicted by about 50%, the relative errors within  the plumes are



comparable to that found in Figures  4-9(a) and (b) at  higher  concentrations.



At low concentrations, the fractional error is much larger in the  areas



between plumes because in these zones the weak fluxes  of material  generated



by truncation error quickly produce  concentration levels that are  comparable



to the ambient values.





     In summary, the cross-sections  of species concentration  give  evidence



of some compounding of error when plumes from one source cross other



sources downwind.  The magnitude of  the error amplification appears to be



species dependent.  Of the species we considered, ozone  and CO showed the



least increase in error while N02 showed the largest change.   In the latter



case the error increased from a 10%  underprediction prior to  plume merger





                                    141

-------
to a 100% overprediction immediately following plume  combination.   The zone



of largest error is confined to a small  area  right  around  the  second  source.



In the next set of concentration plots we will  look further at this phenomenon



of error amplification.





     The first of four sequences of concentration  profiles along  Lagrangian



trajectories is presented in Figure 4-10, pages 172-194.   Sections  (a)-(w)



of this figure compare the predicted and true concentrations of each  of the



23 simulated species along a 57-hour trajectory that  passes through the



center of source e.  This particular trajectory passes  through the  grid



cell adjacent to the left edge of source f approximately  40 hours downwind



of source e.  Figures 4-10(a) and (b) show that the predicted  NO  and  N02



concentrations are within about 10% of the true value during the  time that



the concentration levels are significant, namely during the first 24  hours



of travel.  Figure 4-10(c) indicates that the predicted ozone  concentration



is within 5% of the true value during the first 44 hours,  but  departs from



the correct level by about 10% beyond that point.   This slight increase in



the error level is undoubtedly due to interaction  of plume e with the plume



from source f.





     Figures 4-10(d)-(g), pages 175-178, indicate  that  all four hydrocarbon



species are simulated with accuracies better than  10% over the entire 57



hour travel period.  The paraffin concentration profile in Figure 4-10(e)



contains a rather noticeable perturbation at about the  44 hour point  (i.e.,



day 2, hour 20) where the trajectory passes source f.  The fact that  this



error fluctuation is quite localized supports the  conclusion drawn  earlier



in our analysis of the concentration cross-sections, that the compounding





                                    142

-------
 of  error upon the interception of a plume with another source is a localized



 phenomenon, at least for species like hydrocarbons, NOX, PAN and others that



 are active  in the photochemical process. •





     The profile of CO concentration shown in Figure •i-iO(h), page 179,



 provides evidence that the error compounding phenomenon is associated with



 the concentration undershoot phenomenon that we discussed earlier in our



 analysis of experiments 2A and 2B.  In particular, between hours 18 and 21



 of  day 2, which is the period that the Lagrangian trajectory through the



 center of source e passes the edge of source f, Figure A-10(h)  shows a



 slight negative perturbation in the predicted CO concentration.   The fact



 that the concentration "recovers" to its proper level  downstream of source



 f suggests that the undershoot zone has the character  of a standing wave



 that is locked to the source.  As each air parcel that composes  a plume



 enters the undershoot standing wave that surrounds an  isolated  source, the



 concentrations of all species in that parcel are disturbed from their



 equilibrium values.  The altered chemical reactions that this imbalance



 excites gives rise in the case of some species to new  concentrations that



 are more erroneous than those that entered the undershoot wave.   Most of



 the evidence we have seen thus far suggests that downstream of  the undershoot



 zone, error levels tend to return to their lower, original values.  Moreover,



 ozone,  which is the species of primary concern to us,  is one of the species



that is least affected by the undershoot phenomenon.





     These observations bring us back to the question  of whether a transport



algorithm that maintains positive definite concentrations, i.e., an algorithm



that does not generate the undershoot, would not be preferable  in modeling



applications such as this.  Two responses come to mind.



                                    143

-------
     First, the aberrations that are attributable to our transport algorithm



are locc  zed and are not seriously large.   As we have already noted,  the



model's ability to simulate ozone is practically untarnished by the undershoot



p-henomenon.  A second point is that the methods used in transport  schemes



to prevent negative concentrations may cause serious distortions in the  spatial



distribution and propagation speed of material (see Figure 3-2, pages  63-66)



that are potential sources of major, widespread errors.  The authors are



unaware of any study such as the present one in which a "positive  definite"



transport algorithm has been applied to chemically reactive material.



Therefore, despite the fact that designers  of both the NCAR and the Canadian



regional acid rain models have recently chosen transport schemes of the



positive definite type, there apparently is no quantitative evidence that



algorithms of this type are superior.





     Continuing with our analysis of the Lagrangian profiles, we refer the



reader to Figure 4-10(i)-(w), pages 180-194, for plots of the remaining 23



species.  Since there are no .significant aspects of any of these species



other than the phenomena we have already discussed, we will not elaborate



on any of these results.  We include them for completeness in our demonstra-



tion of the models overall simulation capability.  One counter-intuitive



characteristic of the species profiles shown in these figures is that the



highly reactive free radicals appear to be predicted more accurately than



the less reactive compounds.






     The second sequence of concentration profiles along a Lagrangian



trajectory is given in Figure 4-ll(a)-(w),  pages 195-217.  A major difference



between this trajectory and the one depicted in Figure 4-10 is that the





                                    144

-------
former actually crosses source f whereas the one represented in Figure 4-11



only skirts it  (see the inserts in Figure 4-11 for details).  We find on



comparing each  species plot in Figure 4-11.with its counterpart in Figure



4-10, that the  accuracy of the concentration predictions along this trajectory



is generally comparable or better than that found along the former trajectory.



One possible explanation of this is that the trajectory represented in



Figure 4-lQ passes through the center of source e; and as we have seen the



model is not able to maintain the full amplitude of narrow plumes.  By



contrast, the trajectory represented in Figure 4-11 passes through the



outer edge of source e where both the concentration and the curvature of



its profile are smaller.  The differences in the problems of simulating the



centerline of a narrow plume vs its edge is particularly evident in a



comparison of the alky! nitrate concentration time profiles given in Figures



4-10(1) and 4-11(1), pages 183 and 206, respectively.





     Plots of primary species, such as olefin, paraffin, and CO given in



Figure 4-ll(d), (e) and (h),.respectively, indicate that the fractional



error in the predicted concentrations downwind of the  second source (f) is



approximately the same as that downwind of the first source e.  Species



such as 03, (Figure 4-ll(c)), nitric acid (4-ll(j)), alky! nitrate (4-



11(1)), and a few others exhibit almost no sensitivity to source f, while



others such as  PAN (4-ll(k)), nitrate (4-ll(o)), and hydroperoxyl radical



(4-ll(q)) show enhanced error levels immediately after crossing source f



that subside eventually to their former levels some distance downstream.






     The final sequence of concentration profiles along Lagrangian trajec-



tories is given in Figures 4-12, pages 218-240 and 4-13, pages 241-263.





                                    145

-------
The former describes conditions along a path  that passes  through  the center of



source c and through the outer edge cell  of source f.   Figure  4-13  illustrates



concentrations within the plume from source b.   The errors  apparent  in



these two sets of profiles follow the same pattern as  those we have  already



discussed.  The principle difference is that  the magnitudes of the  errors



tend to increase as the width of the simulated  plume decreases, which we



have already been led to expect.  For example,  we find from Figures  4-10h,



12h and 13h that the fractional error in  the  simulated centerline CO concen-



trations in sources 5, 3 and 2 cells wide are 10%, 18%, and 24%,  respectively,



The corresponding errors in ozone are considerably lower:  4%,  8%  and 9%,



respectively.  These are well within what we  consider to  be acceptable



limits.  It is fortunate that ozone, which is the pollutant of primary



concern to us, is among the species simulated most accurately.  In  contrast,



the predicted concentrations of some of the nitrogen species such as nitrous



acid, nitric acid, alky] nitrate and others are in error  by as much  as  50%



or more in places.  We suspect that these differences in  accuracy reflect



differences in the character of the chemical  reactivity of each species.





     We leave further analysis of the results shown in Figures 4-12 and 13



to the reader.  A brief summary of the chief conclusions  is given in Section



1.
                                    146

-------
      CROSS-SECTION  PLO
      TEST
      DATE
      HOUR
       D_
       O.
       o
       *
       LJ
       o
       2
       o
       o
      o
      o
UNESOURCE EMISSION TE3
                D
             J	I
                    D  D   D:;  di   G  0  til D:;
                                   D
                                                Dn    SOURCE LOCATCN/
                                                U ^	SaATlVE MAGNITUDE
         0.8 1.1 1.3 1.6 1.8 2.1 2.3 2.B 2.3 3.1 3.4  3.6 3.3 4.1 4.4 4.6 4.3
                  DISTANCE  FROM VORTEX CENTER (100KM)
Figure 4-5(a).
    Comparison of predicted (solid curve) and true (dashed)
    CO concentrations in experiment 3A along the cross-section
    indicated in the insert.  (Travel time = 7 hrs from sources
    b, c and e).
                                    147

-------
CROSS-SECTION PLOT

TEST  : LINESOURCE EMISSION TEST
DATE  : 79216
HOUR : 150000
                                                             SOURCE LOCATION/
                                                          ,*	*EUTTVE MAGNfTUDE .
    0.8 1.1 1.3 1.6 1.8 2.1 2.3 2.6 2.9 3.1 3.4 3.6 3.9 4.1 4.4 4.6 4.5
             DISTANCE FROM  VORTEX CENTER (100KM)
Figure  4-5(b).   Same as 4-5(a) except travel  time = 34 hrs
                                148

-------
 CROSS-SECTION1  PLO
      : LJNESOURCE EMISSION TES'
 DATE  : 79217
 HOUR  : 01QOCO
     I    !   I
               D  D   D:  dJ   0  fl  DJ D:i
                            J	L
0
                   SOURCE LOCATION/
                        MAGNITUDE
    0.8 1.1 1.3  1.6 1.8 2.1 2.3 2.6 2.9 3.1 3.4 3.6 3.9 4.1 A.4 4.3 4.3
             DISTANCE  FROM VORTEX CENTER  (100KM)
Figure 4-5(c).  Same as  4-5(a)  except  travel  time = 44 hrs
                                149

-------
CROSS-SECTION PLOT
TEST  :  LJNESOURCt EMISSION TEST
DATE  :  79217
HOUR :  090000
                                                             SC'JRCE LOCATION/
                                                             .ELATWE MAGNITUOE
    O.B 1.1 1.3 1.6 1.8 2.1 2.3 2.6 2.9 3.1 3.4 3.S 3.9 A.i 4.4 4.6 4.9
             DISTANCE FROM VORTEX CENTER  (100KM)
Figure 4-5(d).   Same  as 4-5(a) except travel time  = 52 hrs.

                                '50

-------
     CROSS-SECTION PLOT

     TES~  :  LJNESOURCE EMISSION TEST
     DATE  : 79215
     HOUR :  120000
      X
T
 r
 c
      8:   E
                  CHEMISTRY

                  NEROS
                   D   D  D." d."   0   C  0]  D:=
                                                        SOURCE LOCATION/
                                                            VE MAGNITUDE
         0.8 1.1 1.3 l.B  1.8 2.1 2.3 2.6 2.9  3.1 3.4 3.S 3.9 
-------
 CROSS-SECTION  PLO
 TEST
 DATE
 HOUR
LINESO'JRCE EMISSION TEST
79215
180QOO
            J	L
               D   D   D:; rij  D   D  til  0:;
                                   n
                                                      SOURCE LOCATON/
                                                   -^—RELATIVE MAGNITUDE
     0.8 1.1  1.3 1.6  1.8 2.1 2.3 2.6 2.9 3.1 3.4 3.6 3.9 4.1  4.4 4.6 4.9
             DISTANCE FROM VORTEX CENTER (100KM)
Figure 4-6(b).   Same  as  4-6(a)  except travel time =  13  hours.

                                152

-------
CROSS-SECTION PLOT
TEST  : LINESOURCE EMISSION TEST
DATE  : 79216
HOUR : 060000
o

*  E
2 4^-

Q.  E
             CHEMiSTW

             NEROS
              D   D
                                      0::
                                      _L
D
D  D
                   SOURCE LOCATION/
                -^—RELATIVE
    0.8 1.1 1.3 1.6 1.8 2.1  2.3 2.5 2. 9 3.1 3.4 3.6 3.9 4. 1 4.4  4.6 4.9
             DISTANCE FROM VORTEX CENTER  (100KM)
Figure 4-6(c).  Same  as  4-6(a) except travel time  = 25 hours.

                                153

-------
 CROSS-SECTION PLOT

 TEST  : L1NESOURCE EMISSION TEST
 DATE  : 79216
 HOUR : 150000
 -  5

 O
 T~
 X
  D_
  Q_
r

CHEMISTRY

NEROS
            a  D   D  0::  dJ  0   fl ti.!  D:;
                                      D
                                        Q  0
                                                         SOURCE LOCATION/
                                                         -ftOATiVE MAGNITUDE
     0.8  1.1 1.3 l.S l.B 2.1 2.3 2.6 2.9 3.1 3.4 3.6 3.9 4.1 4.4 4.6 4.9
              DISTANCE  FROM VORTEX CENTER (100KM)
Figure 4-6(d).   Same as 4-6(a)  except travel time  = 34 hours.


                                 154

-------
  CROSS-SECTION PLOT

  TEST
  DATE
  HOUR
      LINESOURCE EMISSION TEST
      79217
      01COOO
T-  3

o


X
      r
  O 3p
  <
  ct:
  LJ
  o
  z
  O
  O
               CHEMISTRY


               NETOS
             D   D   D  a.-  di   !.i  u  D:;  D.~  D
                                                     D  D
 SSURCE LOOTION/
-RELAKYE MACNmjD£
     0.8 1.1  1.3 1.6  1.8 2.1 2.3 2.6 2.9 3.1 3.4 3.6 3.9 4.1  4,4 4.B  4.3

               DISTANCE FROM VORTEX CENTER  (100KM)
Figure 4-6(e).   Same as 4-6(a)  except travel time  =  44 hours.


                                 155

-------
 CROSS-SECTION PLOT
 TES^  : UNESOURCE EMISSION TEST
 DATE  : 79217
 HOUR : 090000
               D   D  It  &   ii  0  03 D::  D
D  D
 SOURCE LOCATION/
-SELATTvt MACNfTUDE
     0.8 1.1 1.3 1.6 1.8 2.1 2.3 2.6 2.8 3.1 3.4 3.6 3.9 4.1 4.4  4.6 4.9
              DISTANCE FROM  VORTEX CENTER  (100KM)
Figure 4-6(f).   Same as 4-6(a) except travel  time = 52 hours.


                                156

-------
     CROSS-SECTION PLOT
     TEST    LINESOURCE EMISSION TEST
     DATE    79215
     HOUR   120000
      O

      X
      CL
      CL

      Ld
      O
      Z
      O
      O

      CNJ
      O
  CHEMISTRY

  NEROS
                   D   D  D:;  dJ   13  13  DJ  D::
                               D
D  D
                                                  SOURCE LOCATION/
                                                 —RELATIVE MAGNITUDE
         0.8  1.1 1.3 1.6 1.8 2.1 2.3 2.6 2.9 3.1 3.4 3.6 3.9 4.1 4.4 4.6 4.9
                  DISTANCE  FROM VORTEX CENTER  (100KM)
Figure 4-7(a).
Comparison of  predicted (solid curve) and true  (dashed)
N02 concentrations  in  experiment 3A along the cross-section
indicated in the  insert.   (Travel time = 7 hrs  from sources
b, c and e).
                                     157

-------
CROSS-SECTION PLOT
TEST
DATE
HOUR
      UNESOURCE EMISSION TEST
      79215
      180000
-  2r
o  f-
Q_
Q_
O
 CE
 J—

 UJ
 O
 •z.
 o
 o
CM
O
    \~
    r
             CHEMISTRY

             NEROS
                                          D          D   0
                                          J	I	J	J.._l.
                                                              SOURCE LOCATON/
                                                           -fe—HELATTYE MAGMfaiDE
   0.8 1.1  1.3 1.6  1.8 2.1 2.3 2.6 2.9 3.1 3.4 3.6 3.9 4.1 4.4 4.6 4.9
             DISTANCE FROM  VORTEX CENTER  (100KM)
  Figure 4-7(b).   Same as 4-7(a)  except travel time  =  13 hours.

                                158

-------
CROSS-SECTION PLOT

TEST  :  LINESOURCE EMISSION TEST
DATE  :  79216
HOUR :  06QOOO
 X   '
 D_
 D_
             CHEMISTRY

             NEROS
 a:
    1-
 LU   ;
 O   j-

 o   L
 O   !
              D   D   D:;  dl  0   IJ  til  D:; a
                                    J	1.

Dn    SOURCE LOCATION/
II -c—RELATIVE MAGNITUDE
    0.8 1.1 1.3 1.6 1.8 2.1 2.3 2.6 2.9 3.1 3.4 3.6 3.9 4.1 4.4 4.6 4.9
             DISTANCE  FROM VORTEX CENTER (100KM)
   Figure  4-7(c).  Same as  4-7(a)  except travel  time = 25 hours.

                                 159

-------
CROSS-SECTION PLOT

TEST  : UNESOURCE EMISSION TEST
DATE  : 79217
HOUR : 010000
 o  T
 x   i-
 Q-   _
 Q.
 UJ
 o
 z
 o
 o

 CN
 O
D  D   D:i  dJ  D   |j D:;  D:;  D
                                                    D  D
 SOURCE LCCATON/
-HELATiVE
    0.8 1.1 1.3 l.S  1.8 2.1 2.3 2.S 2.9 3.1 3.4 3.6 3.9 4.1 4.4 4.6  4.9

             DISTANCE FROM  VORTEX CENTER (100KM)
   Figure 4-7(d).  Same as 4-7(a)  except  travel  time = 44 hours.


                                160

-------
CROSS-SECT! ON PLOT

TEST  :  LINESOURCE EMISSION TEST
DATE  :  79217
HOUR :  090000
-   2

O
T—

X
CL
Q.
CJ
•z.
O
O
CN
O
CHEMISTRY

NEROS
              D   0  D:;  dJ  D   D'tiJ  0,
                             D
Q  D
                                                 SOURCE LOCATION/
                                                -nEUTTVE MAGNfTUDE
    0.8 1.1  1.3 1.6  1.8 2.1 2.3 2.6 2.9 3.1 3.4 3.6 3.9 4.1 4.4 4.6 4.9
             DISTANCE  FROM VORTEX CENTER (100KM)
  Figure  4-7(e).   Same as  4-7(a)  except travel  time = 52 hours.

                                161

-------
      CROSS-SECTION PLOT

      TEST  : UNESOURCE EMISSION TEST
      DATE  : 79215
      HOUR  : 12GOOO
                   CHEMISTRY

                   NEROS
       o 3r
    D   D  D:; dl  !.!   !J  til  D- D
                                                             Dn

                                                             U
CURCE LOCATION/
~ ~ ~ WAGNITlj-Oe
          0.8  1.1 1.3 1.6 1.8 2.1 2.3 2.6 2.9 3.1 3.4 3.6 3.9 4.1 4.4 4.8 4.9
                   DISTANCE FROM  VORTEX CENTER (100KM)
Figure 4-8(a).
Comparison of predicted  (solid  curve)  and true (dashed)
olefin concentrations  in  experiment  3A along the cross-section
indicated in the insert.   (Travel  time = 7 hrs from sources
b, c and e).
                                     162

-------
  CROSS-SECTION PLOT

  TES~    LINESO'JRCE -MISSION ~ST
  DATE    79215
  HOUR   180000
   X
               CHEMISTRY

               NEROS
   o
LJ
O
O
O
LU
_J
O
i
!
. r
nr
i 	 i

n D D D:= riJ D D DJ D:: D Q D -
. . . i . i i i i 	 i.i i. i .. j 	 i - i 1.1 i
                                                               SOURCE LOCATION/
                                                              RELATIVE MAGNITUDE
      0.8 1.1  1.3 1.6  1.8 2.1 2.3 2.6 2.9 3.1 3.4  3.6 3.9  4.1 4.4 4.6 4.9
               DISTANCE FROM  VORTEX CENTER (100KM)
Figure 4-8(b).   Same  as  Figure 4-8(a) except travel  time = 13 hours.


                                163

-------
  CROSS-SECTION PLOT
        UNESOURCE EMISSION TEST
        79217
        010000
                                                        On    SOURCE LOCATION/
                                                        M ^	REU7TVE MAGNITUDE
     O.B 1.1 1.3 1.6 1.8 2.1 2.3 2.6 2.9 3.1 3.4 3.6 3.9 4.1 4.4 4.6 4.9
              DISTANCE FROM  VORTEX CENTER (100KM)
Figure 4-8(c).  Same as Figure  4-8(a)  except  travel  time = 44 hours.

                                164

-------
                    CROSS-SECTION  PLOT

                    TEST   : LINESOURCE EMISSION TEST
                    DATE   : 79217
                    HOUR  : 090000
                    CM  5r-
                     o

                     X
                     Q_
                     CL
                     O

                     !<
                     LJ
                     O
                     ;z
                     o
                     o

                     UJ

                     o
                        4h
                        2-
1-
         CHEMISTOT


         NEROS
                               O
D  D   Eh  rij   13  0  CL! D::  D
D
                                                          SOURCE U3CA710N/
                                                       ~* - REUTT
                                                                                    TTVE MACN(TUDE
                        0.8  1.1 1.3  1.6 1.8  2.1 2.3  2.6 2.9  3.1 3.4  3.6 3.9  4.1 4.4  4.6 4.9

                                 DISTANCE  FROM VORTEX CENTER (100KM)
                   Figure 4-8(d).   Same as Figure  4-8(a) except travel  time = 52 hours.


                                                    165

-------
       CROSS-SECTION PLOT

       TEST  : LiNESOURCE EMISSION TEST
       DATE  : 79215
       HOUR  : 12COOO
          2r
       O


       X
       D_
       Q_
       O
       O
       z:
       o
       O
       <
       Q_
   CHEHiSTKY


   NtROS
              J	L
                     D  D   D:  riJ  D   13  DJ  0:;
                                D
D
                                                   ,;JURCE LOCATION/.
                                                   naATTVE
           0.8 1.1 1.3 1.6 1.8  2.1 2.3 2.6 2.9 3.1  3.4 3.6 3.9 4.1 4.4 4.6 4.9

                   DISTANCE FROM VORTEX  CENTER  (100KM)
Figure 4-9(a).
Comparison of predicted  (solid  curve)  and true (dashed)

PAN concentrations in experiment  3A along the cross-section

indicated in the insert.   (Travel  time = 7 hrs from sources

b, c and e).
                                     166

-------
 CROSS-SECTION  PLOT
 TEST
 DATE
 HOUR
LINtSOURCE EMISSION TEST
79215
180000
              CHEM'SW
                       	'•—/
        D  D   D:;  d'j  U   I.!  m  D:: D
                                                              SOURCE LOCATION/
                                                              	,TYE MACNITJOE
             1   L. .
_J	l ._ L-
                             .L	1.	L ..J	I	i._-
     0.8 1.1 1.3 1.5 1.8 2.1  2.3 2.6 2.9 3.1  3.4 3.6 3.9 4.1  4.4 4.6 4.9
              DISTANCE FROM VORTEX  CENTER  (100KM)
Figure 4-9(b).   Same as 4-9(a) except  travel  time = 13 hours.

                                167

-------
       CROSS-SECTION  PLO
TEST
DATE
HOUR
Cvi

O


X
             LJNESO'JRCE EMISSION TEST
             79216
             060000
       D_
       O_
       LU
       o

       o
       O
       <
       D_
                   CHEMISTRY


                   NEROS
                     D  0   D:  a\   D  D  DJ  0:;  D
                                                    0  0
   SOURCE LOCATION/
-*	RELATE	
          0. 8 1. 1 1.3  1.6 1. 8 2. 1  2.3 2.6 2.9 3.1 3. 4 3.6 3. 9 4. 1 4. 4 4. 6 4. 9

                   DISTANCE FROM VORTEX CENTER  (100KM)
Figure 4-9(c).
          Same as Figure 4-9(a) except  travel  time = 25 hours.

          Insert shows magnified plot of  the  predicted and true PAN
          concentration distributions.
                                     168

-------
       CROSS-SECTION  PLOT
       TEST  : LJNESOURCE EMISSION TEST
       DATE  : 79216
       HOUR  : 150000
                    D   D  D:;  dJ   [I  !]  DJ
SOURCE LOCATION/
     UACNfTUDE
          O.S 1.1 1.3 1,6 1.8 2.1 2.3 2.B 2.9 3.1 3.4 3.6 3.9 4.1  4.4 4.6 4.9
                   DISTANCE FROM  VORTEX CENTER (100KM)
Figure 4-9(d).  Same as Figure 4-9(a)  except  travel  time = 34 hours.
                Insert shows magnified plot of  the predicted and true PAN
                concentration distributions.

                                     169

-------
       CROSS-SECTION PLOT

       TEST  : UNESOURCE EMISSION TEST
       DATE  : 79217
       HOUR : 010000
       CM  2

       O


       X
       Q_
       CL
        LJ
        o
        z
        o
        o
       <
       Q.
                      6-
                    <
                    cr
                    O 3-
                    "z.  :
                    C  :
                    o 2-
   <  »r
   O-  ;

      0^-
            CHEKISTW   *


            NOJOS
             0  Q  Q.. a.
D  0
    D   D
                                           J  Dn  D
H  pi    !!Cy9CJ UX
Ij  y ^_^i>T,-E «»
                 On     SOURCE LOCATION/
                 \\  -*	KaATTVt MAGNITUDE
           0.8 i.i 1.3 1.6 1.8 2.1  2.3 2.B 2.9 3.1 3.4 3.B 3.9 4.1  A.4 4.6 4.9

                    DISTANCE FROM VORTEX CENTER (100KM)
Figure 4-9(e).
Same as Figure 4-9(a) except travel  time = 44 hours.

Insert shows  magnified plot of  the predicted and  true PAN

concentration distributions.
                                      170

-------
 CROSS-SECTION  PLOT

 TEST  : LINESOURCE EMISSION TEST
 DATE  : 79217
 HOUR : 090000
 CM   2r-

 O


 X
  Q_
  Q_
  o


  o:
  i—

  LU
  O

  o
  o
 <
 Q.
CHEMISTRY


NEROS
                               \:
                                                        On     SOUR
                                                        H  « — RaA
                                                              URCE
    0.8 1.1 1.3 1.6 1.8 2.1 2.3 2.6 2.9 3.1 3.4 3.6 3.9 4.1 4.4 4.6 4.9

             DISTANCE  FROM VORTEX CENTER  (100KM)
Figure 4-9(f).  Same as Figure  4-9(a)  except  travel  time = 52 hours


                                171

-------
     TIME TRACK PLOT
     UNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW - 28.25 COLUMN - 35.47
               CHEMISTRY

               ROM
o
                              II 11 I  I L I J
                                        ILfl._l_l!tlJilJtilt!tt_ll
          4   8    12   16   20    24    4    3   12    16   20   24    4    8

                 DAY1	i	   DAY 2	\  DAY 3
                       TIME (HOUR  OF DAY)
  Figure 4-10(a).   Comparison of predicted  (dash-dot) and true NO concentration
                   (solid curve) along a Lagrangian trajectory that  passes
                   through the center of source e, experiment 3A.
                                     172

-------
     TIME TRACK  PLOT

     UNESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW -  29.25  COLUMN -  35.47
,-   2,

O   |

X
CL
Q_
O
LJ
o
~z.
o
o

(N
O
CHEMISTRY


ROM
               8   12    16   20   24


                 DAY 1            I
                            3   12

                              DAY 2
16    20
24   4    8

 I  DAY 5
                        TIME  (HOUR OF  DAY)
  Figure 4-10(b).
   Comparison of predicted  (dash-dot) and true  N02 concentration

   (solid curve) along  a  Lagrangian trajectory  that passes
   through the center of  source e, experiment 3A.
                                     173

-------
  TIME TRACK  PLOT
  UNESOURCE EMISSION SIMULATION TEST
  INITIAL LOCATION OF TRACK : ROW - 29.23 COLUMN - 35.47
                I  i . . I . i  ,  i . i I  L . . I
            8   12    16   20   24

              DAY 1	i
                        8   12    16   20   24    4    8

                          DAY 2	1   DAY 3
                     TIME (HOUR  OF DAY)
Figure  4-10(c).
Comparison of predicted  (dash-dot) and true  ozone concen-
tration  (solid curve) along a Lagrangian trajectory that
passes through the center  of source e, experiment 3A.
                                  174

-------
TIME TRACK PLOT
LJNESOURCE EMISSION SIMULATION TEST
INITIAL LOCATION OF TRACK : ROW - 29.25 COLUMN -  35.47
          CHEMISTRY

          ROM
                          iii t i i i t i  t 1 i i I  I I i i I _i i i i^i""]""
              12   16   20   24
            DAY 1
                                            12

                                          DAY 2
18   20
24
          I  DAY 3
                  TIME (HOUR  OF  DAY)
Figure  4-10(d).
              Comparison of predicted (dash-dot) and  true olefin concen-
              tration (solid curve) along a Lagrangian trajectory that
              passes through the center of source e,  experiment 3A.
                                 175

-------
   TIME TRACK  PLOT
   UNESOURCE EMISSION SIMULATION TEST
   INITIAL LOCATION OF TRACK : ROW - 29.25 COLUMN -  35.47
                 I , I , I I  , , 1 , , ,  I , , , I ,  , , i_i I I  I I I I I I  I I I , , I  I , , , I I  , , I
            8   12   16   20   24

              DAY 1	i
8   12   16   20   24

  DAY 2            |
  4    3

DAY 3
                     TIME  (HOUR  OF  DAY)
Figure  4-10(e).  Comparison of predicted (dash-dot) and true  paraffin
                concentration (solid curve) along a Lagrangian trajectory
                that passes through the center of source e,  experiment 3A.
                                  176

-------
     TIME TRACK  PLOT
     UNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW - 29.25  COLUMN - 35.47
eg
O

x
Q.
CL
    B
I
Ld
O  3

O
O  2
    OLa_L
              I ... I ... I ... I ... I ... I  ... I ... I .,, I  ... I ... I ... 1 ... I .
              8   12   16   20   24

                DAY 1|
  12   16   20

DAY 2
                                                            24    4    3

                                                             .  DAY 3
                       TIME  (HOUR OF  DAY)
   Figure  4-10(f).
                   Comparison of predicted  (dash-dot) and true aldehyde
                   concentration (solid curve) along a Lagrangian trajectory
                   that passes through the  center of source e, experiment 3A.
                                     177

-------
  TIME TRACK  PLOT
  UNESOURCE EMISSION SIMULATION TEST
  INITIAL LOCATION OF TRACK : ROW - 29.25 COLUMN - 33.47
            8    12   16   20   24
                        8    12   16   20   24
                    TIME (HOUR  OF DAY)
Figure  4-10(g).
Comparison of predicted (dash-dot) and true  aromatic
concentration (solid curve) along a Lagrangian trajectory
that passes through the center of source e,  experiment 3A.

                  178

-------
Q_
Q.
<
ct:
UJ
o
2
o
o
o
o
     TIME TRACK PLOT

     UNESOURCE EMISSION SIMULATION TEST

     INfTTAL LOCATION OF TRACK : ROW - 2935
               CHEMISTRY


               ROM
w? 1 ,
4
1
i i 1 i i i I i
8 12
DAY
, , I , ,
16
1
i 1 i
20

, . 1 . .
24
1
1 i t
4

1 .
8

i i 1 i i i 1 i i i 1 t
12 16 20
DAY 2
1 1 1
24
|
i i 1
4
DAY
i i

3
1 i i
8

                       TIME  (HOUR OF  DAY)
  Figure 4-10(h).  Comparison  of predicted (dash-dot) and true  carbon monoxide

                  concentration (solid curve)  along a Lagrangian trajectory

                  that passes through the center of source e,  experiment 3A.


                                    179

-------
  TIME TRACK PLOT
  UNESOURCE EMISSION SIMULATION TEST
  INITIAL LOCATION OF TRACK : ROW - 29.25  COLUMN - 35.47
           8    12   16   20   24

              DAY 1
                          DAY 2
I  DAY 3
                    TIME (HOUR OF DAY)
Figure 4-10(i).
Comparison of predicted  (dash-dot) and true nitrous acid
concentration (solid curve) along a Lagrangian trajectory
that passes through the  center of source e, experiment 3A.
                                  180

-------
o

X
Q_
CL
 -
bJ
O
z:
o
o
     TIME TRACK  PLOT

     UNESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW - 29.23
CHEMISTRY

ROM
                 .I,.,I ... I ... I  ... I ... I ... I ... I ... I  ... I ... I ... I .
              8   12   16   20   24   4    8   12   18   20  24    4   8
                DAY 1
                                 I
                              DAY 2
I  DAY 3
                       TIME  (HOUR  OF DAY)
   Figure 4-10(j).
    Comparison of predicted  (dash-dot) and true nitric  acid
    concentration (solid curve) along a Lagrangian trajectory
    that passes through the  center of source e, experiment 3A,
                                    181

-------

-------
     TIME  TRACK  PLOT
     UNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW - 29.25

-------
cs
O

X
Q_
Q_
a:
Ld
O
Z
O
O
O
CM
X
     TIME TRACK PLOT
     LJNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW - 29.23
    6-
    2-
              CHEMISTRY

              ROM
                                                                       /
                                                                    s
',,,],,
i • i M i
1
i_Lu i 1 u_
8 12
DAY 1
iJ-n
16

20 24
I
i I i i
4
i 1 i
8
i i 1 i i
12
DAY 2
i 1 t
16
' 1 i
20
j 1 i
24
|
i i 1 i i
4
DAY 3
I i i
8
                       TIME (HOUR OF DAY)
   Figure  4-10(m).
                  Comparison of predicted (dash-dot) and true hydrogen
                  peroxide concentration (solid  curve) along a Lagrangian
                  trajectory that  passes through the center of source e,
                  experiment 3A.
                                     184

-------
     TIME  TRACK  PLOT
     LJNESOURCE EMISSION SIMULATION TEST

     INfTlAL LOCATION OF TRACK : ROW - 29.25 COLUMN - 35.47
o>  10

o
x-  9
X
    a
o_
Q_
g
!•
LU
O
Z
O
o
    cP-
                             I I I I I I t  I I I 1
                      1 I I I  1 '-• I I I t  I I I I I I I I  I t I I I
                  12

                DAY 1
16   20
24
  12

DAY 2
16   20
24
8
                                       DAY 3
                      TIME (HOUR OF  DAY)
   Figure 4-10(n).  Comparison of predicted  (dash-dot) and true atomic
                  oxygen concentration  (solid curve) along a Lagrangian
                  trajectory that passes through the center of source e,
                  experiment 3A.
                                    185

-------
  TIME  TRACK  PLOT
  UNESOURCE EMISSION SIMULATION TEST
  INITIAL LOCATION OF TRACK : ROW -  29.25  COLUMN -  35.47
               12   16    20   24
                           12   16    20   24
              DAY 1
                          DAY 2
I  DAY 3
                    TIME  (HOUR OF DAY)
Figure  4-10(o).
Comparison of predicted  (dash-dot) and true  nitrate concen-
tration  (solid curve) along a Lagrangian  trajectory that
passes through the center of source e, experiment 3A.

                  186

-------
  TIME TRACK PLOT
  LJNESOURCE EMISSION SIMULATION TEST
  INITIAL LOCATION OF TRACK : ROW - 29.25 COLUMN - 35.47
                                                       ii i i I I  ! I I I i I  i i
           8   12   16   20   24
8   12   18   20   24
              DAY 1
  DAY 2
I  DAY 3
                    TIME (HOUR OF DAY)
Figure  4-10(p).  Comparison of predicted  (dash-dot) and true  hydroxyl concen-
                tration  (solid curve) along a Lagrangian trajectory that
                passes through the center  of source e, experiment 3A.
                                  187

-------
If
o

X
Q.
CL
g

£
LJ
O
•z.
o
o


-------
     TIME TRACK  PLOT
     LJNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW - 29.23  COLUMN - 35.47
•* '
O
X R
6
S
Q_
5>5
~Z.
— 4
^
£ 3
LJ
O
^M ir->jifvrrTw
• 	 ROM A
/ \
A
r 1
: 1 j
_ [/
- I
/
O
O
X
    1-
              8   12   16    20   24

                 DAY 1            .
                        8    12   16   20   24
         8
                          DAY 2
I  DAY 3
                       TIME  (HOUR OF  DAY)
   Figure 4-10(r).
Comparison of predicted (dash-dot) and true pernitric acid
concentration (solid curve) along a Lagrangian trajectory
that passes through the center of source e, experiment 3A.
                                     189

-------
  TIME TRACK  PLOT
  UNESOURCE EMISSION SIMULATION TEST
  INITIAL LOCATION OF TRACK : ROW - 29.25  COLUMN -  35.47
            CHEMISTRY

            ROM
            8    12   16   20   24
                        8   12    16   20   24
              DAY 1
                          DAY 2
DAY 3
                    TIME (HOUR  OF  DAY)
Figure  4-10(s).
Comparison of predicted (dash-dot) and true alkoxyl  radical
concentration (solid curve)  along a Lagrangian trajectory
that passes through the center  of source e, experiment  3A.
                                   190

-------
O

X
Q_
CL
O

^
a:
Ld
o
-ZL
O
O

CN
O
o:
     TIME TRACK PLOT
     UNESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW - 29.25
CHEMISTRY

ROM
              8   12   16   20   24

                 DAY 1   	i
                            8   12    16   20   24    4    &

                              DAY 2i   DAY 3
                       TIME (HOUR OF DAY)
   Figure  4-10(t).  Comparison of predicted (dash-dot) and true alkylperoxy
                   radical  concentration (solid  curve) along a Lagrangian
                   trajectory that passes through the center of source e,
                   experiment 3A.
                                     191

-------
     TIME TRACK PLOT

     LJNESOURCE EMISSION SIMULATION TEST

     INmAL LOCATION OF TRACK : ROW - 29.25 COLUMN - 35.47
O


X
GL
Q_
o


a:

LJ
o
O


o
CM
ir
CHEMISTRY


ROM
              8    12   16   20   24


                 DAY 1
8   12   16   20   24


  DAY 2
                                                 DAY 3
                       TIME (HOUR  OF DAY)
   Figure 4-10(u).
    Comparison  of predicted (dash-dot) and true alkoxy

    radical  concentration (solid curve) along a Lagrangian

    trajectory  that passes through  the center of source e,

    experiment  3A.
                                     192

-------
  TIME TRACK PLOT
  UNESOURCE EMISSION SIMULATION TEST
  INflTAL LOCATION OF TRACK : ROW - 29,23 COLUMN - 35.47
           8   12   16   20   24
S   12   18   20   24
              DAY 1
  DAY 2
I  DAY 5
                    TIME  (HOUR OF DAY)
Figure 4-10(v).  Comparison of predicted  (dash-dot) and true peroxyacyl
                radical  concentration (solid curve) along  a Lagrangian
                trajectory that passes through the center  of source e,
                experiment 3A.
                                  193

-------
     TIME TRACK  PLOT

     LJNESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW - 29.25
CO

O


X
D_
Q_
I
~z:
LlJ
O
z:
o
O


-------
LJ
O
•z.
O
O
     TIME TRACK PLOT
     LJNESOURCE EMISSION SIMULATION
     INITIAL LOCATION OF TRACK : ROW - 31.61 COLUMN -  37.03
CM   2

O

X
CL
Q.
    QLJ,
                                                        A.
4-1
                              I I I I  I I I
                                 _1 I  I I I I
                                                            -•'••• I "
            12

          DAY 1
16   20
24
  12

DAY 2
18   20
24
                                                                       8
                                                             1  DAY 3
                       TIME (HOUR  OF DAY)
  Figure  4-ll(a).  Comparison  of predicted (dash-dot) and true NO concentration
                  (solid curve) along a Lagrangian trajectory that  passes
                  through the outer most grid cell of source e,  experiment 3A.

                                    195

-------
     TIME  TRACK  PLOT
     UNESOURCE EMISSION SIMULATION
     INHTAL LOCATION OF TRACK : ROW
               31.61
X
Q_
Q_

or
    4-
CHEMISTRY

ROM
                DAY 1
S   12   18   20   24

  DAY 2           ,
                                                                      8
                                                DAY 3
                       TIME  (HOUR  OF DAY)
  Figure 4-ll(b).
    Comparison of predicted (dash-dot) and true N0£  concentration
    (solid curve) along a Lagrangian trajectory that passes
    through the outer most grid cell of source e,  experiment 3A.
                                    196

-------
     TIME TRACK  PLOT

     UNESOURCE EMISSION SIMULATION

     INrTTAL LOCATION OF TRACK : ROW -  31.6.1
,-   2

O

X
Q_
Q_
g
s
UJ
O
:z
o
O
O
CHEMISTRY

ROM
      ...i... i,., i,,, i... i... i... i... i... i... i... i... i... i... i
  12   16   20   24

DAY 1           i
                           3   12   16   20   24
                                                                     8
                                            DAY 2
                                             I   DAY 3
                       TIME  (HOUR  OF DAY)
   Figure 4-ll(c).
    Comparison of predicted (dash-dot) and true  ozone concentration
    (solid curve) along a Lagrangian trajectory  that passes
    through the outer most grid cell of source e, experiment  3A.
                                    197

-------
     TIME TRACK PLOT
     UNESOURCE EMISSION SIMULATION
     INITIAL LOCATION OF TRACK : ROW - 31.61
O

X
Q_
CL
O

1
Ld
O
•ZL
O
O

Ld
_J
O
CHEMISTRY

ROM
                  12   16   20   24    4    3   12   16   20   24    48

                 PAY 1	,	DAY 2            i  DAY 3
                       TIME  (HOUR OF  DAY)
   Figure  4-ll(d).  Comparison of predicted  (dash-dot) and true olefin concentra-
                   tion (solid curve) along a  Lagrangian trajectory that passes
                   through the outer most grid cell of source e, experiment 3A.
                                     198

-------
     TIME TRACK PLOT
     UNESOURCE EMISSION SIMULATION
     INfTlAL LOCATION OF TRACK : ROW - 31.61
O

X
Q.
Q_
O

§3
LU
O  2

O
O
CHEMISTRY

ROM
                                 I . . ,  I , , , I . ,  , I . . ,  I
                                          I,
              8   12   16   20   24

                 DAY 1            i
                            8
  12    16

DAY 2
20
24
8
                                              I   DAY 3
                       TIME  (HOUR OF DAY)
   Figure 4-ll(e).
    Comparison of predicted  (dash-dot) and true paraffin concen-
    tration  (solid curve) along a Lagrangian  trajectory that  passes
    through  the outer most grid cell of source e, experiment  3A.
                                     199

-------
  TIME TRACK PLOT
  UNESOURCE EMISSION SIMULATION
  INfTIAL LOCATION OF TRACK : ROW -
 Ql I  I \Jf\ I I  I I I I., I
                12   16   20   24
                       8   12   16   20   24
              DAY 1
                          DAY 2
DAY 3
                    TIME (HOUR OF DAY)
Figure  4-ll(f).
Comparison of predicted  (dash-dot)  and true aldehyde  concen-
tration  (solid curve)  along a Lagrangian trajectory that passes
through  the outer most grid cell of source e, experiment 3A.
                                  200

-------
     TIME TRACK  PLOT
     UNESOURCE EMISSION SIMULATION
     INITIAL LOCATION OF TRACK : ROW -  31.81
O

X
Q_
    6
CHEMISTRY

ROM
Ld
0 3
O
0 2
n
- l\
I \
- V

                                                         lr\
                                                  "\
                                     4    3   12    16

                                            DAY 2
                                        20
                       TIME  (HOUR  OF DAY)
   Figure 4-ll(g).
    Comparison of predicted (dash-dot)  and true aromatic concen-
    tration (solid curve) along a Lagrangian trajectory that passes
    through the outer most grid cell  of source e, experiment 3A.
                                    201

-------
CL
0_
g

Sc
LJ
O
Z
O
O
O
O
     TIME TRACK PLOT

     UNE50URCE EMISSION SIMULATION

     INITIAL LOCATION OF TRACK : ROW - 31.61
    i\
               CHEMISTRY

               ROM
                                                                 i I i i i  I
8
  12

DAY 1
16   20
24
8
  12

DAY 2
                                 16
                                                        20   24
                                                                DAY 3
                       TIME (HOUR OF DAY)
   Figure 4-ll(h).
Comparison of predicted (dash-dot) and true carbon monoxide
concentration (solid curve)  along a Lagrangian  trajectory
that passes through the outer most grid cell  of source e,
experiment 3A.
                                     202

-------
to
O


X
Q_
Q_
LJ
O
-z.
a
o

(N
o
     TIME TRACK PLOT

     LJNESOURCE EMISSION SIMULATION

     INfTTAL LOCATION OF TRACK : ROW - 31.61
CHEMISTRY


ROM
                 DAY 1
                              DAY 2
I  DAY 5
                       TIME (HOUR OF DAY)
   Figure  4-11(i).
    Comparison of predicted (dash-dot) and true  nitrous acid
    concentration (solid curve)  along a Lagrangian trajectory

    that passes through the outer most grid cell of source e,

    experiment 3A.


                      203

-------
o
X
GL
Q_
      TIME TRACK  PLOT
      LJNE50URCE EMISSION SIMULATION
      INITIAL LOCATION OF TRACK : ROW -
                31.81  COLUMN -  37.03
CHEMISTRY

ROM
                       I ... I ... I ... I ... I
                  12   16   20  24    4    3   12    16
                      TIME (HOUR  OF DAY)
  Figure  4-ll(j).
   Comparison of predicted (dash-dot) and true nitric acid
   concentration (solid curve)  along a Lagrangian trajectory
   that passes through the outer most grid cell  of source e,
   experiment 3A.
                                    204

-------
     TIME  TRACK  PLOT
     UNESOURCE EMISSION SIMULATION
     INOTAL LOCATION OF TRACK : ROW -  .31.61
X
Q_
O
    9r-
    8-
O  3

O
O  2
CHEMISTRY

ROM
          I , .
              8   12   16    20   24

                 DAY 1            |
                            8    12   16

                              DAY 2
20   24    4    8

     I   DAY 5
                       TIME  (HOUR OF  DAY)
   Figure 4-ll(k).
    Comparison of predicted (dash-dot) and true PAN concen-
    tration (solid curve) along a Lagrangian trajectory that
    passes through the outer most grid cell of source  e,
    experiment 3A.
                                     205

-------
     TIME TRACK PLOT
     UNESOURCE EMISSION SIMULATION

     INITIAL LOCATION OF TRACK : .ROW - 31.61 COLUMN - 37.03
O

X
CL
CL
UJ
O
z
O
O
CHEMISTRY

ROM
              8   12   16   20
                                12    16   20   24
                 DAY 1
                              DAY 2
DAY 3
                       TIME (HOUR  OF DAY)
  Figure 4-11(1).
    Comparison of predicted  (dash-dot) and true  alkyl nitrate
    concentration (solid curve) along a Lagrangian trajectory
    that  passes through the  outer most grid cell  of  source e,
    experiment 3A.
                                     206

-------
     TIME TRACK PLOT

     UNESOURCE EMISSION SIMULATION

     INITIAL LOCATION OF TRACK : ROW - 31.«t
cv,
o

X
Q_
CL
O

I.
LU
O
:z
o
o

(N
O
CHEMISTRY

ROM
                  12   16   20   24    4   S    12   16
                       TIME (HOUR OF DAY)
   Figure  4-ll(m).
    Comparison  of predicted (dash-dot) and true hydrogen peroxide
    concentration (solid curve)  along a Lagrangian  trajectory
    that passes through the outer most grid cell  of source e,
    experiment  3A.
                                     207

-------
     TIME TRACK  PLOT
     UNE50URCE EMISSION SIMULATION
     INFTIAL LOCATION OF TRACK : ROW - 31.61 COLUMN -  37.03
o>
O

X
CL   r
CL
O
£
Ld
O  1
~Z.
O
O
CHEMISTRY
ROM
               8   12    16   20   24

                 DAY 1            |
                                                      \
                            &   12    16   20   24
                              DAY 2            ,
DAY 3
                        TIME (HOUR  OF  DAY)
   Figure  4-ll(n).
    Comparison  of predicted (dash-dot) and true oxygen atom
    concentration (solid curve)  along a Lagrangian trajectory
    that passes through the outer most grid cell of source e,
    experiment  3A.
                                     208

-------
O


X
CL

Q_

o:
LJ
o
•z.
o
o
O
     TIME TRACK PLOT

     UNESOURCE EMISSION SIMULATION

     INITIAL LOCATION OF TRACK : ROW - 31.61 COLUMN - 37.03
    1-
CHEMISTRY


ROM
              8
    12


  DAY 1
16    20
8
                                             DAY 2
                                        DAY 3
                       TIME  (HOUR OF DAY)
   Figure 4-ll(o).
    Comparison  of predicted (dash-dot) and true nitrate

    concentration (solid curve)  along a Lagrangian  trajectory

    that passes through the outer most grid cell  of source e,

    experiment  3A.
                                     209

-------
o

X
O
bJ
O
z:
o
O
o
X
     TIME  TRACK  PLOT

     UNESOURCE EMISSION SIMULATION

     INITIAL LOCATION OF TRACK : ROW -  31.61
               CHEMISTRY

               ROM
              8   12   16   20   24

                 DAY 1
                                          8   12   16   20   24
                                             DAY 2
I  DAY 5
                       TIME  (HOUR OF DAY)
   Figure 4-ll(p).
                   Comparison  of predicted (dash-dot) and true hydroxyl
                   concentration (solid curve)  along a Lagrangian  trajectory
                   that passes through the outer most grid cell of source e,

                   experiment  3A.
                                     210

-------
  TIME TRACK PLOT
  LJNESOURCE EMISSION SIMULATION
  INITIAL LOCATION OF TRACK : ROW - 31.61
                      i i i I i  i i I i i i  I i i i I i  I i | i i i  I i
  12   16   20   24
DAY 1           i
  12
DAY 2
18   20
                                         24
                                                                   8
                                                            DAY 3
                    TIME  (HOUR OF DAY)
Figure  4-ll(q).
Comparison of predicted (dash-dot) and true hydroperoxyl
radical  concentration (solid curve) along a Lagrangian
trajectory that passes through the outer most grid cell  of
source e, experiment 3A.
                                  211

-------
     TIME TRACK PLOT
     LJNESOURCE EMISSION SIMULATION
     INITIAL LOCATION OF TRACK : ROW - 31.61
*   8
O

X  7

2
Q.
Q-
o:
z:
Ld
O
CHEMISTRY

ROM
                   12   16   20   24
                       TIME (HOUR OF DAY)
   Figure 4-ll(r).
    Comparison of predicted  (dash-dot) and true pernitric
    acid  concentration (solid curve) along a  Lagrangian
    trajectory that passes through the outer  most grid cell of
    source e, experiment  3A.
                                     212

-------
  TIME TRACK  PLOT
  UNESOURCE EMISSION SIMULATION
  INITIAL LOCATION OF TRACK : ROW - 31.61
o 10
O
^- 9
s^

_ ^^M^M*^^«
^UrUIQTDV
wncMldlnT
- ROM
             DAY 1
                           12
                         DAY 2
                                                16   20   24
DAY 3
                    TIME  (HOUR  OF DAY)
Figure 4-ll(s).
Comparison of predicted  (dash-dot) and true alkoxyl
radical  concentration (solid curve) along a Lagrangian
trajectory that passes through the outer most grid cell of
source e, experiment 3A.
                                  213

-------
     TIME  TRACK  PLOT

     UNESOURCE EMISSION SIMULATION

     INITIAL LOCATION OF TRACK : ROW - 31.61
O


X
Q_
Q_
O  3h
cr
LJ
o
-z.
o
o

CN
O
                     1  I _1. I	I till L J I I ^ 1__ 1 I L^J^ I  I L L J I J^ I I I t I  I I I t	ll	I 1 I 11 i 1 1 |
a   12   16   20   24

  DAY 1            i
                                           8    12   16   20    24
                                             DAY 2
I  DAY 3
                       TIME (HOUR  OF DAY)
   Figure 4-ll(t).  Comparison of predicted  (dash-dot) and true  alkylperoxy
                   radical concentration (solid curve) along a  Lagrangian

                   trajectory that passes through the outer most  grid cell of

                   source e, experiment 3A.
                                     214

-------
     TIME TRACK  PLOT
     UNESOURCE EMISSION SIMULATION
     INITIAL LOCATION OF TRACK : ROW - 51.61 COLUMN -  37.03
o   2

O

X
Q_
Q_
LJ
O

O
O


O
CM
CHEMISTRY

ROM
    Ql i i  i
               ' ' ' ' '  • • ' ' I I  I T
   12   16   20    24

  DAY 1
  12    16   20   24

DAY 2           ,
                                                                       8
                                                                DAY 3
                       TIME (HOUR  OF DAY)
   Figure 4-ll(u).
    Comparison of predicted (dash-dot) and true alkoxy radical
    concentration (solid curve)  along a Lagrangian  trajectory
    that  passes through the outer most grid cell  of source e,
    experiment 3A.
                                     215

-------
tO
O

X
Q_
Q.
g

55
h-
Z
LJ
O
O

-------
CO
O

X
Q.
     TIME TRACK PLOT

     1JNESOURCE EMISSION SIMULATION

     INITIAL LOCATION OF TRACK :  ROW - 31.61
    3-
O
i  2
LJ
O
z
O
O

CN
O

-------
o
     TIME  TRACK  PLOT
     UNESOURCE EMISSION SIMULATION TEST
     INmAL LOCATION OF TRACK : ROW -  36.32  COLUMN
                        40.16
               CHEMISTRY

               ROM
              8   12   16   20   24

                 DAY 1            i
                            12    16   20   24
                          DAY 2
I  DAY 3
                       TIME  (HOUR OF  DAY)
  Figure 4-12(a).
Comparison of predicted  (dash-dot) and true NO concentration
(solid  curve) along a Lagrangian trajectory that passes
through the center of source c, experiment 3A.
                                    218

-------
O


X
Q_
Q_

o
o


CN
O
     TIME TRACK PLOT

     UNESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW - 36.32 COLUMN
                                        40.16
Ld

O  2
           CHEMISTRY


           ROM
1-
                  ^2   16    20   24
                DAY 1
  12    16   20   24

DAY 2           ,
                                                           DAY 3
                       TIME  (HOUR OF DAY)
   Figure 4-12(b).   Comparison of predicted  (dash-dot) and true  N0£ concentration
                   (solid curve) along a  Lagrangian trajectory  that passes
                   through the center of  source c, experiment 3A.
                                     219

-------
     TIME TRACK  PLOT
     UNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW =• 36.32  COLUMN =  40.16
^   3

O

X
Q_
Q_
CHEMISTRY

ROM
                           i i I t i I  I i i t I  i i i I i I  i 1 r _j_.i I i i i ! i  i i i i i  i L I i i  I i I
               8   12    16   20   24-

                 DAY 1            |
                            8   12   16   20   24

                              DAY 2            i
DAY 3
                        TIME  (HOUR OF  DAY)
   Figure  4-12(c).
    Comparison of predicted (dash-dot) and true ozone concen-
    tration (solid curve) along a  Lagrangian trajectory that
    passes through the center of source c, experiment 3A.

                      220

-------
CM
O

X
Q_
CL
O
     TIME TRACK  PLOT
     LJNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW = 36.32 COLUMN =  40.T6
LU
O  1

O
O

LJ
_l
O
CHEMISTRY

ROM
              8    12   16   20    24

                 DAY 1            i
                            8    12   16   20   24    4    3

                              DAY 2           i   DAY 3
                       TIME (HOUR  OF DAY)
   Figure 4-12(d).
    Comparison of predicted  (dash-dot) and true olefin concen-
    tration  (solid curve)  along a Lagrangian trajectory that
    passes through the center  of source c, experiment 3A.
                                     221

-------
     TIME  TRACK  PLOT
     LJNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW -  36.32  COLUMN -  40.16
CM &
O
X
s
Q_
Q_
^ 4
z:
O
j—
t—
!•• •
21
LJ
O 2
—7
^L.
O
O
/M trMI^TTTV
_ -^-^ OntMlw 1 rt T
PHfiii
— —— — • /^'^ UM
•: K\
£r \\
V

:
    1-
Q_
                                                        //
                                                     •v
      ', vr/l , . . i .  , . i , . . i  . , . i . ,  , i , , , i  . . , i , ,  . i , . , i  , , , i .
                                                          i  i I i i i
          4    8   12   16   20   24   4    8   12   16   20   24   4    3
                 DAY 1
                         DAY 2
DAY 3
                       TIME  (HOUR OF  DAY)
   Figure  4-12(e).
Comparison of predicted  (dash-dot) and true paraffin concen-
tration  (solid curve) along a Lagrangian  trajectory that
passes through the center  of source c, experiment 3A.

                  222

-------
     TIME TRACK PLOT
     UNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW =

-------
  TIME TRACK  PLOT
  UNESOURCE EMISSION SIMULATION TEST
  INmAL LOCATION OF TRACK : ROW - 36.32 COLUMN = 40.16

X
Q.
Q_
Z
O
£
C£
I
z:
LJ
O
:z
o
o
1U
9
8
6
5
4

3

«•>
- onrMP
r 	 £\ROM
: / \
1 f\\
\ M
- n
; n
r H
^ v
E
1
:
: J
                 \>
 1-
           8   12   16   20   24

             DAY 1            I
  12    16   20    24

DAY 2      .     i
                                            DAY 3
                    TIME  (HOUR  OF  DAY)
Figure  4-12(g).
Comparison of predicted  (dash-dot) and true aromatic concen^
tration  (solid curve)  along a Lagrangian trajectory that
passes through the center of source c, experiment 3A.

                  224

-------
     TIME TRACK PLOT

     LJNESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW - 36.32
CL
Q_
O
LJ
O
z:
o
O
o
o
               CHEMISTRY

               ROM
                                                         r
           I i I I i  i i I i i  i I i i i I  i i i I i  i i I i i i  I i i i I  i i i I i  . i I i i i I  , , i I ,  . i I ,
          4   8   12   16   20  24    4    3   12    18   20   24   48
                 DAY 1
                          DAY 2
DAY 3
                       TIME (HOUR  OF DAY)
   Figure 4-12(h).
Comparison  of predicted (dash-dot) and true carbon monoxide
concentration (solid curve) along a Lagrangian trajectory
that passes through the center of source c, experiment  3A.


                  225

-------
     TIME TRACK  PLOT
     UNESOURCE EMISSION SIMULATION TEST
     INfTlAL LOCATION OF TRACK : ROW -
* °
O
X
I "
Q-
-
_
—
f\ irUITTTY
PHM
r%UM
g  3
tc
LU
O
Z
O
O


-------
  TIME TRACK PLOT
  UNESOURCE EMISSION SIMULATION TEST
  INITIAL LOCATION OF TRACK : ROW = 36.
       4   8   12   16   20   24    4    8   12   16   20   24   4    8
              DAY 1
                          DAY 2
DAY 3
                    TIME  (HOUR OF  DAY)
Figure 4-12(j).
Comparison of predicted  (dash-dot) and true nitric acid
concentration (solid curve) along a Lagrangian trajectory
that passes through the  center of source c, experiment 3A.
                                  227

-------
  TIME TRACK PLOT
  UNESOURCE EMISSION SIMULATION TEST
  INfTlAL LOCATION OF TRACK : ROW = 36.32 COLUMN = 40.16
            a    12   16   20   24
              DAY 1
8   12    16   20   24

  DAY 2
DAY 3
                    TIME (HOUR  OF DAY)
Figure  4-12(k).  Comparison of predicted  (dash-dot) and true PAN concen-
                tration  (solid curve)  along a Lagrangian trajectory
                that passes through the  center of source c, experiment 3A.

                                  228

-------
     TIME TRACK PLOT
     LJNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW =
CO
O

X
D_
Q_
O

li
LJ
O
;Z
O
O
O
     CHEMISTRY

     ROM
    0Lj-J-L
                                       i i  i I i i t I  i i
4-    8    12   16   20   24

       DAY 1           .
8   12   16   20   24

  DAY 2            !
                                                                DAY 3
                       TIME (HOUR  OF DAY)
   Figure 4-12(1).  Comparison of predicted (dash-dot) and true alky] nitrate
                   concentration (solid curve)  along a Lagrangian  trajectory
                   that passes through the center of source c, experiment 3A,
                                     229

-------
     TIME TRACK PLOT

     UNESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK :  ROW =
CM
o

X

^
D_
o_


2:
    2
    2
bJ
O
•z.
o
o

CN
O
CM
CHEMISTRY

ROM
  12   16   20   24

DAY 1
                                           8    12   16   2Q   24
                                             DAY 2
                                                 DAY 3
                       TIME (HOUR  OF DAY)
   Figure 4-12(m).
    Comparison of predicted (dash-dot) and true hydrogen peroxide
    concentration (solid curve) along a Lagrangian trajectory
    that passes through the center of source c, experiment 3A.
                                     230

-------
05

O


X
CL
o_
     TIME TRACK PLOT

     UNESOURCE EMISSION SIMULATION TEST

          LOCATION OF TRACK :  ROW - 36.32 COLUMN
                            40.16
Ld
O  2
•z.
O
O
O
CHEMISTRY

ROM
                                                             I
                   12

                 DAY 1
        16
20
24
  12

DAY 2
16
20
24   4

 ,  DAY 3
                       TIME (HOUR  OF DAY)
   Figure 4-12(n).
    Comparison  of predicted (dash-dot) and true atomic  oxygen
    concentration (solid curve)  along a Lagrangian trajectory
    that passes through the center  of source c, experiment 3A.
                                     231

-------
  TIME  TRACK  PLOT
  LINESOURCE EMISSION SIMULATION TEST
  INITIAL LOCATION OF TRACK : ROW -  36.32  COLUMN =
              DAY 1
                         DAY 2
DAY 3
                    TIME  (HOUR  OF  DAY)
Figure  4-12(o).
Comparison of predicted  (dash-dot) and true nitrate
concentration (solid curve) along a Lagrangian trajectory
that passes through the  center of source c, experiment 3A.
                                  232

-------
     TIME TRACK PLOT
     UNESOURCE EMISSION SIMULATION TEST
     INmAL LOCATION OF TRACK : ROW = 36.32 COLUMN
                            40.16
O

X
Q_
Q_
O
CHEMISTRY

ROM
          4    8   12   16    20   24    4    3   12   16   20   24    4   8
                 DAY 1
                              DAY 2
DAY 3
                       TIME  (HOUR OF DAY)
   Figure 4-12(p).
    Comparison of predicted  (dash-dot) and true  hydroxyl radical
    concentration (solid curve) along a Lagrangian trajectory
    that  passes through the  center of source c,  experiment 3A.
                                     233

-------
     TIME TRACK  PLOT
     UNESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW =  36.32  COLUMN
                             40.16
rf-

O


X
Q_
CL

LU
O
Z
O
O


CM
O
z:
CHEMISTRY

ROM
      _I_U
         j*4-*..i ' I i i—r-fi i i I  i i i I i  i i I i i  i I i i i 1 i i i I  i j .1 I i -i i I i i  i i i i i  I i i i j L
S   12   16    20   24

  DAY 1
  12    16   20   24

DAY 2
                                                                 DAY 3
                        TIME  (HOUR  OF DAY)
   Figure 4-12(q).
     Comparison of predicted (dash-dot)  and true hydroperoxyl
     radical concentration (solid curve) along a Lagrangian
     trajectory that passes through  the  center of source c,
     experiment 3A.
                                      234

-------
     TIME  TRACK PLOT

     LINESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW = 36.52
o

X
D_
Q_

01

2
LJ
O
IZ
O
O
o
X
CHEMISTRY

ROM
                                               12   16   20    24
8   12   16   20   24

  DAY 1
                                             DAY 2
DAY 3
                       TIME (HOUR OF DAY)
   Figure 4-12(r).
    Comparison of predicted (dash-dot)  and true pernitric
    acid concentration  (solid curve)  along a Lagrangian
    trajectory that  passes through the  center of source c,
    experiment 3A.
                                     235

-------
  TIME  TRACK  PLOT
  LINESOURCE EMISSION SIMULATION TEST
  INITIAL LOCATION OF TRACK : ROW = 36.32  COLUMN = 40.16
           8   12   16   20   24
                           12   16   20   24
              DAY 1
                          DAY 2
DAY 3
                    TIME  (HOUR OF DAY)
Figure  4-12(s).
Comparison of predicted (dash-dot) and true alkoxyl
radical  concentration (solid curve) along a Lagrangian
trajectory that passes through the center of source c,
experiment 3A.
                                  236

-------
  TIME TRACK  PLOT
  LINESOURCE EMISSION SIMULATION TEST
  INITIAL LOCATION OF TRACK : ROW = 36.32
               12

              DAY 1
I 1 1
16
, , I ,
20
i ' I ' i
24
i 1 i
4
i i 1 i ,
3
ii 1 i
12
• i i i i
16
i
20
i i 1 i i i
24
i 1 i i
4
I 1 I
8
                          DAY 2
I  DAY 3
                    TIME  (HOUR OF  DAY)
Figure 4-12(t).
Comparison of predicted  (dash-dot) and true alkylperoxy
radical  concentration (solid curve) along a Lagrangian
trajectory that passes through the center of source c,
experiment 3A.
                                  237

-------
     TIME TRACK PLOT

     LINESOURCE EMISSION SIMULATION TEST

          LOCATION OF TRACK : ROW = 36.32 COLUMN
o   2

O

X
Q_
Q_
O
LT
I—
-z.
LJ
O

O
o


o
CN


    0
                            40.16
CHEMISTRY

ROM
               8   12    16   20   24    4    8   12    16

                 DAY 1           i            DAY 2
                                         20
24   4

 I  DAY 3
                        TIME (HOUR  OF  DAY)
   Figure 4-12(u).
    Comparison  of predicted (dash-dot) and true alkoxy
    radical  concentration (solid curve) along a Lagrangian
    trajectory  that passes through the center of source c,
    experiment  3A.
                                      238

-------
     TIME TRACK  PLOT
     LJNESOURCE EMISSION SIMULATION TEST
     INFTIAL LOCATION OF TRACK : ROW =  36.32  COLUMN
                                          40.16
    5r-
iO
O
X

^  4-
a.
CL
:
             CHEMISTRY

             ROM
             S   12    16   20   24
                      TIME  (HOUR OF  DAY)
 Figure 4-12(v).
                   Comparison  of predicted (dash-dot)  and peroxyacyl
                   radical  concentration (solid curve)  along a Lagrangian
                   trajectory  that passes through the  center of source c,
                   experiment  3A.
                                    239

-------
     TIME TRACK PLOT
     LINESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW = 36.32
(O
O

X
Q_
Q_
O
    c
    5-
    4-
    3-
LJ
O  2

O
O

CM  1
O
CN
CHEMISTRY

ROM
          4    8    12   16   20    24    4    8    12   16   20    24    4    8
                 DAY 1
                              DAY 2
DAY 3
                        1ME (HOUR  OF DAY)
   Figure 4-12(w).   Comparison of predicted (dash-dot)  and peroxy radical
                   concentration (solid curve) along a Lagrangian
                   trajectory that  passes through the  center of source c,
                   experiment 3A.
                                     240

-------
     TIME TRACK PLOT
     UNESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW - 41.04 COLUMN -  43.23
    4|-
O

X



Q_
£
h-
LJ   j-
O
O

o
    1
CHEMISTRY
                                  _t 1 _ * I 1  T ! I ! j  j 1 _!_..* _!__ I  j t * j  II l_ 1 It! I I
               3    12   16   20   24   4    3   12   16    20   24    4   3

                 DAY 1            i           CAY 2            .  DAY 3
                       TIME  (HOUR OF  DAY)
  Figure 4-13(a).   Comparison of predicted (dash-dot) and true NO concentration
                   (solid curve) along a Lagrangian trajectory that passes
                   through the inner edge of source b, experiment 3A.
                                     241

-------
     TIME TRACK PLOT

     UNESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW - 41.04 COLUMN - 43.23
O


X
Q.
Q_
O
cr
LjJ
o
:z
O
o

CN
O
      • •-•• ' t i .1 i i i i i i i _M .1 t ' '  ' i.i
             .         _   .      ..             _
          4   8   12   16   20   24    4   3    12   13
                       TIME (HOUR OF DAY)
  Figure 4-13(b).  Comparison of predicted (dash-dot)  and true N02 concentration
                  (solid curve) along a Lagrangian trajectory that passes
                  through the inner edge of source b, experiment 3A.
                                     242

-------
     TIME  TRACK  PLOT
     UNESOURCE EMISSION SIMULATiON TEST
     INITIAL LOCATION OF TRACK : ROW - 41.04
O

X
Q_
Q_
CHEMISTRY

ROM
          4    8   12   16    20   24    4    3   12   IS   20   24    4    3
                DAY 1
                              DAY 2
i  DAY 5
                       TIME  (HOUR OF DAY)
   Figure 4-13(c).
    Comparison  of predicted (dash-dot) and true ozone concen-
    tration  (solid curve) along a  Lagrangian trajectory that
    passes through the inner edge  of source b, experiment 3A.

                      243

-------
     TIME TRACK  PLOT
     LJNESQURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW -  41.04 COLUMN
                          - 43.23
CNJ

O


X
CL
Q_
g
Si
UJ
o
:z
o
o

LU
_J
O
 CHEMISTRY

^ROM
                                                            1
              3
    12

  DAY 1
16   20
24
  12

DAY 2
16   20
24   4

 |  DAY 3
                       TIME (HOUR  OF DAY)
   Figure  4-13(d).
     Comparison of predicted (dash-dot) and true olefin concen-
     tration (solid curve) along a  Lagrangian trajectory that
     passes through the inner edge  of source b, experiment 3A.


                      244

-------
     TIME TRACK PLOT
     LJNESOURCE EMISSION SIMULATION TEST
     INOTAL LOCATION OF TRACK : ROW - 41,04 COLUMN -  43.23
O

X
8:
O
z:
LU
O
O

C£

CL
    2
    i
                                                                \
              J.J..J I I I I  I 1 I I I I  I I I I I  I I I I i I 1 I. L I 1 I  I. I 1 I I  I J i. I I .1
                                                              J-J.
          4    8   12   16   20   24    4    3    12    16   £0   24    4    3

                 DAY 1            ,            DAY 2           ,   DAY 3
                       TIME (HOUR  OF DAY)
   Figure 4-13(e).  Comparison of predicted (dash-dot) and true paraffin concen-
                   tration  (solid curve) along a  Lagrangian trajectory that
                   passes through the inner edge  of  source b, experiment  3A.
                                     245

-------
  TIME TRACK  PLOT
  UNESOURCE EMISSION SIMULATION TEST
  INOTAL LOCATION OF TRACK : ROW -
                                                            \
                              I i i i I  i i i 1 i i i  1 _L i i i
                                                 i  i i  i i i
       4   8    12   16   20   24    4    3   12   16   20   24   4    3
DAY 1
DAY 2
                                                          |  DAY 5
                    TIME (HOUR OF  DAY)
Figure 4-13(f).  Comparison of predicted (dash-dot) and true aldehyde concen.
                tration (solid curve) along a Lagrangian trajectory that
                passes through the  inner edge of source b, experiment 3A.

                                  246

-------
     TIME TRACK  PLOT
     LJNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW -  41.04  COLUMN
                                            40.23
ro
O

X  7
O
               CHEMISTRY

               ROM
Ld  3
O

O  o
O

O  1


    0
      j i.-f I i i i  I i i i I i  i i I i .1 i I i i i i i i i l i j t l i
          4    3    12   16   20   24   4    3   12   1(3   ZO   24    4   3
     l  	DAY 1	i           DAY 2            ,  DAY 3
                       TIME (HOUR  OF  DAY)
   Figure  4-13(g).
                   Comparison of predicted (dash-dot) and true aromatic  concen-
                   tration  (solid curve) along a  Lagrangian trajectory that
                   passes through the inner edge  of  source b, experiment 3A.
                                     247

-------
  TIME TRACK PLOT
  LJNESOURCE EMISSION SIMULATION TEST
  INfTTAL LOCATION OF TRACK : ROW - *
  Z  	 CHEM1SJRX-

  1	r= WSST
                                                           \
                                                             \!
                                               1_1 i J L i 1 1 I I  I I I 1 I 1  ! 1 1 T
            a   12    16   20   24

              DAY 1           i
                        3   12    18   20   24    4    B

                          DAY 2            i  DAY 3
                     TiME (HOUR  OF DAY)
Figure  4-13(h).
Comparison of predicted (dash-dot) and true carbon monoxide
concentration (solid curve)  along a Lagrangian  trajectory
that passes through the inner edge of source b, experiment 3A.
                                  248

-------
   TIME TRACK  PLOT
   LJNESOURCE EMISSION SIMULATION TEST
   INITIAL LOCATION OF TRACK : ROW -  41.04  CCLUMN -  43.2S
            CHEMISTRY
            ROM
                                                                    !
              DAY 1
  12
DAY 2
                                                 16
24   4    3
 ,  DAY 3
                     TIME  (HOUR OF  DAY)
Figure 4-13(i).  Comparison  of predicted (dash-dot) and true nitrous  acid
                concentration (solid curve)  along a Lagrangian trajectory
                that passes through the inner edge of source b, experiment 3A,
                                  249

-------
     TIME  TRACK  PLOT
     UNESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW - 41.
    7r-
O


X
Q_
Q_
1=  4r-
oi


LU
O

O
O
O
CHEMISTRY

ROM
                  12   1S   20   24

                 DAY 1
                            a   12    16   20   24

                              DAY 2
DAY 3
                       TIME (HOUR OF DAY)
   Figure  4-13(j).
    Comparison of  predicted (dash-dot) and true nitric acid
    concentration  (solid curve) along a Lagrangian trajectory
    that passes through the inner edge of source b, experiment 3A.
                                     250

-------
  TIME TRACK PLOT
  UNESGURCE EMISSION SIMULATION TEST
  INITIAL LOCATION OF TRACK : ROW - 41.04 COLUMN -  43.23
ro
O

X
            CHEMISTRY

            ROM
   i i  i I i 1 i  ^r> r ' ' •  ! ' ' ' '  ' ' ' -i ' -i- i ' ' ' ' '  ' ' ' ' '  ' ' ' ' ' '  '
       4   3    12   16   20   24    4    3   12    16   20   24    4    3
                    TIME (HOUR  OF DAY)
Figure  4-13(k).
                   Comparison of predicted (dash-dot)  and true PAN concen-
                   tration (solid curve) along a Lagrangian trajectory that
                   passes through the inner edge of source b, experiment 3A.
                                  251

-------
  TIME TRACK PLOT
  LJNESOURCE EMISSION SIMULATION TEST
       LOCATION OF TRACK : ROW - 41.04
                                                            \!
                                  I I I : i I  i i I I I |  i ! 1 1 ! I  i i i i j i  i ! l_-i. i , L .1-1
                12   16   20   24    4   3    12   16   20  24    4   &
              DAY 1
                              I
DAY 2
,  DAY 3
                    TIME (HOUR OF DAY)
Figure  4-13(1).  Comparison of predicted  (dash-dot) and  true alkyl nitrate
                concentration (solid curve) along a Lagrangian trajectory
                that passes through the  inner edge of source b, experiment  3A.
                                  252

-------
     TIME TRACK PLOT

     LJNESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW -
    2-
O


X
Q_

Q_
g


o:
i—
^
UJ
O
o
CN   h-

-------
O>
o
X

DC
•z.
Ld
O

8
     TIME TRACK PLOT
     UNESOURCE EMISSION SIMULATION TEST
     INfTTAL LOCATION OF TRACK :  ROW - 41.04 COLUMN - 43.23
    2
    *
                         1 1 i
                            i  i
                                   t  i I I i i -i r
                                                                  I
  12

DAY 1
    16   20
                                24
  12
DAY 2
16    20
                                                             24    4
                                                             .   DAY 3
                       TIME (HOUR  OF DAY)
   Figure 4-13(n).
Comparison of predicted (dash-dot) and true atomic  oxygen
concentration (solid curve)  along a Lagrangian trajectory
that passes through the inner edge of source b, experiment 3A.
                                     254

-------
     TIME TRACK PLOT

     LJNESOURCE EMISSION SIMULATION TEST

     INITIAL LOCATION OF TRACK : ROW - 41.04
*   2

O


X
CL

Q_
g

£
a:
h-
z:
LJ
O
z:
o
O

n
CHEMISTRY


ROM
               8    12   16   20    24
                                                     fl I ! I I I I I  I I j !\! !  I I 1
                 DAY 1
                            3    12    15   20   24    4    3


                              DAY 2	|   DAY 5
                       TIME (HOUR  OF DAY)
   Figure 4-13(o).
    Comparison of predicted (dash-dot)  and true nitrate  concen-

    tration (solid curve) along a Lagrangian trajectory  that

    passes through the inner edge of source b, experiment  3A.


                      255

-------
     TIME TRACK PLOT
     UNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK :  ROW - 41.04 COLUMN - 43.23
o
8   12   16   20

  DM 1
                                24
3   12   18

  DAY 2
                                                        20   24    4    3

                                                             i  DAY 3
                       TIME  (HOUR OF  DAY)
   Figure 4-13(p).  Comparison  of predicted (dash-dot) and true hydroxyl radical
                   concentration (solid curve)  along a Lagrangian  trajectory that
                   passes through the inner edge  of source b, experiment 3A.
                                     256

-------
     TIME TRACK PLOT
     UNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW - 41.04 COLUMN - 43.2S
O

X
Q_
Q_
g
s
LU
O
2
O
O

CM
O
X
CHEMISTRY

ROM
  12    16   20   24

DAY 1
                                           3    12    18   20   24
                                             DAY 2
                                               I  DAY 3
                       TIME (HOUR  OF DAY)
   Figure 4-13(q).
    Comparison  of predicted (dash-dot)  and true hydroperoxyl
    radical  concentration (solid curve)  along a Lagrangian
    trajectory  that passes through  the  inner edge of source  b,
    experiment  3A.
                                     257

-------
  TIME TRACK PLOT
  LJNESCURCE QHSS10N SIMULATION TEST
  INITIAL LOCATION OF TRACK : ROW - 41.04 COLUMN - 43.28
            CHEMISTRY
            ROM
8   12   16   20   24
  DAY 1
                                                               !
                                            12    16   20
                                          DAY 2
  4
DAY 3
                    TIME (HOUR  OF DAY)
Figure  4-13(r).  Comparison  of predicted (dash-dot) and true pernitric  acid
                concentration (solid curve)  along a Lagrangian trajectory
                that passes through the inner edge of source b, experiment 3A.
                                  258

-------
     TIME TRACK PLOT
     UNESOURCE EMISSION SIMULATION TEST
     INfTIAL LOCATION OF TRACK : ROW - 41.04 COLUMN - 43.25
O

O

X
0.
CL
O
bz  *1
Ld
O
^
O
O
O
o:
            I 1 I  I I I I 1 .1  I L I ! I I  1 I I I 1 I I  I I I I 1  I 1 I  I I 1
                   12   16   20   24
                 DAY 1
DAY 2
.  DAY 3
                       TIME (HOUR OF DAY)
   Figure 4-13(s).  Comparison of predicted (dash-dot) and true  alkoxyl radical
                   concentration (solid curve)  along a Lagrangian trajectory
                   that passes through the inner edge of source b, experiment  3A.
                                     259

-------
     TIME TRACK PLOT
     IJNESOURCE EMISSION SIMULATION TEST
     INITIAL LOCATION OF TRACK : ROW - 41.04
o

><  eP
2
Q.
^ 5F-
O  A'
I
I»-
LU
O
o  *
o
O
CHEMISTRY

ROM
               3   12   16   20   24
                        TIME (HOUR  OF DAY)
    Figure 4-13(t).   Comparison of predicted (dash-dot)  and true alkylperoxyl
                    radical concentration  (solid curve) along a Lagrangian
                    trajectory that passes through the  inner edge of source b,
                    experiment 3A.
                                      260

-------
  TIME TRACK  PLOT
  UNESOURCE EMISSION SIMULATION TEST
  INHTAL LOCATION OF TRACK : ROW - 41.04 COLUMN
                        43.23
            CHEMISTRY
                                           12   16   iO
              DAY 1
                          DAY 2
24   4    B

 l  DAY 5
                    TIME  (HOUR OF DAY)
Figure  4-13(u).
Comparison of predicted (dash-dot) and true alkoxy radical
concentration (solid curve)  along a Lagrangian trajectory
that passes through the inner edge of source b, experiment 3A.
                                  261

-------
  TIME TRACK PLOT
  UNESOURCE EMISSION SIMUUT10N TEST
  INOTAL LOCATION OF TRACK : ROW - 41.04
                          I I I I 1 I 1 1 1 I I I I  I I I I I I  1 I I I I I  I 1 I I 1 I  l\ 1 ! I 1 !  I 1
                12   16   20   24    4    3    12   18   20   24    4   3
              DAY 1
                          DAY 2
DAY 3
                    TIME (HOUR  OF DAY)
Figure  4-13(v).
Comparison of predicted  (dash-dot) and  true peroxyacyl
radical concentration  (solid curve) along a Lagrangian
trajectory that passes through the inner edge of source b,
experiment 3A.
                                  262

-------
CO
O

X
Q.
CL
a
bJ
O
~ZL
O
O

CM
O
CN
LT
     TIME TRACK  PLOT
     LJNESOURCE EMISSION SIMULATION TEST
     INfTlAL LOCATION OF TRACK : ROW - 41.04 COLUMN -  40.23
CHEMISTRY

ROM
                       TIME (HOUR  OF DAY)
   Figure 4-13(w).
    Comparison of  predicted (dash-dot)  and true peroxy
    radical  concentration (solid curve) along a Lagrangian
    trajectory that  passes through the  inner edge of source b,
    experiment 3A.

                       263

-------
                                 REFERENCES
Demerjian, K. L. and K. L. Schere,  (1979):  "Applications of a  Photochemical
     Box Model  for Ozone Air Quality in Houston,  Texas.   Proceedings, Ozone/
     Oxidants:  Interactions with the Total  Environment II,  Houston, TX, 14-17
     Oct. 1979, APCA, Pittsburgh, Pa., pp.  329-352.

Demerjian, K. L., K. L. Schere, and J. T.  Peterson, (1980): "Theoretical
     estimates  of actinic (spherically integrated)  flux  and photolytic rate
     constants  of atmospheric species in the lower  troposphere."   In Advances
     in Environmental Science and Technology - Vol. 10,  J.  N. Pitts et al.,
     eds., John Wiley and Sons, New York,  pp. 369-459.

Gear, C. W., (1971):  "The automatic integration  of ordinary differential
     equation".  Communications of  the ACM, Vol.  14, 176-179.

Lamb, R. G., (1983): "A Regional Scale (1000 km)  Model of Photochemical Air
     Pollution: Part 1.  Theoretical Formulation".   EPA-600/3-83-035. 237  pages.

Lamb, R. G., (1984): "A Regional Scale (1000 km)  Model of Photochemical Air
     Pollution: Part 2.  Input processor network  design".  EPA-600/S3-84-085.
     310 pages.

Mahrer, Y. and  R. A. Pielke, (1978): "A Test of an  Upstream Spline Interpolation
     Technique  for the Advection Terms in  a Numerical  Mesoscale Model".
     Mon. Wea.  Rev.. Vol. 106, pp 818-830.

Schere, K. L.,  (1984): Private communication.

Yamartino, R. J., (1984):  Private  communication.

Zalesak, S., (1979): "Fully Multidimensional Flux-Corrected Transport Algorithms
     for Fluids", J. Comp. Physics, 31: 335-362.
                                    264

-------