AN ANALYSIS
                                   OF
                      COMPLEX 1 AND COMPLEX II —
                       CANDIDATE SCREENING MODELS
                                   by
                              John S. Irwin
                                            jt
                             D. Bruce Turner
                    Meteorology and Assessment Division
                 Environments Sciences Reseach Laboratory
                Research Triangle Park, North Carolina 27711
                CNVIROivlENTAL SCIENCE'S l\£SEa!l;H L/BOr;AiW
                    OFFICF OF RESEARCH AND DEVELOPMENT
                    U.S. ENVIRONMENTAL PROTECTION  AGENCY
                RESEARCH TRIANGLE PARK, NORTH  CAROLINA  27711
On assignment from National Oceanic and Atmospheric  Administration,
   Department of Commerce.

-------
                                 DISCLAIMER

     This report has been reviewed by the Environmental 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.
                                  AFFILIATION

     Mr.  Irwin  and Mr.  Turner are meteorologists in the Meteorology and Assess-
ment Division,  Environmental Sciences Research Laboratory, U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina.  They are on
assignment from the  National Oceanic and Atmospheric Administration, U.S.
Department of Commerce.
                                      11

-------
                                  FOREWORD

     One area of research within the Meteorology and Assessment Division is

the mathematical modeling of air quality simulation including photochemical

and meteorological processes.  The Division works to develop, evaluate,

validate and apply models which can accurately describe air quality and the

atmospheric processes that affect the transport and fate of airborne pollu-

tants, on a local as well as global scale.  Within the Division, the Environ-

mental Operations Branch adapts and evaluates new and existing meteorological

dispersion and statistical technique models, tailors effective models for

recurring user application, and makes all models available through EPA's

computer network system and on magnetic tape from the National Technical

Information Service.

     Complex II and Complex I are adaptations of the MPTER model and the

Valley model as proposed by a workshop panel at an EPA Regional Workshop

held in Chicago, February 25-28, 1980.  The panel asked that a brief analysis

be performed to investigate whether a sequential model, capable of accepting

onsite hourly meteorological data could be recommended as a screening model

for estimating worst-case pollutant impacts in complex terrain situations.
                                  K. L. Demerjian
                                  Director
                                  Meteorology and Assessment Division
                                  Environmental Sciences Research Laboratory
                                  Research Triangle Park, North Carolina
                                  October 1981
                                     111

-------
                                 ABSTRACT




     This study, suggested by an EPA Regional Workshop in February 1980, was




conducted as a simple analysis to investigate whether or not a sequential air




quality simulation model, capable of accepting onsite hourly meteorological




data, could be recommended as a screening model for estimating worst-case




pollutant impacts on complex terrain.  The study intercompared the highest




24-h average pollutant concentration values obtained using four algorithmic




air quality simulation models:  Complex I, Complex II, Valley, and Valley-BID.




Complex I and Complex II are sequential (hourly) air quality simulation models




which differ only in their characterization of lateral dispersion.  Complex I




simulates lateral dispersion by assuming a uniform distribution of pollutant




spread over a 22.5° sector centered on the input hourly wind direction.




Complex II simulates lateral dispersion by assuming a Gaussian distribution as




specified by the input Pasquill stability category and the downstream distance.




Valley is the "standard EPA screening model used for estimating worst-case 24-h




concentrations.  Valley-BID is the Valley model, modified to incorporate the




characterization of induced dispersion arising from buoyant plume rise into the




vertical dispersion characterization.




     The models were applied and their results compared for a year's meteoro-




logial data for two different sites.  Various combinations of source release




height and terrain configurations were examined.




     The authors conclude that the Valley-BID (or pencil and paper calcula-




tions using the same assumptions) are most appropriate for screening analyses




for maximum 24-h concentrations resulting from plume impaction on terrain




near the height of an elevated stabilized plume.






                                     iv

-------
                                  CONTENTS
                                                                       Page

Foreword	,	iii

Abstract	iv

Figures	vi

Tables. .-	vii

Acknowledgment	viii

     1.  Introduction 	  1

     2.  Conclusion and Recommendations	5

     3.  Analysis Procedures	11

     4.  Results	19

     5.  Special Topics	29

     6.  Questions for Reviewers.  ..,...„	41

References	43

Appendices

     A.  Additional figures of maximum 24-hour concentrations  ....  45

     B.  Highest and second-highest concentrations  and
           concentrations  for various averaging times  ........  47

     C.  Model estimates by hand  calculations  	  53

-------
                                  FIGURES

Number

  1  Flow vector frequency, hours of stable, and hours of
        calm (Knoxville 1964) 	 16

  2  Flow vector frequency, hours of stable, and hours of
        calm (St. Louis 1976)	16

  3  Example runstream	18

  4  Maximum 24-h concentrations resulting from Complex II
        and Complex I at 1 km for the low source (Knoxville 1964) .  . 21

  5  Maximum 24-h concentrations resulting from Complex II
        and Complex I at 5 km for the low source (Knoxville 1964) .  . 21

  6  Comparison of 24-h maximum concentrations as a function of
        height for the low source	22

  7  Comparison of 24-h maximum concentrations as a function of
        height for the medium source	22

  8  Comparison of 24-h maximum concentrations as a function of
        height for the high source	.22

  9  Quadrants oriented about the direction of maximum flow 	 30
                                       s
  A-l  Maximum 24-h concentrations resulting from Complex II and
        Complex I at 1 km  for the low source (St. Louis 1976)  .... 46

  A-2  Maximum 24-h concentrations resulting from Complex II and
        Complex I at 5 km  for the low source (St. Louis 1976)  .... 46

  A-3  Maximum 24-h concentrations resulting from Complex II at
        1 km for the medium and high sources using the Knoxville
        1964 data	46

  B-l  Second-highest concentrations as a function of
        averaging time from Knoxville 1964 data	52

  B-2  Five highest estimated concentrations in  rank order
        for 1-, 3-, 8-, and 24-h averaging times from Complex  II
        and Complex I using Knoxville 1964 data	52
                                    VI

-------
                                   TABLES

Name                                                                      Page

  1  Point Source Characteristics	13

  2  Matrix of Run Numbers	17

  3  Highest and Second Highest 24-h Concentrations
        Resulting from Complex II and Complex I at 1 km
        (Knoxville, 1964)	19

  4  Highest and Second Highest 24-h Concentrations
        Resulting from Complex II and Complex I at 5 km
        (Knoxville, 1964)	20

  5  Valley and Valley-BID Estimates of Maximum 24-h
        Concentrations ( Hg/m-')	20

  6  Dispersion and Wind Speed Effects Upon Model Estimates	25

  7  Source Characteristics.  . „	0  . 36

  8  Observed and Modeled S02 Concentrations 	 36

  9  Peak to Mean Ratios During Periods of Maximum 24-h Concentrations  .  . 39

      B-1   Highest and Second Highest 3-h and 24-h
              Concentrations  from Complex II (Knoxville, 1964)  	 48

      B-2   Highest and Second Highest 3-h and 24-h
              Concentrations  from Complex I (Knoxville, 1964)	48

      B-3   Highest and Second Highest 3-h and 24-h
              Concentrations  from Complex II (St. Louis, 1976)  ...... 49

      B-4   Highest and Second Highest 3-h and 24-h
              Concentrations  from Complex I (St. Louis, 1976)	49

      B-5   Five Highest Model Estimates for 1, 3, 8,  and  24-h
              Averaging Times (Medium source; Knoxville 1964)	50

      B-6   Five Highest Model Estimates for 1-, 3-, 8-, and  24-h
             Averaging Times  (Medium source St. Louis  1976)   	 51
                                      VII

-------
                              ACKNOWLEDGMENTS

The authors are especially indebted to Tom Pierce and Alfrieda Rankins
who performed most of the computer programming and executions for this study.
The authors are also grateful for the diligent reviews and comments received.
Finally our thanks go to Joan Emory for her secretarial assistance in the
preparation of this report.
                                    Vlll

-------
                                   SECTION 1




                                 INTRODUCTION




     This report presents the results of an analysis comparing the highest




concentrations estimated for various averaging times by four air quality




simulation models:  Valley, Valley with buoyancy-induced dispersion (Valley-




BID), Complex II, and Complex I.  The analysis was suggested by the Complex




Terrain panel of a regional workshop sponsored by EPA and held in Chicago,




February 25-28, 1980.  The panel outlined the characteristics of three of the




models mentioned (Valley-BID, Complex II, and Complex I), which will be




described later.  The purpose of the sensitivity study was to see whether




or not one of the two sequential (hourly) models, Complex II or Complex I,




might provide a more detailed screening technique for use in complex terrain




situations.  (Complex terrain modeling applies whenever the terrain elevation




is above the stack top of a point source.)




     A more detailed screening model would allow a three-phase approach to




the modeling of air quality in complex terrain.  Phase 1 would be a simple




screening procedure (most likely using Valley-BID) to determine whether or




not the source clearly poses an air quality problem or if the potential for




an air quality problem exists at all.  If the simplified screening results




indicate a potential threat to air quality, further analysis (Phase 2) would




be warranted, using a detailed screening procedure.  If these screening




results in turn indicate that a yet more refined analysis is necessary, a




refined model for complex terrain would be used as agreed upon with




the  Regional Office on a case-by-case basis (Phase 3).




                                    1

-------
     Although no accepted model was available then for a more detailed

screening analysis, the panel wanted to make specific recommendations.

thus, they proposed that two sequential models (Complex II and Complex I)

be assembled and a sensitivity analysis be accomplished in order to

examine these models' usefulness for Phase 3 screening.  Whether or not

either of the proposed candidate models would be ultimately selected as the

refined screening model was impossible to predetermine.  Nor was it possible

to predetermine if the proposed sensitivity analysis would provide a

sufficient basis for making general recommendations regarding the use or

non-use of the proposed models in regulatory decision-making.  However, the

working panel thought that the more specific the recommendations, the more

likely the recommended model assembly and sensitivity analysis would be

performed.

     After discussion during the Complex Terrain session, the panel concluded

first that a data base was needed before a refined model could be developed.

The Office of Research and Development within EPA is currently collecting a

data base for use in developing this model.  Secondly, the panel suggested

that the Phase 2 analysis should consist of calculations made using one of

the sequential models to estimate actual measured pollutant concentrations.

     The standard VALLEY model currently characterizes the lateral bounds

of dispersion using  a 22 1/2°  sector; it characterizes vertical  dispersion

using the Pasquill-Gifford dispersion curves.  No changes were recommended

with regard to lateral dispersion.  However, the panel did recommend  that

the  total vertical dispersion  , oz, be modified such that:


                   2    2
                  az   r  az          + (Ah/3.5)2                     (Eq.1)
                           P-G

-------
where      az        = the total vertical dispersion arid is a function of

                       downstream distance,, stability category, and plume  rise

           azp r     - the Pasquill-Gifford vertical dispersion due to

                       ambient turbulence which is a function of downstream

                       distance and stability category

           Ah        = the plume rise which varies as a function of downstream

                       distance until final rise is attained

The resulting model is the VALLEY-BID.

     The sequential model Complex I uses meteorological input as currently

available for CRSTER.  The model determines vertical dispersion as described

for the Valley model.  Lateral dispersion dimensions in this model are bound

by a 22 1/2° sector.  Plume  rise is estimated using subroutine BEH072 or

equivalent.  For stable conditions, the plume remains at a constant level

above mean sea level (as presently employed in CRSTER).  During neutral and

unstable conditions (i.e., Pasquill categories A, B, C, arid D), the plume

follows one-half of the terrain height variation.  The plume is not allowed

to approach any closer than  10 m to the ground.

     Complex II is identical to Complex I except that the lateral dimensions

are given by:


                       °2  =        +  (Ah/3'5)2                         (Eci- 2)
where       av      - the  total  lateral  dispersion  and  is  a function  of stability,
                    downstream distance,  and  plume  rise

            aXP-G  = the  lateral dispersion  due  to ambient turbulence and is a
                    function of Pasquill  stability  category and downstream
                    distance

In  this  second  model,  avp  p and 0zp_r  are specified using the same  Pasquill
                            \
stability  categories.   In  other words,  "split sigmas" are not recommended.

                                     3

-------
     For the analysis, the panel recommended that both sequential models be




executed along with the modified Valley screening model (Valley-BID) in two




terrain situations, a broad valley and a narrow valley.  All three models




should be executed for three stack heights in the two terrains using two




yearly periods of meteorology, one with a relatively uniform directional




distribution, and one with a skewed directional distribution.




     The following section containing conclusions and recommendations, also




summarizes the major comments and recommendations made by reviewers regarding




the use or nonuse of the proposed models.  Section 3 outlines the analysis




used to generate the results presented in Section 4.  Section 5 contains




short discussions on technical issues that are important in assessing the




reasons for the differences in the modeled concentrations.  And Section 6




lists the major questions posed to the reviewers.

-------
                                   SECTION 2




                         CONCLUSIONS AND RECOMMENDATIONS




     The original reason for examining concentrations from Complex I and




Complex II and comparing with Valley and Valley-BID was to choose one of the




two sequential models as a second level (Phase 2) screening technique.




     The algorithm used in Complex I resulted in concentrations less than



those given by the algorithm used in Complex II.  This result was expected




because of the uniform horizontal spreading in Complex I and the narrrower




more peaked Gaussian distribution for horizontal spreading in Complex II.




     A second result of the study was that the values resulting from




Complex I were approximately the same as those obtained from Valley-BID.




     The computational aspects of this study were completed, a draft report




written and circulated among the members of the previously mentioned workshop




panel.  Although the conclusions reached by the authors and panelists are




not unanimous this section attempts to present opinions of both the authors




and other panel members.




     It is the opinion of the authors that sequential models are useful




when the meteorological data entered to the model can be considered represen-




tative of the flow for the particular situation being simulated.




     The sequential models currently available, such  as RAM, MPTER  and




CRSTER, employ one wind-direction to describe the flow  for each hour.   It  is




assumed in modeling the transport that conditions are horizontally homogeneous




and stationary during each hour.  These assumptions are rarely met.  It  is

-------
quite likely that the flow field is rarely horizontally homogeneous.  Vio-




lations of the assumption of horizontal homogenety cause the model to improper-




ly locate the plume relative to the underlying surface.  If the surface is




relatively flat, one can usually assume that the dispersion, relative to the




plume centerline, is independent of the horizontal location of the plume center-




line.  In such circumstances, the modeled concentration values can still be




accepted, recognizing that the modeled locations of the maximum .impacts are




likely in error.  When the dispersion relative to the plume centerline is




dependent upon the underlying terrain and when the dependency is a  function




of the stability stratification (which is quite likely in complex terrain




situations), it is necessary to properly locate the plume each hour relative




to the underlying surface in order to have confidence in the modeled concen-




tration values.




     For a terrain situation where there are gradual rises  in elevation to




heights in excess of the elevation of the stack top but do  not extend to




the general plume heights, a single wind system may give a  fairly good




indication of the flow.



     However, for a situation of a deep valley with sides extending to




elevations beyond the usual plume heights, the wind direction  from  a single




wind system may give reasonable indications of the flow  for upvalley and




downvalley conditions but the occurrence of indicated wind  direction across



valley may poorly represent  the flow.




      There,  of  course,  exist many  topographic situations between  these  ex-




tremes where  it may  be  desirable to  apply a screening  model and the repre-




sentativeness of  the meteorological  data  is very  difficult  to  determine.




      For  situations  where the  representativeness  of the  single  set  of

-------
meteorological conditions is in doubt, the authors recommend the use of




Valley-BID for screening analysis for the 24-hour averaging time.  The




version of the model using 2.6 as the stable plume rise constant should be




used for consistency with other modeling approaches.  Table 8 shows that at




the distances from sources where Valley has been shown to give reasonable




estimates of concentration, Valley-BID gives nearly the same concentrations.




The inclusion of buoyancy-induced dispersion for large buoyant sources has




long been recognized as desirable.  In most cases of modeling over flat




terrain, the inclusion of buoyancy-induced dispersion makes little difference




in the estimated maximum concentration.  However, for nearby receptors in




complex terrain the difference can be considerable. Primarily because it is




good modeling practice (rather than existence of field data for the near




field), the use of buoyancy-induced dispersion is justified.




     The screening calculations may be made with the Valley model using



the buoyancy-induced dispersion option or closely approximated by pencil and




paper calculations where terrain at plume height is being simulated.




Two of the five reviewers essentially agreed with the authors.




     Although two of the five reviewers agree that measurements  at one




point in a valley configuration are not necessarily representative of trajec-




tories of air motion in the vicinity, they recommended the  use of a sequential




model such as Complex  I  (discounting  calms, see  later in this section)  as  a




more justifiable approach than decision-making with Valley-BID alone, because




Complex  I considers onsite meteorological data.  They reasoned that the manner




in which the model performs impact  (within  10 m) whenever receptor elevations




are above plume height,  and the tendency  for a wind system  to sense some cross-




valley  flow could cause  the model to  overestimate concentrations sufficiently

-------
that it could be used as a screening tool in spite of its modeling




deficiencies.  Additional conveniences seen for using one of the models for




screening are that concentrations could be calculated for elevations of




"middle terrain", i.e., above stack top but below plume level, and calcula-




tions can be made for 3-h periods as well as for 24-h periods.




     One reviewer, although leaning toward the position stated in the




above paragraph, felt the issue of calms possibly obscures the screening




analysis sufficiently that it is impossible to decide defensibly on the use




of one of these models as a screening tool.




SPECIAL ISSUES HIGHLIGHTED BY REVIEW




     Most modelers are aware that the meteorological conditions used for




calculation in Valley-BID are not to be taken literally, but are simply a




numerical mechanism for obtaining a value representative of the second-highest




24-h maximum concentration.  The analytical routine for the Valley model "was




not proposed as  a rigorous mathematical description of the physical circum-




stances which pertain to flow about a terrain feature" (Burt and Slater, 1977).




     To disprove the Valley model characterization of the plume geometry




does not eliminate the possible occurrence of plume impingement.  Likewise,



the meteorological scenario employed by the Valley model for screening




analyses is not  meant to represent the conditions that must occur in order




for pl'jme impingement to occur.  To disprove this particular meteorological




scenario does not eliminate the possible occurrence of plume impingement




resulting in concentrations as high as those generated by the Valley screening




analysis.




     Physical characterization of atmospheric flow over complex terrain is




not easily  achieved because the measurements needed to describe the  flow






                                    8

-------
adequately are extremely difficult to obtain.  For instance, great difficulty




is encountered in estimating trajectories of air flow from single point




measurements — a definite obstacle for the use of simple models having




straight line flow with sequential meteorological input.  In addition, the




modeler must also determine whether or not impingement will occur.  This




might be accomplished by considering the height of the streamline separating




flow over an obstacle from flow around the obstacle.




     Calculations from models such as Complex I and Complex II may vary




greatly depending on how atmospheric calms are treated.  For this study,




calms were treated consistent with other modeling practice, i.e., the meteoro-




logical preprocessor output was used directly as input to the models.  Thus




calms were assigned a wind speed of 1 m/s and assigned a direction sector the




same as the previous hour and then randomized to a direction to the nearest




degree within that 10° sector.  Thus a sequential series of calms will all



be assigned the same 10° sector.  Since Complex II is more sensitive to




directional persistence than Complex I, results of comparisons between these




two models might be quite disparate if calms were treated in a different




manner.




     One possible solution suggested by one  of the reviewers to this sensi-




tivity problem would be to eliminate calculation for hours of calm, and




instead to estimate 24-h concentrations as the arithmetic average of the




concentrations resulting  for the non-calm hours.  If less than 18 h resulted,




no 24-h concentration would be reported.  Note that following the above




suggestion  (deleting data for some hours) adds significant bookkeeping




problems to initially rather simple models.




     In response to the suggestion that calms be eliminated and parts of the




analyses then repeated, no plans have been made to accomplish such a reanalysis.




                                    9

-------
     Although modeling for flat terrain is much less sensitive to the treat-




ment of calms than complex terrain modeling is, sensitivity to calms




was raised as an issue for flat terrain modeling in the public meetings




on modeling in October 1980 and was also discussed in the proposed modeling




guidelines in October 1980 (U.S. Environmental Protection Agency, 1980).




     Several reviewers questioned the manner in which the receptor networks




were set up (in all directions at the same distance).  This set up was




possibly so unrealistic as to invalidate the sensitivity analysis since




receptors injrthe direction of the wind rose maxima were not situated




further away.  An attempt was made to examine this factor in the analysis




under Special Topics.  The effect was to reduce the concentrations no lower




than 0.7 of estimated maxima.
                                      10

-------
                                SECTION 3




                          ANALYSIS PROCEDURES




MODEL FABRICATION




     Complex II and Complex I are identical except for the way in which




they model lateral dispersion.  Many of their features, such as buoyancy-




induced dispersion and the ability to change plume height as a function of




terrain elevation and stability, were adapted from the well-documented




MPTER model (Pierce and Turner, 1980); the MPTER source code was used as




the foundation for the Complex II and Complex I models.  In August 1980,




executable versions of Complex II and I were made available on the EPA




UNIVAC computer system.  The Regional Offices were notified of the models'



availability and were requested to test them.  In November 1980, the




Regional Offices observed that the models did not handle receptor locations




above the mean sea level of the final effective plume height in the same




manner as the EPA Valley model.  The Valley model decreases linearly the




concentrations estimated with increasing receptor elevation height (above




the plume level) to zero at and over 400 m above the undisturbed plume center-




line.  Thus, in December 1980 Complex II and Complex I were reprogrammed




with regard to the treatment of the plume centerline and the resulting




surface concentrations due to the terrain interaction with the plume.




     The Valley model was modified to allow certain options, including the




use of buoyancy-induced dispersion in the vertical dispersion parameter,




and the use of the constant of proportionality, 2.4 or 2.6, for estimating






                                   11

-------
the final plume rise during stable conditions.   The latter  value of 2.6

reflects suggestions made by Briggs (1975).   (Thus the Valley model available

in UNAMAP, Version 4, is capable of simulating  the models referred to in this


document as Valley and Valley-BID.)  A working  copy of this modified version of


Valley was further adjusted for the sensitivity study to allow the receptor


locations to be input to the model in the same  manner as the receptor locations
                        !

are input to the Complex II and Complex I models.   This made it easier to achieve


congruency of receptors among the various models used in the sensitivity analysis.


     In the following discussion, when reference is made to results generated


using the Valley model, note that a constant of 2.6 was used for estimating


final plume rise during stable conditions and no induced dispersion due to


buoyant plume rise was included.  Furthermore,  the model was run as recom-


mended to estimate worst case 24-h concentrations, i.e., Pasquill stability

category F; 6 h of occurrence; and wind speed equal to 2.5 m/s.


     When reference is made to Valley-BID, notice that the constant used in


the stable plume rise is 2.6, the same meteorological conditions as with

Valley are used and induced dispersion due to buoyancy is included in the

vertical dispersion.


SOURCES


     The workshop panel had suggested that the stack heights of the point


sources used in the sensitivity analysis be 1/4 of the final effective p*me

height.  Based on sources examined by Mills (1979), the researchers decided


that the suggestion by the panel regarding the relationship between stack

height and plume rise would not have been typical of most power and industrial


plants.  Using Mills  (1979) the low, medium, and tall source characteristics


were selected for the present evaluation and are presented in Table 1.



                                     12

-------
                   TABLE 1.   POINT SOURCE CHARACTERISTICS


Source
type
Low
Medium
Tall
Stack
height
(m)
75
165
335
Stack
diameter
(m)
3
4
13
Exit
velocity
(m/s)
16
38
16
Exit
temperature
(K)
455
425
425
Emission
rate
(g/s)
700
2,750
10,000
Stable
plume rise*
(m)
91
141
231

 •"•Assumed wind speed of 2.5 m/s, vertical temperature gradient of 0.035 K/m,
  ambient temperature of 293 K.


     The light wind stable plume rise estimates presented in Table 1 were made

using the Briggs (1976) plume rise equations as used in Valley, Complex II,

and Complex I.  These plume rise estimates suggest that the ratio of stack

height to final effective plume height during very stable conditions ranges

from 0.45 to about 0.60.

VALLEY WALL PLACEMENT

     Two valley configurations were used in the sensitivity analysis, a

narrow valley and a broad valley.  The narrow valley measured approximately

2 km in total width and the broad valley measured approximately 15 km.  The

intent was to investigate worst-case 24-hr concentration estimates at two

downstream points.  The narrow valley was used to study the effect on concen-

tration estimates when buoyancy-induced dispersion was of primary concern.

Two kilometers was chosen as the width of the narrow valley because the distance

to final rise was generally just less than 1 km (assuming the source was in the

center of the valley).  The broad valley was used to study the effect on concen-

tration estimates where buoyancy-induced dispersion was less important.


                                    13

-------
     For the actual analysis, two concentric cylindrical networks of receptors




were placed around each pollutant source.  The radii of the cylinders were 1 km




and 5 km to simulate a narrow and a broad "valley".  By using these receptor




networks five different terrain configurations could be simulated in each run:




flat terrain plus rising terrain at four different heights.  (These networks




were not to simulate a vertical canyon wall.)




     A numerical processor was written that generated receptor locations for




use as input to Complex II and Complex I.  Initially, 180 receptors were gener




ated at 10° intervals and at 5 heights to form one of the cylinder walls.




The heights were selected subjectively based on the stable plume rise for the




source type.  The maximum 24-h concentrations at each of the receptor locations




were plotted as depicted in Figures 5 and 6, (presented later in Section 4).




An objective analysis was not performed to determine the proximity of receptor




spacing necessary to estimate the maximum 24-h average concentration to within




some known percentage error at the given downwind distance.  However, a sub-




jective analysis was performed which consisted of searching for the maximum




possible 24-h average concentration with receptors in 1° intervals and 25 m




vertical separation between receptors at the 1 km downwind distance.




     The results for the low source suggest that the 24-h maximum concentration




determined using the original receptor grid (10° intervals with 25 m vertical




spacing) was within 20/°o of the highest value determined during the analysis




with -1° receptor spacing.  This result most likely represents-an upper bound




on the uncertainty of locating the maximum 24-h concentration since the low




source had very strong gradients in the  pattern of maximum 24-h concentrations.




In any event, the 20?o error was deemed tolerable for the purpose of judging




variations in maximum concentrations between the models used  in this study.






                                    14

-------
METEOROLOGICAL DATA




     In seeking meteorological data for the sensitivity study, the intent




was to see if skewness in the wind direction frequency caused by terrain




channeling of the wind would affect the model results.  Since the 1964




Knoxville, Tennessee, hourly surface meteorological data were readily




available they were selected for use as the site having a skewed, bimodal




wind rose, typical of sites in narrow valleys.  The 1976 St. Louis surface




meteorological data were selected as representative of a more circular wind




rose site.




     Figure 1 depicts the flow vector frequency relative to a completely




circular wind frequency for the 1964 Knoxville data.  If wind from every




10° interval  (from 10° to 360°) were as equally likely to occur, then




8784/36 = 244 hours would occur for each direction (note that 1964 and 1976




were both leap years).  Therefore the flow vector frequency relative to a




frequency equal  from all directions was determined by dividing the number of




hours of occurrence of each flow vector direction by 244.  The hours of stable




conditions and the hours of calm assigned by the preprocessor to each flow




vector direction are also given in Figure 1.  The hours of stable conditions




follow fairly well the overall frequency distribution.  Figure 2 depicts the




same flow vector frequency  for the 1976 St. Louis data.




     A processor was constructed to convert the meteorological data into a




format compatible  for input to Complex II and Complex  I.  The processor set




the mixing height  to 5000 m; using a mixing height of  5000 m removes the




mixing height as a variable affecting the concentration results due to re-



flection  effects.
                                     15

-------
     I  I  I  I       I  I     I  I     1-1  I  I  I  I  I    I  I   I  1  1  I
 0    20   40   60   80   IDO  120  140  160  ISO  200  220   240  260  280  300  320  340   360
                          AZIMUTH (FLOW VECTOR SECTOR). d*grm

Figure 1. Flow vector frequency, hours of stable, and hours of calm, Knoxville. TN, 1964.
   |   I  |  |  |  |  |  |  I  I  IJ  I  |   I  I  I   I  I  I   I  I  I   I  1  I   M  I   I  I  I   I  I  I
      20   40   60  80   100  120  140  160  160  200   220
                            AZIMUTH (FLOW VECTOR SECTOHI. digrn
 Figure 2. Flow vector frequency, hours of stable, and hours of calm. St. Louis, MO, 1976.
                                      16

-------
MODEL COMPUTATIONS ACCOMPLISHED




     Twenty-four basic runs were made as part of the sensitivity study.




The matrix showing the run numbers is shown in Table 2.  Each of these




runs was for a single pollutant source for 180 receptors, simulating a 1-yr




period with output consisting of the high-five table.  Nearly all runs were




executed in a deferred batch mode of computer operation.  This rendered




costs to be 15% of those that would have been incurred under normal batch




processing.  Processing was generally done overnight.  As many as 4 runs




were executed per night so availability of computer facilities and turnaround




were quite adequate for the study.




     Execution of the runs required 56K of core or less.  Each execution of




COMPLEX II costs between $40 and $46.  Each execution of COMPLEX I costs




between $29 and $34.  Each execution of Valley costs about $0.25.






                      TABLE 2.  MATRIX OF RUN NUMBERS
                      KNOXVILLE MET DATA
                  Complex II    Complex I
    ST. LOUIS MET DATA
Complex II    Complex I
OUUJ. UB
Low
Medium
High
1 km
1
3
5
5 km
2
4
6
1 km
7
9
11
5 km
8
10
12
1 km
13
15
17
5 km
14
16
18
1 km
19
21
23
5 km
20
22
24
                                     17

-------
      Each  run accessed  3 disk  files  and two-disk file  elements.  Figure  3
is an example runstream.  Since many of the runstreams were  similar,  the

runstream  itself was stored  in a disk file element.  This example  was
stored in  WRKSHP.RUN6/XQT.   The compiled  and'.mapped model absolute for this
run  is in  the element UNAMAP.COMPLEXIIABS. .-liThe  hourly meteorological data

for  this example is in  file  FORKNOX64.  The'of.eceptor list is in  file GRHIGH1
For  each run the output was  placed  into a print  file,  in this case PFRUN6.,
so that it could be edited if  further computer processing of output became
necessary.
                      @RUN, D/R 12ADR/70, ACCTNO., EOB. 23.50  -"'."
                      <§> .  RUNSTREAM WRKSHP, RUN6/XQT     HP
                      @SETC,O
                      @ASG, A UNAMAP.
                      @ASG, AGPHIGH1.
                      <°>ASG. A FORKNOX64.
                      @ASG. A PFRUNG.
                      @BRKPT PRINTS/PFRUNG
                      @XQT UNAMAP . COMPLEXIIABS
                      RUNS HIGH 1964 KNX/BNA 5 KM
                      IOPT(25)=1, BID, FINAL RISE. ZMIN=10, 1 DISTANCE. 5 HEIGHTS
                      INPUT BY A. RANKINS (ENVIRONMENTAL OPERATIONS BRANCH), 2/21/81
                      64. 1, 1,366.24,3, 1,0, 1., 1..0.
                      1. 1. 1, 1. 1.0, 0,0, 0, 1, 1. 1, 1. 1, 1, 1, 1. 1,0.0,0.0,0, 0, 1
                      10.. .07. .07. .1, .15, .35, .55. .5, .5, .5. .5. .0. .0, 10.5
                      HIGH STACK  0.  0.   10000.  0.   335.  -425.   13.   16.
                      ENDP
                      @ADDGRHIGH1.
                      ENDREO
                      @ADD FORKNOX64.
                      @8RKPT PRINTS
                      @FR£E PFRUN6
                      @SYM, U PFRUN6... FD04PR
                      @FIN
                                   Figure3. Example runstream.
                                          18

-------
                                SECTION 4




                                 RESULTS   .




     Plots were made initially to depict the variation of the maximum 24-h




concentration at each downwind distance as a function of receptor elevation




height and bearing from the source.  Figures 4 and 5 are examples of such




plots and are typical of the results in general.  Several other examples of




such plots are given in Appendix A.  The maxima in these plots are quite




localized, spanning 10 to 15° in horizontal azimuth and 100 m in the




vertical.  Tables 3 and 4 summarize the maximum 24-hour concentrations




estimated using the Knoxville meteorological data.  Figures 6, 7, and 8




depict these results with the Valley and Valley-BID estimates, for comparison.



Table 5 lists the Valley and Valley-BID estimates.
TABLE 3. HIGHEST AND SECOND-HIGHEST 24-H
SULTING FROM COMPLEX II AND COMPLEX I AT 1
1964).
Source
type

Low



Medium



High


Receptor
height
(m)
200
175
150
125
450
350
275
200
550
525
500
475
Highest 24-hour
concentration, £ig/m3
Complex II
11602.1
12157.6
11744.3
8935.4
14708.9
20895.2
23066.8
2018.6
34997.4
36132.2
36929.2
32371.1
Complex 1
5145.4
5394.9
5251.0
2695.9
6882.3
9761.8
10816.7
1083.5
18534.2
19084.6
19460.2
17194.0
CONCENTRATIONS RE-
KM (KNOXVILLE, TN,
Second highest 24-hour
concentration, pg/m3
Complex II
7019.9
9018.3
8825.8
8252.0
10440.8
14997.5
18174.3
1741.3
26518.9
28410.6
30077.9
26868.7
Complex 1
3759.6
3932.3
3920.4
2253.5
5154.4
7331.7
8201.7
938.3
13983.1
14212.4
14999.4
13537.3

                                     19

-------
TABLE 4.  HIGHEST AND SECOND-HIGHEST 24-H CONCENTRATIONS RE-
SULTING FROM COMPLEX II AND COMPLEX I AT 5 KMIKNOXVILLE, TN,
1964).
Source
type

Low



Medium



High


Receptor
height
(m)
200
, 175
150
125
450
350
275
200
550
525
500
475
Highest 24-hour
concentration, jjg/m-3
Complex II Complex I
1876.7
1977.8
1978.0
1454.8
2845.0
4019.0
4505.4
1180.9
8468.5
8710.6
8852.1
8005.2
652.9
744.8
750.6
557.3
1068.5
1388.9
1 7 1 1 .9
474.1
3202.4
3321.6
3406.6
3113.9
Second highest 24-hour
concentration, ^g/m3
Complex II Complex 1
1270.0
1358.2
1400.0
1037.9
1860.4
2671.1
3234.7
877.2
5786.6
6198.8
6561.0
6052.9
508.3
537.2
555.8
418.8
773.3
1099.7
1237.4
389.7
2369.4
2434.9
2516.1
2359.3
TABLE 5. VALLEY AND VALLEY-BID ESTIMATES OF MAXIMUM 24-H CON-
CENTRATIONS (jug/m3).
                          Valley
                                                  Valley-BIO
      Source
                      1 km
                                 5km
                                               1 km
                                                          5km
      Low
      Medium
      Tall
  7,805
 31,304
112,170
  743
 2,982
10.685
 4,426
12,5??
28.772
 607
2,011
5,187
                               20

-------
                                                                                                                                COMPLEX II
                                                                                                    1OOO        1OOO

   200— 560 668  285  584  656 448 451  538  491  502  372  346 322 658 181  384  894 642 453 795 461  499  844  879/1035\ 512 598/I)601 370 454 510 596 543  600  787  533

   175—599 703  297  617  698 478 483  564  525  537  398  370 338 703 194  410  955 687 484 829 492  533  899  924 1086 540 632 1216  395 486 54S 621 581  641  842  570

   ISO— 616 698  294  603  702 483 499  553  551  553  414  383 346 726 207  407  988 708 507 786 517  S56  927  902M048J 552 625m74J 404 506 559 619 605  662  861  5BS

   125 — 297 392  259  351  372 280 299  276  387  303  241  187 183 380 137  221  484 379 316 427 346  353  50)  476 521 305 345  5B6  224 299 233 353 375  332  417  303
O
CO
                                                                                                                                COMPLEX I
2 200— 278 406 251  357  444 434 284 313 216  177  185  271  284 276 185  236  320  346 364 384 359  265  292  473 402 402 473 /515J 356  347 406 314 284  283  330  295


5 175 — 297 425 259  376  472 461 289 320 230  179  188  284  302 295 197  252  342  370 380 4Q] 374  282  308  494 420 420 495  5391 380  372 434 337 304  302  353  315
< 150—307 410 238  367  475  462 291 299 240 178  168  288  310 304 198  261  354  383 391 383 354  290  313  477 400 400 478 \SZSJ 388  383 447 350 310  295  359  324


*f, 125—  177 204 134  203  241  243 191 173 156 127  124  152  160 155 108  127  177  212 205 219 192  178  202  269 230 200 238  270  198  213 243 212 172  140  207  156

                 I        I        I       I        I        I       I        I        I       I        I        I       I        I        I       I
                 20
                         40
                                60
                                        80
                                               too
                                                                                     200      220     240


                                                 AZIMUTH-FLOW VECTOR SECTOR (degrees)
                                                                                                            260
                                                                                                                    280
                                                                                                                           300
                                                                                                                                   320
                                                                                                                                           340
         Figure 4.  Maximum 24-h concentrations resulting from Complex  II and Complex I  at 1 km for the low source

         (Knoxville, TN.  1964).  (Concentrations are  pg/m3 divided by 10).
                                                                                                                                COMPLEX II
HI  200—  94   98   4b  84  68  64
C/l
<  175— 101  104   48  88  73  68

CO
^  150—104  107   49  88  74  70


U  125—  75   84   38  7t  66  58
<

t-
00
                                                        52 100   28  60  174  98  78  128
                                     55   64  69  46  44  40  79
                                                                                                                                            120
                                                                                                                         78  84 100   95  110  129  101


                                                                                                                         81  87  99   99  114  133  104


                                                                                                                         C2  64  74   76   83  96  77
u
CD
Z
g
^
UJ
_J
UJ

zoo —
175 —




37
39
41
I
3GC

56
59
59
44


34
36
35
1
20

54
57
57



57 56
61 60

1
40

39 43
40 45

1
60

29 25 27
31 26 28

1 1
80 100

39 39
41 42

1
120

38 26
41 28

31 21
I
140

33 45
35 48

26 36
I
160

47 51
51 53

1
180

52 48
54 51

42 38
1
2'JO

38 «2
40 45

1
220

65 55
69 58

1
140

55 65
58 69

1
2GO
70
/71\ 49
I 74 \ 52
^— '
1
280
COMPLEX 1
48 56
51 60

1
300
43 38
46 41

1
3ro
39 45
42 48

1
340
                                                    AZIMUTH-FLOW VECTOR SECTOR (degrees)


         Figure 5.  Maximum  24-h concentrations resulting from Complex II  and Complex I  at 5 km for the low source

         (Knoxville, TN, 1964).  (Concentrations are ^g/m3 divided by 10).
                                                              21

-------


250

225

[.200
1
:
! 175


ISO


12S


too
11
\l\ 1 \
\- \
\ \

i
i '•
i'
- ;• D
1
i
i "
i
1
/•" ° ;
// /
? X
^» o^X.

..-•''/'
,••'•:>' , ,
2 ,03
1 \ 1 1
I
\
\
\
\
\
\
• 1 0
1
I
1
• I km
* / ° OCOMPLE 1 ~~
/ 9COMPLE
/ — VALIE BtO
' • O — — VAIL -
Skm
OCOMPLE i
O BCOMPLE -
— — VAll BID

1 1 1
ID4 10
 Figure 6. Comparison of 24-hr maximum concentrations ai a (unction of height for the low source.
    100
    ISO
              1 km
             O COMPLEX II
             • COMPLEX!
           	VALLEY 8ID
           	VALLEY
              Shm
             O COMPLEX II
             • COMPLEX!
           —	VALLCY BID
           	VALLEY
                                 I
                                                    I
      102                        ID3                        10*                         10s
                                24htMAXIMUM CONCENTRATION. u|/m3

Figure 7. Comparison of 24-hr maximum concentrations as a function of height for the medium source.
    600
    <2Sl—
      103
                                                    I
                                                                               I
                                                                      1  km
                                                                    CCGM?LEX!I
                                                                    • COMPLEX I
                                                                  	VALIEY BIO
                                                                  	VAILEV
                                                                      Slim
                                                                    D COMPLEX II
                                                                    • COMPLEX)
                                                                  	VALLEVSID
                                                                  	VALLEY
                                                                  	1
                                 I04                         10s
                                   24-hrMAXIMUM CONCENTRATION. ;jg/m3
   Figure 8. Comparison of 24-hr maximum concentrations as a function of height for the high source.
                                             22

-------
     How calm conditions were modeled was the determining factor in the




results just presented.  The meteorological-input data used for Complex II




and Complex I were generated by a meteorological preprocessor.   The processor




performs the following functions:  (1) calculates Pasquill stability class




from cloud cover, ceiling height, wind speed, time of day, and time of year;




(2) converts reported wind direction (in 10° increments) to a flow vector




(wind direction ± 180°); (3) converts wind speed to m/s and converts air




temperature to Kelvin; (4) generates a randomized flow vector from the above




flow vector by using random digits to add from -4 to +5 to the flow vector to




create random flow vectors to -1° within the same 10° sector; and 5) inter-




polates twice daily mixing heights to result in hourly estimates of mixing




height.  (Note that mixing height was effectively eliminated as a variable in




this sensitivity study.)  Note also that the treatment of winds reported as




calm on the input is as follows:  the speed is assigned 1 m/s, and the flow




vector remains the same as for the previous hour.  Thus, if a sequence of




hours is reported as calm in the input data to the processor, although the




randomized flow vector is varied over 10°, the winds for this period are




all at 1 m/s and are blowing toward only one 10° sector.  The developers




of the preprocessor thought this procedure was the most reasonable one to




handle calms since any topographic relief at a site might cause drainage




flows with little direction change at nighttime when calms are most pre-




valent.




     For the Complex II model, day 299 of the Knoxville data and days 238 and




320 of the St. Louis data caused the highest estimates of 24-h concentra-




tion.  For the Complex I model, Julian day 299 of the Knoxville data and




Julian day 320 of the St. Louis data caused the highest estimates of 24-h.






                                    23

-------
concentration for all three sources at both distances.  Day 299 had 7 h


of stable, calm conditions in which the wind directions were assigned by


the preprocessor within the same 10° sector.  Of these 7 h, 6 h were


stability F or G and 1 h was stability E.  Day 320 had 7 h of stable, calm


conditions in which the assigned wind directions were within the same 10°


sector.  Of these 7 h, 6 h were stability F or G and 1 h was stability E.


Day 238 had 6 h of stable, calm conditions, of which 5 h were F or G and


1 h was E.  The wind speed at 10 m during these stable, calm conditions was


assigned 1.0 m/s, since the meteorological preprocessor sets the wind speed


to 1 m/s whenever a calm is reported.  Complex II and Complex I extrapolate


the 10 m wind speed to stack top height for use in the plume rise estimates


and for use in the dispersion calculations. Differences between these extrapo-


lated wind speeds and the 2.5 m/s speed assumed by Valley and Valley-BID
              ^

account for most of the differences in maximum 24-h concentrations by Complex II


and Complex I as compared with those generated using Valley and Valley-BID.


     Table 6 attempts to summarize the variation in maximum concentrations


one would anticipate given the four models used in this analysis.  Column


(6) of Table 6 lists the wind speeds at stack top that would be employed by


Complex II and Complex I for calm, stability category F conditions.  The


differences in plume rise estimates (comparing the Valley and Valley-BID


values to the Complex II and Complex I values), reflect the effects of


differences in wind speed.  Note, an ambient air temperature of 293 K was


used for the plume rise estimates.  Inspection of column (4) of Table 6


reveals that by  using a higher wind speed in Complex  II and Complex I, the


vertical dispersion actually becomes less than that used in Valley-BID,


because the buoyancy induced dispersion is proportional to the plume rise.




                                    24

-------
     TABLE 6. DISPERSION  AND WIND SPEED  EFFECTS UPON MODEL ESTIMATES.
(1)
Model
type



Valley





Valley-BID





Complex 1





Complex II



(2)
Source
type

Low

Medium

Tall

Low

Medium

Tall

Low

Medium

Tall

Low

Medium

Tall

(3)
X
(km)

1
5
1
5
1
5
1
5
1
5
1
5
1
5
1
5
1
5
1
5
1
5
1
5
(4)
Vertical
dispersion
(m)
.
13.38
35.71
13.38
35.71
13.38
35.71
29.24
44.17
42.45
53.84
67.34
75.04
28.26
42.12
35.43
47.23
48.89
58.01
28.26
42.12
35.43
47.23
48.89
58.01
(5)
Lateral
dispersion
(m)
t
156.66
783.32
156.66
. 783.32
156.66
783.32
156.66
783.32
156.66
783.32
156.66
783.32
156.66
783.32
156.66
783.32
156.66
783.32
41 36
147.73
47.00
149.27
57.83
153.02
(6)
Stack top
wind
speed
(m/sl

2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
3.0
3.0
4.7
4.7
6.9
6.9
3.0
3.0
4.7
4.7
6.9
6.9
(71
Stable-F
plume
rise
(m)

91
91
141
141
231
231
91
91
141
141
231
231
86
86
114
114
164
164
86
86
114
114
164
164
18)
Xmax (Expected)
Xmax (Valley-BID)
t
1.76
1.22
2.46
1.47
3.85
2.04
1.00
1.00
1.00
1.00
1.00
1.00
.86
.87
.63
.60
.50
.47
3.22
4.62
2.09
3.17
1.34
2.38
(9 1
X max (Calc.)
Xmax (Valley-BID)

1.76
1.22
2.49
1.48
3.90
2.06
1.00
1.00
1.00
1.00
1.00
1.00
1.22
1.24
.86
.85
.68
.66
2.75
3.26
1.83
2.24
1.28
1.71
Vatley and Valley-BID compute O2 dif-
ferently than Complex I and Complex II
because different algorithms are used to
approximate the  Pasquill dispersion
curves.  At  1 and 5 km the algorithms
estimate the vertical  dispersion  para-
meter as:
           Valley      Complex I
 X(km)   Valley-BID    Complex M
   1       13.38         13.95
   5       35.71         33.88
1  The lateral dispersion given is the effec-
  tive Oy for a sector model, computed as
  ( /3Wl6) X, which was found by equat-
  ing  Eq. 5.13 and  Eq. 3.3 of WADE
  (Turner, 1970) and solving for Oy.
     List the ratio of the expected maximum
     24-hour concentration to the Valley-BID
     estimate, for the given source type.  As-
     sumes a 4 to 1 relationship between the
     peak 1 -hour and average 4-hour concen-
     trations. Ratio is computed as:
Xmax (Expected) = 1/(0z0vu) exp{-0.5(10 m/02)2)

max (Vatley-BIO)  l/IOj Oy'uj exp(-0.5(lO m/0^ t2)
     where primed quantities are values ap-
     propriate for Valley-BID calculations.
                                                     25

-------
Decreasing the vertical dispersion tends to increase the maximum concentra-




tions estimated to occur during plume impaction, while a higher wind speed




tends to further dilute the pollutant and thereby lower the estimates.




Column (8) of Table 6 lists the expected maximum 24-h concentrations (as




compared to the corresponding results from Valley-BID for the given source




type) as a function of pollutant source type and dispersion model.  The




assumption made in computing these ratios was that the peak 1-h concentration




is four times the maximum 24-h concentration.  In other words, the worst-case




condition occurs for 6 out of 24 h.  This assumption is the same one used in




the short-term options of Valley and Valley-BID for estimating the maximum




24-h concentration.  Column (9) of Table 6 lists the actual results (simula-




ting 1 yr with the models) corresponding to those listed in column (8).




     The differences in the results listed in columns (8) and (9) for the




Valley model reveal that the actual maxima were not sampled by the vertical




array of receptors used in the model runs.  However, the differences suggest




that the Valley and Valley-BID results are within 2% of the actual maximum




values possible if one would happen to place a receptor directly on the




maximum.  The results listed for Complex I are a bit higher than would be




anticipated.  The primary cause for the differences is that the worst-case




condition occurred for more than 6 h (usually 7 to 8 h with the inclusion




of the hours with calms) during the 24-h period, and the 24-h concentration




was thus slightly greater.  Differences in plume rise caused by using air




temperatures other than 293 K account for most of the other variations from




the expected results.  Since Complex II has  a narrower lateral plume than




that employed in Complex I, it is difficult  to achieve the expected result




that during 6 out of the 24 h the pollutant  should reach the receptor.






                                    26

-------
This usually only occurs to within 1° of the receptor for the 3 to 4 h, but




never for 6 h.  Reducing the frequency of the worst-case condition from 6 to




4 h accounts for most of the variation from the expected results listed for




Complex II.




     Appendix B lists the highest and second-highest concentrations estimated




by Complex II and Complex I for the 3-h and 24-h averaging times.  Also,




tables are given for the medium source for the 5 highest ranked concentrations




for various averaging times.



     These results are in accord with the results discussed here in that




the differences in handling the wind speed and the differences in character-




izing the vertical and lateral dispersion describe the results.  Besides the




variation caused by these differences, the persistence of the worst-case




condition during the averaging time of interest is all that is needed to




summarize the variations seen between the results by the four models.
                                    27

-------
28

-------
                                 SECTION 5




                              SPECIAL TOPICS




     In formulating a position regarding the proposed screening techniques,




limitations must be recognized both in the model physics and in the present




understanding of the processes.  Plume impingement on elevated terrain during




stable conditions is a complex process.  The process can be summarized as a




set of questions:




     o  Will the plume's trajectory reach the terrain?




     o  What are the minimum dimensions of the plume when it




        reaches and impinges on the terrain?




     o  What is the distribution of mass within the plume during



        impingement, relative to the surface of the terrain?




     o  If the concentration resulting during stable plume




        impingement were to result in worst-case (highest)




        3-h and 24-h concentrations, what are reasonable




        values for the ratios of the worst-case 3-h and 24-h




        concentrations to the worst-case 1-h impingement con-




        centrations?
                                    29

-------
ALONG VERSUS CROSS  VALLEY CONCENTRATION ESTIMATES

     Since the sensitivity calculations were made  as  if terrain features

existed at all possible azimuths from the source,  it  was desirable to

examine what the  effect would be upon the maximum  concentrations if terrain

features existed  in only two of the possible four  quadrants of 90° azimuth

each.

     The azimuth  of highest flow vector frequency  determined an axis referred

to as "along"  (see  Figure 9).  This is 60°  for the 1964 Knoxville data and

120° for the 1976 St.  Louis data.  An axis  at 90°  to  the along axis was

referred to as "across."  Quadrants were defined centered on these axes

resulting in two  along quadrants (A and B in Figure 9), and two across

quadrants (C and  D  in  Figure 9).
                         285'
                                                   KNOXVILLE, TN,
                                                    FLOW VECTOR
                                                60°  FREQUENCY
                                                     MAXIMUM
                                               105°
                                 195'
               Figure 9. Quadrants oriented about the direction cf maximum flow vector frequency.
     A tabulation was then made of model  estimates of 24-hour concentrations

for the  24  computer runs by listing the maximum concentration in each  of the

four quadrants defined above.
                                      30

-------
     The "along" quadrants were found to contain the highest 24-h concentra-




tions for both the Knoxville and St. Louis data.  However, these maxima were




in quadrant B, which is opposite the quadrant of high flow vector frequency.




(For the Knoxville data the direction of maximum flow yector frequency was 60°;




the direction of the 24-h maximum concentration was 270°.  For the St. Louis




data the direction of maximum flow vector frequency was 120°; the direction




of the 24-h maximum concentration was 310°.)




     To assess the magnitude of the concentration on the valley sides compared



to maxima in all directions, for valley situations having the two selected wind




distributions, the ratio of the across quadrant maximum to the overall maximum




was determined corresponding to all 24 runs made with the models.  Consider




first the results from the model Complex II.  For the Knoxville data, the




three source sizes had concentration ratios at 1 km ranging from 0.76 to 0.82




and at 5 km from 0.90 to 0.91.  For the St. Louis data, the three source sizes




had concentration ratios at 1 km ranging from 0.80 to 0.83 and at 5 km from 0.84




to 0.86.  Osing Complex I, for the Knoxville data, the three source sizes had




concentration ratios at 1 km ranging from 0.82 to 0.86 and at 5 km from 0.82 to




0.83.  For the St. Louis data, the three source sizes had concentration ratios




at 1 km ranging from 0.71 to 0.72 and at 5 km from 0.71 to 0.72.




     Thus, if locations with these wind distributions had terrain in two




directions instead of all four, and one assumes that straight line flow




could occur as indicated by the wind vane (see caveats on this point else-




where in this document), then estimated concentrations would be between




0.7 and 0.9 of those estimated by runs with terrain in all directions




depending on distance and model used.






                                    31

-------
     A "spin off" of the above analysis is the consideration of lowest


maximum 24-h concentration among the 36 azimuths.  The ratio of this


lowest 24-h maximum concentration (once a year) to the highest 24-h


maximum concentration varied from 0.15 to 0.22 for Complex II and from
                 f

0.27 to 0.36 for Complex I.  Thus, for these wind distributions, the


azimuth with the lowest 24-h maximum had a concentration more than 1/7


that of the extreme maximum.


MODELING OF PLUME TRAJECTORIES


     The sequential models still contain the inherent weakness of modeling


dispersion downstream as straight paths along the input wind direction for


the hour.  In complex terrain this assumption is questionable.  Hence,


whether plume impaction as modeled actually occurs is not known.  Nor is it


known whether or not if the impact were to occur, it would occur as simulated.


One could argue that the characterization of the lateral dimensions as a


pie-shaped 22 1/2° sector is obviously wrong and could not be verified by


actual plume observations.  But then one could also argue that it is


questionable whether a plume can maintain a Gaussian cross-section both


vertically and laterally while interacting with  a sheer cliff or a steep


bluff.


     Because of these arguments, and similar ones not presented, caution


must be exercised in formulating a recommendation for the use or nonuse


of  the proposed models.  These models were not proposed because they  were


considered technically defensible; they were proposed because it was  hypothe-


sized that they would yield conservative estimates of the worst-case  impacts


in  complex terrain.




                                    32

-------
     The question which now must be addressed is whether or not models

employing straight line-of-sight transport can successfully indicate the

duration of and the nonoccurrence or the occurrence of plume impaction.

Note that even a better representation of the plume's path in the vertical

does not resolve the issue.  The algorithms employed to'estimate actual

concentrations merely attempt to estimate the size of the plume and the

distribution of the mass in relation to the receptor.  The real issue is

whether or not a model capable of only straight path transport can estimate

the frequency (and the duration when it occurs) of material reaching each

modeled receptor location?

     This issue will not be solved any time soon.  But it is important to

realize that improper characterization of plume transport in the vicinity

of major terrain features may lead to underestimates of maximum concentra-

tions as well as overestimates.  Consider the problem of an isolated point

source in a narrow valley.  The air flow within the valley is primarily

constrained to be along the valley, but transport will also reflect the

cross valley component of the flow.  Hence the plume might "lay-up" along

one valley wall and remain there persistently despite the fact that a single

point measure of wind direction at the stack might indicate considerable

wind direction variation.

     Hewson and Gill  (1944) discussed the meteorological processes affecting

SQ.2 transport within  a portion of the Columbia River Valley near Trail,

B.C.  Here are some of their conclusions.

     (Pg. 23)

          The effect  of this irregular topography is to produce both
     large and small  eddies.  When the wind velocity is high, these
     eddies can be studied and largely accounted  for.   However, when
     the winds are very light, it is almost impossible  to explain


                                    33

-------
     the variations in wind directions.   Frequently,  during periods
     of light winds,  the smoke leaving each of the several stacks at
     the smelter can be observed going in a direction different from
     that taken by the smoke leaving the others.   For instance the
     smoke leaving the Dwight-Lloyd stack may be  moving up river and
     that from the zinc-plant stack going down river, while the
     smoke from the blast-furnace stack may be moving across the
     valley.

     (pg. 24,25)

          There are two general types of fumigations, diurnal and
     nondiurnal.  The diurnal fumigations occur chiefly during the
     growing  season.   They are found at two periods of the day -
     from 08  to 10 hr. and from 18 to 20 hr.  The nondiurnal fumiga-
     tions show few clearly defined characteristics.   Usually their
     strength is approximately inversely proportional to the distance
     from the smelter	Fumigations of this type may last several
     hours, or under stagnant winter conditions they may persist con-
     tinuously for several days.

     (in discussing the diurnal fumigation pg 116)

    ff;..  The concentration of sulfur dioxide on the west side
   "'>of the valley at about 100 dkm is the interesting feature
    "of this  diagram.  This location agrees very well with that
     anticipated from study of the east-west wind components at
     05 hr.,  shown in figure 48.  From this curve it can be seen
     that the sulfur dioxide will be kept on the west side of the
     valley by the strong components of wind from the east in the
     gas-carrying layer.

Hewson and Gill were summarizing an extensive field program complete with

pibal wind observation, temperature soundings using kites and balloons,

aircraft sampling, and surface monitors (most of which were on the Valley

floor).  Considerable effort was involved, largely due to the complex

and often confusing transport that can result in the vicinity of major

terrain features.  .A single wind instrument coupled with a simple line-

of-sight transport model would not have explained or predicted the observed

transport during the worst-case calm conditions.

     Each complex terrain situation is most likely unique when viewed  in

detail.  Broadly speaking,  similarities may be found with other terrain
                                    34

-------
situations, however, it is impossible to quantitatively estimate the




importance the details of the situation have on influencing the actual




transport. What can be said is that simple models such as Valley, Valley-BID,




Complex II, and Complex I, can characterize the size the plume might have if




it were, by some process, to have reached each modeled receptor location.




Considering the crude approach used to model the transport, it is impossible




 to determine whether or not a plume will affect a particular piece of real




estate.  In other words, models such as Valley, Valley-BID, Complex II, and




Complex I may be able to tell something about the worst-case impacts one




might expect at any given downwind distance, but, since these models can not




model changes in transport due to complex terrain, they can not address




whether or not the impact will occur.




EVALUATION OF VALLEY (1977)




     The Valley model's performance was originally reported by Burt and




Slater  (1977).  Their analysis consisted of a comparison of the Valley




model estimates of second-highest 24-h concentrations with actual monitoring




data.   These monitoring data were the only such data available from sites




on elevated terrain near "singular" polluting facilities.  Burt and Slater's




initial hypothesis  (1970-1972) was that plume impingement during stable




conditions, as simulated in the Valley Model, might well approximate the




potential  second-highest 24-h concentration annually; only later did field




data become available.






     The  source data for that study are given in Table 7, and the model




estimates  and monitoring results are given in Table 8.

-------
                         TABLE 7.  SOURCE CHARACTERISTICS.

Site
name
Crusher





Source
name
Garfield




Stack
height
(m)
124
124
29
32
15
Plume "
rise
(m)
160
140
49
64
15
Emission
rate
(g/sl
3324
2072
202
134
2222
Distance
to receptor
(km)
6.4
6.7
7.0
7.0
7.0
Plume height '
above receptor
(m);
+71
+20
•196
•178
•244


(T/PI(P0/T0) +
1.1145




 Lower-    Garfield
 Lake Point
124
124
 29
 32
 15
160
140
 49
 64
 15
3324
2072
 202
 134
2222
4.5
4.5
4.5
4.5
4.5
 +74
 +23
-193
-175
-241
                                                                                   1.1550
#106


#107


Phelps
Mine

Jones
Ranch

Navajo
Gen. Stn.,
Arizona
Navajo
Gen. Stn..
Arizona
Morenci
Smelter,
Arizona
Miami
Smelter,
Arizona
236


236


183
183

84
61

205


203


111
121

96
83

523


465


5522
8060

815
3105

22.8


23.2


4.7
4.7

2.9
2.9

•97


•169


•41
-31

+43
+7

1.1550


1.2220


1.1282


1.1086


*  For stability F; with 2.5 m/s wind speed at stack top; 2.4 constant used in plume rise formula.
+   Negative indicates undisturbed stable plume final effective height  is estimated to be below receptor
   elevation height.
t  T is station temperature (K) and P is station pressure (mb).  To and Po are standard temperature and
   pressure, 298 K and 1013.2 mb respectively.  Used to convert model estimates to STP or ppm.
            TABLE 8. OBSERVED AND MODELED S02 CONCENTRATIONS.
Site
name
Crusher
Lower-*
Lake Point
#106
#107
Phelps
Mine
Jones
Ranch
Valley*
(2.4)
24801
US

36
25
15490

8610

Valley
(2.6)
2014
0.94

38
27
15902

7820

Valley-
BID
2199
1.08

27
19
11006

6357

Complex 1 Complex II
3527 10966
1.88 5. 57

22 77
15 52
8556 25438

8601 23998

Observed maximum 24-h
Highest Second highest Period
2564
6130
2.66
221
32
30
2547

2042
2642
2473
3130
L20
2J4
19
15
2416

1760
1548
4/73' 1/74
2/74- 1/75
3/75'12/75
V76- V7e
10/74- 2/75
10/74- 2/75
1975

1974
1975
  Underscore indicates concentrations are reported in ppm, otherwise concentrations are reported in £/g/m3
  reduced to STP.
  Valley (2.4) refers to the constant of 2.4 used to estimate the final plume rise during stable conditions. Like-
  wise (2.6) refers to a constant of 2.6 being used in the stable plume rise estimates. Note, Valley-BID,
  Complex I, and Complex II use the 2.6 factor.
  As was done by Burt and Slater (1977), a 3 hour half life was assumed in all model estimates, and for re-
  ceptors above the undisturbed  plume centerline all the models reduced the concentration estimates by
  (401-Za)/400 where Za is the height (m) of the receptor above the centerline.
                                        36

-------
     The results listed for Complex II and Complex I are actually




results generated using the approximation formula outlined in Appendix C.




The intent was to estimate the magnitude of plume impingement concentrations,




not to estimate whether or not plume impingement would occur.  This latter




task was beyond the capabilities of any of the models employed in this




analysis.




     In looking at the differences in model estimates and observed concentra-




tions, note that the emissions data are rough engineering estimates for




all but the Navajo Generating Station.




     It would seem all the models had difficulty for the Phelps Mine and




Jones Ranch facilities.  Perhaps plume impingement at the monitor sites




never occurred or perhaps the emission rates were poorly characterized.




The results indicate that Valley and Valley-BID performed better than




Complex II and Complex I.  As an aside, the Complex II results appear much




better if all the estimates are divided by 2.0.  This suggests that the




overall dimensions of the plume may be well characterized by Complex II,




but that the mass distribution is not quite appropriate.  Complex II might




provide useful results if reflection from 10 m below the centerline in the




concentration algorithm were ignored.




WITHIN PLUME MASS DISTRIBUTION




      Indoubtedly, the most controversial feature of the. Valley model is the




characterization of the mass distribution employed when plume impingement




occurs.  For receptor locations on terrain that extends above the modeled




stable plume rise, the Valley model assumed that dispersion of material down-




ward  from the final effective plume height is inhibited.  This assumption seems




reasonable.  However, controversy arises because impingement concentrations






                                    37

-------
are modeled by Valley as if the plume were restricted from dispersing any




more than 10 m downward from the final effective plume height over the




entire downstream transport.  This effectively reduces the dispersion in




the vertical to a rate 1/2 of that it would have if the plume were allowed




to freely disperse downward (as it is allowed to disperse upward) during




downstream transport.  This causes concentrations to be two times higher




than they otherwise would be.




     It is important to realize that "effective" lateral dispersion, ex-




pressed as the ratio of centerline concentration of the sector-averaged




Valley Model plume to that of the Gaussian plume (see Appendix C) ranges




from just over 4.0 to almost 8.0 during F stability.  Hence, the Valley




model's maximum concentration estimates for receptors at plume elevation




are from two to four times lower than would be estimated using no re-




striction to vertical dispersion and using the Pasquill F curve for the




lateral dispersion.




EXTRAPOLATING TO LONGER AVERAGING TIMES




     The extrapolation of short-term peak concentrations to the longer




averaging times is at best a difficult process.  The basic assumption in




Valley is that in 6 h the entire plume has "swung across" a receptor; no




assumption is made about the horizontal distribution of pollutant, except




for mathematical convenience.  Table 9 lists the peak to mean 502 concen-




tration ratios observed during 1975 and 1976 at several monitoring networks




operated by American Electric Power (AEP):  Clifty Creek, Tanner Creek,




Muskingum, and Gavin-Kyger-Sporn (Mills et al., 1980).  These data would




suggest that the 1:4 ratio is perhaps on the low side but within the ob-




served range.






                                    38

-------
Table 9.  PEAK TO MEAN RATIOS DURING PERIODS OF MAXIMUM 24-H CONCENTRATIONS

Receptor
network
Clifty
Creek
Tanners
Creek
Muskingum
Gavin-
Kyger-Sporn
Year
1975
1976
1975
1976
1975
1976
1975
1976
Peak 1-h/24-h Concentration
during
Highest 24-h Second-highest
concentration 24-h concentration
5.73
2.78
1.69
2.39
. 2.28
5.58
3.25
2.10
2.44
3.76
1.62
3.70
5.09
3.67
3.86
2.05
                                     39

-------
                                  SECTION 6




                           QUESTIONS FOR REVIEWERS




    The following questions were asked of the reviewers  (Team B of the




Chicago Workshop) of the first draft.  Their replies to these questions




are discussed in Section 2. Conclusions and Recommendations.




1.  Based upon the summary information given in Figures 7 through 9, do you



    see any use for Complex II or Complex I as screening techniques?




2.  Does the point that none of these models can be used for estimating




    concentrations and frequency of occurrence at a specific point come




    across clearly?




3.  Do you agree or disagree with our recommendation on the labeling of the




    Valley-BID results as a screening model estimate valid if plume impinge-




    ment occurs?  NOTE:  this implies that Valley-BID is not sensitive enough




    to judge whether or not impingement will occur, only that if impingement




    occurs then, thus and so may occur.




4.  Does the analysis discussed in this draft include all the computations




    that you wanted in the sensitivity analysis?




5.  Are there questions about sensitivity that you believe can be answered




    by the computations made that are either not stated here or that




    insufficient information from the computations are presented so that




    the reader can infer the appropriate answer?




6.  Please give particular attention to review of the recommendations made  in




    Section  2.  On which point or points do you disagree?  Are there additional




    points that should be  included here?




                                     41

-------
                                 REFERENCES

Briggs, G. A.  1975:  Chapter 3—Plume rise predictions.  In:   Lectures on
     Air Pollution and Environmental Impact Analysis,  Duane A.  Haugen,  ed.
     Amer. Meteorol. Soc.  Boston, MA.  296 pp.

Burt, E. W.  1977:  Valley Model User's Guide.  EPA-450/2-77-018, U.  S.
     Environmental Protection Agency.  Research Triangle Park,  NC.  112 p.

Burt, E. W., and H. H. Slater.  1977:  Evaluation of the Valley Model.
     Preprints, Joint AMS/APCA Conference on Applications of Air Pollution
     Meteorology.  Amer. Meteorol. Soc.  Boston, MA.

Hewson, E. W., and G. C. Gill.  1944:  Meteorological  Investigations  in the
     Columbia River Valley, near Trail, B.C.  Part III of Report Submitted
     to the Trial Smelter Arbitral Tribunal (by R.S. Dean and R. E.
     Swain).U.S. Bureau of Mines, Bulletin 453.

Mills, M. T.  1979:  Data Base for the Evaluation of Short-Range Dispersion
     Models.  R-001-EPA-79.  Teknekron Research, Inc.   Waltham, MA.

Mills, M.T., R. Caiazza, D. D. Hergert, and D.A, Lynn.  1980:  Evaluation of
     Point Source Dispersion Models.  Volume II, Appendices (DRAFT REPORT)
     R-030-EPA-79.  Teknekron Research, Inc.  Waltham, MA.  808 pp.

Pierce, T. E., and D. B. Turner.  1980:  User's Guide for MPTER.
     EPA-600/8-80-016.  U.S. Environmental Protection Agency.  Research
     Triangle Park, NC  247 pp.

Turner, D. B.  1970:  Workbook of Atmospheric Dispersion Estimates,
     Office of Air Programs Publication No. AP-26.  U.S. Environmental
     Protection Agency.  Research Triangle Park, NC.  84 pp.

U. S. Environmental Protection Agency, 1980a:  Regional Workshop on Air
     Quality Modeling:  A Summary Report.  Office of Air Quality Planning
     and Standards USEPA.  March 1980 draft.  Research Triangle Park, NC.

U. S. Environmental Protection Agency, 1980b:  Guideline on Air Quality Models.
     Office of Air Quality Planning and Standards, USEPA.  OAQPS Guideline
     Series.  October 1980 Proposed Revisions.  Research Triangle Park, NC.
                                    43

-------
                                 APPENDIX A




               ADDITIONAL FIGURES OF MAXIMUM 24-H CONCENTRATIONS






     Figures A-1 and A-2 are similar to Figures 5 and 6 but are the result




of using the St. Louis meteorological data.




     Figures A-3 gives maximum 24-h concentrations resulting from Complex II




at 1 km for the medium and high sources using the Knoxville data.
                                     45

-------
 E                                                                                                                                  COMPLEX It
—                                                                                                                                1000
UJ  700— 483  494  739  336 254 667 381  479  554  747  807  727  996  69S 67B 990 156 403  245  244  379  646  573  404  586  B50 569 164 305 581  765(1083) 493  496  487  795
C/l                                                     f—v                                               .                        I
<  175 — 517  578  781  360 770 696 40?  S07  581  799  863  7SOjl046|  740 7171035 150 431  257  762  398  690  553  «M  621  908 608 816 372 615  81111341525  531  512  313

^  150 — 545  537  779  367 281 658 399  512  578  813  875  658\1016/  758 687 983 154 423  26?  273  398  714  573  447  604  933 616 834 331 675  B30J1094/ 542  555  579  375
CO
O
                                                                                                                                    COMPLEX I


     200 — 323  299  374  309 225 392 275 393  426  334  491  371  388  381 362 479 303 165  215  2M  2BO  768  777  358  288  355 334 333 736 393/521  545J 327  771  271  771

     175—346  318  390  372 241 410 285 411  447  357 p?A 382  405  405 377 (SoT) 315 168  223  220  294  284  237  375  304  379 357 3S5 752 4081543  4731340  790  781  281

     150 — 363  324  373  304 246 390 264 400  435  3651525/344  388  406 355 482 29? 160  776  224  794  745  238  368  297  391 364 361 758 J93\S33  5S4J341  30?  76?  76?
175—233  20?  181  145  146  195  135 233 734 197 267  166  ?7B  269  735  776  136  1??  138 137 IM 270  158  704  168  708  181

      I        I        I        I       I       I        I        I        I       I       I        I        I        I
     360       70       40      60      80     100      120      140      160      180     700     770      240      760


                                               AZIMUTH-FLOW VECTOR SECTOR (degrees)
                                                                                                                    176
                                                                                                                       171  731  319 302  236  187  152  138

                                                                                                                        I       t       I        I
                                                                                                                       ?BO      300     320      340
          Figure A-1.  Maximum 24-h concentrations resulting from  Complex II and Complex  I at  1  km for the low source
          (St. Louis, MO, 1976).  (Concentrations are pg/m3 divided by  10).
?
tO
<
m
^
o
<
i-
V)
LU
>
o
m
<
z
o
K
<
>
UJ
_j
UJ

200
175 —

150 —








ISO —


60

64

68









1
360
ISO

104 110 147 24 60 30

108 tit 145 22 61 31






©54 55 SS 52 70 43 22 31
54 55 55 52 70 43 22 3!


T 35 i i i i i i i i
20 40 60 80 100 120 140 160 180
COMPLEX
ISO
36 61 94 91 61 86 126 86 1 10 41 88 126 M54 1 71
38 53 101 96 64 91 135 92 116 43 93 134 1 162 76

40 54 105 100 67 91 140 94 122 44 95 138 U63/ >B




COMPI
0~~



1 1 1 1 1 1 1
700 220 240 260 280 300 320
s*
58

no
50



LEX

35
38
40

II
65
70

72
59



|

38
38
38
i

47
SO

5t
39





8
38
38
28
340
                                                    AZIMUTH-FLOW VECTOR SECTOR (degrees)
          Figure A-2.  Maximum  24-h concentrations resulting from Complex
          (St. Louis, MO, 1976).  (Concentrations are pg/m3 divided by 10).
                                                                                         and Complex  I at 5 km for the low source
                                                                                                                             MEDIUM SOURCE
 Ut  4SO— 675 897 388 761  935  598  568  717  621  604 502 442 455 830 253 499 1080  822  567 1032  697  642 1004 1137 1304 669 746 1471  479  596  6S3  787  662  723  994  631
 tO                                                                     /~N                         /	-i
 <  3SO— 9691270 56510901329  856  8221010  907  870 70S 63S 6361195 352 712n552.11SS  8161468  837  931 1446/1621 1852) 954 1067 2090  690  858  9351108  96310401475  909
 m                      ,	.                                           /  1500                   	/      1500                                   --*
 ^  275—11761377 6M 1253(1561} 998 1002 1125 1116 10W 825 772 7331453 412 8291887/1442  9991577  968 1137M761 1824 20«)l068 1196 2307  837 1042 1129 1239 1179 l2tWl714)ll«
 Ti                      ^—^                                           >—'                    V-	—	-^                                   M500
 H  200— 104 144 133 143 159  114  135  112  170  119 102  90  93 137  53  95 167  152  134  161  145  161  184 176  176 129 133 202  102  116  1?7  133  146  118  147  10?

 2
   450—1091 1509 1100 1507 1659 1038 1178 1179  1621 1064 986 806 884 1548 579 1040 1808  1652  1205 1669  1278 1476 1831 1880 1890 1262 1425 2206 1036 1297 1142 1348 1423 1262 1652 1106
 j        I        I       I        I        I        I        I        I       I        I        I        I        I        I       I       I        |        I
 Uj        360       20      40      60       80      100      120     140     160      180      200      220      240     260     280      300      320      340

                                                   AZIMUTH-FLOW VECTOR SECTOR (degrees)




         Figure A-3. Maximum 24-h concentrations  resulting from Complex  11 at  1  km  for the meduim and  high sources

         using  the Knoxville, TN, 1964 data.  (Concentrations are pg/m3 divided by 10).
                                                                    46

-------
                                APPENDIX B




                 HIGHEST AND SECOND HIGHEST CONCENTRATIONS




                                    AND




                CONCENTRATIONS FOR VARIOUS AVERAGING TIMES




     Tables B-1 through B-4 give the highest and second highest concentra-




tions estimated from the two models Complex II and Complex I for two aver-




aging times, 24-h and 3-h.  The azimuth of the receptor relative to the




source,the elevation above the source stack base, the Julian day of the




estimate,and the hour ending the period, if appropriate, are listed with




each estimated concentration.  Tables B-1 and B-2 are for Knoxville data;




Figures B-3 and B-4 are St. Louis data.  Tables B-1 and B-3 are for Complex  II;




Tables B-2 and B-4 are. for Complex I.




     Tables B-5 and B-6 give the five highest model estimates  for four




different averaging times.  Tables B-5 is for Knoxville data;  Table B-6




is for St. Louis data.




     Using data from Table B-5 (Knoxville), the second highest from both




models for each distance as a function of averaging time was graphed.




This is displayed in Figure B-1.




     The five highest concentrations using the Knoxville data  (again




from Table B-5) are displayed in Figure B-2 in rank order for  the four




averaging times for the two models, Complex II and Complex I.
                                    47

-------
TABLE B-1. HIGHEST AND SECOND-HIGHEST 3-H AND 24-H CONCENTRATIONS FROM
                    COMPLEX II (KNOXVILLE, TN, 1964).

Source
Low
24-hour
High
Second high
3-hour
High
Second high
Medium
24-hour
High
Second high
3-hour
High
Second high
High
24-hour
High
Second high
3-hour
High
Second high

Cone.


12157
9018

51053
45054


23067
18174

96704
86935


36929
30078

147174
132200

Azi.


270°
270°

330°
270°


270°
270°
_
330°
270°


270"
270°

330°
270°
1 km
Elev.


175m.
150m.

150m.
150m.


275m.
275m.

275m.
275m.


500m.
500m.


Day Hour


299
22

249 3
299 6


299
22

249 3
299 6


299
22

500m. 249 3
500m.
TABLE B-2. HIGHEST AND SECOND-
22 3

Cone.


1978
1400

8950
7758


4505
3235

20306
17633


8852
6561

40702
34907
HIGHEST 3-H AND 24-H
COMPLEX I (KNOXVILLE,
Source
Low
24-hour
High
Second high
3-hour
High
Second high
Medium
24-hour
High
Second high
3-hour
High
Second high
High
24-hour
High
Second high
3-hour
High
Second high

Cone.


5394
3932

15637
15563


10817
8202

32444
32341


19460
14999

58758
58613

Azi.


270°
250°

120°
320°


270°
270°

120°
320°


270°
270"

120°
320°
1 km
Elev.


175m
175m

150m
150m


275m
275m

275m
275m


500m
500m

500m
500m

Dav Hour


. 299
309

. 207 3
245 3


. 299
22

207 3
245 3


299
22

. 207 3
. 245 3

Azi.


270°
270°

330°
270°


270°
270°

330°
270°


270°
270°

330°
270°
5 km
Elev.


150m.
150m.

150m.
150m.


275m.
275m.

275m.
275m.


500m.
500m.

500m.
500m.

Dav


299
22

249
299


299
22

249
299


299
22

249
299
CONCENTRATIONS

Hour





3
6





3
6





3
6
FROM
TN, 1964).

Cone.


751
556

2170
2165


1712
1237

4953
4943


3406
2516

9972
9958

Azi.


270°
240°

110°
320°


270°
250°

110°
320°


270°
270°

110°
320°
5km
Elev.


150m.
150m.

150m.
150m.


275m.
275,n.

275m.
275m.


500m.
500m.

500m.
500m.

Day


299
350

207
245


299
309

207
245


299
22

207
245

Hour





3
3





3
3





3
3
                               48

-------
TABLE B-3. HIGHEST AND SECOND-HIGHEST 3-H AND 24-H CONCENTRATIONS FROM
                     COMPLEX II (ST. LOUIS, MO, 1976).

Source
Low
24-hour
High
Second high
3-hour
High
Second high
Medium
24-hour
High
Second high
3-hour
High
Second high
High
24-hour
High
Second high
3-hour
High
Second high

Cone.


11343
11163

51218
43741


22476
22138

96691
83855


37693
36559

145823
128851

Azi.


310°
310°

150°
100°


310°
310°

150°
100°


310°
310°

270°
100°
1 km
Elev.


175m.
175m.

150m.
150m.


275m.
275m.

275m.
275m.


500m.
500m.

500m.
500m.

Day Hour


238
320

305 24
302 3


320
238

305 24
302 3


320
238

262 3
302 3
TABLE 3-4. HIGHEST AND SECOND-HIGHEST 3-H
COMPLEX

Source
Low
24-hour
High
Second high
3-hour
High
Second high
Medium
24-hour
High
Second high
3-hour
High
Second high
High
24-hour
High
Second high
3-hour
High
Second high

Cone.


5726
5560

15665
15612


11689
11438

32482
32410


21242
20672

58799
58712

Azi.


310°
310°

360°
360°


310°
310°

360°
360°


310°
310°

360°
360°
1 km
Elev.


175m.
175m.

150m.
150m.


27Sm.
275m.

275m.
275m.


500m.
500m.

500m.
500m.
I (ST. LOUIS,

Day Hour


320
232

235 24
236 3


320
232

235 24
236 3


320
232

235 24
236 3

Cone.


1627
1541

9452
7295


3772
3549

21442
16563


7587
7161

42224
32910
AND 24-H
MO, 1976)

Cone.


801
775

2172
2168


1831
1775

4955
4949


3637
3536

9975
9967

Azi.


310°
310°

150°
100°


310°
310°

150°
100°


310°
310°

150°
100°
5km
Elev.


150m.
150m.

150m.
150m.


275m.
275m.

275m.
275m.


500m.
500m.

500m.
500m.

Day


238
320

305
302


238
320

305
302


238
320

305
302
CONCENTRATIONS


Azi.


310°
310°

360"
360°


310°
310°

360°
360°


310°
310°

360°
360°

5 km
Elev.


150m.
150m.

150m.
150m.


275m.
275m.

275m.
275m.


500m.
500m.

500m.
500m.


Day


320
232

235
236


320
232

235
236


320
232

235
236

Hour





24
3





24
3





24
3
FROM


Hour





24
3





24
3





24
3

-------
TABLE B-5. FIVE HIGHEST MODEL ESTIMATES FOR 1-, 3-, 8-,
     AND 24-H AVERAGING TIMES (MEDIUM SOURCE;
                  KNOXVILLE, TN, 1964).
Averaging
time
(H)
1-H
Highest
Highest 2"d high
Highest 3rd high
Highest 4th high
Highest 5th high
3-H




8-H




24-H




/(b)
(c>/jb)

Complex II
1-km

109044.8
(a) 108163.8
107647.3
107647.3
106309.6
96704.0
86935.4
76877.1
69812.1
58769.4
69133.3
49693.0
39627.6
34317.6
31050.5
(c) 23066.8
(b) 18174.3
14005.0
12486.0
11439.2
5.95
1.27

Complex I
1-km

37717.7
37638.0
37563.3
37560.2
37549.3
32444.2
32341.0
32120.9
32016.4
31677.9
29447.4
22761.4
22279.2
20061.8
17997.1
10816.7
8201.7
8157.8
7656.7
6979.2
4.59
1.32

Complex II
5-km

26072.8
25990.9
25938.4
25938.4
25808.8
20305.9
17633.2
14281.1
12807.7
12236.4
13419.4
8920.0
6186.6
6175.1
6007.6
4505.4
3234.7
2308.2
2259.0
2007.1
8.04
1.39

Complex I
5-km

5406.9
5406.6
5406.2
5405.7
5405.3
4952.6
4943.4
4923.6
4915.1
4884.3
4614.8
3483.1
3386.6
3077.7
2666.2
1711.9
1237.4
1231.5
1176.1
1093.8
4.37
1.38
             (a) Second highest 1-H
             (b) Second highest 24-H
             (c) Highest 24-H
                          50

-------
TABLE B-6.  FIVE HIGHEST MODEL ESTIMATES FOR 1-, 3-, 8-,
      AND 24-H AVERAGING TIMES (MEDIUM SOURCE;
                  ST. LOUIS, MO, 1976).
Averaging
time
(H)
1-H
Highest
Highest 2nd high
Highest 3rd high
Highest 4th high
Highest 5th high
3-H




8-H




24-H




(a)/(b.
/(b)

Complex II
1-km

109189.6
(a) 107851.2
107145.6
106852.9
100405.2
96690.5
83855.0
77653.3
66292.6
65192.6
61749.3
60951.1
43777.2
43707.9
41404.2
(c) 22475.9
(b> 22137.5
16052.3
15519.6
14592.4
4.87
1.02

Complex 1
1-km

37852.9
37661.3
37606.6
37597.1
37565.4
32482.3
32409.6
32308.3
32259.7
31760.0
30459.5
26259.1
24060.8
20882.4
20713.6
11688.7
11437.8
10637.4
8112.0
7987.3
3.29
1.02

Complex II
5-km

26084.9
25958.0
25890.0
25861.6
21845.0
21442.4
16563.0
13485.0
11617.2
10889.8
11103.0
9750.1
6845.9
5860.9
5042.4
3771.7
3549.3
2525.6
2275.9
1778.8
7.31
1.06

Complex I
5-km

5406.1
5405.5
5404.5
5404.2
5404.2
4955.5
4949.0
4940.1
4935.7
4935.2
4795.3
3969.6
3733.9
3173.0
3085.6
1831.3
1774.6
1645.8
1254.3
1205.1
3.05
1.03
              (a) Second highest 1-H
              (b) Second highest 24-H
              (c) Highest 24-H
                          51

-------
 10'
      HIGHEST SECOND HIGH. MEDIUM SO

         Ol km, COMPLEX II

         • I km,COMPLEX I

         D S km. COMPLEX II
         • 5 km, COMPLEX I

     	I	
                                     AVERAGING TIME, hi
Figure B-1. Highest second-high concentrations as a function of averaging time from Knoxville.
TN, 1964 data.
  ID*
5,10*
Sxl O3
     MEDIUM SOURCE. 1 km
 COMPLEX II    COMPLEX!
u. O t-hf        • 1-hr
    I 3 h-        • 3 hr
    , Bht        A Bhf
    > 24-hr       • 24-hr
 Figure B-2. Five highest estimated conceniraiions in rank order for 1 •, 3-, 8-. and 24-hr averaging
 times from Complex II and Complex I  using Knoxville, TN, 1964 data.
                                           52

-------
TABLE B-6. FIVE HIGHEST MODEL ESTIMATES FOR 1-, 3-, 8-,
      AND 24-H AVERAGING TIMES (MEDIUM SOURCE;
                  ST. LOUIS, MO, 1976).
Averaging
time
(H)
1-H
Highest
Highest 2nd high
Highest 3rd high
Highest 4th high
Highest 5th high
3-H




8-H




24-H




/(b)

Complex II
1-km

109189.6
(a) 107851.2
107145.6
106852.9
100405.2
96690.5
83855.0
77658.3
66292.6
65192.6
61749.3
60951.1
43777.2
43707.9
41404.2
(c) 22475.9
 22137.5
16052.3
15519.6
14592.4
4.87
1.02

Complex I
1-km

37852.9
37661.3
37606.6
37597.1
37565.4
32482.3
32409.6
32308.3
32259.7
31760.0
30459.5
26259.1
24060.8
20882.4
20713.6
11688.7
11437.8
10637.4
8112.0
7987.3
3.29
1.02

Complex II
5-km

26084.9
25958.0
25890.0
25861.6
21845.0
21442.4
16563.0
13485.0
11617.2
10889.8
11103.0
9750.1
6845.9
5860.9
5042.4
3771.7
3549.3
2525.6
2275.9
1778.8
7.31
1.06

Complex I
5-km

5406.1
5405.5
5404.5
5404.2
5404.2
4955.5
4949.0
4940.1
4935.7
4935.2
4795.3
3969.6
3733.9
3173.0
3085.6
1831.3
1774.6
1645.8
1254.3
1205.1
3.05
1.03
               Second highest 1-H
              (b) Second highest 24-H
               Highest 24-H
                          51

-------
      HIGHEST SECOND-HIGH. MEDIUM SOURCE

         Ol'in. COMPLEX II

         • 1 km.COMPLEX I

         O S km. COMPLEX II
         • Sim.COMPLEX I

     	I	I
                                     AVERAGINGTIME.br
Figure B-l. Highest second-high concentrations as a function of averaging time from KnoxvilEe,
TN. 1964 data.
  to*
  in*
 s.u'
    MEDIUM SOURCE. 1 km
 COMPLEX II     COMPLEX I

. O 1-hr        • 1-hr
  D 3-hr        • 1-lir
  A Ik'        A I-hr
 Figure B-2, Five highest estimated concentrations in rank order for 1-. 3-, 8-, and 24-hr averaging
 times from Complex II and Complex I using Knoxville, TIN,  1964 data.
                                           52

-------
                                 APPENDIX C




                    MODEL ESTIMATES BY HAND CALCULATIONS




     The results from the Valley and Valley-BID models, when executed in




their screening mode, for estimating worst-case 24-h concentrations can be




duplicated rather simply using a calculator.  The procedure for duplicating




the Valley and Valley-BID results can also be made to approximate the




results from Complex II and Complex I, generated in this sensitivity




analysis.   The computations of plume rise, vertical dispersion, and




lateral dispersion need be computed only once per downwind distance.  Then




equation (C-1) can be used to approximate the various models.  Equation




(C-1) will tend to overestimate the maximum concentrations for 3-, 8-, and




24-h averaging times from Complex II and Complex I as downwind distance




increases beyond 5 km.  However, it would not be that difficult to further




investigate the variation of R as a function of downwind distance and




thereby develop a simple tool for screening analyses in complex terrain




situations.




     The following discussion outlines the  four-step procedure.  The  first




step is to determine the wind speed which is a constant for Valley and




Valley-BID and is a  function of stack height for C^-n^lex II and Complex  I.




The second step is to compute the stable plume rise.  The third step  is  to




compute the vertical and lateral dispersion for the downwind distance of




interest, and the fourth step is to substitute the results into equation




(C-1).  The procedures  lend themselves to programmable pocket  calculators.






                                    53

-------
     The factor R has been chosen to slightly overestimate the sequential



model results for the 1- to 5-km downwind distance range.  Generally, the



results of Equation (C-1) are within 15% of actual modeled results for



Complex II and Complex I.  There is no error as far as Valley and Valley-BID



is concerned, except for those minor differences introduced by the characteriza-



tion used for the vertical dispersion during Pasquill F stability category.



This particular characterization of °z is from Vogtx^ ( /?"77/ t



     If the receptor elevation is above the undisturbed height of the plume,



the concentration resulting from equation (C-1) should be multiplied by



(401-ZR)/400, where ZR is the height in meters of the receptor elevation above



the undisturbed plume.  For instance, say we had a stack height of 100 m, a



plume rise of 150 m, and a receptor elevation above stack base of 275 m.  In



such a case, the receptor would be 25 m above the undisturbed final plume height



and ZR would equal 25 m and H in equation (C-1) would equal 10 m.





     1.  Compute wind speed at stack top during 1-h period of assumed



         worst-case stable conditions.



         A.  Valley or Valley-BID



               u = 2.5 m/s



         B.  Complex II or Complex I



               u = (hs/10)°-55




               where hs = stack height  (m)



     2.  Compute stable plume rise




                                 Vsd2T             1/3

                        5.007     S y    (Tg - T)
                                  U  Ig



          Ah  =     -^                  or
                                     54

-------
   Vs = stack gas exit velocity (m/s)
   d  = stack exit diameter (m)
   T  = ambient air temperature, ~293 K
   Ts = stack gas exit temperature (K)
   u  = assumed wind speed at stack top (m/s)
   Vf = stack gas volume flow (m-Vs), equal to
          U/4))Vsd2
3.  Compute vertical arid lateral dispersion
   A.  Valley
         °z0  = OoVc)  exP (~3-8 + 1'419 lnX-0.055 In2 X)
                 L. • I ,?
         <>y0  = (/27/16)X

           X  = downwind distance (m)
   B.  Valley-BID or Complex I
         °z   = [°ZQ2 + (Ah/3.5)2 ]1/2
         where az   is as given  for Valley
         °y0  = as given for Valley
         Ah   = stable plume rise (m)
   C.  Complex II
         az   = as  given for Valley-BID
         °y   = [ay02 + ^h/3.
         a   = (-0.0029 In! 0.054)X
                                55

-------
     4.   Compute concentration estimate



           Y _   R Q 1D6          '
           \ _ -
           X r  maximum concentration for averaging time (

           H =  minimum separation between plume centerline and receptor,
                no less than 10 m.

           Q =  emission rate (g/s)

           R =  ratio of maximum concentration for averaging time of
                interest to 1-h peak.
                                              Averaging Time (h)

Valley or
Valley-BID
Complex II
Iopt(25)=1
Complex I
Iopt(25)=1
3
Not
applicable
0.90
, 0.95 :
8
Not
applicable
0.60
0.85
24
0.25
0.20
0.35
REFERENCES — APPENDIX C

Vogt, Kurt-Jurgen, 1977:  Empirical investigations of the diffusion of waste
  air plumes in the atmosphere, Nuclear Technology 34, 43-57.
                                    56

-------
                                   TECHNICAL REPORT DATA
                           (Please read Instructions on the reverse before completing)
1. REPORT NO.
                             2.
                                                           3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
 AN ANALYSIS OF  COMPLEX I AND COMPLEX  II
 CANDIDATE SCREENING MODELS
                                                           5. REPORT DATE
             6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
 John S. Irwin  and D.  Bruce Turner
                                                           8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
             10. PROGRAM ELEMENT NO.
             CDTA1D/04-1315 (FY-82)
 (Same as Block 12)
                                                           11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
                                                           13. TYPE OF REPORT AND PERIOD COVERED
 Environmental  Sciences Research Laboratory — RTP, NC
 Office of  Research and Development
 U.S. Environmental Protection Agency
 Rpsearrh Triangle Park. MH 27711	
             14. SPONSORING AGENCY CODE
             EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT

      This  study,  suggested by an  EPA  Regional  Workshop in  February 1980, was
 conducted  as  a  simple analysis to  investigate whether or not  a  sequential air
 quality  simulation model, capable  of  accepting onsite hourly  meteorological
 data, could  be  recommended as a screening model for estimating  worst-case .    .
 pollutant  impacts on complex terrain.   The study intercompared  the highest
 24-h average  pollutant concentration  values obtained using "our algorithmic
 air quality  simulation models:  Complex I, Complex II, Valley,  and Valley-BID.

      The models were applied and  their results compared for a year's meteorological
 data for two  different sites.  Various combinations of source release height.and.
 terrain  configurations were examined.                                     ..      .  '

      The authors  conclude that the Valley-BID (or pencil and  paper calculations'using
 the same assumptions) are most appropriate for screening analyses for maximum 24-h
 concentrations  resulting from plume impaction on terrain near the height of an
 elevated stabilised plume.
17.
                                KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS  C. COSATI Field/Group
13. DISTRIBUTION STATEMENT
  RELEASE TO PUBLIC
                                              19. SECURITY CLASS (TliisReport)
                                               UNCLASSIFIED
                                                                         21. NO. OF PAGES
20. SECURITY CLASS (Thispage)
 UNCLASSIFIED
                                                                         22. PRICE
EPA Form 2220-1 (9-73)

-------
                                                                        INSTRUCTIONS

                1.    REPORT NUMBER
                     Insert the EPA report number as it appears en the cover of the publication.

                2.    LEAVE BLANK                                                -"  '  .      '!    .      ..   "      '.''    '•            • -''••
                                                                                                              '•   •'•         -    '         '     .      I
                3.    RECIPIENTS ACCESSION NUMBER
                     Reserved for use by each report recipient.

 .  ..            4.    TITLE AND SUBTITLE
;•'- ••';-.'•.              Title should indicate clearly and briefly the subject coverage of the report, and be displayed prominently.  Set subtitle, if used. in.smaller  .
 • .'•'•' •""•              type or otherwise subordinate it to main title. When a report is prepared in more than one  volume, repeat the primary title, add volume'
                     number and include subtule for :iie specific title.

                5.    REPORT DATE           .        '
                     Each report shall carry a date indic?*:ng at least month and year. Indicate the basis on which it was selected (e.g., date of issue, da:e of
gy&gpjtf              approval, date of preparation, etc.}.                                                                    .  .,    •   •          ;..,'•'.
•*:--i*M              Give name(s) ii. conventional  order (John R. Doe, ]. Robert Doe. etc.}. . List author's affiliation if it differs from the performing organi-
'" ' _.";;i              zation.                                        '  '                           .'","',           '

    .-  .          8.    PERFpRMING ORGANIZATION REPORT NUMBER
                     Insevt if performing organization w.shes to assign this nuniber.

;. '..•££?$         9-    S't'RI'CnMING ORGANIZATION NAME AND ADDRESS
.  •'"""''".              Give name, street, city, state,  and ZIP code. List no more than two levels of an organizational hirearchy.

. .*4S-;.ey         10.   PROGRAM ELEMENT fMJJI/GFIR
                     Use the program element nu:-iber under which the report was prepared. Subordinate numbers may be included in parentheses.

 :.';:'.  9         '>i   CCNTRACT/GRANT NUMBER
 '.''•;•';•..              Insert contract or grant numbei umier which report was prepared.      ..    ..  .,.-,...	  .            o    .     ,      .  , .-

-Wiiiiij         12.   :">PONSOftlNG AGENCY.NAME ANI.,  ADDRESS, •;,;-;-,}•;.,   i } .-,  , > •; ,, - ,\-| -   -) ;  ;• f ,• \- [ /; -j ;.  •-.,"•-.;:-'   ,-.  ," .   'i ''••;;'•• ,'i i".
 •  •'-.'.?.              Include 7'P code.  ••  -^--••.••••     •     -•   •'   ..•^•.~><    -,-',.    -..   ^.   .-.^  '•  "  ••    '/•  ^   '"  ;    _ ,.:..'  ,•.:.,-'
                13. TYPE 0>F RcPORT AND FER!OD COVERED    .   ,  v-
                    Indicate i.nterim final, etc , and if applicable, dates covsr.ec..
                                     .            .
                14. SfONSO?7iiMG.AGEWCY. CODE	   ,
                    Leave blank.-  •   •• '  '''   •  '  '•'•''-'
                                                                                    - ••• -l.
                                           •'ir'.li  v:i'"(!' ..;.'i."j  J >'.'..-J/   '• ">''" '•>"!'. :!

•!B. SU°PI r-^-EWTARY MOTSS  •'•-•'••     - - •    -      	..,_..
    .E.'itc/ information not included elsew'-iTe but useful, such as:  Prepared i:: cooperatio.•. «-ith, Translation cf. •'••.•sentftd :t conference of,
    , To be p'jblis!i.:.d in, Supersedes, Suppittrients, etc. • ;,,  !0._:   '; ;.  '  ;:".'/,  ":;•'.•     •  ''.   .  ':  i  • .  •    '       .i   . •').    :.l
                     Include a briti (200 words or less) factual sumaiary ( f the most significant information contained in Ihe r«por>\  If the: i'eport contains a,  ."•'
                     significant bibliography ijr lite'ature jiuvfcy, mention it here.

                17.. :KF.Y WORDS AND DOCUMENT ANALYSIS  ';;--  -; -    ' ?.'!.-••- .f r .•;...:..'.;.   jr,;.'  .•••••;•    ~    "   j:' 'i;  •
                    '(-0 DLSCRIKV'JRS • Seiect from th-j 7;ic, ;jrus of Enginee-: -.7, anil .Sczes-vCc  Terms the proper amheriz. c" i-irms that identify the major
                     K.-r-r.r^ rf tlic rrssarch and are suffi oi pl-.y;'- ; -  >;-, r-?i, P •;; i i.''.:«?'os(:=  -vii! l-e c-r-«-rrfere?i«c  .:*Ii :r^-iv.v.ar;- i-'is!'"70"iv.-;. assignments »t,a'. v.'ill fr.llciv
                16.  L-WiRIBUTIONSTATEMENV
                     Denote releasability to the public or iimita&.;! for reasons c\her than security for <;x?.rip!e "Release Unlimited."  Cite any availability to
                     the public, with addrsssanc price. ,'

                19. & 20.  SECURITY CLASSIFICATION
                     DO NOT submit classified reports to the Mt'Jonai Teciinicai Information service.

                21>  NUMBER OF PAGES
                     insert the total number of pages, including this one and unnumbered pages, but exclude distribution li«t, if any.

                22  PRICE
                     Insert the price set by the National Technical Information Service or the Government Printing Office, ii" known.
                     ?A Form 2220-1 (S-V3) (Reverie)

-------