EPA-600/3-83-101
                                           November 1983
     EPA COMPLEX TERRAIN MODEL DEVELOPMENT:
         Third Milestone Report - 1983
                       by
                Thomas F.  Lavery
              David G. Strimaitis
                Akula Venkatram
               Benjamin R.  Greene
             Donald C.  DiCristofaro
                 Bruce A. Egan
   ENVIRONMENTAL RESEARCH & TECHNOLOGY, INC.
696 Virginia Road,  Concord,  Massachusetts   01742
            Contract No. 68-02-3421
                Project Officer

             Francis A. Schiermeier
      Meteorology and Assessment Division
   Environmental Sciences Research Laboratory
 Research Triangle Park, North Carolina  22771
   ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
       OFFICE  OF RESEARCH AND DEVELOPMENT
      U.S. ENVIRONMENTAL PROTECTION AGENCY
 RESEARCH TRIANGLE PARK, NORTH CAROLINA  22771

-------
                      NOTICE

This document has been reviewed in accordance with
U.S. Environmental Protection Agency policy and
approved for publication.  Mention of trade names
or commercial products does not constitute endorse-
ment or recommendation for use.
                       11

-------
                                  FOREWORD
     The Environmental Sciences Research Laboratory (ESRL) conducts
 intramural and extramural research programs in the physical sciences to
 detect, define, and quantify air pollution and its effects on urban,  region-
 al,  and global atmospheres and the subsequent impact on water quality and
 land use.  The Laboratory is responsible for planning,  implementing,  and
 managing research and development programs designed to quantify the relation-
 ships between emissions of pollutants from all types o'f sources with air
 quality and atmospheric effects, and to uncover and characterize hitherto
 unidentified air pollution problems.  Information from ESRL programs and
 from the programs of other government agencies, private industry, and the
 academic community, are integrated by the Laboratory to  develop the technical
 basis for air pollution control strategies for various  pollutants.

     The Complex Terrain Model Development (CTMD) program is designed to
 develop reliable atmospheric dispersion models that are applicable to large
 pollutant sources located in complex terrain.   The first two major field
 studies of this five-year program were conducted during 1980 at Cinder Cone
 Butte near Boise, Idaho, and during 1982 at Hogback Ridge near Farmington,
 New Mexico.  Data from these field studies along with measurements of scaled
 physical simulations performed in the EPA Fluid Modeling Facility are being
 used to quantify the effects of terrain obstacles on stable plume disper-
 sion.  The First Milestone Report presented the performance evaluations  of
 four existing complex terrain dispersion models and described the initial
 development of two proposed models.  The Second Milestone Report  documented
 subsequent improvements to the two new models  and presented detailed  case
 studies to illustrate and contrast various flow situations and ranges of
 model performance.  This Third Milestone Report describes the evolution  of
 the two new models into the Complex Terrain Dispersion  Model (CTDM),  and
 contrasts the performance of CTDM and three existing complex terrain  models
using measurement data from both field studies.
                                 A.H.  Ellison
                                 Director
                                 Environmental  Sciences  Research Laboratory
                                    ill

-------
                               ABSTRACT
     The U.S. Environmental Protection Agency is sponsoring the
Complex Terrain Model Development (CTMD) program, a multi-year,
integrated program to develop, evaluate, and refine practical plume
models for calculating ground-level air pollutant concentrations that
result from large emission sources located in mountainous terrain.
The principal objective of the CTMD program is to develop models with
known accuracy and limitations for simulating 1-hour average
concentrations during stable atmospheric conditions, i.e., to handle
the stable plume impaction problem.   This Third Milestone Report
describes CTDM activities performed from June 1982 through May 1983.

     A Modelers Data Archive (MDA) for Small Hill Impaction Study #1
(SHIS #1) conducted at Cinder Cone Butte, Idaho in 1980 is described.
This archive contains hourly average tracer gas concentrations,  and
hourly and 5-minute average meteorological data estimated for the
height of release on the basis of meteorological data measured at the
150-m tall tower approximately 2 km north of the hill.  A total of  153
tracer experiment hours are contained in the archive.

     The Wrap and Lift modeling components described in the CTMD
Second Milestone Report are modified and combined into the Complex
Terrain Dispersion Model (CTDM).  Performance statistics for two
versions of CTDM are generated with  the 153-hour MDA,  and these are
compared with performance statistics for COMPLEX I, COMPLEX II,  and
Valley for the same data.   Sensitivity of CTDM performance to the
input data is partially evaluated by modeling a total of 14 hours from
two experiments with a revised MDA which reflects the best estimates
of release-height meteorology determined after studying all available
information from those two experiments.  Model performance sensitivity
of CTDM to variations in the meteorology within a 1-hour averaging
period is similarly evaluated by modeling concentrations for 5-minute
intervals with both versions of the  MDA for the 14 hours.  Modifica-
tions to CTDM that are suggested by  the performance of CTDM with the
153-hour MDA as well as with the 14-hour modified MDA are described,
implemented and tested.

     An overview of Small Hill Impaction Study #2 (SHIS #2),  conducted
during October 1982 at the Hogback Ridge near Farmington,  N.M.,  is
presented.  It includes a description of the climatology of the area,
a description of the experiment design, and a discussion of the
phenomena observed, illustrated by a descriptive account of two hours
from the study.
                                  IV

-------
     A preliminary meteorological data set for 17 hours (34 tracer
experiment hours) is presented.   Observed concentrations,  scaled by
the emission rate, are sorted by several meteorological parameters for
these 17 hours, and the resulting patterns are discussed and compared
with the results from SHIS #1.  A comparison of Valley,  Complex I,
Complex II, and CTDM performance statistics for these  17 hours is  also
presented and discussed.

     An EPA Fluid Modeling Facility report describing  a laboratory
simulation of neutral flow and dispersion over Cinder  Cone Butte is
included as an Appendix.  A second Appendix contains a discussion  of
an initial analysis of SHIS #2 acoustic  sounder data and sonic
anemometer data.

     This report was submitted in partial fulfillment  of contract
68-02-3421 by Environmental Research & Technology,  Inc.  under the
sponsorship of the U.S. Environmental Protection Agency.   This report
covers the period June 2,  1982 to June 1,  1983.
                                  v

-------

-------
                               CONTENTS
Foreword	.....:..i..
Abstract	
Figures	
T^les	.- .-	     xiv
Symbols and Abbreviations	     xvi
Acknowledgements	      xx
1.   Introduction		     1
2.   Evaluation of Complex Terrain Models with the
     CCB Data Base	     5
     2.1  SHIS #1 Modelers' Data Archive	     5
     2.2  Complex Terrain Models Evaluated 	    10
     2.3  Model Performance Statistics 	    14
     2.4  Analysis of Model Performance	    16
3.   Development of the Complex Terrain Dispersion
     Model	    36
     3.1  Introduction	    36
     3.2  Description of CTDM(11083)  	    37
     3.3  Case Study Results - Experiments 201 and 210 ....    45
     3.4  CTDM Upgrades:  CTDM (14083)	    93
4.   The Hogback Ridge Experiment  	   106
     4.1  Geographic and Meteorological Setting  	   106
     4.2  Preliminary Flow Visualization Experiment  	   112
     4.3  Fluid Modeling in Support of  the SHIS #2
          Experimental Design  	   117
     4.4  Experimental Design  	   123
     4.5  Preliminary Field Study Results  	   140
5.   Preliminary Modeling of the HER Data Base	184
     5.1  Initial 17-Hour Data Base	184
     5.2  Complex Terrain Models Evaluated .  .  .	193
     5.3  Overall Performance Statistics 	   193
6.   Summary,  Conclusions, and Recommendations for Further
     Study	^  _   203
     6.1  Principal Accomplishments and Conclusions  	   203
     6.2  Recommendations for Further Study  ...  	   203
References   	209
Appendix A:   Laboratory Simulation  of Neutral Plume Dispersion
             over Cinder Cone Butte	212
Appendix B:   Supporting Meteorological  Measurements - Acoustic
             Sounder and Sonic Anemometers  	   252
                                 vii

-------
                                FIGURES
 Number
  3


  4


  5
10
II
12
13
14
 Performance statistics based on residuals  of peak
   observed and peak modeled concentrations for  Valley,
   COMPLEX I,  COMPLEX II,  CTDM(11083),  and  CTDM
   (11083-E) applied to SHIS #1 data  	

 Scatterplots  of observed  and modeled peak
   concentrations for 153  hours of  SHIS #1  data  ....

 Cumulative frequency plot of Co/Cp for Valley
   applied to  SHIS #1 data  ....  	
 Cumulative frequency plot  of  CQ/C   for
   COMPLEX I applied  to  SHIS #1 data   .
                                                                 19
                                                                 21
 23
 24
 Cumulative frequency plot  of  C0/Cn  for
                              "   i *
          COMPLEX II applied to SHIS #1 data	   25
         Cumulative frequency plot of CQ/Cp for CTDM
          (11083) applied to SHIS #1 data  	
        Cumulative frequency plot of CQ/C  for CTDM
          (11083-E) applied to SHIS #1 data  .  .  .  .
        Variation of observed-to-modeled ratios of SHIS #1
          peak hourly tracer concentrations with peak
          modeled concentrations 	  „ .
Variation of observed-to-modeled ratios of SHIS #1
  maximum hourly tracer concentrations with wind speed.

Variation of observed-to-modeled ratios of SHIS #1
  maximum hourly tracer concentrations with u/N  . .  .

Variation of observed-to-modeled ratios of SHIS #1
  maximum hourly tracer concentrations with l-Hc/zr

Variation of observed-to-modeled ratios of SHIS #1
  maximum hourly tracer concentrations with l-H_/z-
                                               L.  i
Variation of observed-to-modeled ratios  of SHIS  #1
  peak hourly tracer concentrations with the product
                                       LZ  ....
          of the turbulence intensities  iy and  i.
Plume dispersion in the region of horizontal  flow
 26


 27



 28


 29


 31


 32


 33



34

38
                               viii

-------
                          FIGURES (Continued)
Number
15
16
17
18
19
20
21
22
23
24
Geometry used to derive the effective radius of
  curvature Re for the portion of the hill above
  Hc, and to incorporate the wind direction
  probability distribution function P (0 ) in the
  Lift model . . .	.,	
Dispersion and flow regions for stratified flow
  around hills 	
Intercomparison of wind direction time-series
  data for Experiment 201, derived from photographs,
  wind measurements at 10 m from Tower B,  and
  interpolated wind measurements from Tower A  .  .  ..

Intercomparison of wind direction time-series
  data for Experiment 201, derived from photographs,
  wind measurements at 10 m and 30 m from Tower B,
  and interpolated wind measurements from Tower A
Intercomparison of vertical turbulence  intensity
  time-series data for Experiment 210,  derived  from
  photographs and interpolated  from Tower A   .  .  .  .
                                                                 41
                                                                 43
                                                                 47
                                                                 48
Definition sketch for computing distance along
  the wind trajectory from the .source	    50

Intercomparison of vertical turbulence intensity
  time-series data for Experiment 201, derived ,from
  photographs by Methods 1 and 3, and interpolated
  from Tower A	    53

Intercomparison of vertical turbulence intensity
  time-series data for two hours of Experiment 201,
  derived from photographs by Methods 1 and 3, and
  interpolated from Tower A.  The "photo" estimates
  are shifted by - 10 minutes	    54

Intercomparison of vertical turbulence intensity
  time-series data for Experiment 201, derived from
  photographs by Method 4, and interpolated from
  Tower A	    55

Average distribution of lidar-inferred oil-fog
  droplet concentrations 800 m downwind of the
  source for Experiment 210, hour 3	    58
                                                                60
                                  ix

-------
                          FIGURES (Continued)
 Number

 25



 26


 27


 28


 29


 30




 31




 32



 33



 34




 35




36
 Comparison of  Pasquill-Gifford az growth
   curves  with  az data from Experiment 201
   photographs   	
                               63
 Comparison of az growth curves with
   data derived from Experiment 201 photographs ....    64

 Comparison of az growth curves with az2
   data derived from Experiment 201 photographs ....    65
 Scatterplots used in computing best-fit p for
  Experiment 201 	
                               66
 Growth of a z with distance from the source
   for Experiment 201, hour 3 	
                               68
Comparison of vertical turbulence intensity data
   contained in the Modelers' Data Archive with estimates
  made by fitting az to the az data estimated
  from the photographs from Experiment 210
                               73
Comparison of vertical turbulence intensity data
  contained in the Modelers'  Data Archive with estimates
  made by fitting oz to the observed aZ2
  data for Experiment 201	   74
Comparison of photo-derived a,
  calculated a, utilizing best-fit  i
     data and
            data
  in Equation 41 with
0.525 and a__ =0.5
Time-series plots of crw and
  iz data from
  Towers B, C, D, E,  and F on CCB compared with data
  from Tower A for Experiment 201  	
Time-series plots of aw and iz data  from
  Tower B, C,  D,  E,  and F on CCB compared with data
  from Tower A interpolated to release  height for
  Experiment 210  	
CTDM(11083-E) performance,  as measured by peak
  concentrations unpaired in space, using the
  Modelers'  Data Archive and the modified Modelers'
  Data Archive for Experiments 201 and 210  . . . . ,
Illustration of the plume deformation expected
  over an isolated "3-D"  hill and the deformation
  by CTDM(11083) 	
76
                               77
                              82
                              92
                                                                96

-------
FIGURES (Continued)
Number
37
38
39
40
AT
42
43
44(a)
44(b)
45
46
47
48
49
50
51
52
53
54
55
56

Illustration of the geometry used to calculate the
horizontal distribution of plume material above H,,.
CTDM(11083-E) and CTDM(14083) Co(peak)/Cp(peak)
residuals for Experiments 201 and 210 with the
Region around the SHIS #2 at the Hogback Ridge . . . .
The SHIS #2 site 	 	

View of the Hogback experiment site from the east . .
View toward the south— east from the Hogback . . . • .
Climatological annual wind rose for Farmington, NM . .
October-December stability E Farmington wind rose . .
October hourly wind roses for four nighttime hours:
2000 0000 0200 and 0400 MST . . 	

Upper air soundings at Four Corners Power Plant
during turnaround meteorological conditions ....
FMF tow-tank flow visualization experiment -

PNM air quality and meteorological monitoring sites
SHIS #2 tracer sampler locations and optical


Raw tower AT from 10 m to 150 m ...........
Hourly CF3Br concentrations 10/12/82,
0500-0600 MDT ...... 	 ...
Pag.
97
104
107
108
109
110
111
113
113
114
115
116
122
125
130
132
134
138
139
144
148

-------
                  FIGURES (Continued)
 Number

 57


 58

 59


 60

 61


 62


 63


 64


 65


 66


 67


 68


 69

 70


71


72


73
                                                        Page

 Hourly CF3Br concentrations  10/29/82,
   0200-0300 MDT	149

 Maximum hourly  x/Q  (usec/m^) vs. l-Hc/zr	   152

 Observed SFg concentrations  for CCB
   Experiment 206, hour  8	153
 X/Q (ysec/nr)  vs. A9/iv
154
 Five-minute exposure from Tower A, Experiment 4,
   10/11/82, 0315 MDT .................  156

 Five-minute exposure from HER south of target area,
   Experiment 4, 10/11/82, 0315 MDT ..........  157

 Five-minute exposure from Tower A, Experiment 4,
   10/11/82, 0340 MDT .................  158

 Vertical profiles of hourly meteorological data,
   Tower A, 10/11/82, 0300-0400 MDT ..........  160

 Time-series of 5-minute wind directions from Tower
   A and Tower C, Experiment 4, 10/11/82  .......  161

 Time-series of 5-minute wind directions from Tower
   A and Tower B, Experiment 4, 10/11/82  ......  .  162

 Hourly SFg concentrations,  Experiment 4,  10/11/82,
   0300-0400 MDT  ...................  163
Along-hill profiles of hourly SFe concentrations,
  10/11/82, 0300-0400 MDT  ..............   164

Hogback samplers by rows ...............   165

Hourly CF3Br concentrations,  10/11/82,
  0300-0400 MDT  ...................   167
Along-hill profiles of hourly
  concentrations, 10/11/82,  0300-0400 MDT  ......   168

Five-minute exposure from Tower A,  Experiment  10,
  10/22/82, 0330 MDT .................   169

Five-minute exposure from HER south of target  area,
  Experiment 10, 10/22/82, 0330 MDT  .........   170
                          Xll

-------
                          FIGURES  (Continued)
Number
74
75
76
77
78
79
80
81
82
83
84
85
86
Vertical profiles of hourly meteorological data,
  Tower A, Experiment 10, 10/22/82, 0300-0400 MDT
Pag.e.


171
Time-series of 5-minute wind directions from Tower A
  and Tower C, Experiment 10, 10/22/82	172.

Time-series of 5-minute wind directions from Tower A
  and Tower B, Experiment 10, 10/22/82	173

Hourly SF& concentrations, 10/22/82, Experiment 10,
  0300-0400 MDT	   175

Along-hill profiles of hourly SFg concentrations,
  10/22/82, 0300-0400 MDT  	   176

Hourly CF3Br concentrations, Experiment 10, 10/22/82,
  0300-0400 MDT	   177

Along-hill profiles hourly CF3Br concentrations,
  10/22/82, 0300-0400 MDT  . .  .	  .' .  .  .   178

Five—minute exposure from HBR south of the target
  area, Experiment 10, 10/22/82, 0425 MDT	179

Hourly«SFg.-concentrations, Experiment ..10,- 10/22/82,
0400-0500 MDT  	 .....   180

Vertical profiles of hourly meteorological data,
  Tower A, 10/22/82, 0500 MDT	181

Time-series of 5-minute wind directions,
  Experiment 10, 10/22/82  . .  .	182

Hourly CF3Br concentrations, Experiment 10, 10/22/82,
  0400-0500 MDT	183
Height section and orientation of idealized HBR
 relative to Tower A as represented in CTDM
                                                                194
                                  xiil

-------
                                TABLES

 Number                                                         Page

  1      Descriptive Statistics and Associated Analyses
           of Model Performance 	  17

  2      Summary of Residual Statistics for SHIS #1 ......  18

  3      Summary of Areas for Improving CTDM(11083) Based on
          Analyses of Experiments 201 and 210	94

  4      Comparison of Model Performance Statistics for
          CTDM(14083) and CTDM(11083-E) Using the
          SHIS #1 MDA	105

  5      Preliminary Visualization Experiment Examples of
          HC Analyses	118

  6      FMF Tow-Tank Experiments	  120

  7      Measured Heights of 10-cm High Streamer at the Upwind
          Base and Crest of the HER Model	121

  8      Instruments,  Heights and  Direct and Derived
          Measures at the Four Towers	127

  9      Definitions of the Measures	129

 10      Primary Sampler Locations  	  135

 11      Additional Samplers  	  .....   136

 12      Number of SHIS #2 Experimental Hours with
          Tracer Gas  Releases and Concentrations  	   142

 13      Summary of Tracer Concentration Data 	   143

 14      Five Highest  x/Q Observed at  HER and CCB	     151

 15      Hogback Samplers by Row	166

16      SHIS #2 Tracer Data Withheld  for  Model  Validation
          Purposes	185

17      SHIS #2 Hours  Selected  for Initial Modeling	187
                                  xiv

-------
Number


18


19 ,



20



21


22


23


24


25


26
                          'TABLES (Continued)
17-Hour Data Base Summary Statistics
188
Comparison of Wind Measurements at 40 m for'1-7':

  Hogback Study Hours	............  190


Summary of Model Input Data for SF5 and
  CF3Br Releases	   191
    •/ .   •>;«  '., "   •         **          .            .

Model Performance Statistics, 17-Hour Data Base  ';.. .  195


Results of Hourly CTDM(E) Simulations  ........  197


Results of Hourly CTDM Simulations . . . .... . . . .  198


Results of Hourly COMPLEX I Simulations  .  .  . . ... .  199


Results of Hourly COMPLEX II Simulations .  .  . . . . .  200


Experiment 11 Five-Minute Average Winds  .  .  ... ... .  201
                                  xv

-------
                      LIST OF SYMBOLS  AND  ABBREVIATIONS
SYMBOL

a(z)
a
am
as
b
 scat
C
X
C.
C
 max
Co(Vz)
d
D
A6
Fr
  H
Y
h
H
H
H
 m
H
Radius of an ideal circular hill in horizontal cross-section
Product of T  • T
            u    h
Angular horizontal plume spread, including meander
Angular horizontal plume spread, without meander
Scattering coefficient
Concentrat ion
Concentration
Instantaneous concentration
Maximum hourly averaged concentration
Observed concentration
Vertical profile of concentration modeled at s = s
Modeled concentration
Distance of the source from the stagnation point
Effective radius of curvature minus the dividing-streamline
height
D modified to simulate stability-dependent horizontal  plume
distortion
Angle between mean plume trajectory and the trajectory over the
hill center
Plume spread angle in the vertical
Error or residual
Froude number based upon hill height
Froude number based upon hill length
Molecular exchange coefficient
Terrain elevation above a "zero plane"
Hill height (crest)
Critical dividing-streamline height
Elevation of the peak observed concentration
Plume centerline height over terrain (stand-off distance)
H-H
                                  xvi

-------
e
ed
e
 o
i , i , i
 x  y  z
L
L
ma
m
 g
N
N
P
p(e)
Q
Q(s,z)
 S
TC,B ,Z
R
 e
Ri
s
s
s
 V
 _ I
a
a
, Wirld direction
.•Direction of the stagnation streamline
 Mean wind direction of approach flow
 Turbulence intensities alongwind, crosswind, and vertical
 Distance beyond s  to the point where T = T •
 Distance beyond s  to the peak terrain elevation along the
 plume trajectory                .  .
 Half-length (horizontal alongwind scale) of the hill midway
 between H and H
 Arithmetic mean  .
 Geometric mean
 Number (in a summation)
 Brunt-Vaisala fr.equency
                           2
 Parameter equivalent to 2y                       .
 Wind direction probability distribution function
 Tracer emission rate                                       '
 Vertical profile of mass flux
 Measure of resolution (arithmetic)
 Measure of resolution (geometric)
 Polar coordinate system
 Effective radius of curvature
 Richardson number
 Distance along plume trajectory
 Distance from source to receptor
 Distance along terrain from impingement point to receptor
 Arithmetic standard deviation
 Geometric standard deviation
 Distance along wind direction from source to terrain at H
                                                          c
 elevation
 Virtual source distance (vertical plume spread)
 Virtual source distance (horizontal plume spread)
 Standard deviation
 Standard deviation of horizontal wind direction            •,•
 Standard deviation of alongwind velocity fluctuations of the
 mean wind
                                  xvii

-------
 w
 zo
 h
  .
 rain
T
 u
u
Uo
X
y
x ,y , z
xR'yR>zR
zr
Standard deviation of crosswind velocity fluctuations  of  the
mean wind
Standard deviation of vertical velocity fluctuations
Crosswind horizontal and vertical standard deviation of
tracer concentrations
Value of a  at s = s
          z         o
a  enhancement factor (terrain factor)
 z
Terrain factor for plume streamline height above ground
Minimum value of T along plume trajectory
Minimum value of T near center of hill
Terrain factor for vertical plume spread
Terrain factor for plume transport speed
Generic wind speed
Approach flow wind speed profile
Generic distance downwind from source
Virtual source distance
Generic crosswind distance from plume centerline
Tracer release coordinates
Receptor coordinates
Plume release height
ABBREVIATIONS
ARLFRO      Air Resources Laboratory Field Research Office
APS         Arizona Public Service Company
ATDL        Atmospheric Turbulence and Diffusion Laboratory
BAO         Boulder Atmospheric Observatory
BFM         Butte Field Model
CCB         Cinder Cone Butte
CTDM        Complex Terrain Dispersion Model
CTMD        Complex Terrain Model Development
EPA         U.S. Environmental Protection Agency
ERT         Environmental Research & Technology
ESRL        Environmental Sciences Research Laboratory
FMF         Fluid Modeling Facility
FMN         Farmington Airport
GC          Gas chromatograph
                                  xviii

-------
 GLC
 HBR
 Hz
 LASL
 LMF
 MDA
 MDS
 MDT
 MRI
 MSL
    3
 us/m
 NAAQS
 NAWC
 NOAA
    3
 ns/m
 ppb
 PPt
 PDF
PNM
PPC
PSD
SHIS
WPL
WSSI
 Ground-level concentration
 Hogback Ridge
 Hertz
 Los Alamos Scientific Laboratory
 Linear mass flow meter
 Modelers' Data Archive
 Meteorological Data System
 Mountain Daylight Time
 Meteorology Research, Inc.
 Mean Sea Level   , *
 Micro-seconds per cubic meter
 National Ambient Air Quality Standard
 North American Weather Consultants
 National Oceanic and Atmospheric Administration
 Nano-seconds per cubic meter
 Parts per billion by volume
 Parts per trillion by volume
 Probability distribution function
 Potential Flow Model
 Public Service Company of New Mexico
 Plume path coefficient
 Prevention of Significant Deterioration
 Resistance Thermometric Device
 Small Hill Impaction Study
 TRC Environmental Consultants,  Inc.
Wave Propagation Laboratory
 Western Scientific Services,  Inc.
                      xix

-------
                              ACKNOWLEDGEMENT S
     Continuing progress in the EPA CTMD program reflects  the  dedicated
efforts of many people.   In particular,  we gratefully acknowledge  the
special contributions of the following individuals to the  Hogback  Ridge
field program:

     •    Ray Dickson, Gene Start, Jerry Sagendorf and Russ Ackerman (NOAA
          ARLFRO) for flow visualization, tracer experiments,  tracer
          analysis, 10-m and 30-m towers, and the meteorological data
          acquisition, communications and analysis system;

     •    Wynn Eberhard (NOAA WPL) for operation of the lidar  system,  and
          Bill Neff (NOAA WPL) for the Doppler acoustic sounder;

     •    Clark King (NOAA WPL) and colleagues at NOAA ATDL for operation  of
          the tethersonde systems;

     •    Norm Ricks (Morrison/Knudsen) for the photography program—and
          especially for his unfailingly good spirits during cold  and  lonely
          vigils high on the 150-m tower;

     •    Bob Lehman, Steve Anderson and John Faust, for site  logistics  and
          operation of the 150-m  tower; and

     •    Dan Godden, for command center coordination and real time data
          analysis.

     Finally, we give special thanks to Frank Schiermeier, Bill Snyder and
George Holzworth (EPA) for their unflagging support and encouragement.
Throughout, they have willingly shared with us the difficult times, and
should share in our successes as well.
                                   xx

-------
                                 SECTION  1

                                INTRODUCTION
     The U.S. Environmental Protection Agency is sponsoring the Complex
Terrain Model Development (CTMD) program,  a multi-year,  integrated program
to develop,  evaluate,  and refine practical plume models  for calculating
ground-level air pollutant concentrations  that result from large emission
sources located in mountainous terrain.  The principal objective of the CTMD
program is to develop models with known accuracy and limitations for simu-
lating 1-hour average concentrations during stable atmospheric conditions,
i.e., to handle the stable plume impaction problem.

     The objectives of the program were described by Holzworth (1980) and
generally follow the recommendations of the participants of the EPA spon-
sored workshop to consider the issues and  problems of simulating air pol-
lutant dispersion in complex terrain (Hovind et al. 1979).  The program was
subsequently designed with a perspective toward modeling and included model
development efforts based on physical modeling, field experiments and theo-
retical work.  The CTMD was designed to proceed from field experiments and
physical modeling for isolated simple terrain features to field measure-
ments/experiments centered on an operational full-scale power plant.

     The program was begun in June 1980.  The first major component was the
Small Hill Impaction Study #1 (SHIS #1), a field experiment* conducted
during the fall of 1980 at Cinder Cone Butte (CCB), Idaho.  CCB is a roughly
axisymmetric, isolated 100-meter tall hill located in the broad Snake River
Basin near Boise, Idaho.  The field program consisted of ten flow visual-
ization (oil-fog) experiments and 18 multi-hour tracer gas experiments
conducted during stable flow conditions with supporting meteorological,
lidar and photographic measurements.
*The prime contractor for the CTMD is Environmental Research &  Technology,
 Inc.  (ERT).  The principal subcontractors for the CCB experiments were
 Western Scientific Services, Inc. (WSSI), responsible for fixed meteoro-
 logical data (towers, instrumentation  and data communication) and North
 American Weather Consultants (NAWC), responsible for the experimental field
 program (tracer and smoke releases, tracer data collection, photography,
 mobile meteorology, and field logistics).  In allied activities, the EPA
 Fluid Modeling Facility (FMF) has provided laboratory fluid modeling sup-
 port; the NOAA Wave Propagation Laboratory (WPL) supported the field pro-
 gram  with a manned lidar system; and TRC Environmental Consultants, Inc.
 provided independent data audits.

-------
      The CCB data were used to  develop  and  evaluate two new, preliminary
 modeling algorithms (called the Impingement and Neutral models) and to eval-
 uate the performance of current EPA complex terrain plume dispersion models
 (Valley, COMPLEX I,  COMPLEX II,  and Potential Flow Model (PFM)).  The
 Impingement model simulates dispersion  during strongly stable flows in which
 plumes remain approximately horizontal  and  impinge on and travel around the
 sides of CCB.  The Neutral model simulates  dispersion during slightly stable
 and neutral flows in which plumes rise  over CCB.  For the model evaluations,
 45 hours were selected from the 18  tracer experiment days.  The meteorolog-
 ical and tracer data were  used  in a "hands-off" mode:  no modifications were
 made to the models to improve their predictions.

      The EPA CTMD First Milestone Report - 1981 (Lavery et al.  1982)
 describes the air quality  models that were evaluated, the evaluation meth-
 ods,  and overall model performance.  The CCB field experiments  and associ-
 ated quality assurance  activities are described, and a series of experiments
 conducted in the EPA Fluid Modeling Facility (FMF) towing tank  to simulate
 the field studies are presented in  an appendix.

      The SHIS #1 tracer gas  source data (emission rates,  locations and
 heights of SF6,  CF3Br,  and oil-fog  releases), concentration data and
 meteorological data (from  six towers, a tethersonde and free balloons) were
 subsequently delivered to  EPA and are available from the EPA Project Offi-
 cer.   A CCB Modelers' Data Archive  (MDA) has also been prepared and is
 available from the Project Officer.  It includes winds and other meteoro-
 logical parameters derived specifically for the SFg and CF3Br release
 heights  as  well  as hourly  average tracer gas emission rates and ground-level
 concentration data.

      The EPA CTMD Second Milestone Report - 1982 (Strimaitis et al.  1983)
 documents work accomplished from fall 1981 through April  1982 to improve  the
 Impingement and  Neutral modeling approaches, to refine the meteorological
 and  tracer  gas data bases, and further analyze the data to relate plume
 behavior and observed meteorological conditions.  Unlike the hands-off eval-
 uations  described in  the First Milestone Report, the approach taken in this
 report was  a case-study analysis.  From the 45 hours used in the first re-
 port,  14 hours were selected for detailed individual case-study to illus-
 trate and contrast various flow  situations and ranges of model  performance.

      The second  report describes improvements to the Impingement and Neutral
 models and  the subsequent  evolution to the Wrap and Lift  models.  It also
 describes several  analyses of the CCB data base, including a comparison of
 turbulence  statistics from tower measurements with those  derived from lidar
 data  and photographs, time-series and spectral analyses of wind  data,  and
 streamline  deflection estimates based on temperature data.

      The Second Milestone Report also contains,  as an appendix,  a report
 prepared by the  EPA FMF.  The report describes a series  of wind  tunnel and
 tow-tank experiments designed to study various aspects of  stratified,
 sheared  flow over  a variety of simple obstacles, as well  as  flow over  a
model  of CCB.  In particular, the dividing-streamline  concept is evaluated
 and shown to be  appropriate for obstacles of differing aspect ratio  embedded
 in stably stratified flows, both with and without  wind shear.

-------
     This Third Milestone Report documents work accomplished from June 1982
through May 1983.  It describes the further evolution of the Lift and Wrap
models into one model, the Complex Terrain Dispersion Model (CTDM).  Two
versions of CTDM and three existing complex terrain plume models (Valley,
COMPLEX I and COMPLEX II) have been evaluated with the entire 153-hour CCB
MDA.  An assessment of the performance of the models shows that a version of
CTDM that simulates enhanced turbulence near the crest and lee of CCB is
significantly better than the other models.

     The various case—study and systematic analyses that have been performed
in support of the model development efforts, as reported here and in the
Second Milestone Report, and the physical modeling studies have led to a
more complete understanding of the dynamics of flows and of plume dispersion
around CCB.  The concept of a dividing-streamline (Hc) has been shown to
be useful in discriminating between essentially horizontal flows and flows
that go over CCB.  A tracer plume released above Hc tended to travel over
the hill, and its plume path was consistent with streamlines in weakly
stratified flow.  The highest ground-level tracer concentrations during this
latter sort of event were observed near the crest or on the lee side of CCB.

     When a plume was released well below HC, it tended to travel around
the, side of CCB, generally preferring one side.  During this situation, the
highest concentrations were measured on the upwind face of the hill or along
the hill sides, depending on how often the plume lay along the stagnation
streamline.  A release at a height near HC  and close to the stagnation
streamline produced the highest concentrations measured during the CCB
experiments. These higher concentrations were generally observed at eleva-
tions near or below the release height.

     An examination of turbulence intensity measurements, photographs and
lidar data has shown  that the vertical plume growth is related to the verti-
cal turbulence intensity and Brunt-Vaisala frequency and is proportional to
the square root of downwind distance to distances of about 1 km.  For dis-
tances farther downwind, the growth rate is uncertain although physical
modeling studies suggest that the growth rate goes to zero after a certain
travel time.  The data measured on the towers located on CCB indicate that
turbulence is enhanced near the crest and on the lee side.  Plume growth in
the horizontal is controlled by high frequency turbulence and by low fre-
quency meanders with  time scales of about two hours.

     This Third Milestone Report also provides an overview of the second
Small Hill Impaction  Study (SHIS #2)* conducted during October 1982 at
*ERT  is continuing  to provide overall CTMD project management and is respon-
  sible for  the model development.  During SHIS #2, ERT was  responsible  for
  field management,  the  150-m and 60-m towers, site logistics, and quality
  assurance.   Through an Interagency Agreement with EPA, NOAA Air Resources
  Laboratory Field Research Office  (ARLFRO) conducted  the flow visualization
  and  tracer experiments, was responsible for the  10-m and 30-m towers,  and
  operated a real-time data acquisition and analysis system.  NOAA WPL pro-
  vided lidar, acoustic  sounders, a tethersonde, and optical and sonic
  anemometers.  TRC  provided independent performance audits.

-------
Hogback Ridge (HER) near Farmington, New Mexico.  The HER experimental
design was similar to the CCB design.  It included tracer and flow
visualization experiments with concurrent meteorological, lidar and
photographic measurements.  The experiment produced approximately 179 tracer
hours and has effectively extended the modeling data base to include flow
and dispersion around a two-dimensional ridge.

     An initial analysis of the HER tracer concentrations and the unedited
meteorological data suggests that the dividing-streamline concept is
applicable to ridge settings.  Hc distinguishes between flows that
surmount the ridge and highly variable flows below.  The highest con-
centrations were measured on the windward side of HER when tracer gas was
released below HC.

     The next field experiment in the CTMD program will be conducted at an
electric generating power plant located in mountainous terrain.  The site
selection process is underway.  Feasibility studies will be conducted at one
or more prospective sites during the fall of 1983.  The full-scale study is
scheduled for the fall of 1984.

     This report consists of five major sections and two appendices.
Section 2 presents the results of the evaluation of the complex terrain
models with the SHIS #1 MDA.  The current version of CTDM and suggested
model upgrades are described and evaluated in Section 3.  Section 3 also
describes case-study analyses of CCB Experiments 201 and 210.  The Hogback
Ridge experiment and its initial findings are presented in Section 4.  The
preliminary modeling of a 17-hour subset of the HER data base with CTDM,
Valley, COMPLEX I and COMPLEX II is discussed in Section 5.  Section 6
presents the summary, conclusions, and recommendations for future work.
Appendix A describes a laboratory simulation of neutral plume dispersion
over Cinder Cone Butte.  Appendix B discusses an initial analysis of the
SHIS #2 monostatic acoustic sounder and sonic anemometer data.

-------
                                 SECTION 2

                 EVALUATION OF COMPLEX TERRAIN MODELS WITH
                             THE CCB DATA BASE
     An initial comparison of modeled  and observed SF^ concentrations
from SHIS #1 was summarized in the  CTMD First Milestone Report.  COMPLEX I
and II, PFM, Valley, and two new algorithms, Neutral  and  Impingement, were
evaluated by means of SFg concentrations from up  to 45 CCB experiment
hours.  Model performance statistics similar to those recommended by Fox
(1981) were used to assess relative model performance.

     During the preparation of the  First Milestone Report, the  Valley model
was the only EPA complex terrain screening  model  recommended  for regulatory
use.  Since then, EPA (Rhoads 1982), in an  addendum to the "Regional
Workshop on Air Quality Modeling:  A Summary Report"  (April 1981), has
recommended a new regulatory approach—a  two-tiered screening approach  for
complex terrain applications.  Valley model calculations  provide the initial
screening.  If these calculations suggest violation of any NAAQS or PSD
increment, then a second-level screening  technique may.be used.  EPA
recommends COMPLEX I as the second-level  screening  technique.

     The Second Milestone Report described  the  evolution  of the Neutral and
Impingement models to the Lift and Wrap models.  Since  that  time,  the Wrap
and Lift algorithms have evolved into the Complex Terrain Dispersion Model
(CTDM);  the meteorological and tracer gas concentration data  bases  have been
refined; and a Modelers' Data Archive (MDA) has been  prepared for all avail-
able  SFg and CF3Br tracer gas release periods.   In all,  153 modelable  tracer-
hours  of data  are now available  from SHIS #1.

     This section contains the results of comparing modeled and observed
concentrations using this expanded data base.   A description of the methods
used  to prepare the SHIS #1 MDA  is followed by  a brief  explanation  of how
Valley,  COMPLEX  I,  COMPLEX II, and CTDM are run with this archive  data.
Model  performance statistics are then presented, and  an analysis of model
performance follows.

2.1   SHIS #1  Modelers' Data Archive

      The MDA developed  from  the  SHIS #1 contains observed 1-hour average
tracer gas  concentration  data, tracer release information,  and  observed and
derived  meteorological  parameters  estimated at release height for each of
the  hours during SHIS #1  in  which  either SF^ or CF3Br tracer gas  was
released.   The method of  estimating meteorological data representative of
 tracer gas  release  heights relies  on a few central assumptions  and is

-------
 objectively applied to all the  data with  few exceptions.  Those assumptions
 and a description of the  procedures are presented in the following sections.

      Because the meteorological data  contained  in the archive are estimates
 of the meteorological conditions affecting the  transport and diffusion of
 the tracer gases, they should be viewed as approximate.  This is par-
 ticularly evident in the  estimated  wind direction.  For example, the wind
 set at the 40-m level on  Tower  A was  partially  operational during
 experiments 205,  206,  210,  and  211.   Some data  from this instrument are
 recoverable,  but winds are estimated  at this level only by making an
 assumption about the difference in  wind directions between 40 m and 80 m.
 At times,  the assumption  may not represent the  real situation.  With the
 scale of  Cinder Cone Butte, a resulting wind direction error of less than
 10° could at  times  cause  the modeled  plume to miss the hill entirely during
 a  period  when the actual  plume  produced significant concentrations on the
 hill.

      Subjective estimates of the most appropriate wind directions as derived
 from evidence of actual plume transport directions are in preparation and
 will be included in a second version  of the MDA when available.

      2.1.1  Tracer  Concentration Data

      Tracer gas concentration data  from the SHIS #1 data base (available
 through EPA)  include data from  1-hour samplers,  co-located 1-hour samplers,
 10-minute  samplers,  and reflection  mast samplers (a subset of the 10-minute
 samplers),  and  they  include "recount" concentration data resulting from
 re-analyzing  a  subset  of  all bag samples as part of the quality assurance
 program.  The data base also provides additional information on the sampler
 location and  the  quality  of each individual sample.

      Tracer gas data contained  in the MDA have been assembled from the  SHIS
 #1  data base.   All values  are reported as 1-hour averages.   These averages
 include concentrations from all  1-hour sampling  bags labeled as  good,  the
 average of  good concentrations  obtained from the two samples collected  at
 co-located  sampler sites,  the average of the two concentrations  obtained
 from samples  included  in  the recount analyses,  the average of good 10-minute
 concentrations  at standard 10-minute sampler sites,  and the  average of  good
 10-minute concentrations  obtained at the foot of the sampling mast locations.
 In  the  case of  the 10-minute samples,  hourly concentration averages are
 included only if not more  than one  10—minute period was missing from the
hour.

      The position of each sampler is included in the archive along with the
 1—hour average  tracer gas  concentration.   The coordinate system  is  a Car-
 tesian  system with origin at the center of CCB,  x-axis  oriented  toward  true
east, and y-axis oriented  toward true north.  The sampler position identi-
 fication code for each concentration is also included.   A map of CCB with
each sampler position is contained  in the  First  Milestone Report.

-------
     2.1.2  Tracer Release Information

     The MDA contains the average emission rate of the tracer gas,  the  polar
coordinates of the position of the source,  the elevation of the  base  of the
source crane, the height of the source above the ground, and the times  at
which the tracer gas was turned on and turned off.

     The emission rate is an average mass release rate (g/s) from the time
at which the release valve was opened to the time at  which it was shut  off.
In some cases, this period of time was less than 1 hour, but in  most  cases
it was several hours.  The start and stop times for the release  are refer-
enced to the beginning and ending times of each experiment hour, respec-
tively.  A start time of -10 (minutes) indicates that the tracer was
released ten minutes before the start of the sampling hour.  An  end time of
-5 (minutes) indicates that the release ended 5 minutes before the end  of
the hour.

     Coordinates of the source position are expressed in the hill coordinate
system, a polar grid centered on Cinder Cone Butte.  The zero height  contour
in this system corresponds to the 3100-ft elevation MSL (944.9 m).  Release
elevations are presented in meters above the ground,  and the elevation  of
the ground at the release position is given as the difference in meters from
944.9 m MSL.  A topographic map of Cinder Cone Butte  can be found in  the
First Milestone Report.

     2.1.3  Meteorological Data

     Meteorological data contained in the MDA differ  from those  contained in
the SHIS #1 data base in that:  all quantities apply  to the release height
of the tracer gas rather than to the height of fixed  instrument  levels;
derived parameters computed from the meteorological data base are included;
and 1-hour averages are constructed.

     A "spline under tension" method (Cline 1974) was used to interpolate
meteorological variables between instrument levels on Tower A.   This  method
produces a linear interpolation when a tension factor of 50 is specified,
and it produces a cubic spline curve through the data when a tension  factor
of zero is specified.  The suggested nominal tension  factor of 1.0 produces
a smooth curve through all data points in a profile.   It does not invent the
cusps and regions of high curvature between data points that are common with
the cubic spline.  After inspecting a number of profiles produced with
tension factors between 0.5 and 3.0, the factor 1.2 was selected.  This
factor produces slightly greater curvature near instrument levels than  does
a factor of 1.0, but it also reduces the magnitude of local maxima/minima
between data points in regions where the vertical gradient of the profile
quantity must change sign.

     Meteorological conditions representative of release height  are assumed
to be equivalent to data taken from the Tower A vertical profile at the same
height above the surface as the height of release,  even though the surface
elevation at the release point generally differs from the surface elevation
at Tower A.  This approximation is consistent with the spatial resolution of
the meteorological instrumentation because the release locations lie  between

-------
1 and 2.5 km from Tower A, because differences in surface elevations between
the base of Tower A and release locations vary between -6.1 m and 1.5 m with
a mean difference of -3.7 m, and because the vertical resolution of wind
measurements above 10 m on Tower A is 30 m or greater.  (Wind sets were
located at 2, 10, 40, 80, and 150 m.)

     The sequence of 5-minute data in the MDA was constructed as follows:

     •    Tower A wind speeds, wind directions, and temperatures contained
          in the refined data base were scanned for missing data.  When
          missing 5-minute values were encountered, they were replaced with
          values estimated by linear interpolation in time.  Only UVW prop
          wind data were used to develop wind information because the F460
          cup and vane instruments were placed only at the 2-,  10-,  and
          150-m levels, while most release heights were between 20 m and
          60 m.

     •    The temperature, the vertical component of the wind speed,  and the
          horizontal wind speed and direction were estimated at release
          height by "spline under tension" interpolation with a tension
          factor of 1.2.   Horizontal speeds and directions were first broken
          into wind components, and the components were interpolated  to
          obtain the wind speed and direction at release height.   The 40-m
          level wind data were incomplete in four of the experiments  (205,
          206,  210,  211), so that interpolation of the winds between  10 m
          and 80 m was necessary.  However, the u-component (east-west) of
          the wind data from 40 m was available,  and the mean winds were
          approximately SE during each of these experiments rather than
          nearly N or S.   By assuming that the directional wind shear was
          small between 40 m and 80 m at Tower A during these four
          experiments,  estimates of the wind speed at 40 m were made  with
          the 40-m u-component and the 80-m wind direction measurements.
          The resulting wind speed profiles indicated that the  speeds
          estimated at 40 m are generally reasonable, and so these speed and
          direction estimates of 40-m level winds  were used in  the
          interpolation procedure for nearly all 5-minute periods during
          which  the  40-m  v-prop was inoperative.   However,  because the wind
          directions near 40 m appear to differ substantially from those at
          80 m during the last hour of experiment  211,  the spline
          interpolation between 10 m and 80 m was  used for this period of
          time.

     •    Turbulence data were estimated at release height by employing a
          linear interpolation rather than the spline interpolation.   The
          turbulence velocity scales CTU,  av,  and aw were obtained from the
          turbulence intensity values  contained in the refined  data base by
          multiplying by  the wind speed.   Unlike wind and temperature data,
          missing  turbulence data were not  filled  in by interpolating  in
          time.   However,  estimates of crw were prepared for those
          experiments in  which one of  the horizontal wind speed props
          malfunctioned at the 40-m level.   In these experiments  (205,  206,
          210, and 211),  reported values  of iz in  the refined data base
          are flagged as  bad because the  reported  wind  speed  is incorrect.
          But  because the w-prop was working,  the  aw values are not

-------
          necessarily deficient.  aw was recovered by multiplying the
          original value of iz by the original value of the wind speed.
          The value of CTW was then interpolated to source height.  A prop
          response correction derived from the work of Horst (1973) was
          applied.  More information on its application to the Cinder Cone
          Butte data can be found in the Second Milestone Report.  No prop
          response corrections were applied to au or av.

     •    The Brunt-Vaisala frequency,  N,  was estimated at source height  by
          interpolating the temperature profile in the immediate vicinity of
          the release height to obtain the local temperature gradient.

     •    The dividing-streamline height,  HC, was obtained from the
          splined profiles of temperature  and wind speed by means of the
          integral formula presented in the First Milestone Report.  A bulk
          hill Froude number was calculated for the layer between Hc and
          the top of the tower, 150 m,  and also for the layer between 2 m
          and 150 m.  The hill height in both calculations is the difference
          between 95 m and the height of the bottom of the layer (either
          HC or zero).

     •    The Pasquill-Gifford stability class was calculated from net
          radiation and wind speed data by means of the method of Williamson
          and Krenmayer (1980).  Wind speeds measured by the cups at the
          10-m level on Tower A (reported  as scalar averages)  and the net
          radiation data were interpolated in time whenever missing values
          were encountered.   The stability class was calculated as a number
          between 1 and 6, where 1 denotes stability class A.   Both the
          stability class and the 10-m wind speed are included in the data
          archive.

     Most 1-hour average data in the MDA were obtained from this sequence of
5-minute average data interpolated to release height.   Only stability class
data were not obtained in this way.  The 1-hour stability class was found
from the 1-hour averages of net radiation  and 10-m wind speed.

     The remaining 1-hour average data in  the MDA were constructed as
follows:

     •    The wind speed and direction were calculated as both vector and
          "scalar" averages.   Two versions of the "scalar"  wind  direction
          were calculated.  One is a scalar average of the 5-minute vector
          resultant wind directions.  The  second is a vector average  of unit
          vectors along each 5-minute vector resultant wind direction so
          that all directions have equal weighting,  as in a scalar average,
          but the averaging is performed with vector arithmetic.

     •    Temperatures and parameters calculated from the splined  profiles
          of the 5-minute temperature and  wind data—N,  Hc,  Fr—were
          simply averaged to provide 1-hour average values.

-------
          Horizontal turbulence data au and crv were computed for
          1-hour periods by adding the contribution due to the 5-minute
          turbulence values to the contribution due to the variability of
          the 5-minute average winds.  Let (o)60,0 denote the total
          1-hour value of the standard deviation; let (0)5 Q denote the
          total 5-minute standard deviation; and let (0)50,5 denote the
          standard deviation of the 5-minute average winds over a 1-hour
          period.  Then
" (a)
                                    (0)
(1)
                  '60,0     x~'60,5  '  x"'5,0
          Note  that  although prop  response corrections were not applied to
          the (0)5 o values, they were implicitly contained in (0)^5
          because the 5-minute wind data in the refined data base include
          corrections for  prop response and wake effects.  Also,  because no
          time  interpolation was performed on the 5-minute turbulence data,
          less  than  a full set of  12 values may be available during some
          hours.  In these cases the average of the (0)5 Q values will be
          incomplete.  The number of (0)5 Q values contained in each hour are
          denoted by N(su), N(sv), and N(SW) in the archive.

          The vertical turbulence aw was computed for 1-hour averaging
          periods in the same way as au and crv, except that the prop
          response correction suggested by Horst (1973) was also  applied.
          Neither the (aw)5 Q nor the (aw)go 5 data used in the
          formula for (crw)60 0 nave tlie correction already applied.
          Rather, the correction was applied directly to (crw)go 0*
          Note  that the construction of (0)50 Q by the above formula is
          exact in the case of CTW, but only approximate in the case of
          au and a
                  V
2.2  Complex Terrain Models Evaluated

     2.2.1  Valley Model

     The Valley model (Burt 1977) is recommended by EPA as the initial
screen in a two-tiered screening approach for complex terrain analyses  in
support of regulatory decisions.  Valley is designed to provide an estimate
of the maximum 24-hour pollutant concentration expected to occur on elevated
terrain near a point source of air pollution in any 1-year period.  This
concentration is computed with a steady-state,  univariate Gaussian plume
dispersion equation, modified to provide a uniform crosswind distribution
over a 22.5° sector and using assumed worst-case meteorological conditions.

     The model assumes that the plume travels toward nearby terrain with  no
vertical deflection until the centerline of the plume comes to within 10  m
of the local terrain surface.  Thereafter,  the centerline is deflected  to
maintain a stand-off distance of 10 m from the terrain surface.   The plume
is considered to impinge upon the terrain at points where terrain height
equals the plume height, and the impingement point used in the calculation
of maximum plume impact is the nearest such topographic point as viewed from
the source.
                                  10

-------
     Worst-case meteorological  conditions are defined by that combination of
wind speed and Pasquill-Gifford (PG) dispersion stability class that pro-
duces  the highest  possible concentration at the impingement point.  For
large  sources of air pollution, a stack-top wind speed of 2.5 ra/sec and PG
stability class F  are recommended as those conditions that will produce the
highest concentrations during stable conditions when plume impingement is
most likely.

     The model estimate  is implied to be a 1-hour average concentration.
The 24-hour average concentration is estimated by dividing this 1-hour
average concentration by four, on the premise that the impinging plume may
affect a specific  point for no more than 6 hours in any 24-hour period.  It
should be stressed that because neither the tracer source release heights,
crane  locations, nor meteorological conditions were equivalently
"persistent" from  hour to hour during the actual experiments, the CCB data
base is not appropriate for testing the Valley model as it is used in
regulatory applications.

     The longest steady averaging period for tracer releases and
concentrations in  the CCB field study was four hours.  Most steady averaging
periods were between one and two hours.  Consequently, only 1-hour Valley
estimates have been compared with maximum observed hourly tracer
concentrations.

     The hourly emission rate has been reduced for those hours in which
tracer gas was released for less than a full hour.   On and off times are
taken from the MDA, and a transit time shift has been included that has been
calculated from the mean hourly scalar wind speed  at source height and the
distance to the center of the hill.

     Release crane positions during the experiments range from a distance of
540 m to a distance of 1,424 m from the center of  the hill.   The local
release heights of the nonbuoyant tracer gas vary  from 15 m to 60 m.   The
distance to the nearest point of impingement is  taken to  be the  difference
between the distance from the source to the center of the hill and the
radius of a representative circle fit to the hill  at the  height  of the
source.  This representative radius is given by (CTMD First Milestone  Report)
           /  \ _ f\~tn     0.38
          a(.z ) = 970 z
             r         r
(2)
for elevations between 10 and 60 m above 944 m MSL.

     Concentration estimates scaled by the  emission  rate are computed by:


          £ = (2.03)(1(P)  ex (_Q 5 (R .  )2)
                 a ux
                  z
(3)
which is derived from Equation 2-1 of the Valley Model  User's  Guide  (Burt
1977).  C is in units of ug/m ,  and Q is in units of  g/sec.  Hs  is the
altitude of the plume centerline above local terrain  (stand-off  distance).
                                  11

-------
     The wind speed is set to 2.5 m/sec, and the stand-off distance  is  fixed
at 10 m to be consistent with the regulatory applications  of  Valley.
However, because the 10-m value is not based on any theoretical analysis,  it
may not be an appropriate scale for the narrow plume configuration of the
tracer experiments (compared to plume sizes encountered with  large pollutant
sources).

     The standard deviation of the vertical pollutant distribution (az)
is calculated from the Pasquill-Gifford-Turner (PGT) formula:
               0.362 x °*55 -2.7
(4)
The constants in this equation are taken from Table 2-1 of the Valley Model
User's Guide.  They are applicable for PG stability class F over a range  of
x between  100 and 1,000 m, which includes all of the impingement distances
in the CCB archive.
                        •  • •   '                           1
     2.2.2  COMPLEX I and 'COMPLEX II Models

     COMPLEX I and COMPLEX II are sequential complex terrain models designed
to estimate 1-hour, 3-hour, 24-hour, and annual pollutant concentrations
resulting from point-source emissions.  Concentrations are estimated at many
receptors using hour-by-hour meteorological data.  COMPLEX I is the EPA-
preferred second-level screening technique.  COMPLEX II has been issued by
EPA for public testing and evaluation.

     Several terrain treatment options (formulated by the Complex Terrain
Team at the February 1980 Chicago Workshop on Air Quality Models) are
available in the COMPLEX models.  The standard option for PG stability
classes E and F is based on the plume behavior simulated by the Valley
model.  This was the option selected for evaluation with those stable hours
in SHIS #1.  (The COMPLEX models also contain a buoyancy-induced dispersion
option, but this feature has not been used in this modeling because the
tracer releases are nonbuoyant.)

     COMPLEX I is a univariate Gaussian plume model with 22.5° sector
averaging in the horizontal.  It uses the PG stability class system-
COMPLEX II differs from COMPLEX I only in its representation of crosswind
plume spread.  Whereas COMPLEX I assumes a 22.5° horizontal sector
averaging, COMPLEX II assumes the familiar crosswind Gaussian profile.  (in
all other respects, the two models are identical.)

     COMPLEX I for stable (class E, F) conditions is essentially an
extension of the direct-impingement Valley model to hourly averaged wind
speed, wind direction, and stability class instead of the assumed worst-case
meteorology; COMPLEX II replaces Valley's 22.5° horizontal sector averaging
with a Gaussian crosswind profile.  For neutral or unstable conditions,
COMPLEX I and COMPLEX II permit different (nonimpingement) terrain
assumptions.  For stability classes A through D, the terrain treatment
allows the plume centerline to rise over terrain features, but at a height
less than  its initial height over flat terrain.  Its actual height at any
point is computed from its initial height, the local terrain height, and  a
plume path coefficient.  The resultant simulated plume height above the
                                  12

-------
local elevated terrain height is always at least half its initial height
when the recommended plume path coefficient is used.

     The evaluation of COMPLEX I and COMPLEX II with the CCB data base  was
performed by embedding the essential computational algorithms of  these  two
models within a computer code called the Butte Field Model (BFM).  The
COMPLEX algorithms were invoked by selecting .specific options for plume path
coefficients, dispersion parameters, sector averaging, and surface
reflection treatment.

     Input data for the models were taken from the MDA.  They consisted of
hourly scalar average wind speed and unit-vector average wind direction at
source height, source location, hourly stability class, receptor  locations,
and receptor heights.  Receptor information was provided for. all  potential
(93) sampling locations.  Although the samplers were designed to  measure
tracer concentrations no closer to the surface than 1 m, no receptor height
offset was used in the modeling because the model resolution is no more
accurate than 1m.

     In addition, the BFM code has been modified to accommodate those
experiment hours that contain fewer than 60 minutes of tracer gas release.
Start and stop times have been computed from data contained in the MDA, and
transport time delays have been estimated from the mean wind speed for  the
hour and the distance from the release point to the hill center.   The mean
hourly speed is used rather than the local 5-minute average speed because
the variability of the 5-minute average speeds is greatest at low wind
speed.  The travel time is likely to extend beyond a single 5-minute period
when the wind speed is low, and therefore some speed averaging occurs over
the time of travel.  The hourly tracer gas release rate is reduced by
multiplying by the ratio of the net period of tracer gas relese to 60
minutes.

     2.2.3  Complex Terrain Dispersion Model (CTDM)

     CTDM is the result of combining the Lift and Wrap algorithms described
in The Second Milestone Report into one model for use in analyzing the  data
from SHIS #1.  Unlike the other models evaluated, CTDM has the ability  to
utilize onsite measurements of turbulence and flow parameters beyond the
mean wind speed and the inferred surface layer stability class.  Version
11083 of CTDM is described in detail in Section 3.2.

     The Probability Distribution Function (PDF) form of CTDM(11083) and
that of CTDM(11083-E) have been chosen for evaluation in this section.  Note
that version 11083-E includes a terrain enhanced az.   The probability
distribution of wind directions was constructed from the 5-minute wind
direction and i  data contained in the MDA.  Other input meteorological
data included hourly values of HC, iz (Horst-corrected), scalar wind
speed, unit—vector average wind direction, and local Brunt-Vaisala
frequency.  Receptor data were provided for all 93 potential sampling
locations, and model estimates of tracer gas concentrations were  computed at
the surface of the hill.

     The model also contains an adjustment for those hours that contain
fewer than 60 minutes of tracer gas release.  As in the BFM simulation  of
                                  13

-------
COMPLEX  I and COMPLEX II, start and stop times and a transport time shift
are computed, but the PDF form of CTDM (11083) adjusts the PDF rather than
the hourly emission rate.  All 5-minute periods that contain fewer than 2.5
minutes of tracer material (after accounting for transport time) are removed
from  the PDF computation.

2.3  Model Performance Statistics

      The evaluation of the several complex terrain models is based primarily
on the analysis of residuals
     CQ -
                                                                          (5)
where CQ is the observed concentration and Cp is the corresponding pre-
dicted concentration.  Both CQ and Cp have been normalized by dividing
by  the tracer gas emission rate.  The specific evaluation measures and
methods follow from the methods discussed in the First Milestone Report ,
from the results of the AMS/EPA workshop on model validation and model
uncertainty (Fox 1981), from the EPRI workshop on validation of plume models
(Londergan 1980), and from the work of Venkatram (1982).

     Two kinds of residual statistics are used,  arithmetic and geometric.
Arithmetic means and standard deviations are particularly useful if the
residuals are normally distributed.   For residuals that are more nearly
log-normally distributed, geometric statistics are more useful.  Log-normal
residuals are transformed into normally distributed residuals by the
transformation
          C1
     In C.
                                                       (6)
     The arithmetic mean and standard deviation of the residuals  are
computed as
m
T 2
e  =
              N
                  (C
                                  C ).
                                   pi
                                   F
(7)
                tl
                I
     Because of the transformation in Equation 6,  the mean and  standard
deviation of transformed,  log-normally distributed residuals become
             = exp
    In (C /C ).
         o  pi
                                                                (8)
           g
               SXP  M
                        (In
         (C /C ).  -
           o  p i
                             In m )2
                                 g
                                  14

-------
The above equations are numerically well-behaved if neither Co nor Cp  is
zero.  However, there are times when even the peak hourly Cp is estimated
to be zero.  Because different models may produce different numbers of
"zero" hours, intercomparison of nig and sg statistics for these models
becomes somewhat imprecise if these zeros are simply removed from the
statistics.

     A second approach is' also used in this report.  When plotted on
log-probability paper, a log-normally distributed set of residuals should
produce a straight line.  If we assume that the residuals for a particular
model applied to the CCB data base represent one sample of the "true"
residual distribution for that model, then small variations from a straight
line plot may be neglected compared to the overall trend contained in  the
sample.  The geometric mean residual (nig), or bias, .is the intersection of
the trend line with the 50th percentile, and the geometric standard devi-
ation (sg), or noise, is found from the slope of the trend line.  Let  m
denote the value of CQ/C  along the trend line for the nth percentile,
and let p denote the number of standard deviations from the center of  a
normal distribution to the point at. which the area under the curve equals
n/100.  Then
                    1/p
                                                                          (9)
           g
                m
                 g
Note that because h may vary between 0 and 100, while the area under half of
the normal curve is at most 0.5, the sign of p will depend on whether m is
less than or greater than mg.
     This method for estimating nig and sg does not suffer from numerical
problems of vanishingly small values of Co or C ,  because c9/Cp
ratios  for these values occur in the tails of the distribution which are
neglected.  Their  influence is accounted for in the 'position and slope of
the trend line.

     A  general discussion of the interpretation of model performance
residuals is contained in the First Milestone Report, in the EPRI workshop
report  (Londergan  1980) and in the paper by Venkatram (1982).  The
characteristics of an ideal model are easily stated.  The modeled
concentrations will nearly match observed concentrations at all points.
Consequently, ma and sa will be nearly zero, or mg and sg will be
nearly  one.  Furthermore, a plot of the residuals versus Cp, or any other
combination of the observed or input meteorological parameters, would show
the  residuals  to be normally distributed in a narrow band, and therefore
uncorrelated with  Cp (or other modeling parameters).  The degree to which
the mean residual  for a particular model differs from the ideal mean and the
degree  to which the standard deviation of the residuals differs from the
ideal  standard deviation are measures of the overall accuracy and precision
of the  model.                                    •

     Beyond these  measures, a desirable attribute of a model is its ability
to distinguish among predictions corresponding to different dispersion and
flow conditions.   This "resolution" depends on the relationship between the
                                   15

-------
  precision of the model and the separation between individual model
  predictions.  Because the precision,  as measured by the  standard deviation
  of the residuals (a(e)),  represents the uncertainty of the model
  estimates, a model will exhibit better resolution if the "distance" between
  individual model estimates exceeds the uncertainty.   The average "distance"
  among a set of model estimates is the standard  deviation of the modeled
  concentrations a(Cp).   Therefore, the ratio of  the  standard deviation of
  the residuals to the standard  deviation of the  model estimates should be as
  small as  possible to maximize  the resolution:
            r = a  (e)/cr  (Cp) «  1.
  If  the  residuals are normally distributed, then
                        (Cp).
  If  the  residuals more nearly follow a log-normal distribution,  then
           rg = In
o (In Cp).
(10)
(11)
(12)
      The analysis of residuals is based on the residuals of (1)  peak
 concentrations paired in time, but unpaired in space,  and (2)  all  con-
 centrations on CCB paired in time and space.  The emphasis on  these  two  sets
 of residuals is based on the necessity of the model to predict the highest
 concentrations* and to simulate properly the dynamics  of the flows and
 turbulence that determine the ground-level concentrations on CCB.  In
 addition to the statistics of the residuals, the residuals are plotted
 against various meteorological parameters to help ascertain biases or
 uncertainties in the model input or formulation.   Table 1 summarizes the
 residual statistics  and  graphical displays used  in the next  subsection.

 2.4  Analysis of Model Performance

      Overall residual statistics developed by comparing model  calculations
 with tracer gas concentrations  sampled at  CCB are  given in Table 2 and
 plotted in Figure 1.   Results are presented for  five models,

           CTDM (11083-E)
           CTDM (11083)
          Valley
           COMPLEX I
          COMPLEX  II

 The  columns  labeled "Peak Concentrations"  in Table 2 summarize  the errors in
 estimating 1-hour maximum concentrations, regardless of location,  over  the
 153  tracer hours.  The columns titled "All Concentrations" summarize  the
mean errors in estimating concentrations observed at all sampling points
 (.paired  in space and time).
*For example, compliance with air quality standards is most  often  estimated
 in terms of the highest,  second-highest  modeled  concentration at  a receptor
 location.
                                  16

-------
    TABLE 1.  DESCRIPTIVE STATISTICS AND ASSOCIATED ANALYSES OF
              MODEL PERFORMANCE
                     •   •  • ,      •                     Scatter
               Means    Variances    Distributions     Plots
                         (C
                                                      vs.  C
 P  ,
C -C
o p
C /C
o p
m
a
m
g
s , r
a a
s , r
g g
                                      Cumulative
                                        frequency
                                                      vs.  C
C   - C.
 °max
               m,,
       Pmax     a
'  r
                                                      vs.  met
cr>   /cn       me       se»  re
 °max  Pmax     &        8   &
                                      Cumulative      vs.  met
                                        frequency  .
met:  u, u/N, iy* iz, l-Hm/zr, l-Hc/zr where ^ is the
      elevation on CCB of the maximum Co,  and zr is the relese height
      of the plume.
                                 17

-------
            TABLE 2.   SUMMARY* OF  RESIDUAL STATISTICS FOR SHIS #1
                Peak Concentrations (unpaired in space)    All  Concentrations
CTDM(11083-E)


CTDM(11083)


Valley

COMPLEX I


COMPLEX II
                  m     s
                   a     a
                                     m
  3.3  27.4  0.90   1.37    3.2   0.99
                   (1.23)  (3.3)

  2.8  33.7  0.88   2.08    6.9   1.18
                   (1.70)  (5.2)

-41.0  25.5  1.63   0.24    3.0   5.52

-15.0  37.9  0.93   0.65    3.8   1.29
                   (0.53)  (3.2)

-60.1  99.7  0.94   0.38    5.0   1.09
                   (0.32)  (4.6)
                                            m     s     r
                                             a     a     a
 0.50  12.7  1.13
-0.51  15.4  1.01
-6.04  21.9  0.99
-5.61  36.9  0.99
Note: (  ) values of m ,  s  are calculated directly  from the
                      &   o
      C /C  ratios, with the C  = 0 values removed.
*Based on the 153 tracer hours from the CCB data base.   Arithmetic
 and sa statistics carry units of 10~°sec/m^ or  us/nr*.
                                     18

-------
m

8.
                3
       Q

       O

       +
0
•*4

•°       —I——
— ia u>  to   u>
X

_i
Q.
2
O
o
•4-
                                      X
                                      UJ
                                      O
                                      O
                       maorv (fi  in  T
                                                                 .  c
                                                                u>  o
                                                                    o
                                                                 .   IB
                                                                CO  «
                                                       CD
























^*.
«
d
^Q

I!







w
CO
00
o
o
&
i.
i.
CO
8
^
~
Q
(-
O








g c
a

=-
	 = '' '
' ^> • '

"*"

, ,' ' '
_ =
*, 1
8 8
+ •-'•• , *
















9 (9(S &&&&>&&

g £?
• CO
*- rM O
Cd t-H
'CD ' — 1
"! PH^
*~v3 )^i
S Q
cd H
•0 "
'> i— i
fa £-1 I— I •
. CD / — \
r- tO X tO
CR rQ W PH:
0 J S
CD PH O
' " V? ^g^ f-l
c. cd O
CD U tO
to PH Ln
• - ^ ,—{
W T*4 ^^ ^~*^
O
: X Cd
CO W 4->
^-f i-J cd
cd PH T3
' 3 S
ro O ' — i
•HO*
to
0 CD « U)
k • f-l >>I-H
' CD rrl
o s I-H en
i O i—l
' rr} ^^
         •I-

         co
         Q

         <3
                CD"

                co

                §
                Q
 —   r\i  .
                                      
8'


CB

S

(S
' CB
S
S"
&

S
(9
S
<9

S

CD
' ©
(M
St
©
*"

CD
b •
1
Snj
&
2.0
••» o
g-H

to
•H
4->
cd
4-1
CD
O
1

-------
     The residual statistics suggest that CTDM(11083-E) has the best  overall
performance.  The CTDM(11083-E) log-normal statistics (mg and sg)  are
closest to the desired value of 1.0, and the bias (ma)  and noise (sa) of
the set of paired concentrations are the lowest of all  the models.  However,
the summary statistics show that the performance of CTDM(11083-E)  is  not
much better than that of CTDM(11083) and COMPLEX I.   CTDM(11083-E) and
CTDM(11083) underestimate peak concentrations while COMPLEX I overestimates
them. For the three models, Sg is greater than 3.0.

     COMPLEX II and Valley simulate the observations much more poorly than
do the other models.  Both models substantially overestimate the peak con-
centrations, and COMPLEX II does a poor job in reproducing the distribution
of the tracer gas concentrations observed on CCB.

     The measure of model resolution identifies Valley as less responsive to
changes in meteorology than the other models,  but the resolution statistics
of the other four models are not significantly different.  One would  expect
CTDM(li083), CTDM(11083-E), and COMPLEX I to have similar resolution
measures on the basis of the scatterplots of Figure 2 discussed below,  but
COMPLEX II would be expected to be inferior.  This dilemma occurs because
the measure of resolution reduces to unity for a model  that produces  a  wide
range of concentration values compared to the range in  observed concentra-
tions, as does COMPLEX II.  Consequently, even though all the models  except
Valley have roughly the same resolution capability,  the three models  with
lower noise measures perform better.

     Scatterplots of peak observed concentrations scaled by the emission
rate (C0/Q) versus peak modeled concentrations scaled  by the emission
rate (C /Q) are presented in Figure 2.  Three of the models show quali-
tatively similar patterns, while two show patterns that are distinctly
different from the rest.
     The Valley model is not designed to use onsite meteorological
measurements, but uses "worst-case" meteorology instead.   Therefore, model
estimates of Cp/Q only depend on the distance from the source  to  the
nearest terrain feature at the elevation of the release.   At CCB, this  leads
to a relatively narrow band of Cp/Q values that is unlike  the  pattern of
the other models evaluated.  We see from the figure that Valley overestimates
most CQ/Q values, but it underestimates the seven largest  CQ/Q values.
This indicates that the standard "worst-case" meteorological conditions
contained in Valley for screening large power plant plumes are probably not
appropriate on the scale of the SHIS #1- at CCB.

     COMPLEX II is the other model with a distinctly different scatterplot
pattern.  This pattern is nearly the opposite of the Valley pattern.  Valley
concentration estimates cover a range much narrower than the range  of obser-
vations, while COMPLEX II estimates cover a range much greater than the
observations.  In both cases, model estimates appear to be poorly correlated
with the observations.
                                   20

-------
             VALLEY
              zra.e-r
         C0/Q ,„.,
COMPLEX!
 200.9
                        oe.aa  ise.ee  see.
                 COMPLEX I
                  286.8-1
                       sa.ee   iee.e
            CTDM(11083)
                                                              .08  288.08
                                                 asa.ea  4tra.ee  ^se.ea sao.ea
                                            CTDM(11083-E)
Figure 2.    Scatterplots  of observed and  modeled peak concentrations
              (C/Q, vs/m3)  for  153 hours of SHIS  #1 data.
                                       21

-------
      COMPLEX I,  CTDM(11083),  and  CTDM(11083-E) display similar patterns of
 scatter in that  the  range of  estimated and observed peak hourly concentra-
 tions are nearly the same,  and  the visual correlation between observations
 and estimates is much better  than that indicated by the Valley and
 COMPLEX II patterns.  Among these three models, COMPLEX I is biased toward
 overestimation;  CTDM(11083) and CTDM(11083-E) are somewhat biased toward
 underestimation  of C0/Q values  of less than 100.

      The geometric statistics for log-normally distributed residuals were
 obtained from the cumulative  frequency of occurrence plots shown in Fig-
 ures 3 through 7.  These plots  indicate that peak residuals from none of the
 models are strictly  log-normal, but they also show that all of the distribu-
 tions are nearly log-normal if  the "tails" of the distribution are ignored.
 These tails are  generally made up of C0/Cp values in which either Co or C
 is  nearly zero.

      Besides estimates  of mg  and  sg, these frequency plots may also be
 used to estimate the frequency witn which peak model estimates are within a
 factor of 2 (or  any other factor, for that matter) of the peak observed con-
 centrations.   22% of Valley estimates, 41% of COMPLEX I estimates, 21% of
 COMPLEX II estimates, 28% of  CTDM(11083) estimates,  and 44% of CTDM(11083-E)
 estimates are within a  factor of  2 of the peak observed concentrations.
 These figures  indicate  that CTDM(11083-E) and COMPLEX I ought to have better
 performance statistics  than the other models.   However, the cumulative fre-
 quency distributions and the other residual statistics show that all the
 models produce a substantial  amount of noise in their simulations.  There is
 clearly room for improving the reliability of these  models intended for use
 in  complex terrain settings.  There is a discussion of suggested improve-
ments  to  CTDM  and their basis in Section 3.

      To help understand the noise in the model calculations,  the residuals
based  on  the peak concentrations have been plotted against Cp and several
meteorological parameters.   These plots show where a model might be doing
 comparatively  better or worse, thereby indicating areas for improvement.
Figure  8  shows Co/Cp vs. Cp/Q for CTDM(11083),  CTDM(11083-E),  COMPLEX
 I,  and COMPLEX II.  A similar figure for Valley has  not been presented
because Valley model concentration estimates fall into such  a narrow range.

      The  figure  shows that many of the underestimates from CTDM(11083)  and
CTDM(11083-E) occur for very low values of Cp/Q,  and that  the tendency for
underestimation decreases with increasing Cp/Q.   The degree  of overesti-
mation also decreases with increasing Cp/Q.   These observations  indicate
that these models perform relatively better under those conditions that
produce  the higher concentration estimates.

     COMPLEX I and COMPLEX II do not  display this  trait.   These  models  tend
to  overestimate  the observed peak concentrations  to  nearly the same degree
for all Cp/Q, although the  number of  underestimates  decreases  with in-
creasing Cp/Q.

     Scatterplots of C0/C_  vs. wind  speed are  given  in Figure 9  for
CTDM(11083), CTDM(11083-E3,  COMPLEX  I and COMPLEX II.   There  is  considerable
                                  22

-------
                                                             oa
                                                               o
                                                             o
(9
05
«D
O
                                                                                           Ul
                                                                                           IS
              \
                   V
                         .^c
                 09
                 eg
                                                                                                         T*     H,
                 CO
                 in
                                                                                                                     OS
                                                                                                                     at

                                                                                                                     8
                                                                                                                     a
                                                                                                                     s

                                                                                                                     10
                                                                                                                     a>
                                                                                                                     o>
                                                                                                                     a>
                                                                                                                     in
                                                                                                                     a»
                                                                                                                     

                                                                                                    s
                                                                                                                    . OB
                                                                                                                     m
                                                                                                                             g-
                                                                                                                             cr
                                                                                                                             05
                                                                                                                             CD
                                                                                                                             £
                                                                                                                             3
                                                                                                                            O
                                                                                                                                       o
                                                                                                                                       •p
                                                                                                                      •H
                                                                                                                      t— l
                                                                                                                      S-
                                                                                                                      &
                                                                                                                      x
                                                                                                                      (D
                                                                                                                       PH
                                                                                                                      U

                                                                                                                       O
                                                                                                                                       q-i
                                                                                                                                       o
rH  tfi




X O



0>  M
    LO
                                                                                                                      M-l  oj

                                                                                                                       (U  ni
                                                                                                                       > "d
                                                                                                                      •H
                                                                                                                      +-> i-l
                                                                                                                       Cti ^
                                                                                                                                          CO
                                                              a.
                                                            O
                                                                     23

-------
   5.M
   .82-
   .01 •
                                7

                                        V

                                                                          2.88
     • 01    .1    .512   5  18   28  30 43  SB 63 70  6C   08  05   98 89 00. S  00.0   00.80

                             Cumulative Frequency (%)

Figure 4.   Cumulative  frequency plot of  C0/CD for COMPLEX  I  applied to
             SHIS #1 data (153 hours).
                                       24

-------
 1.80-
                        7
                                 i
                                      L
                                             7
                                                  7
    .01     .1   .512   5   18  28 38 48 58 68 78  88  OB  SS  86  88 00.5  08.8  B8.08

                            Cumulative Frequency (%)

Figure  5.   Cumulative  frequency  plot of Co/Cp for COMPLEX  II  applied
             to  SHIS #1  data (153  hours).
                                     25

-------
268.09











Cn/CD
o p









gg-



.18-



.es-











































































































































/
































/































i
1































/
/





























ft
l>
/






























1













I



















/

































/

































/




























/
/
1 1
/
7


















r
I :
/





/
/
/
/
































/

































/






























/
f






























/

































/
'































1
f

































- see.aa








- sa.ee





p





2.ee
1.08




.50

.28

Figure  6.
     1  2   S  IB   20 30 48 50 68  78 SB  0885  08 00 00.5  08.0   00.00

                Cumulative Frequency (%)

Cumulative frequency plot of C0/Cp for  CTDM (11083)  applied
to SHIS  #1 data  (153 hours).
                         26

-------
                    I
                  t
 • 81    .1   .512   5   IB  2B 30  « S3 68  7B 88   OB  85   98 BS 00.S  90.0   04.00

                        Cumulative Frequency (%)

Figure  7.    Cumulative frequency plot of Co/Cp for CTDM (11083-E)
             applied  to SHIS #1 data  (153 hours).
                                  27

-------

uT
CO
§
T™
i
Q
\5




















*»* ^


*









j_i^
CO
§
O


















t

















*



f 4


* *
* +
* ** * *
* * *• * * *
*, " • <£*!****« **.***. **


*• *v - '
ott
X
o
o




















-


•
* . '
» , *
^ *
•« '•,***
\i *£#$**'*• *•* 7** *
• • • Ul m
n N -
*j
X
o
o
I
s
I

z
s

I
"
* (T
S p


3 S,
15 a
s XQ
1 °

s
s

s
'i

$
"a













3
S

g
'$
s
'S


s
'i


3 — .
i «E
1 S-
5
1 °
§
5
i
•§

g
'a

*




i
Q-
2
O
o





















•• + t
B
D











^
X
LU
COMPL

















it
>



* -
* .

+

*


+
*


*
* ^
* * +
+
•*• +
* * *
* + 4-
t ** *

+ * *
* *+ + +
*++ *+ **+*+•*• + *
V. *.***** ^ :\ *
+ * * * *•'!**** + + * ^> *
**i* *+*• ,.
a a 01 in m —

oa
X
o
o





















t
*
* * . *
+ i-
»»*' *»>* *(.» ,.
. »*' *.*»**
t*V^* ::*"**** * * '
* »* >?>*; „ * §• -* » ** * * * t
Q.
O
o
O
3
i
1

i
s
1

S

S CY^
•| E
X
tn
s 3
% O ^
a. a)
S ° ft
s
*=

'a co
ffi
• UD
S
o
s
a tn
o
•H
« ^J
. • cd


(D

0)
O
e
i
0
•p
1 1
s rci
0)
« fH
"a a)
.« "o
MH
o
s
i pj
0
. -H
* -^T" j_l
- « ^3 ^"^
S E rt
• i 'S
^ a >
g Q
"i
OO
S (U
-I g
1 -H
r~T-t

s
"s
s
'










t/)
PI
o
•H

nj
ij
PI
O
O
0


-------
LU
* CO
* * 00
. ' O

+ + *— *
5
, * t a
* * 0
* *
* *
* ***
++
** ***
•
•*• + *
* *
* * % * *
+ *
* * *
* + + * *
* * * *******„*
* ***
** * * * ** + * *
***+ * * * *
** * ******


.****'
-t-
* ***** * *
*

« «••»". -.
*
 ++++
* * + *
++++ * +
** +. / * +
. + +
* * +* +•

, * * .*

* * *****


S S S ? " "
S

—

s


.1
.8 §
" S
•H
.8 — «

E S
.8 "= ^H
W 4^: •
/ — \
g CO ^3
' ^ ^

.8 
' ' .. t
	
CO
• 00
o
,_
I
* , ^~
'* •*• *
++ * -f
+
h + t +
+ *+ *
•*• -fr
' + ' +
* + * ' *
* + *• +
+
+ *

* + +- *
* i *
* ** »*
* * * ,* ** *
****** *
4-fc "*" +
* ** * *****
. * *» » *
*** * *
+ * +
+
•. k k k ". °\ '-.
a.
. CJ
-
S
a

s
ei
s
°

§

g 	
S In
£
-? 3
u

.i


8
m
8
$
8
'
CD S
i ^ d%

/ ' '" ; "' •" "i
a.
* * * v 1
» CJ
+ * +
* * * * *

» »* * * * *
****** * * *
+
+ +
+ * +•
* *
* * **
» * * *
* ** *
*****
+ + +• 4-4-
* ** *
**+ + *
***** * * *
*******
** * **** * **
* *-
** *
+ * * *
i i i i 	 1 	 1 	 • —
a at m ui m — ,
a.
O
= fn 0
CD S
2 CO O
• rQ O
O
fn
g 4-4 CD
d O O
Oj
S OP
• -H
P X
s Cti i— 1
* -H f-t
^ o
J « ^" ^
§
S3 .
01

£ CD

bO
S .H

I
i

^o >
" • o
29

-------
 scatter  in all of the plots, but some trend can be seen in the patterns.
 CTDM(11083) exhibits a distinct bias toward underestimating observed peak
 concentrations for wind  speeds in excess of about 5 m/sec.  Because Hc is
 probably small (or zero) compared to the source heights when source-height
 wind speeds are as great as 5 m/sec, this suggests that the Lift component
 of CTDM(11083) is underestimating the amount of plume material on the
 surface  under the more nearly "neutral" flow conditions.  When crz is
 enhanced as in CTDM(11083-E), the figure shows that much of this bias at
 larger wind speeds is reduced, although it is not eliminated.  Therefore,
 the Lift component may need revisions to account for the combined effects of
 flow distortion and plume dispersion more realistically.

     COMPLEX I exhibits a bias towards overestimating peak observed
 concentrations at the lower wind speeds.  COMPLEX II appears to exhibit the
 same behavior, except that a few large underestimates also occur at light
 wind speeds.  The overestimates for light winds may be the result of using
 PGT CTy values in COMPLEX II, and 22.5° sector-averaging in COMPLEX I
 (e.g., see Figure 12 in  the CTMD Second Milestone Report).  At very low wind
 speeds,  the wind direction often underwent large variations at CCB.  The
 22.5° sector within COMPLEX I may underestimate the plume meander in these
 conditions, and thereby consistently overestimate concentrations on the
 hill.  COMPLEX II would also certainly underestimate the meander, but its
 narrow Gaussian plume might also nearly miss the hill at times,  thereby
 producing both the underestimates and the overestimates indicated in
 Figure 9.

     Scatterplots of CQ/Cp against other modeling parameters also
 reflect  the patterns just described.  For example, u/N (Figure 10), the
 ratio of the mean wind speed to the Brunt-Vaisala frequency,  distinguishes
 between  the more "stable" and the more "neutral" hours, and the patterns  of
model performance are similar to those discussed above for the plots against
 wind speed.  l-Hc/zr (Figure 11), where zr is the plume release height,
 orders model performance in the near-neutral limit when 1-Hc/z- is greater
 than approximately 0.5,  and in the very stable limit when l-Hc/zr is less
 than zero.  Figure 11 indicates that CTDM(11083) and CTDM(11083-E) are most
 prone to overestimate peak concentrations when Hc exceeds 0.5 zr, but is
 less than 1.5 zr, and CTDM(11083) generally produces underestimates for
 HC less  than 0.5 zr.  The bias toward overestimating peak concentrations
 with COMPLEX I increases as HC increases.

     Figure 12 contains scatterplots of Co/Cp against l-Hm/zr, where
 Hm is the elevation of the peak observed concentration.  This plot shows
 that CTDM(11083) tends to underestimate peak concentrations when they occur
 appreciably above the release height—yet another indicator of more nearly
 "neutral" flow.  Enhancing az in CTDM(11083-E) tends to lessen this
 trend.  The corresponding plots for COMPLEX I and II (not shown here)  show
no discernible pattern.

     Figure 13 contains a plot of Co/Cp vs.  the product of the cross-
wind vertical and horizontal turbulence intensities  for CTDM(11083)  and
 CTDM(11083-E).  (The other models do not use these data).   Large turbulence
 intensity products indicate a relatively large dilution of plume material-
 The figure indicates that modeled and observed peak concentrations most
                                  30

-------
Ill
ci
00
o
5
Q
0
*

+
*


*
•f
* + +
+
+
++• ^
+• **• *
**,* ** *
* *
•*• + •*" + 4- +
****** **
* * * *** V, /* *
* *+* *„*..*.* *:****,* * *
f+* H-* ***
 / '* { * *
i — i 	 ( 	 1 — i 	 , 	 . —
X
'i
1
s
'1
s
•s §
S
•H

1 g
E S
•; z i-:
D
en
°. t — i

C/D
• '4-1
1 0

.J J •
cd 3
•S *
2 rj
TTJ 4-J
H
0 0 0 4J
•P cd
i j_)
"Td -P
•t & fi
^ • > 
+* * *
* . f
+
* * *
. ++ **^
* + * +
. + , *
* + + •«• +,
. ++ +
* * * *
,** * *
+ +*+ ++*•«- »
* +*** „*+**
* * »,*** * •**, * * ***
t * /»,**. ** *
. „ ^
O
i
g
1
X
i
t
I

i
i
*• I
1 x
3
I

S
i

I

g
s
a
i

X
a!
I


*
»
*

»
»
» ** *
** *
+ f
* + + ^
* *• + +. *
v * %r **" ** **
* * » »* » * ,.*
*+ *++* + **+ + **+
**** > *"**** *,*,** * * * *
oa
'I FH 0

-------
                                                                                                 CO
                                                                                                 n
                                                                                                 ffi
                                                                                                 co
                                                                                                 • H  ffi
                                                                                                 +J   I
                                                                                                  tD -H
                                                           o
O
                                                           O
O  C
S  O
 I  -H
O  4J
                                •s,
                             .8
                                                                                   O
                                                                                       .S
                                                                                                     -P
                                                                                                     O
                                                                                                  O  -P
                                                                                                  (D
                                       32

-------
    3.8


    2.8
        CTDM(11083)
                                *   *
    3.8-

    2.8-
C0/CP
        CTDM(11083-E)
  Figure  12.    Variation of observed-to-modeled ratios of  SHIS  #1
                maximum hourly tracer concentrations with l-Hm/zr.
                (Hm is the height  of the observed concentration.)
                                 33

-------
I8.B-
N
2.8-
i.a-

.«-
.3-
'P


.1 -

.8-
.1
IB.8-
3.8-
2.8-
i.e-
.5-
.3-
•*
.1 -


.8-
.e
•V
f
+ +
k'
+1?
+
++
+ •<
+
+ +
•h
+
*
*+

+
+

18

f+
+
i;
ir
rf+*
%4
>+
V*
+
+
+
+
*

10

+
+
+
+
+ +
% ++ ' +
* + +
+
•f +
•
+
+ +
+
+


+
+
CTDM(11083)
.85 .18 .IS .2
iyiz
+ +
+
+ +
t / * + •
+
+ +
++ + +
+
+
+

- , CTDM(11083-E)
.es .ie .is .21
iyiz
Figure 13.   Variation of observed-to-modeled ratios of SHIS #1 peak
             hourly tracer concentrations with the product of the
             turbulence intensities i  and iz.
                                34

-------
nearly agree when the plume is well-diluted.   When the dilution is  much
weaker, the plume is more compact,  exhibiting considerably less meander.
Under these conditions, the peak modeled concentration is very sensitive  to
plume path assumptions, wind direction,  and postulated flow distortion/plume
dispersion effects.  This sensitivity is illustrated in the figure  by  the
large scatter for low values of iyiz.  The figure also shows that the  bulk
of the CCB data falls into this category.
                                 35

-------
                                 SECTION 3

                     DEVELOPMENT OF THE  COMPLEX TERRAIN
                          DISPERSION MODEL (CTDM)
3.1  Introduction

     CTDM is a point source plume model that incorporates  several concepts
about stratified  flow and dispersion over an isolated hill.  The emphasis
to date has been on including those phenomena that are  thought  to be  impor-
tant in controlling the magnitude and distribution of plume concentrations
across the windward portion of CCB.  However, although  the formulation of
modeling concepts contained in CTDM has general  inherent applicability, cer-
tain details make explicit use of the geometry of CCB to  simplify the code.
Consequently, CTDM should be viewed as a research code at  this  time rather
than a code suitable for regulatory applications.

     We expect CTDM to evolve during the CTMD project.  As more data  are
analyzed from SHIS #1, and as data from SHIS #2 are analyzed, the specific
formulations will change as will the range of phenomena contained in  the
model.  These changes will be treated as updates to CTDM  rather than  the
basis for new model nomenclature.

     CTDM evolved from model codes described in the first  two Complex
Terrain Model Development (CTMD) milestone reports.

     Two preliminary models were described in the First  CTMD Milestone
Report.  The Impingement model corresponds to low Froude-number flows in
which a plume remains horizontal as it impinges on and  flows around the
hill.  The Neutral model corresponds to moderate or large  Froude-number
flows in which a plume goes over the top of the hill.   A number of  refine-
ments were made to these two models, and these were described in the  CTMD
Second Milestone Report.  These refined versions of the Neutral and Impinge-
ment models are referred to as the Lift and Wrap models.

     Since then the Lift and Wrap models have undergone additional  refine-
ments and have been combined into one model; now CTDM performs  both kinds of
computations whenever the dividing-streamline height,  H^,,  is non-zero.
This section of the report provides a description of how Lift and Wrap were
modified and combined to form version 11083 of CTDM.  The description is
followed by a detailed study of the data from two SHIS  #1 experiment  days.
CTDM upgrades suggested by the analysis of these data are presented,  and  the
performance of this latest version of CTDM is evaluated.
                                   36

-------
3.2  Description of CTDM(11083)

     CTDM treats the flow around an isolated hill as though the flow takes
place in two discrete layers.   Flow below the critical dividing streamline
(1^) is restricted to travel in horizontal planes around the hill.   Flow
above HC is allowed to rise over the top while undergoing some degree of
distortion consistent with weakly stratified flow over a sphere.   That is,
Lift modeling assumptions are employed for plume material which is  released
or which diffuses above H,,, and Wrap modeling assumptions are employed for
plume material which is released or which diffuses below HC.  Because
streamline patterns above and below Hc are markedly different,  the  plume
is assumed to separate into two independent segments where the Hc surface
intersects the windward face of the terrain.
     The segment initially below Hc spreads around the sides of the ter-
rain.  As the plume grows with distance beyond the point of separation,  how-
ever, its vertical profile is no longer Gaussian (or reflected Gaussian) in
shape.  This is because the material above Hc at the point of separation
is no longer available to mix below HC.

     The segment initially above Hc continues to travel up and over the
hill, but the plume material that had been below HC at the point of sep-
aration has been displaced by the solid terrain surface.   The vertical pro-
file of the upper plume segment therefore is no longer a simple reflected
Gaussian distribution; rather, it is a reflected segment of a Gaussian dis-
tribution profile.

     3.2.1  The Wrap Component of CTDM(11083)

     Because observations indicate considerable meander of the horizontal
wind during a typical hour, we assume that the hour can be divided into
several quasi-steady periods.  We also assume that the average angular
spread of the plume, as, corresponding to these periods is small com-
pared to the total meander, am, during the hour.  Concentrations are
assumed to occur on the hill's surface only when a portion of the plume  is
directed along the stagnation streamline; therefore, the maximum concentra-
tion occurs at the stagnation point.

     During any single quasi-steady period, denoted by the subscript i,  the
concentration at the stagnation point A (see Figure 14) will be non-zero
only if the mean wind direction during the period lies within otg^/2 of
the stagnation streamline (9^ = 9^ ± as£/2).  If this concentration is
denoted by C^(9,9d), then the average concentration, Cmax, at A over an  hour
is given by
     c    = J c.(e, e,) p(e) de
      max   'id
            O
(13)
where P(6) is the hourly probability density function of wind directions.
If the wind speed, vertical plume spread, and horizontal plume spread angle
are nearly constant among quasi-steady periods during the hour, then the
only non-zero contributions to the integral in Equation 13 arise for wind
directions within ± as/2 of the stagnation wind direction.  Furthermore, if

                                  37

-------
              Receptor
Source:  (xr, yr, zr)
                              Stagnation Streamline
  Figure  14.    Plume  dispersion in the  region of horizontal flow.
                                38

-------
the concentration distribution within the plume during each quasi-steady
period is assumed to be uniform,  and if the spread due to plume meander
during the hour is much greater than as, so that P(e) is nearly constant
within the interval ots, then
     C    = C.   f
      max    i
P(9)d6 - C. P(9_) a
          ids
                                                                         (14)
where
     C. =
      i
          /2Tra U a d
              z o s
                                                                         (15)
In Equation 15, az, UQ, andas represent averages for the relevant hour.

     From the CCB experiments we do not have a continuous record of 9(t)
as a function of t; only 5-minute averages of 9  and iy are available.
To use this _information, we assume that.9  is uniformly distributed in  the
interval 2 S3dQ about the 5-minute averaged value 9,  and we set 
-------
 sources,  and integrate the response function or Green's  function  from the
 level surface to Hc.   Consequently,  the "vertical" distribution of plume
 material  lies along the surface of  the hill,  and reflection occurs only
 along the plane  z = 0.

                   P(8.)  Hc C (d,z)  -0.5((z_  - z)/0')2
      „/•_,.      •>       d  -of         R        z
      C(d+s,z)  = •- S           (e
           a,R
                      a  o
/2ir
-------
                                                     H
 Section
   (§)
(enlarged)
View
                                         Undistorted Plume

                                         Receptor (XR/ yR, ZR)
Section
                                                    View  (A
                                  Source: (xr, yr, zr)
 Figure  15.    Geometry used to derive  the  effective radius of
               curvature Re for the portion of the hill above Hc.
               and to incorporate the wind  direction probability
               distribution function P(6) in the Lift model.
                                 41

-------
By assuming  that the flow is axisymraetric, the relationship between the
undistorted plume and the distorted plume depends only upon the section of
arc  through  the receptor (Section A, Figure 15).  In Figure 15, the angle
for  specifying the probability factor is
     6 - arcsin  (yp/s) + e0

where 6O is the mean wind direction.
                                                                    (21)
     The  third feature of the Lift model as incorporated into CTDM(11083)  is
the treatment of the evolution of the vertical distribution of plume ma-
terial.   As drawn in Figure 16, plume material below Hc is removed from
the profile at the Hc impingement level, and the remaining distribution of
material  above HC continues to diffuse with full reflection from the hill
surface.  To obtain concentration estimates beyond the point (so) where
Hc intercepts the hill along direction 6,  the distribution of material
above HC  at so is treated as a set of point sources,  and the total con-
centration at downwind receptors is found by integrating the response func-
tion for each point source from Hc to infinity.
„,    „,   r" CQ(S>2)
C(s,9,0) = J   — —
          H   /2ir a1
                                                      dz.
                                                              (22)
As in subsection 3.2.1, the solution to this integral produces a series  of
product terms of error functions and exponentials,  and
     (a')2 - (a (s))2 - (a (s ))*
       2       2          Z  O
                                                                   (23)
The distribution of material at s0 is given by
               -0.
           • 2ir U a (s )
                02  o
                          2  - z)/o  )
                           r       z
                                               -0.5((z
0/0  )'
                                            + e.
                                                                   (24)
     3.2.3  Dispersion Parameters

     When the PDF form of CTDM(11083) is selected,  only the dispersion  para-
meter for plume spread in the vertical is needed.   Background  information
and analysis methods employed in developing the az  expression  used  for
CCB data are described in Section 3.3.  We find that values of az cal-
culated from the expression
iz(s
                     N(s
                                  )/(U02y2))
               (25)
with 2y2 = 0.55, fit the CCB data quite well.   N is the Brunt-Vaisala fre-
quency and sv is the virtual source distance associated with source-
                                  42

-------
         REGION 2
Figure 16.   Dispersion and flow regions for stratified flow
             around hills.
                            43

-------
induced plume spread.  If the PDF form is not selected,  the dispersion para-
meter for horizontal plume spread is assumed to be given by
ay = iy(s
                 s'v)
(26)
where s'v is the virtual source distance based on the rate of plume  growth
in the horizontal.

     3.2.4  CTDM(11083-E) and CTDM(ll083-E)-5

     The structure of the basic version of CTDM(11083)  was discussed in the
preceding sections.  Additional versions of CTDM( 11083) have been developed
to test the model sensitivity to alternate hill-effect  assumptions and
alternate meteorological data resolution.

     CTDM(11083), as described in subsection 3.2.2,  presumes that streamline
distortion and the accompanying plume distortion for flow over the top of
the hill (Lift) do not cause any increase in diffusion  over that  which is
expected at similar downwind distances over level terrain.  The plume cen-
terline approaches closer to the hill surface, but the  vertical extent of
the plume is proportionally reduced, so concentrations  at the surface remain
unchanged.  This model for flow over the top is most appropriate  if  enhanced
diffusion due to the distortion of the streamlines can  be neglected,  and if
the flow over the hill is no more turbulent than that away from the  hill.

     The effects of increased diffusion (aided by the plume's approaching
nearer the hill surface) can be approximated in CTDM by enhancing the rate
of growth of az over the hill, without explicitly bringing the centerline of
the plume nearer the surface.  The effect of this approach on the vertical
distribution factor is similar to the terrain factor approach contained in
the COMPLEX models and the Neutral model evaluated in the CTMD First
Milestone Report.  But instead of viewing the factor as a change  in  the cen-
terline height, we view the factor as a change in the relative size  of the
plume.  However, because az increases, dilution also increases, which
reduces the centerline concentrations in the plume.   Therefore, this treat-
ment is not equivalent to the terrain factor approach,  particularly  as
az approaches the height of the plume centerline over the terrain.
     The a --enhancement approximation is made by altering Equation  23:
             (a (s))
                                                                         (27)
where
     T = PPG
           for
                     ZR- zr
     T = 1 - (1 - PPC) (zn - H )/(z  - H )
                         R    c    r    c
                                                                   (28)
                              for
                              z  <  z
                               R    r
                                  44

-------
PPG is the "plume path coefficient."  This definition for the factor T is
the same as that used to define terrain factors in the COMPLEX models.  This
enhanced-az version of CTDM(11083) (denoted as CTDM(11083-E)) has been
run on the CCB data base with PPG set equal to 0.5 in order to evaluate the
importance of increased dispersion over the hill.

     A second alternative CTDM(11083) utilizes the sequence of 5-minute
meteorological data, rather than 1-hr average meteorology supplemented by
the calculated PDF for wind directions. , In this version, the 5"-minute
horizontal distribution of plume material  is assumed to be Gaussian in form,
characterized by
     a  = i(5-min)
      Y    Y
(29)
All other model formulations remain the same.  Because 1-hour average con-
centrations are constructed from calculated 5-minute average concentrations,
variations in wind speed dilution, turbulence intensities,  and the dividing-
streamline height are explicitly modeled in addition to the distribution of
5-minute average wind directions.  Running this version (denoted by append-
ing (5)) of CTDM(11083) on the CCB data base will test the  importance of
including short-term variations in Hc, u, and turbulence intensity in the
modeling of SHIS #1.

3.3  Case Study Results - Experiments 201 and 210

     The model evaluations discussed in subsection 2.4 indicate that some
improvement is gained in the performance of CTDM if az is enhanced over
the hill.  However, much "noise" in the residuals remains.   One possible
source of noise is the meteorological data used to drive the model esti-
mates.  The MDA described in subsection 2.1 represents an objective best
estimate of the meteorological conditions at source height  derived from data
measured at Tower A alone.  However, alternate meteorological data for model
input might be developed for some experiments if all available data are
studied.  Note that Tower A was located 2.3 km to the north of CCB, approxi-
mately one to three kilometers away from the tracer release locations.

     We have begun an extensive case-study of the SHIS #1 data to develop
all available information on the transport and diffusion of plume material
at CCB.  In this process, the consistency among the data will be tested, and
the performance of various modeling assumptions can be better evaluated.
Because the modeled concentrations are very sensitive to the trajectory of
plume material and the rate of diffusion of that material to the surface, we
are particularly interested in evaluating (1) the correspondence between the
sequence of 5-minute average wind directions interpolated to source height
on the basis of Tower A measurements and the smoke plume trajectory as
recorded by observers' field notes and photographs; (2) the characteristics
of the plume trajectory near the hill; (3) the correspondence between the
rate of plume growth in the vertical as observed in photographs and the
measured vertical turbulence intensity on Tower A; (4)"the  relationship
among turbulence intensity measurements on the hill and those at Tower A;
and (5) the form of the growth of az upwind of the hill as  derived from
                                  45

-------
 a combination of lidar data and photographs.   Results  of  these evaluations
 performed on data from Experiments  201 and  210 are  summarized in this sec-
 tion.

      3.3.1  Plume Trajectories

      Key views of the oil-fog plume that are  important in estimating plume
 trajectories from photographs are those taken from  behind the release crane
 looking along the plume trajectory  and those  taken  from atop CCB looking
 back toward the release crane.   For those views from behind the crane, wind
 directions are most easily estimated when the plume passes over some
 recognizeable portion of the  hill.   Any horizontal  displacement of stream-
 lines when the plume is close to the hill is  subjectively taken into
 account.   For views from the  hilltop,  wind directions  are most easily esti-
 mated when the plume is seen  to pass over one of the camera positions on
 either peak,  or when the plume  trajectory "passes through" a known landmark
 on the ground away from the hill (such as a turn in a  road).  On the basis
 of these  considerations,  wind direction estimates from the photographs are
 least accurate when the plume misses the hill to one side or the other.
 Fortunately,  the most useful  information during these periods is that the
 plume is  off to the side of the hill,  and is  therefore having very little
 impact on tracer concentrations in  the  sampler array.

      In Figure 17,  photo-derived estimates of wind  direction at the source
 for each  available  5-minute period during Experiment 201 are compared with
 the MDA wind  direction estimates and also with the wind direction measure-
 ments  at  the  10-ra level of  Tower B,  which is  located on the south peak of
 CCB.   The first  2-hour segment  of data  displays good general agreement in
 wind direction,  but  the period  between  2000 and 2200 displays more scatter.
 In particular,  the  various  wind direction estimates would lie nearer one
 another if  the Tower A estimates were shifted by approximately 10 minutes
 between 2020  and 2030.  Because the  plume was released 952 m northwest of
 CCB, the  release point is nearly as  close to Tower A as it is to  Tower B.
 Furthermore,  because the  wind speed  exceeded  3 m/sec during the interval,
 events  at Tower A would be  swept across the source area within about 5
 minutes.  Therefore,  the  discrepancy between  the Tower A data and the infor-
 mation  inferred from oil-fog plume  photographs and measured  at  Tower B is
 probably  due  to  spatial as well  as  temporal inhomogeneity in the  flow.   For
 modeling purposes, it  is evident that Tower A wind directions do  not accu-
 rately  depict plume  transport for this period.

     Figure 18 displays the time-series of 5-minute  average wind  directions
 for Experiment 210 as  estimated from photographs,  as interpolated to release
 height on the basis of Tower A measurements,  and as  measured  at the  10-and
 30-m levels of Tower B.  These results indicate that the 10-m level  of Tower
 B is most representative of the oil-fog plume  transport directions for the
 three hours of available data from this experiment.   Tower B  wind direction
 data from the 30-m level are generally within  5-10°  of  the "photo" direc-
 tions for the remainder of the experiment.   The wind directions inferred
 from Tower A data correspond well with the  "photo" directions at  times
 during the experiment but are  off by up to 20° at  other times.  This  rela-
 tive lack of correspondence is probably the  result of relying on 80-m  level
wind direction data to supplement the incomplete wind data for  the 40-m

                                 46

-------
    ^

    §
o   ^
O  
                  
                                                       S

                                                       LA
                                                                   CD
                                                                   CD
                                                                                                                            B
                                                                                                                            o
                                                                                                                         SU
                                                                                                                     PH  0
                                                                                                                     nj  oi  f-i
                                                                                                                     •P  0  13
                                                                                                                     ni  fi  5)
                                                                                                                     13     nj
                                                                                                                          "  0
                                                                                                                     to
                                                                                                                     0 -H
                                                                                                                     (D
                                                                                                                            13
                                                                                                                            fi
                                                                                                                     0  fti3
                                                                                                                     S  oj  0
                                                                                                                     •H  f-i  -p
                                                                                                                     •P  bO cti
                                                                                                                         O  .-<
                                                                                                                     Pi  -P  O
                                                                                                                     O  O  ft
                                                                                                                    •H
                                                                                                                     P  ft
                                                                                                                     O
                                                                                                                     0  B
                                                                                                                     fH  0
                                                                                                                    •H
                                                                                                                            0
                                                                                                                     r!
                                                                                                                     •H
                                                                                                                            T3

                                                                                                                            g
                                                                                                                        •H  pa

                                                                                                                    <4-t  0  fH
                                                                                                                     O 13  0

                                                                                                                     c   •> o
                                                                                                                     O i-l  H
                                                                                                                     tn o
                                                                                                                    •H CM  g

                                                                                                                     nj -P  f-t
                                                                                                                     ft  S  Mn

                                                                                                                     IIs<
                                                                                                                     CJ -H
                                                                                                                     f-l  fH  O  f-l
                                                                                                                     0  0  i-H  0
                                                                                                                     P  ft     S
                                                                                                                     C  X  +J  O
                                                                                                                    h-1 w  nj E-1
                                                                                                                     0
                                                                                                                     !H
                                                                                                                     g)
                                                          47

-------
          WD (deg)
          158.01
          140.0-
          138.8-
          120.0-
          110.0-
          188.0'
                      1       2      3

                      Photographs (oil-fog)
                      Modelers' Data Archive
7      8
  Time (hr)
          WD (deg)
          150.01
          140.0-
          130.0-
          128.0
          118.8-
          100.0 liiiiiiniiiii
               0       t
                      Photographs (oil-fog)
               	Tower B @ 10 m
               	 Tower B @ 30 m
7      8
  Time(hr)
Figure 18.    Intel-comparison of wind direction  time-series data for
               Experiment  210, derived from photographs,  wind measurements
               at 10'and 30  m from  Tower  B,  and interpolated wind
               measurements  from Tower A.   ..                   .,   '
                                        48

-------
level on Tower A.  Modeling for this experiment must therefore consider the
wind directions measured on Tower B and those inferred from the photographs.

     3.3.2  Vertical Turbulence Intensity and Initial Plume Growth

     Experiment 201

     Values of the vertical component of turbulence intensity, iz,  have
been estimated; from available photographs by several methods and compared
with iz values measured on Tower A.  Photographs used in the analysis for
Experiment 201 are wallet-size enlargements from 35-mm film.  Lengths
measured on each photo are converted to angles by the formula:
      a = arctan (&('mm)/113 mm)
                                                                (30)
where'A is the measured length, and 113*mm is the effective focal length
for the enlargement (50 mm •  2.26).  The enlargement factor of 2.26 was
estimated by comparing the size of the dominant features on the film and the
enlarged print.

     These angles are converted to full-scale distances by means of the
geometry depicted in Figure 19.  The., length C is the known distance between
the camera position and the release crane; A is the distance between the
release crane and any point downwind along,the wind trajectory;  B is the
distance between this point and the camera position; and the sum a^ +
a 2 is. the total angle a subtended by A as viewed from the camera
position.  Because this angle can be large, the small angle approximation to
Equation 30 is not .used, and all distances along the film plane are measured
from the center of the image.  Hence, two measurements are needed for a.
The angle 3 is obtained from the wind directiolfand the angle' frbm the
camera position to the release crane.  With these definitions, the  distance
downwind of the source is
     A —
  C sin (ex)
sin (180-a-3)
                                                                         (31)
Similarly, the thickness of the visible oil-fog plume is obtained from the
photograph as an angle, 
-------
                                     Wind Vector
                                             \
                                                   Source Position
                                           B
Camera Location
                                                                  Centerline of
                                                                  the Photograph
                                                                      Evaluation
                                                                      Point
       Figure 19.   Definition sketch for computing distance along the
                    wind trajectory  from the source.
                                       50

-------
      zl
a /x = E/2V/J .
 z        '
                                                                         (33)
     Method 2.  This approach utilizes the same principle as  Method  1, but
instead of measuring the spread angle at the source,  we measure  the  plume
thickness, h, at several distances along the plume and set the spread angle,
E, equal to the maximum value of h/x.

     Due to the uncertainties introduced in using the MDA wind directions
for the respective 5-minute periods,  photo-estimated  directions  are  used to
obtain h(x); and instead of assuming a top-hat distribution to relate E to
iz, we have assumed a bi-linear "triangular" distribution because the
smoke density is not uniform across the plume.  In this method,  the  vertical
intensity of turbulence is given by

            °           _                                               (34)
               max
     Method 3.  Due to uncertainties in defining the edge  of  the plume and
in relating the az estimates to a Gaussian distribution, Method 2 was
judged to be of marginal utility.  Method 3,  used in place of Method 2,
applies the technique of Gifford (1980)  to relate observed plume dimensions
to equivalent az estimates, assuming that the plume spread is Gaussian.
This approach provides an estimate of a_ for each observation of h when
                                       £i
a maximum value of h is specified for each plume.

     Ideally, the plume should be photographed against  a uniform back-
ground.  It should be uniformly illuminated,  and it should be visible at
distances large enough to define hmax.  Most of the nighttime photographs
from Experiment 201 were taken with a fairly uniform background, but the
lighting is not uniform and hmax is generally not clearly  established.
However, most of these photographs show a slowly-growing- plume at the larger
distances downwind, so we have set hmax = h(xmax) + 1 (rounded to the
nearest whole meter) for those cases in which hmax is not  clearly attained
within the field of view of the photograph.
     With this assumption, each value of h measured on the  photographs gives
rise to an estimate of az for an effective Gaussian distribution of
plume material.  The turbulence intensity is then calculated  with
     i 0 = (CT /x)
      z 3     z   max
                                                             (35)
Equation 35 once again assumes that the vertical growth of  the  plume with
distance (time) is linear in the initial stages  over which  az/x should
attain its maximum value.  Due to the assumptions required  for  this method,
it should provide "upper bound" estimates for iz.

     Method 4.  This approach makes use of the estimates of az  derived
by means of the Gifford technique,  but it obtains iz estimates  by matching
the observed values of a,, (t) to
     az(t) = awt/(l
           Nt/p)
                                                                        (36)
                                 51

-------
at  t
            ax-
     Equation 36 is the model for az that  was  used  in  the
Second Milestone Report.  In that report, p was a free parameter that was
set equal to  1.5 on the basis of comparing az estimates with az values
obtained by the WPL lidar system at CCB.
to solve for p with crw = iz3*U and az(t)
approximately 0.6.  Consequently,
                              However, when Equation 36 is inverted
                               az  (Gifford), p  is  found to be
z (Gifford)
                             NX
                                  o.s
      'z4
                x
                     '
                 0.6u
                                                                         (37)
                                       x ^ 0.5 x
                                                max
where xmax is the,farthest downwind distance at which a value of h can be
estimated.                                                 ;

     Figure 20 (top panel) shows the correspondence between izi and the
30-m iz value interpolated from Tower A data.  The intensities nearly
agree during the first two experiment-hours and during the sixth hour,  but
show significant differences during the fourth and fifth experiment-hours.
In these two hours, izj_ values appear to be either too small  or too
large.  It should be pointed out, however, that the larger values of izi
occur during periods of significant horizontal plume meander  and that this
horizontal "smearing" could appear as an increase in the vertical dimension
of the plume in the photographs.

     The 30-m interpolated iz value from Tower A is also compared to iz3
in Figure 20 (bottom panel).  The iz3 values are moderately greater than
the Tower A values during hours 18, 19, and 23.  But they agree better dur-
ing hours 21 and 22, with the notable exception of the peak at 2100.   Al-
though the peak in the "photo" iz's may have an artificial component as
has been speculated above, it is unlikely that the peak should be entirely
absent.  In fact, it is tempting to shift the plots in time so that the
fluctuations in the two data sets nearly coincide.  Figure 21 shows the
result of shifting the "photo" iz values by -10 minutes.

     Several of the peaks in hours 21 and 22 nearly line up in Figure 21.
Therefore, it seems plausible that  the turbulence field near  the smoke  plume
lags the turbulence measurements on Tower A during this period of time.  We
saw evidence for this 10-minute phase lag in the wind direction time-series
plotted in Figure 17.  Plots (not shown here) of aw and iz from 2,  10
and 40 m on Tower A and from 2 and 10 m on Tower B further support  the
interpretation that the wind shift and the vertical turbulence occur uni-
formly later near the hill than at  Tower A.  If this is correct,  then the
photo iz values are roughly 2 to 3 times the Tower A iz value during
this part of Experiment 201.

     The time-series of iz4 values is plotted in Figure 22 alongside those
Tower A iz values interpolated to the release height.  It is  seen that the
iz4 estimates tend to inflate the larger values in the iz3 time series,
but change very little otherwise.  The lower part of Figure 22 compares
iz4 with the iz values measured at  10 m and 40 m on Tower A.   The com-
parison suggests that "actual" iz values at plume height could have been
underestimated in the MDA during the first two hours of Experiment  201, and
by similar reasoning they could also have been underestimated during  those
                                  52

-------
 11.0
 io.o
  9.0
  8.0
  7.0

  6.0-
)
  5.0
  4.0-
  3.0-
  2.0-

  1.0-
      0.0-
            —•»•— iz-| (photo)
            	 Modelers' Data Archive
   17.0
16.0	
15.0-
14.0-
13.0-
12.0-
11.0-
10.0-
 9.0-
 8.0-
 7.0-  /
 6.0-  '
 5.0-
 4.0-
 3.0-
 2.0-
 1.0-
     0.0
       17.0
                  18.0
                             19.0
                                        20.0
                                                  21.0
                                                             22.0
                                                                    23.0
                                                                Time (hr)
                 iz3 (photo)
                 Modelers' Data Archive
                  18.0
                             19.0
                                       20.0
                                                  21.0
                                                             22.0
                                                                   23.0
                                                                Time (hr)
Figure 20.
           Intel-comparison of  vertical turbulence  intensity  (iz)
           time-series data for Experiment  201, derived  from
           photographs by Methods 1  and 3,  and interpolated from
           Tower A.  (Circles denote  periods  of considerable
           horizontal meander.)
                                     53

-------
 11.0
 10.0
  9.0
  8.0

  7.0

  6.0
)
  5.0

  4.0-

  3.0-

  2.0-

  1.0-
      0.0-
        20.0
16.0
15.0
14.0
13.0
12.0
11.0
10.0
 9.0
 8.0
 7.0-
 6.0-
 5.0-
 4.0-
 3.0-
 2.0-
 1.0-
      0.0
                                             —4— iz-| (photo: shifted by-10 min)
                                                    Modelers' Data Archive
                                         21.0
                                                                         22.0
                                                                      Time (hr)
                                                   iz3 (photo: shifted by-10 min)
                                                   Modelers' Data Archive
                                                                  »	
       20.0
Figure 21.
                                        21.0
                                                                        22.0
                                                                     Time (hr)
          Intel-comparison  of vertical  turbulence intensity (iz)
          time-series  data for two hours of  Experiment  201, derived
          from  photographs by Methods  1 and  3,  and  interpolated
          from  Tower A.   The "photo" estimates  are  shifted by
          - 10  minutes.
                                       54

-------
         16.0-
         15.0-
         14.0-
         13.0-
         12.0-
         11.O-
         .I 0.0-
          9.0-
     iz(%) 8.0-
          7.0-
          6.0-
          5.0-
          4.0-
          3.0-
          2.0-
          1.0-
iz4 (photo)
iz (Modelers' Data Archive)
          Q.O.
            17.0
         27.0-
 18.0
           19.0
         24.0-

         21.0-

         18.0-

         15.0-

         12.0-

          9.0-

          6.0-

          3.0-
          0.0-
               	h— iz4 (photo)
iz (Tower A @ 10 m)
iz (Tower A @ 40 m)
                      20.0
                                 21.0
                                           22.0
   23.0
Time (hr)
            17.0
                      18.0
                                            20.0
                                                      21.0
                                                                 22.0
                                                      23.0
                                                   Time(hr)
Figure 22.    Intel-comparison of vertical turbulence intensity (iz)
               time-series  data  for Experiment 201,  derived from
               photographs  by Method 4,  and  interpolated  from Tower A.
               (Circles denote periods  of considerable horizontal meander.)
                                         55

-------
 periods of time in which the iz  value at  40 m was nearly zero.  However,
 the interpolated tower values appear appropriate for  the remainder of hours
 21 and 22,  with the exception of the,peak iz  (photo) values.

      Experiment 210

      Processed lidar data are available for two of the hours in Experiment
 210,  and these data are used to  supplement the photographic data in evaluat-
 ing the representativeness of vertical turbulence intensity estimates at
 plume height  (MDA).   Photographic and lidar coverage of experiment-hour 3 is
 the most extensive for this experiment, although photo-documentation of the
 initial stages of  plume growth near  the source is generally less extensive
 than that during Experiment 201.  Consequently, much of the analysis has
 been performed on  experiment-hour 3.

      Unlike the previous use of  SHIS #1 lidar data, we have looked more
 carefully at  the processed data  to evaluate how well calculated second
 moments of  the backscatter distribution represent an effective az for
 the  oil-fog plume, and how well an estimate of the 1-hour az
 by
calculated
          = a
                  JE
            (38)
 agrees  with the crz  for a  superposition of each individual distribution
 during  the hour.  Note that CTZ is the average of the squares of the
 calculated second moments of the individual distribution profiles during the
 hour, and  OH is the variance of the plume centroid height during
 the  hour.   Recall that much of our previous evaluation of the proposed
 a 2 growth  function  used in CTDM was based upon the 1-hour crze values
 obtained from the lidar data using Equation 38.

     In this new analysis, best-fit oz values are obtained from the pro-
 cessed  individual scan lidar data arrays by following these steps :

     1)    Data arrays are scaled from relative bscat intensity to actual
           bscaf

     2)    Data arrays are integrated in the crosswind direction to yield a
           crosswind-integrated vertical profile  of bscat.

     3)   A Gaussian envelope is  fit to the vertical profile by the method
           of  least  square errors.   The envelope  is that of a reflected
          Gaussian.   The location of the centerline is  set at the centroid
           height,  and the surface is set to the  height  of  the terrain
          beneath the centroid.   The fitting procedure  provides both the
          centerline bscat value  and the az for  the best-fit curve.

Step 2 takes any apparent rotation or tilt  of the  plume in cross-section
 into account.
                                  56

-------
     Best-fit cz values for extended periods are obtained from the ver-
tical profiles resulting from step 2 above in the following way:

     •    A vertical grid-point spacing is set up to match the greatest ver-
          tical resolution available in each scanning plane.

     •   , An average surface elevation is specified for each  scanning plane..
           scat
                       are interpolated to the vertical grid  and  summed
          over all scans within each plane.

     •    A Gaussian envelope is fitted to the resulting vertical distribu-
          tion, and best-fit az and maximum  bgca^ values are  obtained.

     The crosswind-integrated/time-integrated vertical  profiles of  the plume
upwind of the hill are remarkably Gaussian.   An example vertical profile  is
shown in Figure 23.  The best-fit CTZ values  for the individual means are
generally not much different from the second moments of the distribution.
The average ratio of the best-fit a.
individual scans for hour 3 are:
                 to the second moment  for  the
scan plane
ratio
211°
.90
225'
.92
236C
.95
244'
.92
251'
.84
264°
1.06
Those for hour 6 are:
scan plane
ratio
211°
1.14
225°
1.20
236°
1.00
244C
.86
251°
1.14
264°
1.23
     In each case, the 236° plane had by far the most  samples.
     Focussing on the 236° scanning plane,  the  ratio of  the best-fit 1-hour
az to the effective az (second moment plus  centroid) is  found  to be
0.77 in hour 3, and 0.94 in hour 6.  These  results tend  to suggest that the
effective az values used in the CTMD Second Milestone Report may be
overestimating the vertical size of the plume.

     Photographs of the plume taken from the  side  of the plume trajectory
are available from two locations during experiment-hour  3.  Because lidar
data are available for this hour,  measurements  of  optical smoke plume thick-
ness from these photos can be compared with the best-fit az values ob-
tained from individual lidar scans.  Although the  lidar  data are not
5-minute averages of the plume as are the photographs, the plume appears
very coherent during this hour, which suggests  that a comparison of the two
measurement techniques may be instructive.

     Estimates of plume thickness are 'obtained  from the  photographs using
methods similar to those discussed above.  The  major departure from that
methodology involves projecting the negative  image of each frame onto a
sheet of graph paper to obtain the plume dimension and position data.

     Lidar-az data are compared with photo-h  (thickness) data only at
those points where the lighting is sufficient to fully illuminate the plume.
                                  57

-------
                                                                                                                             4J -H
                                                                                                                              P-l P^
                                                                                                                              O  X
                                                                                                                              f-i  W
                                                                                                                                 U
                                                                                                                             0
                                                                                                                             •H  O
                                                                                                                            <4-(  O
                                                                                                                            •H O
                                                                                                                             4-> O
                                                                                                                             3 00
                                                                                                                             -P   O
                                                                                                                             t/1  'H
                                                                                                                            •H  4J
          <»  ro
          CCD
Oj  0)
M  O
  S
>  O
                                                                                                                                   O
    O  CN
         Q CO
CD
                                                                                                                            O
                                                                                                                            bO
                                                              58

-------
 For photographs taken at  camera position  0-11, these points are close to the
 source,  about  50 to  150 m downwind.  For  photographs taken at camera posi-
 tion 0-10,  these points are  near the base of the hill  Ov< 600 m downwind),
 and also in the col  area  when  the, plume traveled over  the top of CCB and
 through  the illumination  of  the  draw tower lights.  By taking only those
 lidar and photo .data coincident  in time,  and' within 50" m"d£ stance s'we' find
 that.the six available ratios  of h/az average 3.69 with a variance of
 0.24;  Due  to  the criteria for strong illumination, the photo-h data ought
 to  fully repre'sent total  plume thickness, and this may be why the variance
 is  so small*

      If  we  had assumed a  top-hat distribution for the plume material, then
 the ratio of h/az would be 3.46.  If we had assumed a semi-circular dis-
 tribution,  h/
-------
                                	, ^/
                                                    CO
                _ — —^
                                          	, _/
                                	I
                                                                     •H
                                                                      X O
__ ^s
                                 •*.\^
                                                                      C "d
                                                                      0)  
                                                                      0 -H
                                                                     •H  fH
                                                                              4)
                                                                      0) O H
                                                                     i — I i — I
                                                                      3 CM  g
                                                                     £>      O
                                                                      f-l  -P  f-l
                                                                      3  c m
                                                                     P  (D
                                                                          S -d
                                                                     r-H -H  CD
                                                                      Oj  fH -P
                                                                      (J  (D  OJ
                                                                     •H  Pnr-1
                                                                      
-------
     Conclusion

     Vertical turbulence intensity data contained in the MDA generally  "fol-
low" the observed behavior of the oil-fog plume.   However,  there are periods
in which these data do not appear to be representative of .turbulence at the
release site.  Excluding these periods,  there remains an apparent underesti-
mation of iz in the archive data compared to rates of initial plume growth
obtained from photographs.  Some of this may be due to the  neglect pft
source-induced turbulent mixing.  Results obtained after including source-
induced turbulence are discussed in subsection 3.3.4.

     3.3.3  Plume Growth in the Vertical

     Because CCB surface concentrations can be very sensitive to  the size of
az, the form of crz(x) is an especially important topic for  evaluation.   In
particular, we have evaluated the form of az(x) proposed in the CTMD Second
Milestone Report by analyzing the growth of photo-derived and lidar-derived
az values with distance between the release position and the upwind base
of CCB.                    ,             ,                           •

     Experiment 201

     Plume thickness values obtained from photographs are converted  to  esti-
mates of 0Z by using the Gifford technique and a top-hat distribution
assumption.  Because processed  lidar data are as yet unavailable  for Experi-
ment 201, there is no check on which of these methods is more appropriate.
Also, although the analysis of  lidar data from Experiment 210 showed  that
az ^ 3-7 h, this result may not be appropriate for Experiment 201
because the  lighting conditions are substantially different in these  two
experiments.  Consequently, this analysis of the vertical growth  of the
plume with distance will  continue to focus on both methods.  Note that  the
top-hat method is expected to underestimate actual crz values with in-
creasing distance downwind because the "edges" of the plume are lost.   Also,
the  Gifford  technique is  expected to overestimate actual az values with
increasing distance because the true maximum in apparent plume thickness is
not  usually  seen  in  the photograph, and because the plume brightness  may
decrease with distance due to the light scattering properties of  the plume
rather  than  due to dilution.

     Vertical  turbulence  intensities estimated from photographs are referred
to by izi, iz3, and  iz4,  and these have the same meaning as in the
previous  subsection.  az^ values are computed from plume thickness data
by  assuming  that  the plume material exhibits a top-hat distribution:
      a  .  = h/2/3
       zl
                                                                         (39)
 az2 values are computed  from h  data by means of the Gifford technique
 (only azi and az2 are  used  in the  subsequent analyses).  Vertical
 plume spread is also computed from the PGT dispersion parameter formulas for
 the most appropriate stability  class as contained in the MDA.  These are
 simply denoted by oz(PGT).
                                   61

-------
     Figure 25 displays the ratio of azi and az2  to az(PGT) versus downwind
advection time.  The ratios are plotted as  logs to facilitate "factor of 2"
types of comparisons.  It is evident that the aZ2 estimates follow a
growth law more similar to that contained in the  crz (PGT) curves than do
the az± estimates.  The analyses also suggest that az(PGT) values would
match the photographic estimates better if  they were based on a more refined
estimate of vertical turbulence intensity than the surface stability class.
Even so, the PGT curves tend to overestimate the  rate of growth of the ver-
tical spread with time as derived by the Gifford  technique.

     Figure 26 contains comparisons  of  azl (top hat) against az values
calculated by means of Equation 36 with p set equal to 1.5.  As in
Figure 25, the ratio of azi to az is plotted as a log versus advection
time.  Two versions of crz (calculated)  are presented: one uses iz]_»
the other uses iz3«  These two estimates of iz tend to bracket those
values interpolated from Tower A data.   The results of the comparison show
that the functional form assumed in  Equation 36 causes the plume to grow too
rapidly with advection time,  compared to the development of az obtained
with the top-hat assumption.   Futhermore,  iz^ tends to align az^ with the
calculated values better than iz3 because the iz^ values are also derived
from the top-hat distribution assumption.

     Figure 27 indicates that Equation 36 performs better when az2 (Gifford)
is viewed as the more accurate description of the plume spread.  The upper
portion of the figure compares aZ2 with crz calculated by means of Equation
36 with iz = iz3 and p set equal to  1.5.  The distribution of ratios is
centered at approximately 1.0, and the rate of growth of aZ2 beyond
approximately 30 seconds travel time tends  to follow that prescribed by
Equation 36.

     Because Equation 36 appears to  simulate the  overall growth trend in
°z2 beyond t = 30 sec during Experiment 201,  we have used the a22 and iz3
data to re-evaluate the value of the parameter p.

     By solving Equation 36 for p,

       _      Nt                                                        (40)
     p , 	.	
         (i ut/a ) -1
           z    z
                                                        2
and a least squares fit to the scatterplot  of (iz3ut/az2) -1 versus Nt
produces a value of 0.6 for p.  In the  notation of Pearson et al. (1981), p
s 2y , and so a value of 0.6 produces a f of 0.55.  The theory of Pearson et
al. (1981) indicates that YJ  a molecular exchange coefficient, could range
from 0.1 to 0.8.

     Using this new value of p, we have recalculated the iz^ estimates
described previously.  Recall that the  matching actually takes place at the
measurement point nearest to 0.5 t^x,  where t^x is the measurement
point farthest from the release crane.   A re-evaluation of p was made with
o~£ and iz4 as a consistency check,  and the results are displayed in
Figure 28.  The upper plot contains  all data, and the lower contains only
those data that were not used to match iz4 to az2-  The value of p is
nearly the same in both plots; p (upper) = 0.567, p (lower) = 0.566.

                                  62

-------
  ln(£rz1/
-------
     ln(crzl/iz-|uf)
     z.eee
                                                           Time (sec)

Figure 26.    Comparison of az  (calc.)  growth curves with az-j_ data
              derived from Experiment 201 photographs.   (az(calc.)
              iz uf, f = t/(l + Nt/p)°-s).
                                  64

-------
     In (crz2/iz3uf)
     2
     In (crz2/iz4uf)
                                                                  Time (sec)
     2.886
     1.758 -

     t.see-

     I.2S0-

     1.888-

      .758-
      .sae-


      .258-
      .258-


      .SB8-
     -.750-


    -1.803-


    -1.2SB-


    -I.50B-


    -I.7SB-


    -2.ee
p=0.6
                                                                  Time (sec)
Figure 27.    Comparison of  az(calc.)  growth curves with  az2 data
               derived from Experiment  201  photographs.   (az(calc.)
               izuf,  f =  t/'•
                                      65

-------
(iz4Ut/o-z2)2-1
                                                 ia.8      12. e
(iz4Ut/crz2}2-1
                                                                   14.
                                                                   Nt
        (iz4 fit points removed)
    • B        2.8       4.8       E.B       8.8       IB.8       12.8       14.8
                                                                  Nt
 Figure 28.    Scatterplots used in computing best-fit  p for
                Experiment  201.
                                  66

-------
     These analyses increase our confidence in estimating az by Equation 36.
We see that the growth trend is similar to those Gifford technique estimates
of az derived from the photographs.  And we also see (Figure 22) that
the  iz values inferred by matching the photo oz values to Equation 36,  with
p =  0.6, at the .measurement point nearest 0.5 tmax are generally "reason-
able" compared with iz values measured on Tower A.

     • Experiment 210                      '"....'                       ,

     The analysis of the growth of CTZ with distance during Experiment
210  focuses on experiment-hour 3 because the link between vertical plume
thickness as measured from the photographs and az as calculated from the
lidar data is best for this hour.

     The relationship among the photo—estimates of az,  the lidar estimates
of az, and the curve CTZ(X) computed with Equation 36 (p = 0.6)  and the
photo-derived iz estimates are summarized in Figure 29.  In this figure,
the  "plus" symbols represent the photo-estimates of az; the "box"
symbols represent the lidar-az values;  and the solid line is  the com-
puted az(x) curve.  The figure suggests the following:

     •    The computed az (x) curve corresponds well with most  of the
          data points out to 600 to 800 m from the source.

     •    For those periods without photo-estimates of  az,  simply
          multiplying the interpolated  iz data from Tower A by  1.37 (the
          reciprocal of 0.73) places the calculated crz(x) curve in the
          neighborhood, of the lidar-az  points.

     •    When the plume passes through the col,  the apparent size of the
          plume is on the order of one-half the value expected  in the
          absence of the hill.

     •    Although the az(x) curve appears quite reasonable in  general,
          the correspondence with the other data would  improve  somewhat  if
          the iz value were increased,  and if  the curve grew  more slowly  !
          than with distance to the 1/2 power  beyond 200  to 300 m.

     These results indicate that the form of Equation 36  is appropriate  for
estimating the dispersion of elevated smoke releases during Experiments  201
and  210, at least over distances on the order  of 500 m  to 1000  m.   Unre-
solved, however, is the question of whether the square-root distance  growth
continues well beyond 1000 m at CCB.  This far-field growth law is not of
critical importance for the purpose of  estimating concentrations over CCB,
but  it will become more important in estimating concentrations  for full-
scale power plant situations.  For the  present, we shall  use  the formulation
of Equation 36,  and re-interpret our estimates of iz by including a
measure of source-generated turbulence  in, the  analysis .

     3.3.4  az Tuning Parameters

     The preceding discussions of the CTZ analyses assumed that  iz and
p could be viewed as free parameters  in fitting observed  estimates  of az

                                  67

-------
2  S
                                                o
                                                      §  f
                                                      i
                                                                     o  o
                                                                                                                                 o  >
                                                                                                                                 •P  (D
                                                                                                                                 pj id
                                                                                                                                 0)   I
                                                                                                                                 •H  cd
                                                                                                                                 H T3
                                                                                                                                 CD -H
                                                                                                                                     CD \o

                                                                                                                                 p  o d
                                                                                                                                 CD
t/>  10 c

CD  N
                                                                                                                                 ^
                                                                                                                                 6  cu
                                                                                                                                 O  > ^
                                                                                                                                 fH  -H  +
                                                                                                                                 4-1  ^ rH
                                                                                                                                     
-------
69

-------
70

-------
  to  the other variables in Equation 36.  But Equation 36 does not explicitly
  contain the effect of source-induced turbulence on the development of a
  with distance.  This omission could be responsible in part for discovering
  that the photo-derived iz estimates are generally greater than iz values
  contained in the MDA.

      When the source-induced turbulence is included as a "virtual source
  distance" (xv),  Equation 36 is rewritten as
      a (x) = i (x + x )/(! + N (x + x )/u 2Y* )°'5
       ^    ,   Z      V               V
                                                                         (41)
 so that there are three free parameters.   xv is not treated as  a parameter
 in itself, however.   Instead,  the initial mixing is characterized by an  ini-
 f~ 1 a 1 rr    T.T!-» o >• d
tial crzo,  where
      azo = az(x = 0)
                                                                        (42)
 Fixing the value of azo therefore allows one to  calculate xv by means
 of Equation 41.
      A Monte Carlo parameter optimizing scheme  is used  to  select the
 combination of azo,  iz,  and  y  that provides the best fit of Equation 41 with
 the estimated az  data.   When this  is  applied  to each 5-minute period of
 Experiment 210-hour 3, y  is  found  to  vary  from 0.16 to  0.76, with an
 average value of  0.4.  The range in y reflects the range in the shape of
 oz(x)  contained in Figure 29.  The smaller values of y apply to a  (x)
 curves that level out more at  large x, and the larger values apply to az(x)
 curves with a greater rate of  growth at large x.  Due to  the uncertainty in
 the az estimates  in Figure 29  and the possible influence of the hill on
 the plume near the upwind base of the hill, no clear trend in y could be
 identified in the data to construct a model for y.  Therefore, we choose
 to  fix y  at a representative median value.  On the basis of the analyses
 presented in subsection  3.3.3, and the generally good agreement between the
 data in Figure 29  in which crz(calc.) is based on p = 0.6,  y is chosen to be
 0.525  (and  p  is now  0.55).
°zo and
     Fixing the value of y forces the optimizer to adjust the values of
       iz in fitting Equation 41 to the data.   Using the data for
Experiment 210-hour 3 once again, azo varies from 0.3 m to 1.7 m.   Eight
of the twelve values are less than 0.75,  and these average 0.48 m.   The
best-fit value of iz is not very sensitive to the value of CTZO when azo
is in the 0.3-m to 1.7-m range.  For example,  if 0.5  is substituted for
1.21, the best-fit iz value changes from 0.014 (1.4%) to 0.016.
Therefore, most of the variability in azo whenazo is appreciably different
from 0.5 during this experiment-hour goes toward improving the goodness-of-
fit.  A representative value for azo appears to be 0.5 m.

     All of the experiment-hours of Experiment 210 produce similar  ranges of
y and azo.  When y is fixed at 0.525,  and azo  is fixed at  0.5  m, the
best-fit values of iz generally follow the measured i_ data.   Differences
                                                     £i
                                  71

-------
 between the best-fit  and  measured  iz data are found to result from the
 following  factors:
      •     waves  -  the  plume had pronounced waves in it at times, and best-
           fit  iz's were approximately 1.3 times the measured iz's.   How-
           ever,  this difference is reduced considerably if we assume that
           the  photographs  show a triangular distribution rather than a top
           hat.

      •     horizontal spread - during some periods of considerable horizontal
           meandering or spreading of the smoke plume,  the vertical  size of
           the  plume is overestimated.

However,  there are periods in which the photographic data are genuinely dis-
similar to the Tower A turbulence intensity estimates  at the plume  height.
These occur when the tower iz values are less than 0.008, or when the
photos show periods of enhanced vertical spread.  Figure 30 contains the
time-series plots of the interpolated iz data from the MDA and the  best-
fit iz data from the preceding analysis.  Those periods in which an undis-
puted discrepancy exists between the two iz values are indicated by
circling  the best-fit iz data points.

      The  analysis of plume spread data from Experiment 210 indicates that:
     •    the oz formula with y = 0.5 and azo *> 0.5 is appropriate for
          moderate wind speed (3-8 m/sec) experiments such as 210, when the
          oil-fog is generated by the TIFA thermofogger; and

     •    turbulence intensities interpolated to the release height from
          Tower A data are generally appropriate, although the lower values
          (less than 0.008) appear consistently too small.

     The same analysis has been applied to the data for Experiment 201.   In
this case, we have taken the "az-£* estimates of plume spread derived
from the photographs with the Gifford technique.  Preliminary tests of fit-
ting several randomly selected 5-minute time periods with y, azo,  and
iz as free parameters produced several estimates of y between 0.3  and
0.6.  On the basis of the overall fit of 0.57 (subsection 3.3.3)  for Experi-
ment 201 and on the results found in analyzing Experiment 210 data,  y was
set at 0.55; azo was set at 0.5; and best-fit values of iz  were
obtained.
     Figure 31 shows these estimates of iz compared to the iz data
contained in the MDA.  Between 2020 and 2145,  the archive wind speed, Brunt-
Vaisala frequency, and iz data are shifted in an attempt to account  for
the apparent lag between the Tower A data and the plume and Tower B  data.
Data for 2020 are persisted for 10 minutes,  and the remaining data in the
period are shifted in time so that the 5-minute period ending at  2035 is
filled with data from 2025, and so on.   The  photo-derived az data are
fit using Equation 41 with the shifted u and N data,  and the resulting best-
fit i2 data are compared against the shifted MDA iz data in Figure 31.
                                  72

-------
                                                                    t	. _ _\
CO
(1)
+•»
(0
E
Q.
CD

D,<
O Q
si

I  I
•—  L.
~  3
o  to
~  CO
r-  CD
<*  E

E  E
                                                             ir
                                                             iL

                                                                                                  O  tr
                                                                                                  co  £.
                                                                                                       i—(
                                                                                                          •H  -H  43
                                                                                                           erf  m  ft X
                                                                                                          P     erf .-i
                                                                                                           S  X fn CD
                                                                                                           O  4^  bo !>
                                                                                                           O     o -H
                                                                                                              0  P P
                                                                                                                 O erf
                                                                                                          P  nj
                                                                                                              in  CD
                                                                                                           X CD  43
                                                                                                           P  -P  P

                                                                                                           w  S  *
                                                                                                           f3  -H  O
                                                                                                           CD  P  .fn
                                                                                                                    (D
                                                                                                                    f-i
                                                                                                          P
                                                                                                          S
                                                                                                          •H
                             CD  P P
                             O  -H erf
                             f3
                                                                                                                •H  O
                                                                                                                 P  S
                                                                                                                 I/)  CD
                                                                                                           erf  erf     fn
                                                                                                           O  p   tsi-H
                                                                                                          •H  erf  b  CJ
                                                                                                          P  Q     •—'
                                                                                                           f-i     0)
                                                                                                           CD  -   43
                                                                                                          >

                                                                                                           S  O  CD
                                                                                                          •H    r-f  g
                                                                                                           fH  CD  erf  -H
                                                                                                           erf 43  cj  f-i
                                                                                                           fl, +j >—r 
-------
                                                          i
                                                                -a  .
                                                          »-     CD  •!->
                                                                 C  -M
                                                                •H  -H
                                                                 0)  <+H
                                                                 O  ,0
                                                                 ctj
                                                                     CD  fi
                                                                         CD
                                                                 ^  S  -
                                                                 cd  S -H
                                                                -P
                                                                             _
                                                                             bo
                                                                     ni
                                                                 w   B      fn
                                                                 PI  -H   H  oi
                                                                 (D   +->   O  i— I
                                                                 4->   10  'W
                                                                    ^3  +J  0)
                                                                 CD  -P  cti  >
                                                                 O  -H  TJ  -H
                                                                 C  S      •(->
                                                                 (D      CN cti
                                                                r-H    t>  CD
                                                                rQ  -H
                                                                •P
                                                                    r
                                                                     O  CD  <+H
                                                                        >  O
                                                                        CD
                                                                 cd   ed
                                                                 O  4->  r
                                                                •H   cd   o
                                                                •P  Q
CD -
                                                                            
-------
      Aside from the major isolated peaks in the best-fit  iz  time-series,
 the best-fit estimates lie reasonably  close to  the MDA  iz values.  This
 boosts our confidence in modeling  az at  CCB with Equation 41 and the
 turbulence data interpolated to the release height.   However,  this analysis
 has also indicated that the Tower  A data are not always representative of
 conditions at the release.  The smoke  plume undergoes considerable vertical
 growth at times when the archive data  would indicate  more modest growth and
 when low values of turbulence intensity  are recorded, the smoke-plume growth
 does not reflect this.

      An assessment of how well the calculated crz data fit the  photo-
 derived oz data can be obtained by plotting the  photo-derived a
 values against  the calculated az values.  Figure 32 shows the result of
 doing this for  Experiments 201 and 210.   The fit is uniformly best for the
 smaller values  of az because  these are closer to the  release point, and
 are therefore controlled most by the best-fit iz value.   Beyond this
 region,  the shape of the growth curve  is more strongly influenced by y,  N,
 and u.   Here the fit is not so good, but most of the  calculated values lie
 well within a factor of two of  the photo-derived values.  These figures,
 taken with Figures 30 and 31,  indicate that y =  0.525 and azo = 0.5
 are reasonable  representative values for Experiments 201 and 210.

      3.3.5  Turbulence Over GCB

      Experiment  201

      Five towers were operated  on  CCB during SHIS #1.   Tower B (a 30-m
 tower) was  placed  on the  south peak, and Towers C,  D,  E, and F (10-m towers)
 were placed near the 70-m height contour on the northeast, southeast,  south-
 west, and northwest  sides of CCB, respectively.   Vertical turbulence  data  on
 CCB for  Experiment 201  are  available at the 2-m level  of Towers C,  E,  and  F,
 at  the 2-m  and 10-m  level on Tower D, and at the 2-m,  10-m,  and 30-m  level
 of  Tower B.   The 2-m and  10-m level data from these towers,  the turbulence
 data  from Tower A  at the 2-m, 10-m and  40-m levels,  and  the  Tower A turbu-
 lence data  interpolated  to  the  smoke-release height (30  m) are plotted in
 Figure 33.

      The data interpolated  to smoke height are taken from the MDA.  Other
 Tower A data and the Tower B data are taken from the full  data  base, and a
 prop-response correction factor is applied (as discussed in  CTMD Second
Milestone Report).  Data from Towers C, D,  E, and F are  taken from the full
 data  base, adjusted  for prop response,  and then adjusted for the inclination
of the mean 5-minute wind to the horizontal. The latter adjustment is an
 approximate method to obtain the turbulence intensity  in the direction
perpendicular to the mean streamline of the  flow:
              + v
      w
           _2
           U
                     W
                                                                        (43)
     This formula is approximate  in that  the'full expression for 
-------
      Sigma-z Observed (m)
EXPERIMENT 201
                                                           25.a        38.0

                                                       Sigma-z Calculated (m)
      Sigma-z Observed (m)
EXPERIMENT 210
           .8    2.8
                            e.e    e.e    IB. a   12.
                                                                .e   2e.e
                                                       Sigma-z Calculated (m)
Figure  32.   Comparison of photo-derived crz data and  calculated CTZ
              utilizing best-fit iz data in Equation 41  with
               7=  0.525 and crzo = 0.5.

                                    76

-------
                                              *l
                                                                                      r""
                                                                                            "S"
                                                                           05
                                                                                                              tq o
                                                                                                                 CM
                                                                                                                r\

                                                                                                              Q -P
                                                                                                                 S
                                                                                                                * 0)
                                                                                                              U S
                                                                                                                    o o
                                                                                                                    E-> M-l

                                                                                                                    e <
                                                                                                                    o
                                                                                                                    h f-l,
                                                                                                                    <4H (D

                                                                                                                    CTi O
                                                                                                                    •P E-i
                                                                                                                    nj
                                                                                                                    •U S
                                                                                                                       O
                                                                                                                      tslf-l
                                                                                                                    • H <+H

                                                                                                                    ^ Cti
                                                                                                                    e +j
                                                                                                                    cd cd
E

O
g>
'<»


.8

1  E  E
(/) CM 01


< O U-
I
u>
                  «n
                                             V
                                             > .

                                           r>-
                                             r    i
                                                                                                          t-   O
                                                                                                                    rH CtJ

                                                                                                                    PH &


                                                                                                                    t/1 O
                                                                                                                    0) O
                                                                                                                    •H
                                                                                                                    10 U
                                                                                                                     I
                                                                                                                    « a
                                                                                                                    6 o
                                                                                                                    •H
                                                                                                              •H
                                                                                                              tL,
                                                    77

-------
                                                       . <*>
                                                                                                                      «M
                                            
                                                                                                                             oa
                                                                                                                                 .
                                                                                                                                ra
                                                                                                                             O  O
                                                                                                                             S  <
                                                                                                                             cd  o
           E  E
          Q  UJ
£
5
 E "g
o  E
                                                     (D  O
                                                    • H
                                                     f-l  PQ
                                                     
-------
79

-------
 are u'w1 - 0, and u'w1 = auaw.   Because we expect  u'w'  to be non-zero, we
 have chosen the latter limit.   This implicitly assumes  that a portion of the
 measured crw arises from horizontal  fluctuations being turned upslope
 (downslope) by the hill surface.  Note that w is corrected for prop response
 on the basis of the wind tunnel data presented in  the CTMD Second Milestone
 Report.

      Tower F is the truest "windward" tower on CCB for  Experiment 201.  The
 vertical turbulence intensity,  iz,  at 2  m  varies between 0.01 and 0.03
 (1.0 and 3.0%),  with one peak as  great at  0.06.  The Tower A iz data
 interpolated to  30 m varies between 0.01 and  0.06.  In general,  however,  the
 turbulence intensity at 2 m on  the  windward face of CCB is considerably less
 than that at 30  m on Tower A, and it  is  substantially less than that at 2 m
 on Tower A,  which averages about  0.06.   Consequently, these data do not show
 an increase  in vertical turbulence  intensity  on the upwind slope.

      Of further  interest is a peak  in the  Tower F  data  record at 2100 MST.
 This is about  10 to 15  minutes  after  a peak in the Tower A data.   This is
 further evidence of the turbulence  shift seen in comparing the plume photo-
 graphs  with  Tower  A turbulence  data,  as  reported in subsection 3.3.2.

      The Tower E (at a  windward location for much  of Experiment  201) ver-
 tical turbulence data at  2 m are similar in magnitude to the  Tower F data.
 However,  these data exhibit a greater mean value and less variability.
 Still,  the turbulence intensity and 
-------
      In conclusion,  the vertical turbulence data over CCB during Experi-
ment  201 indicate  little, if any, increases in plume diffusion along the
upwind  face  of  CCB.  Increased mixing could occur near the surface of the
crest of CCB, especially during those periods when the vertical mixing away
from  the hill was  relatively weak, and increased mixing would certainly
occur in the lee close to the surface.

      Experiment 210  '

      The oil-fog plume was released on the southeast side of CCB for the
duration of Experiment 210.  Therefore,  Tower D is primarily a windward
tower;  Tower F  is  a  leeward tower; and Towers C and E are along the side of
the hill.  Vertical turbulence data from these towers and from Tower B atop
the south peak of  CCB are compared with data from Tower A, estimated for the
plume release height, in Figure 34.

      The vertical  turbulence intensity (iz) from the 2-m height on Tower D
is generally slightly greater than that  at the release height on Tower A,
although there are periods when the fluctuations are out of phase.   The
intensity at D is approximately 2% (+1%)  after 0200.   Data from 10  m on
Tower D  are similar.

      In  the wake (Tower F), there is a greater difference between the 2-m
and 10-m data, and also between these data and the turbulence intensity
interpolated to release height with Tower A data.   Vertical turbulence
intensity at 2 m is 2 to 3 times that at  10 m after 0200,  and both  are
several  times the  turbulence intensity at Tower A after 0200.  During the
first hour, however, the Tower A turbulence exceeds that at Tower F.

      Over the top of the south peak of CCB, the turbulence at 2 m exceeds
that at Tower A nearly all the time,  and  the turbulence at 10 m,  although
less  that that at 2 m,  also frequently exceeds that at Tower A during the
shortened 3-hour record.   This suggests  that the estimates of turbulence
intensity from Tower D may be underestimated by setting u1 w1 = CTUOW.
Consequently, the turbulence data from Experiment  210 indicate that turbu-
lence levels may increase over much of the hill, causing increased  plume
diffusion.

      3.3.6  Implications for Modeling

     The preceding analyses show that some degree  of  modeling error could
arise in simulating Experiments 201 and  210 due to using inappropriate
meteorological data as  model input.   To  test the significance of this  pos-
sibility, CTDM(11083),  CTDM(ll083)-5, CTDM(11083-E),  and CTDM(11083-E)-5
have been run on a new version of the data for the SF6 release height  (at
or near  the oil-fog smoke release height).

     The  following changes have been made to the MDA in preparing a modified
MDA for Experiment 201:

      •    1705-1800 - All data are unchanged from the MDA.
                                  81

-------
82

-------
83

-------
84

-------
           1805-1900 - Best-fit  iz data are substituted when available, and
                      a  lower bound of 0.020 (2.0%) is placed on the MDA
                      values, reflecting the lack of any best-fit iz data
                      near or below 0.020.

           2005-2020 - Photo wind directions are substituted when available,
                      and iz is set equal to 0.035 as a representative
                      lower bound.
           2025-2030 - Photo wind directions and best-fit i_ values are
                                                          £1
                      substituted, and all other meteorology is persisted
                      from 2020.

           2035-2200 - Photo wind directions and best-fit iz values are
                      substituted when available, and all other meteorology
                      is shifted by 10 minutes (e.g. 2025 data are used at
                      2035).  iz is restricted to be no smaller than 0.020.

           2205-2300 - Best-fit iz values are substituted when available,
                      and all other meteorology is unchanged except that
                      iz is restricted to be no smaller than 0.020.
     Note that whenever the best-fit iz exceeds 0.080,  its value is
reduced by 25%.  This is done because the photo-derived az values at
these times are thought to be overestimates,  either due to horizontal
meander or the approximations required to apply Gifford's technique to these
data.
     The following changes have been made to the MDA in preparing a modified
MDA for Experiment 210:

     •    0000-0020 -  All data are unchanged from the MDA.

     •    0025-0100 -  Photo wind directions are substituted.
          0305-0800 -
          0105-0300 -  Photo wind directions and best-fit iz data are
                       substituted when available;  the missing wind  direc-
                       tion data are filled-in with data from 10 m on Tower
                       B, and iz from the MDA are restricted to be no
                       smaller than 0.015, as reflected in the lowest best-
                       fit estimated for data from Experiment  210.

                       Same as for the period 0105-0300 except that  the
                       average of the archive wind  direction and the 30-m
                       Tower B wind direction are used to fill gaps  in the
                       photo wind directions.

     Experiment 201, Hour 18

     The peak observed concentration is 458 ppt,  and this occurred on the
windward face of the hill at release height (30 m).   The wind  speed  varied
from between 6.0 and 7.4 m/sec during the hour, and the dividing-streamline
height was no greater than 2 m (essentially zero).

                                  35

-------
     Because the meteorology is not altered for this hour,  the modeled  esti-
mates are unchanged.  CTDM(11083) estimates a peak concentration of  178 ppt,
and CTDM(11083-E) estimates 169 ppt.  Both estimates place  the peak  value  at
the location of the observed peak concentration.   The 5-minute version of
CTDM(11083-E) estimates a peak concentration of 173 ppt,  located high on the
windward face.

     The observed concentration is underestimated because the  plume  size is
"too large," causing too great a dilution.   A second calculation made on the
basis of films taken from the peak of CCB and aerial photographs fares  some-
what better.  The size of the instantaneous core  of the plume  is estimated
from the aerial photos; iy is estimated by assuming a top-hat  distribu-
tion; the plume core is assumed to be circular; both ay and az are
assumed to grow linearly with distance; the center of the plume is allowed
to touch the surface; and the frequency of this impact at a receptor near
the camera location on the north peak is estimated from the film.  The
resulting estimate lies between 380 ppt and 440 ppt, while  the observed con-
centration is 465 ppt.  This exercise indicates that even the  best represen-
tative meteorological data measured at Tower A can lead to  substantial
errors in estimated concentrations.

     Experiment 201, Hour 19

     The peak observed concentration is 401 ppt,  located  near  the
windward peak of CCB.  Peak modeled concentrations are:

     •    34 ppt  - CTDM(11083) with MDA, located just leeward of the crest

     •    94 ppt  - CTDM(11083) with modified MDA,  located  near the  windward
                    peak

     •    94 ppt  - CTDM(ll083)-5 with modified MDA, located near the
                    windward peak

     •    482 ppt - CTDM(11083-E) with MDA, located on the  windward  face
                    near the release elevation

     •    605 ppt - CTDM(11083-E) with modified MDA, located on the
                    windward face near the release elevation

     •    389 ppt - CTDM(ll083-E)-5 with modified MDA, located on the
                    windward face near the release elevation.

Using the best-fit iz data, which are generally greater than the MDA
values, leads to greater concentrations.  In the  case of  CTDM(11083), this
improves both the location and magnitude of the peak value.  For
CTDM(11083-E), however, this change causes increased overestimation.  When
5-minute averages of the meteorological data are  used, CTDM(11083-E)-5 per-
forms much better in estimating the peak concentration magnitude, although
the placement is too low on the windward face. Note that the  second highest
observed concentration (363 ppt) is also located  near the release elevation
on the windward face of the hill,  so the model estimate of  the location of
peak concentration is not as bad as it might first appear.

                                    86

-------
     Experiment 201, Hour 21

     The peak observed concentration is 246 ppt, located slightly above the
release elevation.  Peak modeled concentrations are:

     •    900 ppt - CTDM(11083) with MDA
     •    723 ppt - CTDM(11083) with modified MDA
     •    689 ppt - CTDM(11083)-5 with modified MDA
     •    797 ppt - CTDM(11083-E) with MDA
     •    569 ppt - CTDM(11083-E) with modified MDA
     •    851 ppt - CTDM(ll083-E)-5 with modified MDA.

All of these estimates are located midway up the windward face of the hill.
From the distribution of observed concentrations and  the oil-fog plume
photographs, it appears that the plume preferentially curved around one side
of the hill.  The band of greatest observed concentrations lies along
approximately the same elevation as the peak modeled  concentration.  The
hourly mean Hc is 23 m, or 7 m less than the release  height.   This sug-
gests that the plume is undergoing a greater horizontal deflection than what
is allowed by the Lift module.

     The modifications made to the MDA produce a significant improvement in
the model estimates, especially in CTDM(11083-E).  Perhaps an altered ver-
sion of the model which shifts more of the plume to the side of the hill
will reduce the size of the peak modeled concentration  as well as improve
its location.

     Experiment 201, Hour 22

     The peak observed concentration is 437 ppt, located slightly above the
release elevation (the same receptor as last hour).   Peak modeled concentra-
tions are:

     •    1595 ppt - CTDM(11083) with MDA
     «    1064 ppt - CTDM(11083) with modified MDA
     e    1144 ppt - CTDM(ll083)-5 with modified MDA
     •    1081 ppt - CTDM(11083-E) with MDA
     •    605 ppt  - CTDM(11083-E) with modified MDA
     «    963 ppt  - GTDM(ll083-E)-5 with modified MDA.

All of the modeled peak concentrations are located near the top of the wind-
ward face.  The distribution of concentrations and the meteorology are quite
similar to that from the preceding hour.  The dividing-streamline height is
now computed to be 26 m, or only 4 m below the release height.

     The modified MDA again produces a substantial improvement in the peak
estimated concentrations.  As with hour 21 however, more horizontal deflec-
tion seems to be needed.  Also, the shift in Hc closer  to the release
elevation has increased both the observed and the modeled concentrations.
This sensitivity to Hc implies that the resolution in calculating and
incorporating HC could account for much of the remaining discrepancy in
the peak concentrations.
                                  87

-------
     Experiment 201, Hour 23

     The peak observed concentration is 304 ppt,  located at the windward
base of the hill.  Peak modeled concentrations are:

     •    923 ppt - CTDM(11083) with MDA
     •    695 ppt - CTDM(11083) with modified MDA
     •    430 ppt - CTDM(11083)-5 with modified MDA,  located 10 m
                    above the observed peak
     •    405 ppt - CTDM(11083-E) with MDA
     •    315 ppt - CTDM(11083-E) with modified MDA
     •    433 ppt - CTDM(11083-E)-5 with modified MDA.

All of these estimates, except that for GTDM (11083)-5,  occur on the same
face of the hill as the peak observed concentration,  but at an  elevation of
approximately half the hill height.  The mean dividing-streamline height is
now I m above the release height, but the range in H,, over the  hour
extends from 24 m to 40 m.  Even with such a range in an important modeling
parameter, the inability of the model to produce  the  peak observed concen-
tration well below Hc is probably real.  It  may  be that the Wrap module
needs to be re-evaluated, or that additional dynamics (e.g., some type of
upwind vortex flow) need to be introduced to bring more  material to the sur-
face at the base of the hill.

     Experiment 210, Hour 1

     The peak observed concentration is 28 ppt, and few  others  are appreci-
ably greater than zero.  The peak occurs at the upwind base of  the hill, but
the second highest concentration (26 ppt) is found low,  off on  the side of
the hill ("side" of the hill as opposed to either the leeward or windward
faces of the hill).  The peak modeled concentrations  are:

          163 ppt - CTDM(11083) with MDA
          175 ppt - CTDM(11083) with modified MDA
          145 ppt - CTDM(ll083)-5 with modified MDA
          56 ppt  - CTDM(11083-E) with MDA
          59 ppt  - CTDM(11083-E) with modified MDA
          59 ppt  - CTDM(ll083-E)-5 with modified MDA.

All of these occur on the side of the hill above  Hc,  which exceeds the
30-m release height by 12 m.  It appears that the Wrap module does not pro-
ject enough material along the stagnation streamline, and that  the Lift
module allows too much of the material away from  the  stagnation streamline
trajectory to impact the hill.  Changes to the MDA only  modify  the wind
directions, and these appear to have a small impact on the modeled concen-
trations.

     Experiment 210, Hour 2

     The peak observed concentration is 395 ppt,  located well below the
release height on the south side of the hill.  With a release height of 57 m
and an Hc of 35 m, az must be large and the frequency of winds  along
                                  88

-------
 the  stagnation  streamline must be high for the Wrap Module to produce so
 large a concentration.  Peak modeled concentrations are:

     •    45 ppt  - CTDM(11083) with MDA, located at the center of
                    the hill
     •    226 ppt - CTDM(11083) with modified MDA, located high on the lee
                    side
     •    146 ppt - CTDM(ll083)-5 with modified MDA, located near 35 m on
                    the windward face
     •    162 ppt - CTDM(11083-E.) with MDA, located high on the windward face
     •    386 ppt - CTDM(11083-E) with modified MDA, located on the south
                    peak
     •    311 ppt - CTDM(ll083-E)-5 with modified MDA,  located near the
                    release height on the windward face.

 The new wind directions and vertical turbulence data make a substantial dif-
 ference in model performance.  Although the location of the peak concentra-
 tion is not close to the location of the peak observed, CTDM(11083-E)  comes
 close to the observed magnitude.   With 5-minute input data, CTDM(11083-E)-5
 underestimates the peak value by a larger margin, but the location is  im-
 proved considerably.

     Experiment 210,  Hour 3

     The peak observed concentration is 297 ppt,  located at the base of the
hill in the lee.  The peak modeled concentrations are:
     *
     9
          0 ppt   - CTDM(11083) with MDA
          2 ppt   - CTDM(11083) with modified MDA
          42 ppt  - CTDM(H083)-5 with modified MDA, located low on
                    the lee side
     •    16 ppt  - CTDMU1083-E) with MDA, located midway up on the
                    side of CCB
     •    148 ppt - CTDM(11083-E) with modified MDA, located high on
                    the lee side
     •    166 ppt - CTDM(11083-E)-5 with modified MDA,  located at the
                    crest.

Once again, the modifications to the input data are extremely important  in
modeling this hour,  but equally important is the "enhancement" in
CTDM(11083-E).  Together, the observed concentration on the crest of the
hill is virtually matched,  and the concentrations high  on the lee side are
approached.  What seems to be missing in CTDM(11083-E)  is the observed
streamline depression in the lee caused by lee waves.

     Experiment 210, Hour 4

     The peak observed concentration is 132 ppt,  located high on the lee
side of the hill.  Peak modeled concentrations are:

     •    1 ppt  - CTDM(11083) with MDA
     •    0 ppt  - CTDM(11083) with modified MDA
     «    3 ppt  - CTDM(ll083)-5 with modified MDA
                                 89

-------
     •    74 ppt - CTDM(11083-E) with MDA, located high on the lee side of
                   CCB
     •    21 ppt - CTDM(11083-E) with modified MDA, located high on the lee
                   side of CCB
     •    37 ppt - CTDM(11083-E)-5 with modified MDA,  located high on the
                   windward face.

It is not at all clear from these results that the modified input data are
any more representative.  CTDM(11083-E) seems to work  better with the MDA.
However, a modified enhancement algorithm that would push the peak enhance-
ment leeward of the crest might improve the results of CTDM(11083-E)-5.
Because photo-derived wind directions are available for less than half of
the hour, and because no photo-derived i? data are available, the modifi-
cations made to the MDA are largely persistence-guided guesses.

     Experiment 210, Hour 6

     The peak observed concentration is 134 ppt, located midway-up on the
lee side of the hill.  Two samplers on this part of the hill actually
recorded 134 ppt, but the next highest concentration is 67 ppt located on
the south peak.  Peak modeled concentrations are:

     •    0 ppt  - CTDM(11083) with MDA, and CTDM(11083), CTDM(11083)-5 with
                   modified MDA
     •    10 ppt - CTDM(11083-E) with MDA, located half-way up on the lee
                   side
     •    7 ppt  - CTDM(11083-E) with modified MDA, located at the same
                   receptor
     •    15 ppt - CTDM(ll083-E)-5 with modified MDA,  located at the same
                   receptor.

Little can be said about why the modeling is so inaccurate at this time.
More study is needed to understand the evolution of the observed concentra-
tions.

     Experiment 210, Hour 7

     The peak observed concentration is 212 ppt, located just leeward of the
south peak.  Peak modeled concentrations are:

     •    69 ppt  - CTDM(11083) with MDA, located at the center of CCB
     •    18 ppt  - CTDM(11083) with modified MDA, located at the center of
                    CCB
     •    45 ppt  - CTDM(ll083)-5 with modified MDA, located just leeward of
                    the crest
     •    166 ppt - CTDM(11083-E) with MDA, located high on the windward face
     •    151 ppt - CTDM(11083-E) with modified MDA, located high on the
                    windward face
     •    195 ppt - CTDM(ll083-E)-5 with modified MDA, located high on the
                    windward face.

     The revised input data have a relatively minor impact on the modeled
concentrations compared to the enhancement of CTDM(11083-E).  It appears

                                  90

-------
from the distribution of concentrations that CTDM(11083-E)  could  do better
with a modified enhancement factor algorithm that  would  shift  the primary
impact area to the lee side.

     Experiment 210,  Hour 8

     The peak observed concentration is 384 ppt,  located near  the base of
the hill in the lee.   The peak modeled concentrations  are:

     «    12 ppt  - CTDM(11083) with MDA,  located  near the  observed peak
     •    5 ppt   - CTDM(11083) with modified MDA,  located  near the observed
                    peak
     •    98 ppt  - CTDM(11083)-5 with modified MDA,  located near the
                    observed peak
     •    114 ppt - CTDM(11083-E) with MDA, located low  on  the side of the
                    hill
     •    112 ppt - CTDM(11083-E) with modified MDA,  located near the
                    observed peak
     *    232 ppt - CTDM(11083-E)-5 with modified  MDA, located low on the
                    side of the hill.

The modified input data have the greatest  impact on the  performance of
CTDM(ll083-E)-5.   The peak is now within a factor  of 2 of the  peak observed
concentration, but the location is shifting away  from  the observed location
in the lee.  It is difficult to judge what types of model modifications
might improve these results because no photo-derived estimates of iz or
wind direction are available.  Modifications made  to the MDA consist of
averaging the MDA wind directions with those from the  30-m  level  of Tower B.
This is based on the correspondence of these data  with photo—derived wind
directions during previous hours.

     Summary

     Some mechanism for increasing concentrations  estimated by Lift over the
hill appears to be necessary.  The CTZ enhancement  algorithm in CTDM(11083-E)
apparently improves the model performance, but  a more  carefully structured
algorithm is needed to better resolve when and  where an  "enhancement" should
occur.  For the present, CTDM(11083-E) can be used to  evaluate the impact of
modifying the input data for Experiments 201 and 210.

     The ratio of the peak observed concentration  to the peak  modeled con-
centration (CTDM(11083-E)) is plotted in Figure 35 to  compare  the two sets
of input data.  The results of using the actual 5-minute sequence of data
(CTDM(ll083-E)-5) are also included to evaluate the importance of simulating
the coupled variability in wind speed, Hc, and  turbulence intensity.  Out
of this set of 12 experiment-hours, 2 hours show no significant change (one
of these had no modifications to the MDA), and  1 hour  shows a  worse model
performance when the best of either the 1-hour  or  the  5-minute version of
CTDM(11083-E) with the modified MDA data is compared to  the 1-hour version
of CTDM(11083-E)  with the MDA.  Therefore, the  overall model performance
statistics will probably improve when the  detailed evaluation  of  the mete-
orological information from all available  sources  is complete.
                                  91

-------
    .
              •f


              1
  55
      *9
  uT 21uj
  CO CO CO
    §co oo
    o o
          555
          a a a
T"

 i
3
-i—r~


 u5   V
                                                                                                o
                                                                                               •H  O
                                                                                                O -H CS
                                                                                             O "^ ^
                                                                                             >-  cd fn
                                                                                             CM  CD <;
                                                                                               rQ Ctj H
                                                                                                  Q (D


                                                                                                (!) -  X
                                                                                                fn  PJ
                                                                                                3 fH
                                                                                                t/) 
-------
     Model performance is not consistently improved by simulating each
5-minute period instead of simulating hourly concentrations with hourly-
averaged meteorological data.  At times, the 5-minute simulation improves
the distribution of concentrations even though the peak estimate is de-
graded  This'behavior may reflect an inadequacy in the model formulation,
and a subsequent improvement' in- the model could lead to even better model
performance'.1               ..:.>•       .
     Comparing mode-led- and observed concentrations with the two sets-of
input data has indicated areas in which CTDM should be changed to better
reflect the observed 'plume behaviour.  Chief among these is the use of a
terrain-driven mechanism similar to that contained in CTDM(11083-E).   Other
areas for model improvement are summarized in Table 3.  Because most of the
hours studied in Experiments 201,-arid 210 are primarily "Lift" hours,  we have
focused bur attention on Lift-related improvements.  These are described in
the. next-subsection.-   •-•    •     :   .;      .,. '

3.4  CTDM Upgrades:  CTDM (1408 3)'   (

     The case-study results for Experiments 201 and 210 reported above
indicate that:
                            !•....                             ,
     •    Equation 41 should be used to estimate az,  with y ^ 0.5,
and a
               zo
                    0. 5 m.
     •    Turbulence may increase within 10 m of the hill surface.

     •    The transition from Wrap,to Lift at Hc is too "severe," espe-
          cially when the mean wind direction is not directed near  the stag-
          nation streamline.

     •    The effects of terrain on the path, distortion, and diffusion of
          the plume over the hill must be explicitly formulated within Lift.

CTDM(11083) already incorporates the first finding, but CTDM must be reform-
ulated to incorporate the remaining three.  At this stage,  we will  treat the
increase in near-surface turbulence as an implicit factor in the plume dis-
tortion and diffusion formulation.

     This section contains a description of the new formulations for the
Lift module, an evaluation of the model performance characteristics for
Experiments 201 and 210, and an overall evaluation of the relative  model
performance when CTDM(14083) is applied to the MDA.

     3.4.1  Lift/Wrap Transition Upgrade

     CTDM(11083) treats all flow beneath Hc as two-dimensional,  horizontal
flow and all flow above Hc as "neutral" flow.  Because the  Lift  module
deforms the horizontal distribution of plume material above H  as if the
flow were part of axial flow over a sphere,  material just above  HC  travels
up and over the hill while that just below is advected round the side.   This
produces a zone of zero concentrations between these two plume segments.
                                  93

-------
         TABLE 3.
Experiment-Hour
SUMMARY OF AREAS FOR IMPROVING CTDM (11083)
BASED ON ANALYSES OF EXPERIMENTS 201 and 210*
           Area  for Model Improvement
    201-18
    201-19
    201-21
    201-22
    201-23
    210-1

    210-2

    210-3

    210-4

    210-6
    210-7

    210-8
      Non-Gaussian formulation
      Unclear
      More horizontal deflection in Lift near H
                                               c
      More horizontal deflection in Lift near HC
      Wrap assumptions near base of hill
      More horizontal deflection in Lift, less in
      Wrap near H
      More horizontal deflection in Lift, less in
      Wrap near H
                 C
      Better terrain effect adjustment in Lift with
      lee wave and wake  sensitivity
      Better terrain effect adjustment in Lift with
      lee wave and wake  sensitivity
      Unclear
      Better terrain effect adjustment in Lift with
      lee wave and wake  sensitivity
      Unclear
*The need for a terrain effect in Lift similar to that in
 CTDM(11083-E) is seen in all hours.
                                 94

-------
     Figure 36 is a schematic illustration of how this abrupt transition
between the Lift and Wrap modules differs from what one would expect  on the
basis of tow-tank studies (e.g., Snyder and Hunt 1983).  When the bulk of
the plume material is near or below Hc, we would expect a significant
horizontal deflection of much of the material above HC.  This deflection
would be greatest for plume segments to the side of the stagnation stream-
line.  Material close to the stagnation streamline, but above Hc,  would
experience relatively little horizontal deflection.  The structure of
CTDM(11083) allows little horizontal deflection for all plume material above
Hc because the effective radius of curvature controls the deformation.
Stratification is not accounted for.

     This departure of the modeled plume distribution from that which is
expected not only produces a qualitatively incorrect distribution of  surface
concentrations, it also can lead to quantitative overestimates of the peak
concentration.  Below Hc, the plume is physically deflected to the side of
the hill so that concentrations on the hill are solely determined by  the
probability that plume material lies along the stagnation streamline.   If
the plume spends little time near the stagnation streamline,  the concentra-
tions on the surface are considerably less than those under the "center" of
the plume.  Above HC, CTDM(11083) shifts the plume very little in the
horizontal so that the "center" of the plume rides over some portion  of the
hill.  For plumes close to Hc in height but away from the stagnation
streamline, the "center" of the plume would be expected to deflect mostly
around the side of the hill instead,  like material below Hc.   Hence,
CTDM(11083) would overestimate peak expected concentrations.

     The effective radius .of curvature can be altered in the Lift  module to
increase the horizontal displacement of streamlines with increasing strat-
ification.  Recall that an offset D was defined to be the difference  between
the effective radius of curvature,  and the height of the hill above HC.
If D is nearly zero, the horizontal shift of plume material away from the
stagnation streamline would approach that of the Wrap module,  but  material
close to the stagnation streamline  would still rise along the surface  over
the hilltop.  The parameter D therefore provides a convenient mechanism
within Lift to smooth the transition between the two module components when
the release height is close to Hc.

     Figure 37 illustrates the geometry.  The example shown in the figure
illustrates how the method shifts the plume material farther  from  the  hill
center as D1 becomes much less than D.  Vertical profiles of  plume material
are weighted by the fraction of time that the wind blows along the effective
direction to the receptor, rather than the actual direction to the receptor
from the source.   Therefore the PDF for small angles from the stagnation
wind direction controls the magnitude of concentrations over most  of  the
hill.

     This geometric method for shifting plume material is just a convenient
mechanism to obtain the desired results.  The central assumption for
actually controlling the degree of redistribution is the parameter D1.
Because we wish to retain the basic structure of CTDM for the time being,  D'
has been designed to respond to the release height of the plume relative to
Hc.  If the plume is released below Hc,  then we set D1  equal  to some

                                  95

-------
                                                                O
                                                                                                    CO
                                                                                                    CO
                                                                                                    00
                                                                                                    o
                                                                                                                          S  B
                                                                                                                          O  ^
                                                                                                                         •H  O
                                                                                                                          •P <4-l
                                                                                                                          n!  
                                                                                                                          (D
4-1   I
 o  to
                                                                                                                          •P  O  Q
                                                                                                                            tn  t-<
                                                                                                                          3 -H  CJ
                                                                                                                              cd  ,0
t>o
                                                                 96

-------
                                                                                                                                0)
                                                                                                                                 O
                                                                                                                                 si
                                                                                                                                • H
                                                                                                                                 4)
                                                                                                                                ^
                                                                                                                                 •P

                                                                                                                                 
                                                                                                                                 Oj
                                                                                                                                 rH
                                                                                                                                 3
                                                                                                                                 
                                                                                                                                     O
                                                                                                                                 Tj  f>
                                                                                                                                 CD   cd
                                                                                                                                 t/5
                                                                                                                                 3 r-i
                                                                                                                                     Oj
                                                                                                                                 X-H
                                                                                                                                 ^  M
                                                                                                                                 •P  0
                                                                                                                                 0  4J
                                                                                                                                 s  S
                                                                                                                                 o  e
                                                                                                                                 +->
                                                                                                                                  o  o
                                                                                                                                  c   c
                                                                                                                                  o   o
                                                                                                                                      10
                                                                                                                                     -H
C
O
                                                                 97

-------
 minimum value so that all plume material above H,, at the impingement
 circle is  distributed over the hill surface by streamlines close to the
 stagnation streamline, regardless  of how large crz is.  If the plume is
 released above Hc, D1  is  set  to some value between D and the minimum value
 of D1, depending on  the ratio of the release height to Hc.  Because we
 have no theory or physical modeling results as yet to formulate an expres-
 sion  for D',  we have chosen a simple weighting expression:
           D1  = D1  .   +  [1 -  (H  /z  )  ]D
                 mm          c  r
                                          for
z  >
 r
(44)
 This expression has  the  desired  limits when zr is equal to Hc and much
 greater  than Hc, and it  shifts to the neutral flow limit more rapidly than
 a  linear interpolation formula.

      3.4.2   Terrain Effects Upgrade

      CTDM(11083-E) incorporates  a measure of the effects of terrain by
 increasing the  size of az over the hill.  This enhancement is viewed as
 a  combination of streamline distortion, diffusion, and increased turbulence
 over the  hill.   The enhancement  factor itself was taken to be the inverse of
 the  "terrain correction" factor  used in the COMPLEX models.  A better
 formulation  is  needed to relate  the enhancement to the individual effects of
 streamtube contraction,  speed changes, increased mixing, and lee wave
 effects.

      The  upgrade to the method of simulating various terrain effects is
 included  in  the Lift module.  As discussed in Section 3.1,  the plume is
 split where  the  dividing-streamline of the flow meets the hill (so).  If
 the hill  surface were flat beyond so,  the surface concentration would be
                   2Q(s ,z) P(9)
                                                                         (45)
                Ho
                 c
where
     a-2 - ar2(s) - 0,,2(sri)
      «5     £t        fj   \J

Q(so,z) is the vertical profile of mass  flux at  s0:

     Q(s0,z) = C0(s0,z)U0  .

The vertical profile of concentration is given by  Equation  24.
                                                                         (46)
                                                                        (47)
     If the terrain is not flat beyond so,  the  flow adjusts to the ter-
rain, which changes the closeness of approach of  streamlines to the hill
surface.  The terrain also changes the speed of the flow,  the turbulence in
the flow, and the trajectory of the plume over  the  hill surface.  Assuming
that the trajectory adjustments are already accounted  for  as in subsection
3.4.1, denote the net effect of plume height changes between so and s in

                                  98

-------
 Equation 45 by T^z-H,,).   Denote the net effect  of travel time,  stream-
 tube  contraction,  and  turbulence  changes  by Taaz',  and  denote  the
 net effect  of flow  speed changes, by  TUUQ.   The  diffusion  of  each
 elemental plume segment' in the vertical profile  at so is modeled as a
 point  source released at  a net height of T^(z-Hc),  into  a  flow of speed
 Tuuo,  with  a diffusion rate  characterized  by  Taaz'  over  the  dis-
 tance  between so and  s:
      C(s,9,0)  = 2
_., •>  .    C (s ,z)u
p(e)  /     oo    o
 s   H  /2iF T a '  I U
      c      a z   u o
 Substituting  for  C0(s0,z),
      C(s,9,0)  =
                                                           dz
(48)
2P(e)
s
Q l
2fra U T T a
CO
.
/Th(z-Hc)\»
\T /2a 'I-
e \a z>
Iz -z
1 r
l/Ta
_e\ z
                         zo o  a u z  H
                    z +z
                     r
              + e
                       ZOi
                              dz
                                                                         (49)
                    ^     /
     If the flow behaved as purely two-dimensional potential  flow,  then
and  Tu would be  inversely related:   TU =  l/Th.   Because  the  flow is
unlikely to behave as purely two-dimensional potential  flow, set

     TU " a/Th

     T 5 Th/Ta

Then the solution to Equation 49 is
     C(s3e,0) =
p(e)
s
1
1
4- a
QT
/lirU ctff "
0 Z
/T(z +H )\ 2
r c \
/To " )
V z /
/T(z -H )\2
1 r c
l/2a " 1 z -H a '
r \ Z /(T+arf * C Z \
Lc (1+crf /2a a" }
zo z
z +H o '
n - P r c z ^
                                             a  '
                                          zo  z
where
                                                                         (50)
                                                                         (51)
     a I2 = a 2(s) -
      z      z
                      zo
     a "2 = a 2(s) - (l-T2)a   2
      Z      Z              ZO
                                                                        (52)
                                  99

-------
         2 —
                (s ) .
       zo     z   o
     Equation 51 takes on a more familiar form when HC  is  set  to  zero:

                                          2
                                    Tz
     rf « n n) - 2P(6)
     C(s,e,0) -- —
                                                                         (53)
                          o  z
The terrain modifications enter through two factors:  a  and  T.   T appears
only in combination with trz".  From Equation 52,

                                                                        (54)
                     + a  2
                        zo
 so  that  the  factor T is seen to modify the rate of growth of az between
 s0  and s, much like, the az enhancement factor of CTDM(11083-E).  If
 o is  set  to  1.0 as in the two-dimensional flow limit, then the terrain
 effect is explicitly a relative increase in dispersion, presumably related
 to  the combined effects of streamtube distortion (enhanced concentration
 gradients) and increased turbulence.  If a is set to T, then the net
 "centerline" concentration in the plume remains unaltered by the terrain,  so
 that  the  terrain effect is explicitly a displacement of the plume toward the
 surface.

      The  dependence of a and T on the parameters controlling the flow is
 not known for certain.  Approximate expressions for the vertical deflection
 of  streamlines in uniform weakly stratified flows (Frjj>l,  but Frj
-------
is reached, we assume that T remains at this value  due  to  a  combination of
streamline adjustments (Smith 1980)  and wake turbulence.

         is defined to be                                          ,
               H-H
                                                                         (55)
where Li/2^Hc^ ^s the half-length of the hill midway between HC  and  •>
the top of the hill, H. FrH is the Froude number based  on the height of
the hill, calculated for the flow above HC.   The distance from the
impingement point (or the base of the hill if HC =  0) to  the greatest ter-
rain elevation along the wind trajectory from the source  to  the  effective
receptor is denoted as LQ.  The distance from the impingement point to the
point where T first attains its minimum value for this  wind  direction is
                  -2.3 Fr
     L = L /(I - e
          o
                                                                  (56)
When Fr-r is much greater than 1,
                           L =  L,
                                      'o'
When FrL is equal to 0.3,
L = 2L .  The coefficient of FrL contained in Equation 56 is  chosen to
place the minimum value of T at the leeward base of the hill  (H =  Hc) when
    attains its minimum expected value (FrL = 0.3).   For receptors  at
distances between s0 and L,  T is obtained by linear interpolation:
                1 - T
     T = T .   +
          mm
                     mm
                   (s   + L  -  s)
                               (57)
    sol
               s 1 so
                      + L
where s is the distance to the receptor,  and so is the distance to  the
impingement point (or the base of the hill if Hc = 0).   T is  set equal  to
1.0 for receptors with s less than so,  and T is held at
tors beyond so + L.
                                                  mm
                                                      for recep-
     The value of T^.^ in Equation 57 should be different for a plume  seg-
ment that passes toward the side of a hill than for a plume segment  that
passes directly over the peak of a hill.   To include this effect,  let
equal To over the peak of a hill, and
      mm
min
                    - T0) (y/a(Hc))2
                                                                  (58)
        if
                      a(Hc)
where y  is the cross-wind position of the receptor in a coordinate system
with x aligned with the mean wind direction, and a(Hc) is the approximate
radius of the hill at an elevation equal to HC.

     Because T/ct is a measure of the amount of dilution experienced by the
modeled  plume as T becomes different from 1.0, we would expect a  to be
approximately equal to 1.0 when the plume is very close to the surface com-
pared to 0Z, and we would expect a to be approximately equal to T when

                                  101

-------
 the plume  is  far  from  the surface compared to az.  When the plume is
 close to the  surface (i.e., close to Hc), much of the decrease in T will
 probably result from increased diffusion because the flow distortion places
 much of the plume within a region of increased turbulence.  When the plume
 is well above the surface, the convergence of streamlines towards the
 surface will  place a good deal more of the "tail" of the plume close to  the
 surface and within the zone of increased turbulence, but the majority of the
 plume will remain above and will not be subject to a marked increase in
 mixing.

     These concepts have been tested by running the new version of the model
 (CTDM(14083)) on  the data for Experiments 201 and 210,  with a set equal  to
 T.   CTDM(14083) incorporates Equations. 44 and 51, and TQ has been set to
 0.5.  By using a.  similar enhancement scale,  the new formulation of the Lift
 model can  be  compared with CTDM(11083-E).  Because a appears outside the
 exponential functions, a comparison of the observed and modeled concentra-
 tion patterns will indicate how a should vary.  This comparison has been
 made with  the use of the modified MDA.
     The ratio of the net release height to the size of az at  the
impingement point ((zr - Hc)/azo) is found to order the behavior of a, and
our expectation of how a should vary is confirmed :

     •    a -\. T when (zr - HC)/CTZO = 3.14, 3.91, 5.17,  8.50

     •    a *> I when (zr - Hc)/azo = -0.20, 0.33, 1.00, 2.00

Note that four of the hours are not pertinent to this evaluation of a.
The comparison also indicates that some plume dilution is  needed immediately
on the windward face of the hill when (zr - Hc)/azo is  less than 0(1.0),
that Equation 57 works well for T when (zr - Hc)/azo is greater than 0(1.0),
and that TQ = 0.5 is generally of the right magnitude when (zr - Hc)/azo
is less than 5.0.  For (zr - Hc)/azo much greater than  5.0,  T  should be
controlled exclusively by the streamline depression.  If we take the minimum
approach distance from potential flow theory for flow over a sphere, then
the lower bound for T0 should be approximately 0.3.

     These findings have been incorporated into CTDM(14083).   When
(zr - Hc)/crzo is less than 1.0,  T has been set equal to TQ at  all
receptors and a has been set equal to 1.0 to simulate dilution of plumes
very close to Hc as they are swept over the hill in a turbulent boundary
layer.  Whenever (zr - HC)/CTZO is greater than 3.0,  a has  been set
equal to T to simulate the importance of plume displacement relative to
dilution.  Linear interpolation is used for a  when  (zr  - Hc)/azo
is between 2.0 and 3.0.  And To varies linearly between 0.5 and 0.3 for
(zr - Hc)/azo between 5.0 and 10.0.  TQ is 0.5 below 5.0,  and  0.3
above 10.0.
     3.4.3  CTDM(14083) Performance Evaluation

     CTDM(14083) was applied to the 12 hours  from Experiments 201 and 210
for which modified input data were developed.   The  results are displayed in
                                 102

-------
Figure 38 in terms of the ratio of the peak observed concentration to  the
peak modeled concentration,  unpaired in space.

     CTDM(14083) is compared with CTDM(11083-E)  with both 1-hour  and
5-minute data in the figure.  When the best results  for either the 1-hour or
5-minute version of the two models are compared, version 14083'does better
than in version 11083-E in 3 of the,hours;  it does about as well  as version
11083-E in 5 of the hours; and does worse in the remaining 4  hours. The
performance of CTDM(14083) tends to be worse than that  of CTDM(11083-E) for
those hours when the peak concentration occurs  on the windward face of the
hill not far above Hc.  In these cases, the enhancement structure in ver-
sion 11083-E begins to increase the dilution of  the  plume dramatically (see
Equation 27), while the structure in version 14083 requires a significant
growth in az with distance before the enhancement can produce a similar
effect (see Equations 52 and 54).  Several hours in  Experiment 201 appear to
require this dilution.  This deficiency in the new version of the Lift
module emphasizes the need to consider the effect of the hill on  the growth
of oz upwind of the point where Hc "intersects"  the  hill.  The tow-
tank experiments of Snyder and Hunt (1983)  document  significant vertical
motion in impinging streamlines below Hc,  and observers during SHIS #1
frequently reported a more "diffuse" plume  when  the  oil-fog plume was
directed toward the hill, below H,,.  For releases well  above  H-.,  CTDM
(14083) generally does better in locating the peak concentration  correctly,
even though the magnitude may not be estimated better.

     CTDM(14083) has also been tested with all  153 hours of data  in the
MDA.  The results are listed in Table 4 along with those from CTDM
(11083-E).  These results are similar to those  contained in Figure 38  in
that no great improvement is evident in the performance of the new version.
The bias has moved marginally closer to the ideal, and  the noise  has
increased marginally.  A slight improvement in the resolution measure  is
also noted.  These results indicate that a significant  improvement in  CTDM
will require a careful evaluation of the structure of a and T, the inclu-
sion of a framework for modifying flow and turbulence assumptions in the
Wrap module, and detailed evaluation of the data contained in the MDA  on a
case-hour by case—hour basis.
                                  103

-------
104

-------











w
Jfi
H
CO
0 O
H H
CO CO
M &
H
H ftS
CO I
en
w oo
0 0
5 c?
3
0 g

M O
s
Q
iJ E?
n
O '"»
S C""!
00
fj-j ^^
0 
-------
                                  SECTION 4

                        THE HOGBACK RIDGE EXPERIMENT
4.1  Geographic and Meteorological Setting

     The Hogback Ridge  (HER) was selected as the site for the second Small
Hill Impaction Study (SHIS #2).  HBR is located in the northwestern corner
of New Mexico, about 15 miles west of Farmington.  It lies on the semiarid
Colorado Plateau near the western slopes of the San Juan Mountains
(Figure 39).  Three rivers—La Plata, Animas, and Chaco—drain into the San
Juan River near Farmington.  The terrain features of the area include
occasional isolated ridges (e.g., HBR), isolated promontories (e.g.,
Shiprock), and low mesas and plateaus.  The area is characterized by a
sparse vegetative cover of desert shrubs and grasses.

     SHIS #2 was conducted in the environs of an approximately 1.5-km long
section of the ridge north of the San Juan River (Figure 40).  Here, the
Hogback is oriented NNE-SSW and rises about 85 m above its base elevation.
HBR extends from just north of the SHIS #2 experimental area to about 8
miles south of the San Juan River.   The ridge splits where the San Juan and
Chaco Rivers flow westward and forms separate "hogbacks."  The two southern
sections, separated by the Chaco River, are shown in Figure 41 as viewed
with a telephoto lens from the SHIS #2 site; the plume from the Four Corners
plant is passing over the ridge.  Waughan Arroyo is located just east
(upwind) of the experimental section.  Farther east a series of irregular
mesas, arroyos, and surface coal mines extend all the way to Farmington.
Figure 42 presents a view of the experimental section of HBR from the east,
and Figure 43 shows the area east of the section, as viewed from the top of
HBR.

     Because of substantial reserves of coal and the adequate water supply
from the San Juan River, two major electric generating facilities operate in
the region.  The San Juan Power Plant is located just east of the experi-
mental area,  and the Four Corners Power Plant is located south of the San
Juan River.  Public Service Company of New Mexico (PNM)  and Arizona Public
Service Company (APS),  the respective operators of the two generating
stations, have sponsored several meteorological and air quality measurement
programs.  The National Weather Service operates a weather station at the
Farmington (FMN) Airport.   Consequently, there was a wealth of meteorologi-
cal data available during  the site  selection and experimental design phases.
                                  106

-------
               ^r-    'jA«

'- "?i,!*?'*   &-•   -   " "'^?OT^?^y^ v
jie** .    - '"" .>> %& _  r^ %,^v -^
   asiL
S^\""
TV-'- -N(i '•  • -*•» !r*rru: P S^*^*  %'^/&*WR*;


&•:

                                                                                                             0)
                                                                                                             t>0
                                                                                                            •a
                                                                                                             o
                                                                                                             (D

                                                                                                            •6
                                    (S)
                                    =tt=
                                                                                                            en
                                                                                                             4-1
                                                                                                             O
                                                                                                             f-i
                                                                                                             oj


                                                                                                             O
                                                                                                             •H
                                                                                                             M
                                                                                                             CD
                                                                                                             a
                                                                                                             3.
                                                                                                             •H
                                                         107

-------
Figure 40.   The SHIS #2 site.  Contour interval 20 m.
                          108

-------
                                                       §
                                                       c/}
                                                       0)
                                                       o
                                                       X!
                                                       3
                                                       o
                                                       CD
                                                       •H
                                                       •H
                                                       a.
109

-------
                                               0
                                               
-------
                                                                       •a
                                                                        O
                                                                       33

                                                                        CD

                                                                       •P

                                                                        S
                                                                        O
                                                                        (D
                                                                        1


                                                                       4->


                                                                        O
                                                                       (D
                                                                       i
                                                                       O
                                                                       •M
                                                                       •H
                                                                      to
                                                                       0)
                                                                       f-l


                                                                       bO

                                                                      •H
                                                                      BH
111

-------
      The meteorological  measurements  taken by  PNM,  APS  and  the National
 Weather Service provide  detailed  information on the dispersion climatology
 of the HER region.   Most of the available  information has been analyzed and
 summarized recently by Moore et al. (1981).  Figure 44  shows two wind roses
 derived from FMN data.   The first rose  shows the  distribution of winds
 during 1976;  the second  shows the distribution of Oct-Dec winds for
 stability "E" conditions.   Figure 45  shows October  hourly FMN wind roses for
 four  nighttime hours—2200,  0000,  0200, and 0400.   Evidently, the wind
 typically "turns around" (Crow 1975)  from  westerly  to easterly during the
 night as  the  drainage from  the  San Juan Mountains is established.

      Drainage periods occur often during the summer and fall months and
 produce stable easterly  flows toward  HBR at night.  The drainage and
 turnaround days are characterized by  light winds all day.  Cooling in the
 evening and night results in strong surface inversions.  Dense air flows
 from  the  mountains  down  the river valleys  and  the Chaco wash.  Stable air
 flows toward  HBR along the  San  Juan River  valley, around Pinon Mesa from the
 northeast and occasionally  from the southeast.  Figure 46 (taken from Moore
 et al.  1981)  illustrates a  typical summer morning downslope drainage flow
 situation.  Figure  47 shows representative early morning upper air soundings
 during easterly drainage situations.  The  deep stable layer, which is
 uncoupled from the  synoptic westerly  winds aloft, continues well into the
 morning until  convective turbulence destroys the inversion,  and the flow
 near  the  ground is  again coupled  with the  synoptic winds aloft.

      In summary, the available meteorological data suggested that frequent
 stable easterly winds occur at  night  during the fall months.  This was the
 principal  reason for selecting  HBR as the site for SHIS #2.   Other reasons
 included:   (1)  HBR  is the dominant terrain feature  in the area of interest;
 (2) the area  is  easily accessible and has electric power available; (3)  PNM
 provided  records of meteorological data taken on and near HBR and was
 willing to provide the data  in  real-time during the experiment;  and (4)  the
 Bureau of  Land Management,  which manages the HBR and the area to its east,
 granted permission to use the ridge.

 4.2   Preliminary Flow Visualization Experiment

     A preliminary flow visualization study was conducted at a section of
HBR just south  of the power transmission lines  (see Figure 40) during the
period June 6 - June 11,  1982.  Scientists  from ERT, EPA/Terrain Effects
Branch (TEB), EPA/Fluid Modeling Facility (FMF),  NOAA/Wave Propagation Lab
 (WPL)  and the Los Alamos Scientific Lab  (LASL)  participated.

     Meteorological data were obtained using two tethersondes,  pibals and
fixed meteorological sensors on three  towers (two  10-m  towers  on the  Hogback
and a 200-ft tower east of the arroyo) operated by PNM.   Smoke was  released
using smoke candles and canisters  suspended from a 600-ft^ blimp  at a
 location approximately 150 m east-southeast of  the road along  the base of
ridge and 300 m south of the major transmission lines crossing the  ridge.
                                 112

-------
                                5.3
                    7.0
19.7
                                              11.3
T=Q28-°
   Figure 44(a).    Climatological annual wind rose for Farmington,  N.M.
                    (2927 valid data points,  3-hour NWS data for  1976).
                        2.8
                             2.0 2-0 2.8
          12.2[
         28.4
                 7.2
                         2.8
                                2.4
                                                   Percent Occurrence
                                       0.0      8.0      16.0      24.0     32.0
                                            ,1-3  8-12,
                                               4'7    13-18  19-24   Over24 (mph)
    Figure 44(b).   October-December stability E Farmington wind  rose.
                                       113

-------
                                                                           Percent Occurrence
                                                                 0.0     8.0     16.0    24.0    32.0



                                                                  ^"—^  4'7    13-18  19-24  Over 24 (mph)
               Faimington. New Mexico — Airport (FAA) October (1977-19811 Hourly Rose - Hour (LSI) = 22
               Farmington. Now Mexico — Airport (FAA) October (1977-1981) Hourly Rose - Hour (LST) = 24
                                                                  313,5
               Formmgton. New Mexico — Airport (FAA| October |1977-1981) Hourly Rose - Hour (LST| = 02
               Farmington, New Mexico — Airport (FAA) October (1977-1981) Hourly Rose - Hour (LST) = 04

Figure  45.     October  hourly  wind roses  for  four  nighttime
                    hours:   2200,   0000,  0200,  and  0400 MST.
                                          114

-------
                                                                      090
                                                   i i *i i I i i i i i  i i i i
                    710
720        730
   ERSTING
                                                      740
750
                                                                     020
Figure 46.   Summer morning downslope drainage flow.   (Taken from Moore et al,
             1981.)  The squares (ffl)  indicate the San Juan and Four Corners
             power plants.  The SHIS  #2 site is just  west of San Juan at UTM
             coordinates 722 E, 4074  N.
                                     115

-------
                                                 111 Tl-t H H H H+H UlU U HiU IU III1111111111
                                                                         I I I III I II I II I I I M
                                                                                       in     o
                                                                                          uj  2^
                                                                                          a:
                                                                                          ^  a
                                                                                          h-   36
                                                                                          tr   *
                                                                                       a a:  o
                                                                                          UJ
 o
 CM
CO      CO       <*•      CM       O       03       ID
«-1       4-1
                                                                                                        O rt
                                                                                                        O If)
                                                                                                           O
                                                                                                           -H
                                                                                                           •H
                                                                                                        to  -d
                                                                                                        c  o
                                                                                                        •H  O
                                                                                               3  cti
                                                                                               o  o
                                                                                               Ul -H
                                                                                                  bO
                                                                                               H  O
                                                                                              •H rH
                                                                                               nj  O
                                                                                                  ^i
                                                                                               !-H  O
                                                                                               CD  CD
                                                                                               4)
                                                                                               ^

                                                                                               M
                                                                                              •H
                                                                                              PH
                                                    116

-------
     WPL operated a tetherson.de near the blimp to document  meteorological
conditions near the source.   LASL operated a tethersonde east of Waughan
Arroyo to document conditions upwind of the source.   In addition, EPA
released pibals and neutrally buoyant ballons.

     The flow visualization experiments were documented with 35 mm cameras
and 8 mm movie cameras.   The meteorological data and  the photographs were
analyzed to aid in the SHIS #2 experimental design.   Specifically, the  flow
visualization experiment helped in the following areas:

     «    selection of the section of HBR for SHIS #2,
     •    location of the roads for the release crane,
     •    location of the tracer gas sampler locations,
     9    location of the meteorological towers and other meteorological
          sensing devices, and
     •    location of the lidar.

     In addition, photographs and tethersonde data were analyzed  to test the
critical dividing-streamline height (Hc) concept for  the Hogback.  Table 5
summarizes three experiments during which plumes released above H  went up
and over the ridge, and plumes below the HC stagnated and tended to remain
horizontal.  The release height is denoted by zr.

4.3  Fluid Modeling in Support of the SHIS #2 Experimental  Design

     During September 1982,  a series of wind tunnel and towing tank flow
visualization experiments were conducted at the EPA FMF to  provide input to
the SHIS #2 experimental design.  The studies were designed to investigate:

     *    plume height above the surface over the hill  crest and at the
          upwind edge of the hill,
     •    apparent size of any plume deformation upwind of  the hill,
     •    lee wave importance and structure, and
     •    sensitivity of the plume trajectory to "wind  angle."

This information was then used to (l) guide the design  of the smoke and
tracer gas release protocols at HBR, and (2) help select  sampler  and camera
locations.

     Two models of the Hogback were used—one for the tow-tank and a second
for the wind tunnel.  Each has a scaled radius equivalent to 800 m, and the
northernmost point is the prominent peak just north of  the  high tension
power lines near PNM station 105.  The ends of the models were extended to
the walls of the test chambers.  Peak elevations above  the  base plate did
not exceed 14 cm (84 m,  scaled).  Each model was covered with approximately
3 mm size gravel.

     Two tests were made in the wind tunnel.  One test  was  done with the
ridge perpendicular to the flow, the other with the ridge rotated by 30°.
Because the flow was very turbulent (boundary layer about l~m thick) with
the 'trip fence in place, the fence was removed to gain  a better measure of
plume centerline height.
                                  117

-------
            TABLE 5.  PRELIMINARY VISUALIZATION EXPERIMENT
1.
11 June
      z.
 1982
. = 40
                      EXAMPLES OF Hc ANALYSES
                  and 60 m at 0539MDT
          Hc = 37 m at 0456 (WPL #21 ascent)
               10 m at 0528 (WPL #21 descent)
                6 m at 0503 (LASL #35 ascent)

     Observer comment:  Smoke  largely made  it over  from both release
     heights.

     Photos:   Narrow plumes definitely above HC.

    11 June  1982
          zr  = 60 m at 0620
          Hc  =  8 m at 0557 (WPL #22  ascent)
               39 m at 0626 (WPL #22  descent)
                1 m at 0603 (LASL #36 descent)

     Photos:   Narrow plume  above Hc

    10 June 1982
          zr - 15 and  35 m  at 0650
          Hc = 30 m at 0635 (WPL #17 ascent)
               19 m at 0650 (WPL #17 descent)

    Observer  comment:  The 15-m level is very diffuse and the  35-ra
     level maintains some integrity and it has some definite direction
     to it.  The  15-m  level release is going right down to the  ground
    within a  few tens of meters from the source.  Smoke from the 35-m
    release seems  to be approaching the hill slope but getting up and
    over.  The 15-m smoke  release is involved in all the hillocks at
    the bottom of the Hogback but seems to be getting up and over the
    ridge.  It does not appear to be blocked.

    Photos:  Hr is between 15 and 35 m.
                                 118

-------
     In the first test, smoke was released 10 cm above the floor of the
tunnel.  The centerline appeared to be about 9-11 cm above the surface of
the model near the base of the hill, and about 6 cm above the surface at the
crest (l cm = 6 m) .  Therefore, the plume maintained its initial height
upwind of the hill and approached the crest surface at nearly half its
initial height.  This is approximately consistent with what would be
expected in potential flow.  The plume was forced to speed up over the crest
of the ridge, but no region of deformation could be seen upwind of the hill.

     In the second test, the flow angle produced little additional effect on
streamline height but did produce a small plume path deflection.  Although
the deflection was difficult to quantify, it was estimated to be about 8 cm
(about 1/2 the visible plume width).  Because the smoke was released about
800 m from the crest (scaled), this deflection (^50 m, scaled) was less
than 4°.

     Eight individual tests (Table 6) were run in the tow-tank.  The first
three were run with Froude numbers of 0.8,  1.0, and 2.0 and with a wind
direction of 118° (wind flow perpendicular to the ridge).  In the cases of
Frjj =0.8 and 1.0, tow speeds were set to produce Hc values of 0.2 H and
0.0 H.  Consequently, bulk Froude numbers were probably less than 1.0 in
both tows, due to the mixed layer near the surface of the model.

     Dye streaks were released at heights of 2, 6,  10, 14,  18, 22,  25,  30,
35, 40, and 45 cm at the upstream edge of the baseplate.  These heights are
referenced to the water surface.   The dye streamer released at 2 cm was
mixed in a stagnation region upstream of the ridge during the first tow.

     Tows 4 and 5 had Frjj =1.0 and 2.0,  respectively, based on the linear
density profile above the surface mixed layer (no mixed layer was present
during the first tow at Frji = 1.0).   Dye release heights were switched  to
2, 10, 18, 25, 35, and 45 cm.  Dye released at 10 cm Ov60 m, scaled) rose
to 15 cm above the surface at the upstream edge of the Hogback face in both
tows and dropped back to 9 cm and 10 cm (Fr = 1.0,  2.0) at  the crest except
near the end of the fourth tow (Figure 48)  when the dye passed over the
crest at an elevation of about 3  cm.  Table 7 summarizes the measured
heights of the'10-cm streamer for five experiments.

     In experiments 6 and 7 the model was rotated 10°  and retowed at
Frjj = 1.0, and 2.0.  Streamline deflections at 2 cm were small in both
tows.  A rough estimate puts it at 2-3°.   A final experiment 8 was  done with
    = » (neutral conditions) .
     The FMF flow visualization experiments suggested that  during
unstratified conditions the streamline patterns  are  similar to those
expected from potential flow theory.   The plume  should approach  HBR at  its
initial elevation upwind of the ridge, and flow  over the  crest at nearly
half its height.  During the weakly stratified simulated  conditions,  the
tow-tank experiments suggested a rise in streamlines near the upwind  base of
the hill and a fall over the crest to near or slightly lower than its
                                 119

-------
                   TABLE  6.  FMF TOW-TANK EXPERIMENTS
Tow No.
 HOG 1
HOG  2
HOG 3
HOG 4
HOG 5



HOG 6



HOG 7

HOG 8
Direction
  118C
  118'
  118'
  118°
  118'



  108C



  108°

  108°
 0.8
 1.0
2.0
1.0
2.0
1.0
2.0
  Remarks

 Slight  shifting  at  lower  levels
 toward  south, but all streamers
 over ridge top.  Strong
 contraction of streamers observed
 on  lee  side.  Slow motion below
 hill top.

 More pronounced  shift of lower
 level streamers  to  south, but for
 shorter time period.  Wider
 spreading on lee side.  Rotor
 observed at 4-5 hill heights on
 lee side.  Lower streamers upwind.

 Straight narrow plumes.   Less
 contraction in lee.

 Dye released at 2,  10, 18, 25,
 35, 45  cm.  Stack visibility
 good.  Fairly wide diffusion in
 lower levels.  Blockage  noted
 upstream.

 Plumes somewhat wider.  Distance
 from crest to 1st max. in lee
 wave ^ 4. 9 m.

 Lower plumes twist or deviate to
south upstream consistently,
straighten out downstream.

Slight S-shape in trajectory.

More plume meandering.  Wider
dispersion.
                                 120

-------
TABLE 7.  MEASURED HEIGHTS  OF/THE  10-cm HIGH'STREAMER AT THE
          UPWIND  BASE AND CREST OF THE HER MODEL
                   Height  (cm)  at
                   Upstream Base
     HOG 4  FrH =1.0

                    14
                    15
                    15
                    15
                    15
                    14
                    12
  Height (cm) at
       Crest
8.5
9
9
8
7
5
3
     HOG 5  FrH=  2.0
                    15
                    15
                    15
                    15
                    15
9.5
10
11
10
9.5
     HOG 6  FrH =1.0

                    12
                    14
                    14
                    14
                    15
8.
9
9
8
8
     HOG 8  FrH =
     Wind Tunnel
                    13
                    10
                                  121

-------
a.  Side view
b.  Top view
     "Flow"—i
Figure 48.   FMF tow tank flow visualization experiment - Hogback Ridge
             (wind direction = 118°, Fr = 1.0).
                                  122

-------
Initial height.*  Low-level releases near or below Hc are expected to
experience considerable mixing.  Wind direction variability will  not  be
significantly amplified.

4.4  Experimental Design

     The SHIS #2 at the Hogback was designed to obtain meteorological,  flow
visualization, and tracer gas emission and concentration data in  the
vicinity of a nearly two-dimensional ridge in order to enlarge the modeling
data base and to provide a good basis for testing,  evaluating,  and refining
the modeling concepts developed from the CCB data base and the various  FMF
experiments.  The experimental methods of the SHIS  #2 were similar to those
used and tested at CCB.  The experiment was conducted at the Hogback  from
October 5 - October 29, 1982 and included:

     •    Releases of two tracer gases (SFg and Freon 13B1 (CF3Br)) and
          oil-fog, using a mobile 150-ft crane and  a tower as source
          platforms;

     •    Fixed meteorological measurements:
          -    a 150-m tower instrumented at ten levels,
          -    a 30-m tower instrumented at five levels,
          -    a 10-m tower instrumented at three levels,
          -    a 60-m tower instrumented at two levels,**
               two monostatic acoustic sounders,
               a doppler acoustic sounder, and
          -    three optical crosswind anemometers;

     •    Two tethersondes:
          -    one operated at source elevation to document meteorological
               conditions representative of the source,  and
               one operated to measure vertical profiles of meteorological
               parameters upwind of the ridge;

     •    Ground-level tracer gas concentrations;

     •    Lidar measurements; and

     •    Photographs and videotapes.
*The initial rise (e.g., from 10 cm to 15 cm) near the base of the HER
 model was not expected.  Initial analyses of the SHIS #2 photographs
 suggest the occurrence of this phenomenon in the field,  although the
 results are currently obfuscated by the thermal plume rise of the
 oil-fog plume.

**Tower P was operated by PNM.  PNM maintained temperature and
  temperature difference sensors on the tower.  ERT installed  two
  cup-and-vane sensors for SHIS #2.  In addition, PNM operated a
  network of instruments in the area,  including two 10-m  towers on the
  Hogback.  The PNM data were displayed in real-time at the ERT
  command post.

                                  123

-------
     After three smoke visualization experiments,  11 combined tracer and
flow visualization experiments were performed.   The three initial
experiments were conducted to (1) understand the autumn weather conditions
at HBR (the preliminary experiment was conducted in June),  (2)  gain
experience working at the HBR site, and (3) finalize the release and
sampling protocols.  During the 11 tracer and simultaneous  flow
visualization experiments, SFg and CF3Br were released, sampled and
analyzed for concentrations on the ridge.  The meteorological data were
archived and displayed in real-time by a system of onsite minicomputers.
The real-time information on ambient meteorological conditions and the
flexibility of releasing oil-fog and tracer gases at a wide variety of
heights and locations allowed the field managers real-time control of the
experiment in the selection of source positions to obtain useful information
for model development purposes.  The real-time meteorological feed-back was
supplemented by near real-time lidar observations and an approximately
48-hour turnaround time on the photographs.  Figure 49 illustrates the
layout of the SHIS #2.

     The SHIS #2 participants and their principal responsibilities were*:

     •    ERT
          -    overall CTMD project management
               field management and control (selection of experiment times,
               release locations, heights, etc.)
               the 150-m and 60-m meteorological towers
          -    site logistics (power, roads, security, weather forecasts,
               etc.)
          -    quality assurance

     •    NOAA ARLFRO
          -    tracer gas releases, sampling and analysis
          -    oil-fog generator
          -    the 10-m and 30-m towers
               telemetry and meteorological data archive and display system

     •    NOAA WPL
          -    lidar
          —    two sonic anemometers and data acquisition system at the
               150-m tower
          -    two monostatic acoustic sounders
          -    one doppler acoustic sounder
          -    three optical anemometers
          -    one tethersonde

     •    Morrison-Knudsen Company (under subcontract to ARLFRO)
          -    oil-fog plume photographs (five positions)
          -    video tapes of the plume
*The details of the SHIS #2 are presented in the "Work Plan for the
 Small Hill Impaction Study No. 2," ERT Document P-B348-620,
 September 1982.

                                  124

-------
                                                    LEGEND
                                                   A 500' Tower A
                                                   A Tower B
                                                   O Tower C
                                                    1 Tracer Release Pt. No. R-80
                                                    2 Tracer Release Pt. No. 203
                                                   • 3 Tracer Release Pt. No. 21 5
                                                   * 4 Tracer Release Pt. No. 21 6
                                                    5 Tracer Release Pt. No. 111
Figure  49.
SHIS #2 field  experiment layout.

            125

-------
     •    NOAA ATDL (under agreement with ARLFRO)
          -    one tethersonde
               video tapes of the plume during the first three experiments

     •    TRC Environmental Consultants (under subcontract to ERT)
          -    external quality assurance audit

     4.4.1  Meteorological Measurements

     Meteorological Towers

     Four meteorological towers were operated during the SHIS #2.   ERT
installed and operated the 150-m and 60-m towers.  ARLFRO installed and
operated the 10-m and 30-m towers.  Tables 8A through D list the instru-
ments, heights, and direct and derived measures* for the four towers.
Table 9 provides definitions of the measures.  Approximately 242 5-minute
and 71 1-hour measures were archived in real-time.  ARLFRO was responsible
for the data collection and display system (see Section 4.4.2).  The 150-m
tower was also outfitted with a tracer gas and smoke release platform and
winch assembly to allow releases to an elevation of approximately  75 m.

     NOAA WPL provided two sonic anemometer systems that measured  u,  v, w,
and temperature at a high frequency (20 Hz).  The sonic systems were located
at the 40-m and 5-m levels of the 150-m tower.  WPL also provided  a separate
data logging system to archive the sonic data.

     The purpose of Tower A was to characterize the approach flow  and  the
meteorological conditions at the release heights.  Tower B (30 m)  was
located west of the main road toward the base of the Hogback to document
changes in flows and turbulence intensities in this region of highly dis-
torted flow.  Tower C (10 m) was located at the windward side of the top
of.the Hogback to document the wind, temperature, and turbulence field below
10 m over the hill crest.  The 60-m Tower P was operated by PNM.  ERT in-
stalled two wind sensors on it to help document the approach flow  well
upwind of HBR.  PNM also operated other meteorological instruments in the
area (Figure 50), data from which were available in real-time during SHIS #2
as half-hour averages.

     Tethersondes

     Two tethersondes were operated to measure temperature, horizontal wind
speed and direction, pressure, and humidity.  One sonde was operated by ATDL
to characterize meteorological conditions within a few meters of the source.
A second sonde was operated by WPL to obtain vertical profiles east of the
ridge.  The WPL sonde was operated near Tower B until October 15 and near
the doppler sounder through the end of the experiment.  Data were  recorded
*Measure is used here to indicate a direct measurement (e.g.,  the
 u-coraponent of the wind) or a calculated parameter (e.g.,  turbulence
 intensity).
                                 126

-------
en
                                                 X    X    X    X         X    X    X    X    X
 5  S
 Crf CTl
 H .-i
 co \o
 2: r-i
CQ
  „
erf
    cd
                                                                          XXX
S  I
 I  O
O  H
00

s
CQ



,-^
g
^-^
u
00
•H
HI
33


en
CO
0.
o

Ca


s-^
n
(0
P.
3
O



"3 2


H <


0)
0)
VJ
3
CO
W
0)
S
                                                           co    en    H
                                                                          H
                                                                          1=1
                                                                               OT   CO   M
en  o
S  vD
1-H  i-H

<;  ca
 — CO
W *-s  > 0 ° H

X - x" Pn H
O a co CQ EH H Q
X
X


%

CO

sT
CO
X
X






a
CO
                                                                                       g
                                                                                       CO
                                                                                                              XXX
                                                                                                                  H   H
                                                                                                                  3   co
                                                         127

-------
 co

    co
 CO vO
 D-,
    CD
    CD
 00
CO /-v
E- J
2: to
E- co
CO \0
o

OS
    CD
CJ
CO
CO
                                        128

-------
                   TABLE 9.   DEFINITIONS OF THE  MEASURES

U, V, W     Vector averaged wind components - props
UX, VX      Vector averaged wind components - cups
S, D        Scalar mean wind speed and direction - props
SX, DX      Scalar mean wind speed and direction - cups
TF          Fast Thermistor Temperature
T           Slow RTD Temperature
DT, TC      Slow RTD AT, and calculated T (T(2m)* + DT)
SU, SV, SW  Turbulence Scales - sigma-u, sigraa-v, sigma-w
SD, SDX     Sigma-theta (props),  sigma-theta (cups)
UW, WT
ST
U'W', W'T'
Standard Deviation of TF
m/s
m/s
m/s, degree
ra/s, degree
°C
°C
°C
m/s
degree
     2
(m/s),  m/s °C
°C
* Or 1 m for the 30-m tower and 9.1  m for the  60-m  tower.
                                   129

-------
Figure 50.   PNM air quality and meteorological monitoring sites,
                             130

-------
 on digital  cassette  tape  for processing.  Real-time paper copies were
 available to help guide the experiment.

     Monostatic Acoustic  Sounders

     Two monostatic  acoustic sounders provided data from which the vertical
 profile of  turbulence and stability of the planetary boundary layer above
 the sounder were inferred.  The two sounders were operated around the clock
 and the data were recorded on facsimile charts.  This information was useful
 for experimental guidance by showing the development in time of the vertical
 structure of the lower atmosphere, particularly the arrival of air masses
 with different characteristics of temperature profile, shear, and turbulence.

     One monostatic  sounder was operated near Tower B upwind of the Hogback.
 A  second sounder was operated in the vicinity of the 150-m tower, about
 800 m east  of the ridge.

     Doppler Acoustic Sounder

     The doppler acoustic sounder measured the vertical profile of the
 horizontal  wind velocity in the planetary boundary layer above the sounder.
 Vertical resolution  was 15 m with adequate signals to a height of typically
 1 km.  The  doppler acoustic sounder was installed about 2 km east of the
 targeted segment of  the ridge in the vicinity of the Shiprock Substation
 (Figure 49).

     Optical Crosswind Anemometers

     An optical crosswind anemometer measures the path—averaged wind across
 the line-of-sight path between a transmitter and receiver.   Three were used
 in  the SHIS #2 (Figure 52).

     4.4.2  Meteorological Data System (MDS)

     NOAA ARLFRO provided a real-time Meteorological Data System (MDS) to
 acquire, process,  display, and store data.   Figure 51  depicts the component
 structure of the system.  Operating continuously during each experiment,  the
MDS sampled the 86 meteorological  sensor inputs,  calculated  the derived
measures, and displayed selected parameters and profiles.   Data from the
 10-m, 30-ra  and 60-m towers were transmitted by ARLFRO  radio  links.   The
 150-m tower data were transmitted  by shielded signal cable.

     Additional meteorological data were available from the  PNM stations  in
 the area.   The stations of particular relevance are located  on top of the
Hogback; station 103 is about  5 km SW of the experiment area,  and station
 105 is just to the north of the experiment  area.   The  PNM data were available
as half-hour averages,  which were  intercepted from the  PNM RF communications
 links at the 60-m tower and telemetered by  ARLFRO's radio  link to the
command post near the 150-m tower,  where they were displayed on a line
printer at each half-hour scan.
                                  131

-------



CO
ft
S o
W 0
1 g
10 10
JT3
c
•s **
5 n
g- ^X
ra -5
2 §
S «
D j=
CJ to
Z 0>
i 1



•g I"
*S i- •
to o O
c w
E o 15
••"S
w







CD
D>
2
§
0
c
o
1
O
O)
11
0 8
2 g
1? 3
S W
1 1














S S
to iH.
' o
* & 8
S 2 I
M w t;
S o 2
i s "
•la"
E 2 .=
If) W 

UJ •? In 1 1 Q. Q. 0 U. CM ^r *£ «£ ^•^ C jg U) 5 CO fx N. 9 5 0) M 00 z .2 c Q. \ co CM r 1 DC =™ 1 1 S .H J ' » ^ g1 1 1H £ * 2 ca .a c >. s f JS <= .2 .2 « S /^\"\\ sx "O — j §• / CQ \ \\ W ro ' ' ' to 2 1 U 5 % § s O I II "^ .. a> o , 00 / // != co " •£ \3/ J/ .g 5^3 ,s 1 1 O — 1- O 'a. O5 CO 0 S» ^Q IO -o X *- * CO m I sand indices tter diagram •^ o to w (0 O i? s 03 'IT "5. S •o -o -5 § ^ » 0) W « o li 0. 1 ^ " \ Q. \ 0 \ ° 1 X to G O •H •H •H Cti Cd 4-J cd bO O O M O tu S bO 132


-------
     4.4.3  Tracer Gas Release, Sampling and Analysis

     Tracer Release System

     Tracer releases were made from the 150-m tower and from locations on
the three roads east of HBR.  The roads were culverted, graded and graveled
to support the 150-ft crane that lifted the oil-fog generator and the tracer
gas tubes.  Two tracer gases,  SFg and CF3Br, were released at different
heights from the tower or the boom of the mobile crane.  The SFg release
was co-located with the oil-fog release and dispersed from a common nozzle.
Occasionally, the SFg and oil-fog were released from the tower while the
CF-jBr was released from the crane at a different location.

     The SFg and CF-^Br tracer gases were stored in individual compressed
gas cylinders at ground level.  Piping carried each gas through a linear
mass flow meter (LFM) system to the point of discharge into the atmosphere.
A time history of each tracer release was used to describe the rate and
quantity of release of tracer.  The LFM measured and displayed the rate of
gaseous tracer discharge via real-time digital display, the total amount of
gas discharged via a digital counter, and the analog output voltage directly
proportional to the flow rate.  The voltage was logged and monitored on a
strip chart recorder.  Pre- and post—test release weights of gas tracer
cylinders were measured by certified scales.  Beginning and ending times of
tracer release and the time and character of any deviations from the design
rate-of-release were logged.

     Tracer Sampling

     Tracer samples were collected in 2-liter Tedlar bags at about 110
locations on the ridge.  ARLFRO operated 125 samplers during each approxi-
mately 8-hour experiment.  Twenty samplers were used to get 10-minute
averages at five locations (4 samplers x 12 bags = 48 10-minute samples for
each of the five locations).  The remaining 105 were used to get 1-hour
samples.  Two of the 1-hour samplers were operated on the 30-m tower and one
on the 10-m tower.

     The bag samples were collected by means of modified EMI AQSIII or
similar type of air sampler.   Each sampler used 12 separate pumps,  bags,
and external tubes to draw in ambient air to fill the individual 2-liter
Tedlar bags.  The system was battery powered and electronically programmed
in function and timing.  Time was set and maintained by a crystal-controlled
digital clock accurate to within 1 minute per month.   Beginning and ending
sampling times for the individual (sequential) whole-air samples were
controlled by this clock.  The actual local time (MDT)  for beginning of the
sampling sequence (in each unit) was preprogrammed during servicing by
sampling team technicians within about 20 hours prior to the start of each
experiment.

     The sampler locations (Figure 52) were selected by ERT scientists on
the basis of the June preliminary flow visualization experiments,  the FMF
wind tunnel and tow-tank simulations, and the meteorological data collected
by PNM.  Tables 10 and 11 summarize the characteristics of the sampling
grid.  The sampler locations were selected to provide the widest horizontal

                                  133

-------
                                                                   tfl
                                                                   PL.
                                                                   CD
                                                                   O
                                                                   O   •
                                                                   O  r—*
                                                                  i-H  CD

                                                                   in  cd
                                                                   CD  S
                                                                   CD  v—'
                                                                   CJ
                                                                   ni  C_3
                                                                  =tt=

                                                                  cn  oa
134

-------
        TABLE  10.     PRIMARY SAMPLER LOCATIONS


A.  Four Primary Sampler Rows  on Windward Face of HER

     •    #1 is ^10 m above  the base  of Tower B

               16 locations  centered  (approx) on Tower B (60-series
               stakes)

               80-m horizontal spacing covering 1200 m


     •    #2 is ^25 m above  #1

               17 locations  centered  (approx) on Tower B (60-series
               stakes)

               40-m horizontal spacing out  to 200 m, 'v-SO-m spacing
               to approx 460 m'from center;  total range is 920 m.


     •    #3 is 1-25 m above  #2

          -    21 locations  centered  (approx) on Tower B (60-series
               stakes)

               40-m horizontal spacing covering 800 m


     •    #4 follows "crest";  mean height  is •v-lO m above #3

               22 locations  centered  (approx) on Tower B (60-series
               stakes)

          -    nominal 40-m spacing covers  860 m

B.  Three Secondary Rows on  "Windward" Face of HER

     •    #1 lies among hillocks  at base  of HER

          -    5 samplers centered on Tower B

          -    middle is at  Tower B;  adjacent 2 are atop hillocks near
               the road; 2 "ends" are in  low areas adjacent to these.

          -    row covers 280  m


     •    #2 lies between primary rows #1  and #2 (-vl3 m above #1)

          -    4 samplers centered (approx) on Tower B

          -    100-m spacing covers 300 m


     •    #3 lies between primary rows #2  and #3 (>v.l3 m above #2)

               4 samplers centered (approx) on Tower B

          -    100-m spacing covers 300 m

G.  Two Lee-Side Rows

          -    6 samplers in each row

               centered near Tower B  (60-series stakes)

          -    100-m spacing along and between rows

          -    covers 500 m along each row

          -    covers out to 280 m beyond crest


                               135

-------
                    TABLE 11.  ADDITIONAL SAMPLERS

A.  Co-located Samplers:  two locations along center of  grid: one on
     primary row #1, and one on primary row #3 (10 m and 60 m above
     the base of Tower B, respectively).

B.  10-minute Samplers:  five locations: one at grid center along
     primary row #2; two along primary row #1,  80  m to either side of
     center; two along primary row #3, 280 m apart,  centered 20 m
     south of grid center.

C.  Background Samplers:  one on the east side of  Waughan Arroyo; one
     near the doppler acoustic radar near the substation on the high
     ground east of the east arroyo.
D.
"Edge" Samplers:   one to the north  of  the grid on the road up to
 the top of HER beyond the lidar; one  to the  south where the San
 Juan River flows through a gap  in  the ridge.
E.  Elevated Samplers:   one on Tower C at ^  8m;  two on Tower B at
     * 14 and 28 m.
                                  136

-------
coverage near the base of HER and the highest resolution between half way up
the ridge and the top.  Three primary rows followed the height contours,  and
the fourth row defined its crest.  Samplers were deployed less densely in
the lee of the crest.

     Tracer Analysis System

     The Tedlar bag samples were analyzed for SFg and CFgBr by electron-
capture gas chromatography.  Eight automated adaptations of 1972 Lovelock
prototype gas chromatographs were used to set up a gas lab (Figure 53) in
Farmington.  The functional steps of the analysis are shown in Figure 54.
The laboratory analysis process had a sample bag check-in, an assignment  to
a particular GG and subsequent analysis, and the output of an area propor-
tional to gas tracer concentration.  These areas, along with measurements of
pressure, temperature, and physical constants,  permitted calculations of
tracer concentrations.  The calibrated response of individual GCs was checked
by injections and analyses of "known standard"  reference mixtures of  the
tracer gases instead of the unknown sample bag  mixtures.

     4.4.4  Flow Visualization

     Oil-Fog Releases

     The generation and release of oil-fog for  lidar, visual,  and photo-
graphic observations were accomplished by the injection of corvus oil into
the exhaust of a small turbine engine.  The rate of oil injection,  and the
ensuing fog density, were coordinated by the field operations  director.
Fifty-five gallon oil drums and pumps provided  oil to the generator.   The
rate of pumping was monitored by a liquid rate—of—flow meter and was
adjusted by means of a valve.  The operator recorded the start and end
times, the indicated oil flow rates, and any alterations.

     Lidar Sampling

     The plume-mapping lidar (or laser radar) scanned vertically through  the
plume at several distances downwind from the source to produce three-
dimensional data on the fraction of light backscattered to the lidar  by
particles within the plume.  Plume position,  dispersion,  shape,  and proximity
to terrain will be calculated as a function of  time from processed data.
The lidar transmitter is a frequency-doubled Nd:YAG* laser emitting pulses
in the green portion of the spectrum at rates up to 10 Hz.  Vertical  scans
through the plume can be completed at a typical rate of two per minute.   The
slower ruby laser used at SHIS #1 was installed as a back-up in case  of
unanticipated failure of the Nd:YAG laser.
*Nd:YAG denotes a laser using an Yttrium-Aluminum-Garnet  crystal  doped
 with Neodimium.
                                  137

-------
GC
1



GC
8


GC
3



GC
4


     Graphics \
      Scope  J
                        I
                     SP4000
                    Integrator
                       P/E
                      7/16
                    c±>
                     5x5 MB
                     Hard Disk
CON

CRT
 1
9T
800/1600
cpi



9T
800/1 600
cpi
      Plotter
        Devices available but not online during data acquisition
Figure 53.    Gas  chromatography data acquisition system.
                       138

-------
         Samples
            of
         Standard
         Mixtures
                          Computer
                         Storage and
                         Calculation
  Audit By
Independent
  Standard
  Mixtures
Figure  54.    Tracer  gas analysis system.
                    139

-------
     The lidar recorded data on 9-track digital magnetic tape  for later
processing.  Some real-time information was available  on plume behavior for
experimental guidance.  The lidar was located about 2  km NNE of  the  tracer
release area (Figure 49).

     Photography Program

     Morrison-Knudsen Company, Inc., as subcontractor  to NOAA  ARLFRO,  was
responsible for all photographic work for SHIS #2.   The work consisted of
continuous video tape of each experiment and photographs taken every five
minutes from each of five positions.  The photographs  were taken during all
periods of smoke release, regardless of tracer release or other  project
considerations.  Morrison-Knudsen was responsible only for photography and
video taping; ARLFRO operated the oil-fog generator, and ERT provided a.
60-inch carbon arc lamp to illuminate the plume.  The  arc lamp was located
just north of the doppler acoustic sounder (Figure 49), east of  the  main
arroyo.

     Three photography positions were fixed and two were roving.  One of the
fixed positions was the 120-m level of the 150-m tower.  The other two fixed
photographers were positioned to the north and south of the release  east of
the base of the ridge.  The two roving photographers chose positions that
allowed photography of the plume near the ridge top or of other  meteorologi-
cal phenomena not in view from the fixed positions.

     The photographic archive consists of 11 binders of color  slides covering
experiments 1 through 15, one binder of supplemental slides, and 24  two-hour
video tapes.  During periods of smoke release, and in  the absence of other
technical difficulties, 60 photographs were taken each hour (five photo-
graphers each taking 12 photographs per hour).  The photography  program is
summarized as follows:

     •    five 35-mm cameras with data backs to document time  of photo, etc.
     •    color film (Kodak Ektachrome ASA 200)
     •    one videocamera equipped with Starlight-scope
     •    locations:
               120-m level of the 150-m tower
               north of the release area east of the ridge base
          -    south of the release area east of the ridge base
          -    two roving photographers to get best pictures of  plume
               dynamics
     •    five-minute exposure (12 photos/hour/camera) at night; shorter
          exposures in sunlight but same frequency
     •    turnaround - slides within 48 hours
     •    lighting - carbon arc lamp revolving once every 30 seconds.

4.5  Preliminary Field Study Results

     4.5.1  Summary of HBR Data Base

     The three flow visualization and il combined flow visualization and
tracer experiments at HBR have produced an extensive data base that  covers a
wide variety of dispersion conditions and concentration patterns for

                                  140

-------
modeling purposes.  SFg was released (Table 12) during 91 experimental
hours and CF^Br during 88 hours.  The tracers and oil-fog were released at
seven different locations during the course of SHIS #2 at elevations ranging
from 5 m to 75 m.  The release locations (Figure 49) varied from 333 m
(R-80) to 800 m (Tower A) from the crest of the ridge.

     Table 13 summarizes the number of hours with tracer releases and tracer
sampling for each SHIS #2 day.  The experiments typically started around
midnight and lasted until after dawn.  Approximately 80 1-hour average SFg
and CF3Br concentrations were produced every hour (Table 13).   More than
7300 1-hour average surface-level concentrations of SFg and of CFgBr
were sampled and analyzed during SHIS #2.

     Three samplers were suspended above the ground from the shorter
meteorological towers; two hung at approximately 14 and 29 m on the 30-m
tower at the base of the ridge, and one at about 7.5 m on the 10-m tower on
the crest.  These samplers yielded more than 230 additional 1-hour concen-
trations of either tracer gas.  At each of five surface sampling sites,  four
sampler boxes were set out to gather 10-min samples on which to base assess-
ment of concentration variability during periods less than 1-hour long.
Approximately 1600 10-min samples of each tracer gas were analyzed.

     ARLFRO has prepared the SHIS #2 tracer gas data base.  A tape of
concentration data and isopleth maps (hard copies and slides)  have been
given to the EPA Project Officer.  After he witheld a randomly selected
subset of the data base for independent model testing (see Section 5.1),  a
tape and the maps of the remaining tracer data were given to ERT.   This
"learning" tracer gas data base has been used for the analyses presented  in
the remainder of this report.

     As discussed in Section 4.4, the tower meteorological data were archived
in real-time via the MDS.  During the first few experiments of SHIS #2,  it
became apparent that the meteorological data contained "noise."  Subsequent
investigations during and after the experiment have suggested  four types  of
noises:  (l) large "hits" or "spikes" which produced apparent  full-scale
voltages from the instruments, (2) "cross-talk" or "skips" between MDS
channels such that signals from one channel were picked up by  another
channel,  (3) a high frequency, apparently  random low-voltage oscillation
that was attributed to 60-cycle AC, and (4) an apparent 2- to  4-sec  noise.

     Figure 55 shows a plot of raw, unedited,  1-sec Tower A AT data  (in
counts) taken between 2200 and 2300 on October 22,  1982.   The  excursions  to
0 and 4095 are evident.  It is also evident that signals  from  one  channel
are being picked up by another channel.  This channel-skipping appears as
vertical lines within the band of traces on the plot,  the lines being
composed in this case of skips from channels 72 to 73,  73 to 74,  etc.  in
general,  although some scans, like those near minute 31,  involve only
channels 77 through 79 or some other subset of channels.   Skipping is
particularly frequent in minutes 53 to 55  of this hour.
                                  141

-------
         TABLE 12.   NUMBER OF SHIS  #2  EXPERIMENTAL HOURS WITH
                    TRACER GAS RELEASES AND CONCENTRATIONS
                            Tracer  Releases
Experiment
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Date
5
6
7
11
12
12-13
14
14-15
20
22
22-23
24
25
25-26
28-29
Oil-Fog
4
5
6
9
7
8
7
9
5
10
9
9
1*
10
11
SF6
-
-
-
7
7
8
7
9
5
10
9
9
-
9
11
CF31
-
-
-
6
7
8
7
9
5
10
9
9
-
8
10
                          Tracer Concentrations
                              SF,.    CF^Br
                                                      12

                                                      12
7
7
9
7
9
5
11
10
10
6
7
9
7
9
5
11
10
10
                                       12

                                       12
        TOTAL**
110
91
88
99
98
 *Experiment 13 was terminated due to unfavorable weather.
**There are more hours of concentration data than there  are of
  tracer release because samples obtained after the tracer releases
  were terminated were analyzed.
                                  142

-------
TABLE 13.   SUMMARY OF TRACER CONCENTRATION DATA
                 SF
Experiment Date One-Hour
4
5
6
7
8
9
10
11
12
14
11
12
12-13
14
14-15
20
22
22-23
24
25-26
15 28-29
Subtotals
523
497
695
527
659
393
842
798
765
936
930
7565
One-Hour
Tower
13
21
18
18
23
10
23
28
23
26
28
231
10-min
108
103
120
95
189
62
165
215
134
214
209
1614
One-Hour
487
496
696
526 -_
666
396
839
795
764
919
918
7502
One -Hour
Tower 10-min
13
21
18
20
23
10
23
28
23
26
28
233
101
103
119
82
186
62
168
211
135
214
209
1590
   Total SV, 9410
Total CF3Br 9325
              GRAND TOTAL 18,735
                    143

-------
m

N
8
                                                                                          o
                                                                                          LO
                                                                                          rH
                                                                                           a
                                                                                          o
                                                                                           S
                                                                                           o
                                                                                           o
                                                                                           o
                                                                                           I
                                                                                          LTl
                                                                                          LO
                                                                                           CD
                                                                                          • H
                                                                                          PL.
                                         sjunoo
                                           144

-------
      The  "hits" and "skips" can be characterized as follows:

      •    When a hit strikes a channel (between 1 and 16) on a multiplexer/
          A-D board, it usually drives the counts to 0 (0.000 Vdc) or 4095
          (5.120 Vdc).  This hit will then appear on all subsequent channels
          (up to and including channel 16) on that board for that scan,
          although sometimes a hit that first appears as a 0 will be a 4095
          in subsequent channels.

      •    There are two subclasses of hits, whose characteristics are suffi-
          ciently different that they may be manifestations of different
          phenomena.
          -    The hit may not drive the counts to either limit but to some
               number like 19 or 23,  which recurs in subsequent channels and
               may happen more than once in a few minutes.  This sort of hit
               might result from dropped bits in the A-D conversion.
          -    The hit may show a gradual change from a "good" value on one
               channel to bad values  on subsequent channels,  finally settling
               on a limit value (0 or 4095) after a few channels in the  scan.

      •    A skip seems to put the datum from channel n + 1 (occasionally n +
          2) into channel n.  Once a channel skip occurs, all subsequent
          channels also skip to the next channel on that board.   Channel 16
          appears not to skip to channel 1 but to give correct values.

      Due to these patterns of hits and skips,  channel 1 yields the "cleanest"
data  on each board; channel 16 has the most hits, and channel 15 the most
skips.  As much as 30% of the data on channel 15 may be affected by these
errors for some periods.

      Both ARLFRO and ERT staff tried to alleviate the noise problems during
non-experiment hours at the Hogback.   Between experiments 8 and 9,  the six
16-channel multiplexer boards were reconfigured so that none served instru-
ment  outputs from more than one tower,  and additional  grounding of the
components of the control data collection system was done.   These changes
seemed to remove most  of the problem  with the  60-HZ noise and the 2- to
4-sec noise, which may in fact have been different ways of  looking at the
same  phenomenon, but did not eliminate the big spikes  and channel cross-talk
or skipping.

   •   ARLFRO and ERT scientists continue working on the hit  and channel-
skipping phenomena.  Preliminary methods have  been developed  and implemented
to filter the hits and skips.   The effects of the 60-HZ noise are impossi-
ble to remove from the data.   A procedure for  isolating the 2- to 4-sec
noise has been developed and tried on several  periods  of data,  but no method
has been developed for removing this  noise from the data.   The latter two
types of noise have little effect on  the average meteorological  data but  may
degrade the turbulence statistics  and covariances.

     A first attempt to remove the hits and skips  has  been  made  by ARLFRO
staff using data collected during  the 17  hours  selected  for initial modeling
and analysis (see Section 5.1).   ARLFRO filtered the hits and skips,  applied
                                 145

-------
non-cosine response correction factors developed in 1978 by Dr.  John F.
Clarke* to the prop data, and converted the counts  to  engineering units.
They have not yet investigated the 60-HZ and 2 to 4-sec noises,  nor  have
they "recovered" any of the cross-channel data.   Substantial work remains to
refine and complete the tower meteorological data base.

     Of the remaining SHIS #2 measurements, the WPL and ATDL tethersonde
data have been processed and delivered.  The WPL data  include  tables and
plots of profiles of pressure, relative humidity, mixing ratio,  wind speed,
wind direction and potential temperature up to altitudes of about 375 m.
The ATDL data include tabulations and plots of 15-minute averages and
standard deviations of wind speed, wind direction,  and the vector wind
components, u and v, from the tethersonde flown near the smoke source.  The
WPL monostatic acoustic sounder data have also been delivered  as photographs
of the facsimile output from the sounder located near  Tower A.  The  wind
speed data from the three optical anemometers have  been received from WPL as
10-min averages displayed in both graphical and tabular forms.  The  archive
of photographs is also complete.  No doppler acoustic  sounder  nor lidar data
are yet available.  Preliminary 20-minute average sonic anemometer data,
excepting heat and momentum fluxes, are available from the real-time output
for the 17 hours selected for preliminary analysis  (see Section 5).   WPL is
currently reducing the sonic data.

     In summary, the following data are currently (i.e., as of 31 May 83)
available from the HER experiment:

          SFg and CF3Br concentrations and hourly average emissions  data;
          Preliminary, edited tower meteorological  data for the 17 hours;
          Unedited tower meteorological data;
          Tethersonde data;
          Monostatic acoustic sounder records;
          Optical anemometer data;
          Preliminary sonic anemometer data (taken from the real-time output
          in the field); and
          Complete photographic archive.
*These  corrections were developed from wind tunnel experiments with
 modified R.M. Young wind component transmitters rather than with the
 Cliraatronics  instruments used at CCB and HER.  The Young instruments
 have a vertical offset of several inches between the U and V arms,
 which  reduces the interference of one with the other, whereas the
 Climatronics systems do not.  Furthermore, the Young instruments were
 modified by a plumb-bob leveling device that altered the flow
 uniquely.  Non-cosine corrections were developed for the Climatronics
 instruments by ERT for application to the CCB data-base (Greene and
 Heisler 1982) but were based on less extensive experimentation than
 Clarke's.
                                  146

-------
     4.5.2  Overview of SHIS #2 Results

     The flow visualization and tracer experiments at the Hogback have
produced an extensive (still evolving) data base.   The real-time experimental
selection of release locations and heights based on the MDS  output produced
concentration patterns for a wide variety of dispersion conditions.   These
data will be used to evaluate and refine the modeling approaches developed
with the CCB data base.  Initial analysis of the 17-hour data base suggests
that the concept of a critical dividing-streamline height is appropriate  for
stable flows toward HER.  Releases below HC dispersed in the "blocked"
flow and produced peak ground-level concentrations (GLC) on  the  windward
face of the ridge.  Releases above Hc dispersed in a flow that went over
the ridge and produced peak GLC's near the ridge crest and on its lee side.

     Figure 56 shows isopleths of hourly CF^Br concentrations (x/Q)
measured during experiment 5 (October 12, 1982) between 0500-0600.   A band
of concentrations above 50 psec/m  was observed from the ridge base to  near
its crest.  The highest x/Q of 387 psec/m  was measured about 15 m uphill
(sampler 105, see Figure 52) from the intersection of Hogback Highway and
Tower Road.  The tracer gas was released at location 203 at  a height of
15 m.  The calculated Hc was 47.9 m.  Note that the CF3Br dispersed up
and over HBR while producing the maximum GLC at a  ridge elevation roughly
equal to the release height.  This seems to be different from the CCB
results, which show that releases below Hc tend to disperse  along the
sides of the hill rather than up and over.  An initial analysis  of the  HBR
data suggests that tracer gases released below Hc  are often  eventually
transported up and over.

     Figure 57 shows CF-jBr concentration (x/Q) measured during experiment
15 (October 29, 1982) between 0200 and 0300.  The  tracer gas was released
from the 40-m level of Tower A, above the calculated Hc of 24.4  m.   Only  a
few non-zero GLC's were measured, and these occurred on the  lee  side of the
ridge.

     Roughly eight concentration patterns were observed during SHIS  #2:

     1.   Tracer gas distributed front and back of HBR
     2.   Primarily front side
     3.   Primarily lee side
     4.   Along upwind base
     5.   Near top
     6.   Near top and on lee
     7.   Spots
     8.   Nothing

Figure 56 exemplifies pattern 2 and Figure 57 exemplifies pattern 3 (or 7).
Other examples are given in subsection 4.5.4.
                                  147

-------
                                                       LO
                                                       o
                                                       CQ
                                                        to
                                                       pt,
                                                       X
                                                      fi.
143

-------
                                  "*""* "t"
 ,**

%>-
*
 t
 >
 •>»
i-..
        4
        f
      .«**
                _ i

                0
                                   t
                                   t
                                   i
                                   t
                                   •
  00

   
                                                       I
                          "•* •
                         I

                                                       O  i!  M  11
                                                       o — • CM ro
                                                          l\
                                                                  1
                                       \l
                                                                          CO
                                                      O
                                                      _J
                                                     t uj

                                                   CO U,
                                                                             
                                  CO
                                149

-------
      4.5.3  Comparison  with  SHIS #1 Results

      The essentially two-dimensional character of the Hogback and the
 three-dimensional  character  of CCB dictate the behavior of the flows that
 affect  dispersion  of tracer  gases released upwind of these terrain features.
 Despite these very basic differences, an analysis of the CCB data base and
 an initial analysis of  the 17-hour HER data base show the importance of the
 critical dividing  streamline in discriminating between types of flows and
 subsequent dispersion conditions.  At CCB, releases below Hc tended to
 remain  horizontal  and travel around the sides of the hill.  At HER, releases
 below H,., tended  to stagnate, producing high GLC on the windward side, but
 then were  often* eventually  transported up and over the ridge.  At both CCB
 and HBR, releases  above Hc flowed up and over.

      At this  time  only  the 17-hour HBR data base has been analyzed and
 modeled, so any  results are  preliminary and not conclusive.   However,  it is
 interesting to compare  some  of the results from HBR with those from the CCB
 experiments.  Table 14 summarizes the five highest x/Q and the associated
 meteorological conditions observed at HBR and CCB.  Generally higher x/Q
 were measured at the ridge.  The higher concentrations at HBR were measured
 at elevations at least as high as the release height, while at CCB the
 higher  concentrations were frequently measured at elevations lower than the
 release height.

      Figure 58 shows a plot  of maximum x/Q vs. l-Hc/zr for the entire
 CCB data base and  for the 17-hour HBR data base.  The higher x/Q at CCB
 tended  to  occur when zr«sHc, while at HBR the higher x/Q were measured when
 zr^c'  Since the  ridge is basically two-dimensional, the higher values
 could be explained by the highly variable and generally stagnant flow below
 Hc  and, if  the average wind were directed toward the ridge,  the tracer gas
 would be transported directly to a sampler near the release  elevation.
 Flows below Hc at  CCB tended to remain essentially horizontal with some
 uphill  transport and generally preferred one side of the hill.  Figure 59
 shows a concentration (ppt) map for CCB experiment 206,  hour 8 (the hour
 with the highest x/Q of 164).  The pattern of concentrations is fairly
 horizontal  and is  elongated northward along the north side of CCB.   Compare
 this  pattern with  the patterns in Figure 56 and Figure 77 (in the  next
 subsection).  At HBR there seems to be eventual uphill transport for flows
 even below  Hc.

      The higher concentrations at CCB occurred when the transport  wind
 direction was close to the stagnation streamline.   Figure 60 shows  a plot of
 the CCB x/Q vs. A9/iv,  where A6/iv is the ratio of the wind  angle  between
 the actual  wind direction and the stagnation wind direction  divided by  the
crosswind turbulence intensity.  A large value implies  that the tracer gas
 plume would have little chance of "hitting" the hill.   The data show that
the peak GLC values occurred when A6/iy^l.   Small GLC values occurred  when
A6/iy>5.0.
*Not all releases below Hc were transported over the ridge.   If  the
 winds near the release had a substantial component parallel  to  the
 ridge, then the tracer stayed below the crest.

                                  150

-------
VJ C
(U O
1-) .1-1
a, u
g cd
w a)
w


C* CO CM CO O
CM CM oo m CM



                                                            in   in    o   o    o
                                                            CM   CM    CO         CO
                                                                                                      CO
                                                                                                      cd
                                                                                                      erf
M
o
                           CM    00   

                                                                                                      fc
                                                                                                      CD
                                                                                                      CO
                                                                                                      ,0
                                                                                                      o

                                                                                                      Cf

                                                                                                      X
w
PQ
0

Cf
X
H
CO
Cd
ffl
Cj>
r— 1
PC

s
v_x o 'n m o
H CM i-l i-l CM
N

^-v
CO
e

o
co
3.





r~- co r-^ r~-
r-t OO 00 ^O

o
CM







^j.
0
-
-------
o
                                                             + SF6
                                                             * CFgBr
                                     1-Hc/zr
                         a.   17-hour HER data base.
  138.8

  120.0-

  118,0-

  108,0-
n"
^ Oi.O-

a so.o-
•54
   SB.8-

   <0,0

   30.0-

   20.0-

   10.0-
                                                            _£*_
                                      -.SB
                                     1-Hc/Zr
                              b.   CCB  data base.
Figure 58.    Maximum hourly x/Q

                             152
                                                       vs. l-Hc/zr.

-------
                                              • » r = 500 m
                                                                N
Figure 59.   Observed SFg concentrations (ppt) for CCB Experiment 206,
             hour 8 (0700-0800 MST).   Source: r = 595.9 m,  6= 123.6°,
             net height = 29.5 m, Q = .062 g/s.
                                153

-------
                                                                          §
to
 g
             w u
             + *
                                                                     * *
                                                               *•*
'V

                                                                     •*-•*).
                                                                         .i
                                                                         .8
                                                                                rt
                                                                                CQ
                                                                                CJ
                                                                                CJ
                                                                                bO
                                       154

-------
      Large concentrations  were observed at HBR whenever the average wind
 direction approached  the ridge, even for releases below HC.

      In summary,  there are several similarities and differences among the
 results from the  two  experiments:

      •     Observed  peak x/Q values were greater at HBR than at CCB;
      0     Hc  discriminates between the flow regimes;
      •     At  CCB  the  peak  GLC values occurred when z^ HC ; at HBR the
           peak GLC values  occurred when zr 
-------
                                                 00
                                                 rn
                                                  CD
                                                  CD  LO
                                                  VO
156

-------
                                                                   oJ
                                                                   CD
                                                                   fn
                                                                   oj

                                                                  •p
                                                                   CD
                                                                   00
                                                                   ^
                                                                   cd
                                                                  •P
                                                                   3  E-
                                                                   O  Q
                                                                   w  S
                                                                  CQ
                                                                  33
                                                                   o   •>
                                                                   fn  OJ
                                                                  M-l  00

                                                                   CD  i-H
                                                                   CD  4->
                                                                  •M  S
                                                                   3  CD
                                                                   C  S
                                                                   6
                                                                   i
                                                                  CD
                                                                  H

                                                                  C>0
                                                                  •H
                                                                  PH
157

-------
                                                 
-------
     Figure 64 shows vertical profiles of hourly average wind direction*,
wind speed, and temperature derived from the Tower A measurements.   On the
basis of this information and measurement from the photographs,  the oil-fog
plume rose to about 60-80 m and was transported toward HER in the ESE winds.
Time-series of 5-minute average winds measured on Towers A, B, and  C, and
5-minute winds estimated from the photographs are plotted in Figures 65 and
66.

     A map of one-hour SFg concentrations is given in Figure 67.  Figure 68
gives the along-hill (approximately crosswind) distribution of hourly SFg
concentrations for nine sampling "rows."  Figure 69 identifies the  row
plotted by each symbol in Figure 68 and Table 15 defines the rows by their
sampler numbers.  There is a wide area of relatively high (greater  than
5.0 usec/m ) concentrations over most of the ridge.  The peak values
were measured near the ridge crest and on its lee side.

     Figure 70 gives a map of hourly CF3Br concentrations,  and Figure 71
shows the along-hill distribution.  The peak values occurred south  of Desert
Drive at an elevation of 5470 ft (row 4), an altitude about 10 m higher than
the release elevation.  Note that the wind direction at 40  m in Figure 64  is
consistent with the observed concentration pattern.

     Experiment 10 (0300-0500)
                                                    ,.<
     The morning of October 22, 1982 also experienced an easterly drainage
wind.  The drainage became. well established just after midnight,  and by 0300
ESE winds were directed toward HBR at about 2 m/sec.  The oil-fog,  SFg,
and CF3Br were released from Tower A — the oil-fog and SFg at 50 m,  the
      at 30 m.
     Figures 72 and 73, photographs taken from Tower A and from the ridge
crest south of the target area,  show a coherent plume going up and  over HBR
at 0330 MDT.  The second 5-minute exposure (Figure 73) shows some plume
material diffusing to ground-level near the ridge crest.   The profiles of
hourly meteorological variables are given in Figure 74.   Time-series of
5-minute wind directions are given in Figures 75 and 76.

     According to the photographs and the measured wind directions,  the
oil-fog (and SFg) plume dispersed steadily up and over HBR south of the
light cross.  The plume rose to an altitude above HC (50.1 m) as it was
*The profile plots of wind direction, wind speed,  and temperatures
 were produced in the following manner.   The 5-minute temperatures and-
 the prop wind speeds and directions were used to  compute 1-hour mean
 temperatures, scalar wind speeds,  and unit vector u [sin WD]  and unit
 vector v [cos WD] wind direction components.  This was done for each
 instrument level of Tower A [2,5,10,20,30,40,60,80,100,150  m].   A
 "spline under tension" method was  used to interpolate the meteorologi-
 cal variables for every 5 m between instrument levels on Tower  A.
 The splined u and v unit vector components were used to derive  the
 unit vector wind direction.
                                  159

-------
               M  O r-
                  	  O
                  CM X
                                                                                                                          o  o
                                                                                                                          o  o
                                                                                                                          rH  K5
                                                                                                                          o  o
                  T
-5-
5   S
                                                                                                                -8
                                                                                                                      •C  CD  CN
                                                                                                                       6  4->  00

                                                                                                                      TJ  a  •-!
                                                                                                                      2  §
                                                                                                                          o  fi
                                                                                                                             CD
                                                                                                                          w  g
                                                                                                                          CD  -H
                                                                                                                          i-H  fH
                                                                                                                          •H    ^
                                                                                                               O  I
                                                             160

-------
        Tower A
        	60m
            80m
        	100m
            Photo-Derived (oil-fog)
                                                    tine (Hour)
WD (degree)
 270
        Tower C
             ,2m
            - 5m
        	10m
             Photo-Derived ;(oil-fog)
                                                    time (Hour)
 Figure 65.    Time-series of  5-minute  wind directions  from
                Tower A and Tower C  (Experiment  4, 10/11/82,
                0300-0400 MDT).
                             161

-------
8
         CO   G>
                             CD   tD   G»
ffl

•^ ia   is    is    is    is    io
   N-   *    —    flO    tO    PJ
_ W   M    (M    —    —    ^
                                                                                                  £   E
                                                                                                  o  o
                                                                                                  t-  CO
                                                                                                                  I
                                                                                                                  o
                                                                                03    CO   03
                                                                                                   09   G)
                                                                                                                         00
                                                                                                                       LO
                                                                                                                           -p
                                                                                                                       
-------
                                 *»*
                                  ess
                        X


                        UJ
    os> o in
    {D*-«
                                               fO
                                            CSJ
                                                       o
                                                      •H
                                                      ed
                                                      S
                                                      CD
                                                      O


                                                      O
                                                      O
fl,
CO
                                                      f-l

                                                      I
                                               U,
                                               CO
                                               CM
                                               oc
Cs.!
163

-------
                               CO  E



                               "~  X
                                                                                                       to
                                                                                                         B
                                                                                                         C
                                                                                                         o
                                                                                                        •H
                                                                                                        •P
                                                                                                         cd
                                                                                                        C
                                                                                                        r    •»•    m
                                                                                                 L§
                                                                                                 h r-
                                                                                      M-l
                                                                                    ^ O

                                                                                    S. 10
                                                                                       
-------
CO

 o
cc
1
cc
g
                                             1
                                            DC
        00


        1
        cc

        Q
OT


 O
CC

 *
                                                                                                        o
                                                                                                       tn
                                                                                                       ^
                                                                                                       
-------
Row
7





8
                   TABLE  15.   HOGBACK SAMPLERS  BY ROW
          Samplers
               1-5, 702,  703
               101-106, 711-714, 109, 809,  721-724, 112-118
               107, 108, 110, 111
               201-205, 207, 208, 210, 731-734, 211, 213, 214, 216-220
               206, 209, 212, 215
               301-307, 741-744, 308-313, 751-754, 314-319, 810
401-417, 419-422, 701
501-506
               601-606
                                  166

-------
                                                             t-(
                                                             •—I

                                                             o
                                                              I
                                                             •H
                                                              o
                                                              •H
                                                              4->
                                                              oJ
                                                              fn
                                                              •p
                                                              C
                                                              CD
                                                              O
                                                              a
                                                              o
                                                              o
                                                              fin O
                                                              U O

                                                               XO


                                                               3 o
                                                               o to
                                                              33 o




                                                              o


                                                               0>




                                                              is.,
167

-------
                                                                               I
                                                                                                                                                      to
                                                                                                                                                       ft
                                                                                                                                                        O O
                                                                                                                                                       CQ O
                                                                                                                                                        K50
                                                                                                                                                       Xo
                                                               ro   CM   —
S
                                                                           1
                                     O   <5>   CD   O)
                                                           Q>  CD   CD     u>   *•   m   0
                                                                                      168

-------
                                                               CM
                                                               OO
                                                               CN
                                                               CM
                                                                •P



                                                                I
                                                               •H
                                                               tu
                                                                CD


                                                                I

                                                                6
                                                                o
                                                                CD
                                                                fn

                                                                10
                                                                O
                                                                   •

                                                                C P
                                                                • M Q
                                                                s s
                                                                 I
                                                                co o
                                                                > to
                                                                •H tO
                                                                PI-I O
                                                                 CD
                                                                 f-i
                                                                 a
                                                                •H
169

-------
                                                    CQ
                                                       OO
                                                    0 T-l
                                                    P!  B
                                                     I  
-------
                ID  O r-
                 II  II   II
                  o T ~
                Xu.li.
     
                                                                                                                                    1
 ctf  Q
 4-J  5*.
 ctf
*"Ci  ^^
    O
<—I  "=t
 cd  o
 o   i
•H  O
 t>00
 O  tO
t—(  O
 o
 ^   «
 o  CNI
 •?
                                                                    17-1

-------
278
248-
218-
188-
150-
128
 00
 68-
 38
  8'
-30-
-BB-
        Tower A
        	60m
        	 80m
        	100m
        —•»— Photo (oil-fog)
 -00-
WD (degree)
 278-1
                                                      time (Hour)
 248
 210
 188-
 158
 I28-|
  08
  ea-
  se
   e-
 -38-
 -68-
       Tower C
       	2m
       	5m
       	10m
       —+— Photo (oil-fog)

                                                     tin* (Hour)
 Figure 75.    Time-series  of 5-minute wind directions  from
                Tower A and  Tower  C (Experiment  10, 10/22/82,
                0300-0400 MDT).
                              172

-------
o> '. _
C P S £r'*"r
1 g 1 = '
 S 4 N 
-------
 transported  toward  the ridge.  The resulting pattern of hourly average
 concentrations is shown in Figure 77.  Distributions of along-hill concen-
 trations  are given  in Figure  78.  The highest concentrations were measured
 in  the  lee and near the crest (one sampler). Relatively low concentrations
 were measured on the windward face.

     The  winds below Hc were quite variable (see Figure 76) and the
 resulting CFgBr concentration pattern (Figure 79) reflects the variability.
 During  the beginning of the hour, the winds were from the SSE, producing the
 CF3Br concentration peak along the north edge of the sampler grid.  A
 secondary peak was  produced by the easterly winds that occurred during the
 second half-hour.  The along-hill distributions of Freon concentrations are
 given in  Figure 80.  Two values above 5.0 ysec/m  were measured—one
 near the  crest (row 7) and one at the base of HER.

     The  easterly drainage wind persisted throughout the next hour.   The
 SFg and oil-fog release was changed to 70 m; the CF3Br release stayed at
 30 m.  Figure 81 is a photograph of the oil-fog plume taken from the crest
 south of  the target area at 0425 MDT.  It shows more lifting of the  plume  as
 it goes up and over the ridge crest.   Figure 82 shows the distribution of
 hourly ground-level SFg concentrations.   The highest values were measured
 on the lee of HER.

     The  vertical profiles of hourly winds and temperatures are given in
Figure 83.  The winds at 70 m remained from the ESE,  while there was  a  layer
 of more southerly winds centered around 30 m.   The time-series of 5-minute
wind directions are given in Figure  84.   Again,  the  CF3Br concentration
pattern (Figure 85)  reflects the variability in wind directions—a primary
peak near the north end of the target area produced  by the SSE winds  and a
secondary peak at the end of Desert Drive.
                                 174

-------
                                                                          CN
                                                                          CN
                                                                           CD
                                                                          CD
                                                                          o
                                                                          •H
                                                                          S-i
                                                                          •p
                                                                          s
                                                                          (U
                                                                          o

                                                                          o
                                                                          Pi  O
                                                                          co  o
                                                                          i— i   i
                                                                          ^  o
                                                                          3  o
                                                                          O  K5
                                                                          X  o
                                                                          CD
                                                                          1-1
                                                                         •H

                                                                         fin
175

-------
                            coin
                                                 I
                                                             §    §
                                      <\J    ^-   —    ^    ^    _
                              g     ^
                                                       £
i
i    i
                                            CO
                                            00
                                      
-------
                                                           CM
                                                           OO
                                                           CM
                                                           CN
                                                            e
                                                            
-------
                                              I  S
                                                g   ^
                                                                                          .--•"*
                                                                                   ''''
                                                                                                     8
                         g
f.ia
                    ffl
                                           & i)
                                                           "S	E"
                                                      g
15	S"
       §     m
                                                                                                           •P
                                                                                                           nJ
                                                                                                           t-t
                                                                                                           C  E-
                                                                                                           O  Q
                                                                                                           oa  o
                                                                                                           XO
                                                                                                           o   •>

                                                                                                              00
                                                                                                           0)
                                                                                                           o o
                                                                                                           IX) 
-------
                                                 CD

                                                 cd

                                                •p
                                                 CD
                                                o




                                                II
                                                w
                                                   LO
                                                Di CM
                                                CQ -5t
                                                ffi O

                                                a  -
                                                o cj
                                                !H oo
                                                
-------
                                                 00
                                                 w
                                                 o
                                                 C/D O
                                                 33 o
                               rr
180

-------
      in o  to
      ii ii  n
s
u>
s
<\f
1    §
8    S    -5
s
m
                                                                   is
                                                                   (M
s
                               g    g
                                    -5-
                                    m
                                                                               Is-  O

                                                                                  2
                                                                               «  3

                                                                               .|
                                                                        H
                                                                        Q
  O
T—I LO
cri o
u  i
•H O
boo
o n-
i-H O
o

O CM
CD 00

(D CM
g CM

XO
                                                                                  U)
                                                                                  1
                                                                                     O O
                                                                                    ^3 •-<

                                                                                    4H 4->
                                                                                     O C
                                                                                       O
                                                                                     w 6
                                                                                    •H
                                                                                    4-1
                                                                                     O
                                                                                       m
                                                                      
                                                                      H O
                                                                      CD H
                                                                               s  i
                                                                               ~  5
                                                                                     bO
                                                                              -§


                                                                              -a
                                                                               §
                                        •181

-------
          Tower A (props)
            20m
            30m
            40m
                                                  tine (Hour)
WD 
-------
                                                             CM
                                                             OO
                                                             CM
                                                             CM
                                                              •P
                                                              rt
                                                              CD

                                                              •H
                                                              f-t
                                                              
                                                              CTS
                                                              (D
                                                              O

                                                              O
                                                              O
                                                              pa
                                                              fin O
                                                              U O
                                                              3 O
                                                              O •<*
                                                              DC o
                                                              LO
                                                              oo
                                                               bO
                                                              •H
                                                              O-,
183

-------
                                  SECTION 5

                  PRELIMINARY MODELING OF THE HER DATA BASE
     The SHIS #2 at the Hogback has enlarged the data base with which to
evaluate and refine the models developed with the SHIS #1 data base and to
extend them to a two-dimensional ridge setting.  This section describes the
first evaluation of existing models with the HBR data base.   CTDM( 11083) and
GTDM(11083-E), modified for the geometry and topography of HBR, ": and COMPLEX I
and COMPLEX II were run on an initial 17-hour data base that was  selected to
represent a variety of dispersion conditions.  Model calculations were
compared to observations, and the performance of the three models was
evaluated and compared to the SHIS #1 results.

5.1  Initial 17-Hour Data Base

     5.1.1  Selection of Case Hours for Model Evaluations
     The 91 hours of SF5 release and 88 hours of CF3Br release were  examined
to pick hours for model evaluation.  The first step was the selection of  a
set of the tracer data to be withheld by the EPA Project Officer for
independent model validation purposes.   The selection procedure involved
tossing (actually spinning) three coins.  All experiment hours for which  the
unedited wind direction at source height was between 60° and 170°  were
candidates; there were 66 such hours.  If the first coin toss for each hour
yielded three of a kind (prob = 1/4), a second toss was made to determine
whether SFg, Freon, or both would be withheld.  If the first toss  was not
three of a kind (prob = 3/4), the next hour was investigated.  If the second
toss again yielded three of a kind, both tracers were withheld (prob ?= 1/4 x
1/4 - 1/16).  Two heads and one tail on the second toss meant SFg  was
withheld; one head and two tails meant Freon was withheld.   The probability
of either event is 1/4 x 3/8 = 3/32.

     The results- of this exercise were that SFg was withheld for six hours,
Freon for eight, and both for three, so that 20 tracer hours (Table  16) have
been selected.  There were 86 tosses of the three coins; 20 (23%)  were three
of a kind vs.  25% expected; 27 (31%) were 2H,1T vs.  37.5% expected;  and 39
(45%) were 1H,2T vs. 37.5% expected.

     For the remaining 159 hours of tracer data, a pattern  was assigned to
each hourly SFg and CF3Br concentration map based on the eight patterns
discussed in subsection 4.5.2.  After examining the hourly  concentration
patterns, photographs, log book comments, and unedited meteorological data
for each experiment hour, a set of 42 hours was selected.  A priority was
then assigned to each of the 42 hours (6-GOOD, 21-MEDIUM, 15-LOW)  by

                                  184

-------
     TABLE  16.   SHIS #2 TRACER DATA WITHHELD FOR MODEL  VALIDATION PURPOSES
Exp . Hour
No. MDT at
WD* and WS**
SF& Release Ht.
SF, to be Withheld
4
5
9
10
12
15
Freon to
5
10
10
11
11
12
15
15
Both to
6
6
9
06-07
00-01
06-07
00-01
05-06
08-09
112
67
62
Var
101
102
0.7
3.9
4.6
0.6
3.0
1.8
Release Ht.
SFg Freon
30
40
40
30
75
50
20
25
20
20
50
40
H ***
60
26
31
62
40
-
Location Fr(H,J
216
203
203
203
.A
A
•"!•«••—
2.25
1.83
1.90
Not yet
2.68
—
Fr(2m)
'0.54
0.70
0.46
available
0.89
-
be Withheld
06-07
01-02
09-10
00-01
04-05
08-09
05-06
09-10
168
87
116
106
100
111
80
125
1.4
1.0
1.9
0.6
2.3
3.0
1.1
1.4
30
30
70
40
50
50
50
50
25
20
30
20
25
40
40
40
76
36
11
50
36
16
-
-
203
,203
A
203
A
A
A
. A
0.40
0.86
1.70
2.23
2.01
1.51
-
-
0.26
0.06
1.77
0.52
0.78
2.29
-
-
be Withheld
00-01
05-06
08-09
114
88
112
2.2
2.0
2.5
40
35
40
30
25
20
22
35
19
111
203
203
0.95
2.32
1.90
0.32
1.07
1.05
  *Unedited wind direction estimated for  SFg  release height.
 **Unedited wind speed (m/s).
***Calculated from the unedited  meteorological  data.
                                       185

-------
 analyzing concentration patterns and other information to select a variety
 of  situations.  The final selection of 17 hours, presented in Table 17, was
 based on obtaining a good distribution of concentration patterns, meteoro-
 logical  situations, release locations, and release heights for the modeling.
 Table 18 summarizes various statistics based on the preliminary meteorolog-
 ical  data for the 17 hours.

      5.1.2 Construction of Model Inputs

      Tracer Data

      The release crane position, the time and duration of tracer releases,
 the emission rates of each gas, and the height of the tracer release above
 the local surface were recorded in release logs maintained by ARLFRO during
 the experiment at HER.  The survey data were used to determine the locations
 of  the release point, which were expressed in the hill coordinate system,  a
 polar grid (r,6,z) centered at Tower A.   The base elevation used in this
 system is 1600 m.

      Plume Rise

      The oil-fog generator produced enough heat flux to cause a thermally-
 induced  rise in the oil-fog and co-located SFg plumes.   A first attempt
 was made  to calculate the effective release height of the SFg plume for
model  input by using a method devised by Halitsky (1961)  for single-camera
 measurements of smoke plumes.

     Analyses of the wind direction time series and wind speed and tempera-
 ture profiles were also used to guide the estimate of the plume rise.

     Wind Data

     A "spline under tension" method, as discussed in Section 2.1,  was  used
 to  interpolate wind speeds and directions between instrument levels on
 Tower A.  The 5-minute prop wind speeds  and directions from Tower A were
 first broken into wind components,  and the components were interpolated to
 obtain the horizontal wind speed and direction at plume height for each
 5-rainute period.

     For  those 5-minute periods in which photographs taken from the 150-m
 tower were available,  wind directions estimated from the  photo images were
 substituted for the splined wind direction data.   These photo-derived wind
directions could only be estimated  for the SFg release  because only the
SFg was emitted with the oil-fog.

     A scalar average of the 5-minute interpolated wind speeds was  used to
form the 1-hour mean wind speed.  A 1-hour vector average of unit vectors
along each 5-minute wind direction was used to form the 1-hour mean wind
direction.
                                  186

-------
TABLE 17.  SHIS #2 HOURS.SELECTED FOR INITIAL MODELING
Exp.
4
4
5
5
6
8
8
8
10
10
11
11
12
12
14
15
15
Date /Time
11/0300-0400
11/0400-0500
12/0100-0200
12/0500-0600
13/0100-0200
15/0100-0200
15/0200-0300
15/0600-0700
22/0300-0400
22/0400-0500
23/0200-0300
23/0500-0600
24/0200-0300
24/0600-0700
26/0300-0400
29/0100-0200
29/0200-0300
Release
Location
216
216
203
203
111
215
:215
215
A
A
203
A
A
A'
203
R80/A
R80/A
MSTQ.'
30
40
30 ,
20
30
10
10
30
50
70
40.
50
75
75
40
20
20
: Pattern
IE
IS
6C
IE
6S
1C
5C ,
IN
6N
3C
5/4C
6S
7
7
IS
IE
IS
z (CF B
20 -
30
15
15
2-0
5
. '5
30
30
30
20 ..
25
40
25
20
40
40
r) Pattern
2S
2S
2C
2C
. 1C/N
2E
4C/S
IN
' 4C
2N
1E/S
IS
IN
8
2E
4C/7
3/7
                         187

-------
            TABLE  18.   17-HOUR DATA BASE  SUMMARY STATISTICS
1.   Number by release location
216/215
10
2. Number
< 10m
4
3 . Number
<. 20m
2
4. Number
(>

5 . Number
17 Hours
#
%
Entire Data
#
%
by

by

by
S
120
9
203 Hi A R80
local release height
11-20 21-30 31-40 41-50 > 50
98733
height of source above the ground
21-30 31-40 41-50 51-60 > 61
2 8 6 2 14
wind direction
E N
°) (120-45) (<45)
19 6
by concentration pattern
1

13
38
Set
67
41
234567

723242
21 6 9 6 12 6

52 3 11 4 16 7
32 2 7 2 10 4
                            188

-------
      Critical Dividing-Streamline Height,(H )        •,      ..  .  •


      The prop wind speed and temperature profiles from Tower A were used to
compute the  5-minute values of HC by means of the integral formula presented
in the CTMD  First Milestone Report, and  first suggested by Sheppard (1956).
These 5-minute values were then averaged to 1-hour values.  The effective
height of HBR was taken to be 85 ra.

      Brunt-Vaisala,Frequency (N)         '

      For each 5-minute period, the Brunt-Vaisala frequency (N) was estimated
at plume height by evaluating the temperature change along the splined
temperature  profile in the immediate vicinity (zr..i 0.1. m) of the plume
height to obtain the local temperature gradient.  These 5-minute values were
then  averaged to 1-hour values.

      Stability Class

      The Turner dispersion stability class was calculated from cloud cover.
and wind speed data.  The cloud cover was estimated from photographs and the
wind  speed was a 1-hour scalar average measured by the props at the 10-m
level on Tower A.

      Turbulence Parameters

      A comparison between the 5-m and 40-m turbulence data from the sonic and
the prop systems (Table 19) at the same "elevation proved that the two systems
were  inconsistent.  Due to uncertainties "with proper prop response correc-
tions, the sonic data were used to calculate the turbulence parameters.

      One-hour values of u, v and aw obtained from the NOAA/WPL sonic
anemometers on Tower A at 5m and 40 m were used to derive the vertical
intensity of turbulence (ig)  at plume height.   The sonic  data were
linearly interpolated to the plume height if this height was  less than 40 m;
otherwise the 40-m iz was used.   Sigma-w (aw)  at plume height was calculated
by multiplying the interpolated sonic iz data with the Tower  A propeller
wind  speed interpolated at plume height by means of the "spline under
tension" algorithm used in the MDA (see Section 2).  This methodology
preserves the sonic iz data within the model.   Five-minute average sigina-v
data  were set equal to the 1-hour sigma-w data for input to the PDF form of
CTDMU1083 and 11083-E) .

      5.1.3  Summary of Model Input Data  .

     Table 20 lists for each tracer gas all of the inputs  t'o  the models
evaluated with SHIS #2 data in this report.  The time and date for each  of
the 17 experiment-hours  is followed by the  release location (r,9),  the
emission rate (Q), the release height above the local ground  .elevation,  the
estimated plume rise,  the effective plume elevation above  the zero contour
(1600 m,MSL), the critical dividing-streamline height (H^,  the hourly
average wind speed and direction,  the Brunt-Vaisala frequency '(N),  the
Turner Stability Class,  and sigma-w (aw).

                                  189'-

-------
                                                   IT! CM    0*

                                                   CM 1-1    r-l
                                                                     O i-l  CM
                                                                        VO  CM
                                                                            CM     r-<  i-l
                                                                                              O CM
                                                                                                                          •* CM
                        00
                       .§
                                                  CM O
                                                                    CO O  O     VO CM    VD O    COCM
en  en
                                                                    oo
                             -3- o>
                                                                                   r~- o
                                                                                      co
                                                                                              o co
                                                                                                            oo
                                                                                                        en  co
                                                  -* O
a
I
I
                                                  03  03
                                       o  o
                                                  r-l  O
                                                                                  C3 O
                                       O  r-4
                                                  CM  O
                                                                    O CM  ,-1    00
                                                                                                                          O O
                                       O i-t
                                                                    Or-Hr-l     OO
                                                                                             O  r-4     i-HCO
                                                                    OOO     OO     OO     OO
                                                                    •   •    •
                                                     in    r-l     r-l CM  VO
                                                                                   II
                                                  O O
                                                                       - _   _
                                                                    OOO     OO
                                                                                  J.CM
                                                                                  o o
                            00
                                                                                                                                                           a)  a)

                                                                                                                                                           « B-5
                                                                                                                                                             m
                                                                                                                                                           cd CM
                                                                   00
                                                                                                                                                s.
                                                                           190

-------
                                                             to   to   to  'to   to   to   Q   to
                                                                                                                    to   W   to   to '  to   t=*   to
                          en   en   r-i
                                                                                           cd
                                                                                                r-*   vo   co   o   o   m   o

                                                                                                en   CM   T-I   cM-en   r-4   t~t
                                                                            TO   CM   CO
                                                        O   O   O   O   O   O   O
                                                                                                                    en   m   o   m  -'-d-   co   co
                                                                                                                    c-   01   **   i-*   oi   en   m
                                                                                                                    ,-1   CM   CM   -t   o
      a
    < CO
          in  m    r-i

          CM  r-(    en
CM   vO   CM   tn

i—i   en   en   m
                                                                            o   op
                              CM   00   CM
                                                                           -*   -d-   •*
                                   co   in   c?\   \o
                                                                                                                   vo   d-   CM   
-------
              •rf 0
              A r-t
               a o
                                         O  \D  O   en   CM  CM  CO
  (U
  §
                    O  CO   ,-t

                    00  CO   O
                    CM  *-i   en

                    O  o   o
  o
 o
              01 •

              ra u
                                 O  r-~  o*\
 M


  CO
             < a


             •a ^>

             •»*1
             3 a
              n -o
              u 01

              > a
             •< eyj
CO

erf

s
 «   .8



I S3
 0) 3 AJ

S5 >,
Ed   CJ
                                            co  en  oo  r^.  en  en
                                            CM  -^  CM  en  
en
m
o


o


o


o


0

en
00
u-i
•a-
o
o

o
o

0
o

0
o

o
o

o
o

0
en
en
CO
S
•»
o
o

o
o

o
o

o
o

                                                        192

-------
5.2  Complex Terrain Models Evaluated

     Valley, COMPLEX I, COMPLEX II, CTDM(11083), and CTDM(11083-E) have been
run on the preliminary 17-hour data base for SHIS #2 at HER.  Valley,
COMPLEX I, and COMPLEX II required no modifications for application to HBR,
aside from a change in the sampler grid data.  CTDM, however,  required
several changes.

     CTDM needs terrain data to select stagnation streamlines, impingement
points, and to identify -which receptors are subject to Wrap-type calcula-
tions.  For CCB, the polar coordinate system is fixed at the center of the
hill, and a general terrain profile is specifed by
                    = 970 h~°'38
(59)
where h is measured in meters from the 944.9 m MSL height contour.   For HBR,
the polar coordinate system is centered-at Tower A, and the hill profile
perpendicular to the ridge from Tower A is assumed to extend to infinity
along the local tangent to the ridge.  The hill profile is fit by the
equation
          h = -.42 (807-r) (l-exp(-253/(807-r))) + 91'
(60)
where r is measured from Tower A along the perpendicular to the ridge,  and  h
is the terrain height above 1600 m MSL.  Note that Tower A lies 807 m from
the ridge top, at an elevation of 1603.6 m ASL.

     Figure 86 summarizes the HBR geometry contained in CTDM.   The height
contours across the sampler grid on the eastward face of HBR define a tangent
plane oriented approximately 2,7° from truet north.  The ridge is considered
to be two-dimensional.  For Lift computations,  the losbal radius of curvature
is set to infinity, and no streamline deflection is allowed in the horizon-
tal.  For Wrap computations,  the stagnation streamline is defined  to follow
a 117° wind, and the distance to a receptor is  the sum of the  distance to
the receptor-height contour along 117°, and the  distance measured  along the
ridge from this point of intersection to the receptor.  If a receptor lies
on the lee side of HBR,  then no Wrap computations are made because the layer
below Hc is assumed to remain on the windward side of the ridge.   Receptor
heights are adjusted to equal the dividing-streamline height for performing
a Lift computation at lee-side receptors below Hc.

5.3  Overall Performance Statistics

     The data listed in Table 21 were used to run CTDM(11083),  CTDM(11083-E),
Valley, COMPLEX I and COMPLEX II.  The model calculations were  compared to
1-hour average SFg and CFgBr concentrations taken from the 17-hour HBR
data base.  Table 21 summarizes, .the residual' statistics.  The  results are
based on both tracer gases (17 hours each).   Columns labeled "Peak
Concentrations" refer to the highest CQ/Q and Cp/Q values unpaired in
space, but paired in time, and the columns labeled "All Concentrations"
refer to values paired in space and time.  Tables 22 through 25 give the

                                  193

-------
Figure 86.
Height section and orientation of idealized HBR relative to
Tower A, as represented in CTDM.
                                  194

-------
w



PQ





Q

Pi
 P-.
a
               CO

               §
              •I-l
               4J
               cd
               n
               4J
               a
               Hi
               u
               e
               o
              o
              CO


              g

             •t-4
              fi
              (U
              O

              §
              o
in

•  *

.
co
CM
O
CO
co
i— i
VO
CM
•
O
CO
o
r-l
m
CM
S.X
/-N
1-1
o
^~s


CN
I-l
r-l
r-l
•
O
o£
•
o
^ — \
^x
/--N
0
•
0
-**~s


                                                                          CO
                                         vO
                                                                          M
                                                                          I
                                                        CM

                                                        in

                                                        CM
                                                                                        VO

                                                                                        vO
                                                                                        CM
                                                         X


                                                         §
                                                         CU
                                cd
                               >

«

Q)
^
o
S
(^
co
a>
3
r— 1
cd
^
o
u
ft,
o

(J


0)
4J
cd
r-l
3
U
r-l
td
o

M)
CO

m co
0 0)
-- 3
CO r-l
3. Cd
I4_l
O c)0
CO
CO
4J T3
•H a
a «
3
bO
cu E
cd co
,£ V
4-1
cd o
rJ C
CU
td
CO x— s

cd
S v*^


r-l CM

CO
CU
4-1
O
i3
CD o
•J3 >
4-1 Cd
too
O
4-1 >\
a r-i
*r-t t
as
cd w
r-i o
cd
CO
CU CO
CO -r-t
cd &
V H
0)
ca
13 0)
0 g
r-l -r)
4-J
CU
4J «
cu cu
CO 60
3 M
cd cd
0 r-l
cu
.£> to
cd
cu 5
M

w'x
cd v»x
CJ b3
g h
^

ctf^^*«
H 0
1—i " •
cu a

4J CU t"l
^ O
a o
0 <4-i (3
cu cd

II5
4-J ^^
4-1 0.
cd • *^ c*_^
cu ^-*
r-l tlO
CU C O
4-1 cd
cu cu
.Q 4J 60

co o cd
CU f r-l
O co
13 cd
4-1
>*i cd (>,
cu cu
, 1 I 1
r-l rH r-l
c« -H cd
^ ^c ^

*
                                                                    195

-------
 results of  the 34  individual hourly simulations for all the models except
 Valley.   It should be emphasized that these results are very preliminary
 because the meteorological measurements used to derive the model input need
 further editing  and refinement.

     The  residual  statistics show that CTDM(11083) and CTDM(11083-E) under-
 estimate  peak CO/Q values by roughly a factor of 2.0.  On the other hand,
 Valley and  the COMPLEX models overestimate the peak observed values by
 roughly a factor of 4.0 or more.  ma for the space/time-paired concentra-
 tions is  lowest  for the two COMPLEX models.  The noise and resolution
 statistics  are large for all the model calculations.

     The  tables  of the results for each hour show that the overall perfor-
 mance statistics are generally dominated by extreme values.  For example,
 Table 23  shows that for seven of the 34 tracer-hours the ratio of the peak
 CQ  to the peak Cp  is greater than 9.0.  The input wind directions and
 modeled concentrations for these hours suggest that CTDM (11083-E) largely
 missed the  hill.   Similarly, Table 25 shows that COMPLEX I completely missed
 the hill  for six of the hours.

     Several of  the problem simulations occurred for wind directions nearly
 parallel  to HBR  and for releases below the calculated Hc.   Further analysis
 of  the variability of the 5—minute average wind directions for a few hours
 suggests  that CTDM should be reformulated for releases below Hc at HBR.
 For example:

     •    Is the concept of a single stagnation streamline wind
          direction of 117° relevant for these highly variable flows?

     •    Should the PDF form or Gaussian form of CTDM be applied?

     •    How should the recirciilation of pollutant material from the
          south  (or north) of the sample grid be handled by the model?

     •    In addition to the data refinement issue, are the Tower A data
          representative of the plume dispersion conditions,
          given  the highly variable nature of the flow below Hc?

     Experiment  11 (0200-0300) exemplifies some of these concerns.   Estimates
 of  5-minute winds  for the release height and from the 5~m, 10-m and 30-m
 levels of Tower -B  are listed in Table 26.  The wind data estimated for the
 release height indicate that the CF^Br plume should have impacted on HBR
 for only about a 10-15 minute period during the middle of the hour.  The
 Tower B 10-m data  suggest a more prolonged impact, from about 0215 through
 0245 (see Figure 56).   In any event,  the CF3Br plume was probably swept
 from the south side of HBR to north of the sampler grid during the course of
 the hour, so that  the samplers were influenced by plume material for a
 considerable part of the hour.

     The PDF reflects the wide range (18.2 - 225.0°) of wind directions
measured on Tower_A;  hence,  CTDM(11083)  and CTDM(11083-E)  produced very  low
 concentrations (Cp =0.6 and 0.2,  respectively).   The COMPLEX I and
 COMPLEX II models  produced relatively high concentrations (Cl = 50.3 and

                                  196

-------
              I
              o
                       (\j o vy.
                                                                '^rO— ih-
                                                                       «-< -«      ru    ru »•-
                                                                                                               »-< 53-
             to


              c

              i
              o
                       tn j"> in x>
                                                inN-ssr\j3-ruot«liin
                                   -< ru
                                                       *-< *-4 "-o'*-* ^ in  co
                                                       c      it
                                                                                        o  -o  IB »M

to
o
          X  X
          (C  (D
          X s:
          o  c.
          o o
                             «n co  >^  in •*"* co in co  co
                                                             ,£> CO i
                                                             «• AJ
                                                                      m  —i      -t
                                       ru AJ •-«    ru     —i       -*     ru
                                                                                                     oo
                                                                                                     ru
                                                                                                                  ru
3

I-H
to
              X
              (0
              a
             CJ
                       Ajcoo-4!>r«-»^<«a-Ln
                                                                o- zt *-> rvj
ru »
ru r\j
                                                                                                        (M
                                                                                                                        »i    in     ru
00   r~>
O   tU
•-<   CO
S
e
u
     CQ
     a
             to
i   I     'o
I   g
U-,   O
o   tc
      I
CO   1^       X
CO
                       53" *-« o O* O*  c** «••* "^ r*^ f\j ^  ^ ?y r^ rO ru o

                       ru-*o.'M''M:1\jru^or^'~txi-*co—««o-*>o
                       o-»^rux»H(n—«-oin!o»-*oin
                                X> if
                                       o
                                       ru
                                             oooruru^'-io^in
                                             5t ^, ^, _, ^,        nj ru in
                                                                                  ^jox>--  o o
                                                                                  -o o> ru
CNl
CNJ


W

CQ

ft
            i  O
            I U
                          a~  v *o o r>-  ^

                          —•  «»• in «-« co  ru  i
      ^5 3^ O^  ^5 ^^ ^^* in 2Q
     >»o«a**ea
      ru ^< ru        =r in co
                      ^) in  ^o -^ c^ ~~*
                      o  r\j ^ 'XJ
                      —i  ^    ru
                                         4 ~-
                                                                                  AJ    ru c-
                                                                                                        «4    «-« a-
                                                                                                                              ru
                                                                                                                              ru
                                                                                 ru     it ^>
                                                                                 cu     ru oo
                                                                                                 >-<•-«•«-    ooru^r     o
             .
                                                                            CQ
                                                                 197

-------
             o
                                                                          «•«       ru
                                                                                      CO
           lo
           I I
           I o
           10
                       ru in o r^. oc -* i

                       f~ ru ru •** i  ft
                          —i  i ru
                                                 I      i
ro     I  I  in
                                                                          *-«      *~*  I  ^-* GO
o o ru r-- ru >c  I  *-<
-* ~o  i  i   i         i
          (D (C
          £ Z
          O C.
          O CJ
                                              ru    ru     ru —«    r*Y
                                                                                   m    --  *-• -<
                                                                       •^  *H      0s ^ ru  ru t^
                                                                       ru  ru      i-> ru >o  «^ «•<
ro
eo           /-»
O   r^      Q.
r-t   W     /T
r-l   CO     ".
*-*  <     tX
                                — ru ru     xr o -i ru
                                                                    •-« r^ in
                                                                                                                      rvj
S   °
D!   Crf
           i a
           to
                             in     »^ ">
                                                                                         so  ru
                                                                                                                       53-  «-• ru
O,   3
O   Q

     f-      x
3   «—'
                                                                                   ^-i     oo  cc ru     ^-» =3- •-«
to
OJ
hJ
03
                                                                        in co      t»»  »H ru i
                                                                           -H      ru     ru
                                                                                                                                ru
                                                                                                                                ru
            10
                       orufiru     ^ru^ru'-truru
                       -4 ~*    ru
                                                                                   ru     ^» ^ OD  ^< —< h»    ooru^1    o
                                                                        •^ -^      ru    ru so
                                                                                                                •H  >£>
                                                                                                                                ru
              a b
                   \o
                                                                  198

-------
               i
               o
                         •oomoo-eef».^aoooin?rac
                                                                          o o
             I  a
             lo
             I  i
             I  o
             10
                                                   til
                                                                                                                          it
 en
 O
 i—i
           X X

           1C (5
           c a
          ? i j_j
                                                                                     o
                                                                                     o
                         a- o-
                         c^ c?*
                                                                                                         &•
                                                                                                         o-
       &
       o*
 S
 i— i
 en
              a
             o
nj co
i-«.O'.
r\i
OO
                                                                                                                     f\J

                                                                                                                     (\J
X

h-4
a,


I
      «
              0
              u
]—1   Q£j

O   CQ
K   K

tL.   Crf
o   no
     O
en   a
E-    i
                       •~«-*r\l>-i    —t        ix^^,,^     .^
                                                                                                                        »H f\)    t\l
en
BJ
crf
m
i-4
              c
             CJ
                                                                                                                                  rg
                                                                     *-ii'j"'.  nj
                       =>-vj~n-M
                                                                        •^ <*~       r\j    ^i  ,c ®
                                                                        •^ —i       r\j    AJ  oo
                                                                                                                 •^ -C
                                                                                                                                 o
                                                                                                                                 nj
              , t/J
             LlJ
                                                                              CQ

                                                              ru ru a-  in,tn ptT
                                                                  199

-------
                              o f\j o  ^* o zf o f^ '3O oo  -0 ^ ^r ^  r*»
              i
              o
             u
                       K1
                                 «O ^^ O* ^"* NO fy  *"^ *O O^  NO *™* t^ *™^  O1*
                                 -*•     ,TfUK1»HNO»-ifUrOrU    CO  «•«
              a       o *-i co »•«
                                                           rvi r»- o &• t^  in
            II        NorvjNONi     irururur-^o«-iNO5TJOru
            to         iii-i>H         I   I =r  —< —«  I —<  I      l—i
            to                I                   I   I   I     I
                                                                                                                                        ru
                                                                                                                                         l
                                                                                                                            II
00

o
          X  X
          N3-ooor~Ok


              2T        r^-NOiro     j^coNoo.somor^»r**-oxNO3'
              O        ON. QN,  ^^ ^p^     ^^ ^N JQ  gjj  f^ f^. ^Q  -Q  ,plf n^ f^  ^Q

              tj        inrutoro     ir»-"fUenNico«-ti-ii-<    o
                                                                                                           OO
                                                                                                           ru
                                                                                                                                        ru
                                                                                                                                       i   •

                                                                                                                                        NO
X   U3

3   3
O—   CQ

O   •<

     ^
>2   Q

Ctf   Cc£

O   E
^3   —^

b   3
O   Q
              a
              o
            I a
                       O *F«« O 5T »•< O^ ^ *•*
                                                                     ro
                       5T ^^  3O

                       tO «T  >O
^^ ^^ ^O
ru *\ «-<
                                                        —< ru KI ru
                          cotJL^lO•^«-^ooxoxln^u'7•coof^J'^Jco
                       NO  cr o o-
                       -H  »-< ru
mNO=TO-ocoi-iNoruo
       ^- -H ru     •rt       ru
                                                                                       N-< ru
                                                                                                                                        NO
                                                                                                                     rvi
                                                                                                           f\j
J   ^-i      x
3   "^^      (D
tO            2
              C
s
                       ••* r*-  NO rO i-*


                       O sO  9- -O in
                                                     in NO in oo i^ o
                                                  NCoruruNT»^o^-in
                                                  ,*,*_,_        ru ru  in
                                                                                    ••H     coeC-fV     •^^•^^STNO    >-<     »i
in
CN
tu

ca
                       un  ^^ ^r "O ^s  ^^ 5T~ ^^ ^D  ^^ ^^ ^*

                       t---<-Ttni-'r')rut>»r'iru«-i(\j
                                                                  -« N> in  CD
                                                                     w in cc       r- •-  ru i
                                                                           •^       ru     ru
                                                                                       ru
                                                                                       ru
                       o  ru MI ^u
                       1-1  i-<    ru
                                           ru=rru^ruru        ^ —•
                                                                           ^*«      f^     •*^ -O flC *•*  *^ ^^    ^^  ^^ ^U -3"     ^?
                                                                           •»^      (\i     AJ CC                 *-»  >O           f\I
                   \o
                                                                               pa
                                                                                 to
              oJ
                                                                   200

-------
O
O
CO
o

o
o
CM
O





CO
a
0
>-i
CM

PQ

a)
u
O
H





in
g 0
O vO
O
. j





CM
6 o
O —
i™H G\
o

CM
,_,
H o
m vo
r-l

r-
o
^
•*
vo

-------
 62.4,  respectively)  because of  the  narrower simulated crosswind distribution
 of the CFBr about the mean direction.
      It is obvious that the CTDM(11083-E)  calculations would have better
 simulated the observations  if  the Tower B  10-m winds had been used.  Photo-
 graphs and visual observations of the  oil-fog plume during other experiments
 suggest that during highly  variable wind situations aged plume material is
 often transported from the  south (or the north) over the sampling array.
 This  is probably  what  is happening during experiment 11 (0200-0300).  This
 type  of "sloshing" phenomenon  is not handled by the current models and
 should be considered during subsequent model development efforts.

      Because of the uncertainty  in modeling the CF3Br data with the wind
 directions based  on Tower A data interpolated to release height,  a second
 series of model runs was made  with the bi-variate Gaussian form of
 CTDM(11083-E) , assuming that the mean wind was directed from the  source to
 the area of  greatest observed  concentrations on the ridge.   These changes
 tend  to place  the peak of the horizontal distribution closer to the
 stagnation wind direction assumed in the model for material below Hc,  and
 they  place the peak of the  distribution for material above  Hc at  those
 receptors  with the  greater  observed concentrations.

      As expected,  these changes cause peak modeled concentrations  to
 increase considerably in many cases.   The net effect can be seen  in the
 overall performance statistics.  When the results of modeling the  CF3Br
 data with  the bi-variate Gaussian form of CTDM(11083-E)  and the modified
 mean wind  directions are combined with the previous model results  for  SF6,
 the mean bias  (mg) of the residuals  of  peak concentrations  is nearly 1.0,
 although the noise  remains  large (sg  =  4.3, sa = 75.6).   The  resolution
 improves considerably:  ra (peaks) drops from 4.6 to 1.5;  ra (all
 concentrations) drops from 4.7 to 2.2;  and rg (peaks) drops from 1.6 to
 1.2.   This exercise serves to demonstrate that there is  plenty  of latitude
 for improving  the performance of CTDM at HBR.

     In summary,  the results of the  initial HBR modeling  are tentative.
 None of the models reproduce the observations very well  and no  model is
better than any other.   Several new  issues  must  be considered in simulating
 dispersion below l^ at  the Hogback.
                                  202

-------
                                  SECTION 6              .

         SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS FOR FURTHER STUDY,
      This Third Milestone Report  has  documented  the  evaluation of several
 complex terrain models  using  the  complete SHIS #1 MDA.  The model performance
 measures suggest that CTDM(11083-E) is  significantly better than the other
 models.   The report  also  describes in detail the further progress made to
 date in analyzing the SHIS #1 experiment-hours and in  improving CTDM.  It
 provides an overview of the SHIS  #2 at  the Hogback and describes the pre-
 liminary results from this field  experiment as well  as the modeling of a
 17-hour subset  of the HER data base.

 6.1  Principal  Accomplishments and Conclusions

      The Small  Hill  Impaction Study #2

      The Hogback Ridge  field  study has achieved its objective of extending
 the modeling data base  to include a detailed set of meteorological data,
 tracer  concentrations,  lidar  data and photographic data from a two-dimen-
 sional  ridge site.   The field program has produced a set of about 179
 tracer-hours  for model  testing, evaluation and refinement.

      An initial  analysis  of the SHIS #2 data base suggests  that the concept
 of  a critical dividing-streamline height is appropriate for stable flows
 toward  HER.   Tracer  gas released below Hj. disperses in a highly variable,
 "blocked" flow and produces relatively large ground-level concentrations  on
 the windward  face  of the  ridge.  Tracer gas emitted above HC disperses  in
 a flow  that travels over  the ridge and produces peak ground-level concen-
 trations near the  ridge crest and on its lee side.

      Like SHIS #1, SHIS  #2 has verified the basic concepts  of  the experi-
mental design.   The release of gaseous and visible tracers  from a mobile
crane or fixed tower, using real-time  meteorological  data to guide the
 selection of release locations and heights,  has resulted  in a  data base that
covers a wide variety of dispersion conditions  and  concentration  patterns.

      The meteorological  data from four towers  were  archived in real-time via
an onsite system of minicomputers.  During the  first  few  experiments, it
became apparent  that the archived data contained  noise.   Several  subsequent
modifications to the data system  were  successful  in reducing this noise.
ARLFRO and ERT scientists are currently working on  eliminating the noise
from all of the archived data. Initial  editing suggests  that  most of the
noise can be removed although there are subtle  high frequency  effects that
                                 203

-------
 need further investigation.  The latter effects will have no influence on
 the mean meteorological data but may degrade the turbulence statistics and
 covariances.

      The SHIS #1 Modelers' Data Archive (MDA)

      A Modelers' Data Archive (MDA) has been prepared and submitted  to the
 EPA Project Officer.   It is intended to be used by modelers in  simulating
 concentrations at Cinder Cone Butte and is available from the Project
 Officer.   The MDA consists of meteorological data that were objectively
 interpolated from the Tower A measurements specifically  for the heights of
 release of the SFg and CF3Br tracer gases.   It  also  includes hourly emission
 rates and ground-level tracer gas concentrations.  The MDA is continually
 evolving to include the meteorological  data that are found to be most
 representative of dispersion conditions at plume height.   As plume photo-
 graphs, lidar data, and meteorological  data are analyzed  further, the MDA
 will be supplemented and changes will be submitted to the Project Officer.

      The  Complex Terrain Dispersion Model  (CTDM)

      The  Lift and Wrap models reported  in  the Second Milestone Report have
 been combined into one model, the Complex  Terrain Dispersion Model.  CTDM
 represents  the first  step in developing a practical  regulatory model.   It
 was designed to combine both "Wrap" and "Lift"  flow  and dispersion assump-
 tions so  that a plume near Hc would exhibit  characteristics  consistent
 with both,  and it was designed to provide  a  better method  for treating
 surface reflection of plume material in complex  terrain.  Two versions of
 CTDM were prepared for testing:   CTDM (11083) and  CTDM (11083-E).  'E'
 denotes an  enhancement to the effective size of az over the  terrain.
 This enhancement is similar in some respects to  the  "terrain factor"  used in
 other models such as  COMPLEX I and  COMPLEX II.

      Comparative Model Performance  Evaluations

      CTDM(11083-E), CTDM(ll083),  Valley, COMPLEX I and COMPLEX II computa-
 tions of  1-hour average SFg  and  CF3Br concentrations were compared with
 measured  values from  153  SHIS #1  tracer hours.  The model performance
 measures  indicate  that  CTDM(11083-E) performs significantly better than the
 other models.   However, both CTDM(11083-E) and CTDM(11083) tend to under-
 estimate  the peak  observed concentrations on average.  Yet the bias of the
 residuals based on predicted and  observed concentrations paired in space and
 time is lowest  for  CTDM(11083-E).

      Valley and the two COMPLEX models  tend to overestimate the peak  observed
 concentrations  on  average, and have higher biases than the two CTDM models.
 The  noise and resolution  statistics  for all the models are large and,  in
 short,  there is  still  substantial room for model improvement.

      Twelve hours  from Experiments  201 and 210 were used to test certain MDA
modifications suggested by detailed case-studies, and also to gauge the
 importance of simulating the 5-rainute sequence of meteorology in obtaining
1-hour average concentration estimates with CTDM(11083-E).  The  results
                                  204

-------
 indicate  that the MDA modifications could significantly improve overall
 model  performance, but  that simulating each 5-minute period of an hour does
 not consistently improve the hourly concentration estimates.

     Investigations of  Plume Growth

     Detailed analyses  of the Tower A turbulence intensity data, lidar data,
 and photographic data collected during CCB Experiments 201 and 210, as well
 as comparisons of iz data with lidar measurements from the other experi-
 ments, have  shown that  the following expression for az adequately
 simulates plume growth  upwind of CCB to distances of about 1 km:
     a  —
      z
              a t
               w
               t/2TT)°-5
                   Lt
where TL = y2/N.

The data suggest that y ^ 0.5, which is consistent with the Pearson
et al.  (1981) estimate that y» a molecular exchange coefficient,  could
range from 0.1 to 0.8.

     CTDM Modifications and Testing

     The case-study analyses of SHIS #1 Experiments 201 and 210 have
suggested three principal areas for modifying CTDM(11083):

     •    An enhancement in the size of the effective az over the hill
          is required for the part of the plume that diffuses or  is released
          above Hc.

     •    The form of this enhancement must recognize the relative importance
          of increased plume dilution and decreased streamline heights above
          the surface, and it must include the effects of lee waves.

     •    A more nearly continuous transition between Wrap  and Lift
          computations is needed near Hc.

     These model modifications were formulated and implemented to create
CTDM(14083).  Because a portion of the formulation was designed on the basis
of the results of case-study analyses of Experiments 201 and  210,  CTDM(14083)
was tested on the complete 153-hour MDA as well as on the twelve  hours from
Experiments 201 and 210 so that an overall comparison of model performance
with CTDM(11083-E) could be made.   The resulting measures of  model perform-
ance (ma,  sa,  ra, mg,  sg,  rg)  were nearly the same for CTDM(11083-E) and
CTDM(14083) so that CTDMC14083),  when using the initial MDA,  is not a
significant improvement over CTDM(11083-E).   However,  the comparison using
the twelve case-study hours indicated that the placement of peak  concentra-
tions is improved significantly,  even though there appears  to be  a need to
include some degree of plume dilution upwind of the hill in both  the Lift
and Wrap modules.
                                  205

-------
     Preliminary Modeling of the SHIS #2 Data Base

     CTDM(11083), CTDM(11083-E), Valley, COMPLEX I, and COMPLEX II calcula-
 tions of  1-hour average SFg and CF^Br concentrations were compared with
 concentrations obtained for 17 SHIS #2 hours (34 tracer-hours).  The per-
 formance evaluations show that none of the models simulates the observations
 very well.  CTDM(11083) and CTDM(11083-E) underestimate the peak observations
 by roughly a factor of 2.0; the other models overestimate the concentrations
 by about a factor of 4.0.  The performance statistics for all models show a
 great deal of noise.

     Several questions arise from an analysis of the model simulations:

     •    How relevant is the concept of a single stagnation streamline  (as
          in Wrap)?

     •    Does the recirculation of tracer material from the south (or
          north) of the HER target area affect measured concentrations?   If
          so, how can this phenomenon be modeled?

     •    How can meteorological data be selected to represent the highly
          variable flow below Hc?

     These questions must be addressed in the subsequent model development
 efforts.  It also should be emphasized that the modeling results are
 preliminary because of the problems with the tower meteorological data.

 6.2  Recommendations for Further Study

     Further analysis of the SHIS #1 and SHIS #2 data bases are required to
 improve and test CTDM and to extend it to ridge and other terrain settings.
 These analyses must be supplemented by new fluid modeling investigations.

     6.2.1  The SHIS #1 Data Base

     The case-study approach to analyzing the CCB data should continue.
Experiments 201 and 210 have provided insight on refining the Lift modeling
 approach.  Because the higher x/Q values measured at CCB occurred on the
 windward face, renewed emphasis should be given to analyzing the Wrap
modeling approach with data from experiments like 206 (see CTDM Second
Milestone Report).  The case-study approach will also help update the entire
MDA by investigating photo- and lidar-derived wind directions and turbulence
 intensity data.

     More theoretical work and data analysis are needed to refine the
modeling of dispersion close to and above Hc.  Numerical potential flow
modeling and physical modeling will help in formulating streamline distor-
 tion effects.  The case-studies will help modify and evaluate the resultant
 formulation for a and T in the Lift module, and also the resultant changes
 to the simulation of streamline distortion and plume dilution in the Wrap
module.
                                  206

-------
      CTDM must also be modified with a. view towards  simulating flow and
 dispersion around terrain features more  complicated  than CCB.  We have seen
 that successful modeling of  the SHIS #2  data will'require some additional
 reformulation of such basic  components of  CTDM as the  stagnation streamline.
 What terrain height should be  used in calculating Hc for less ideal terrain
 shapes?  How does the "Hc surface" vary  over rolling terrain between the
 meteorological measurement site and the  major  terrain  feature being .modeled?
 Similar questions must be posed and answered within a  larger framework for
 CTDM so that the refined modeling  concepts  contained in  it can be transferred
 to other terrain geometries.

      Closely allied to these modifications  of  CTDM are considerations of how
 CTDM,  as developed for SHIS  #1 and #2, will  be  applied to the upcoming full-
 scale study.  Certain components of CTDM will  be considered more important
 on this scale than others.   For example, with the relatively slow growth of
 az for elevated plumes in stably stratified  atmospheres, az may be
 wholly determined by the initial buoyancy-dominated growth near the source
 so that the proper specification of nonbuoyant  plume growth over distances
 of 1 to 2 km,  which is so important  in understanding the SHIS #1 data,  may
 have little relevance.   Similarly,  certain aspects of  the formulation of
 Lift designed to  simulate concentrations over the crest and in the lee  of
 major terrain features may assume  considerably  greater importance than the
 Wrap formulations  if the plume  is  expected to be near or above terrain  height
 much of the time.   The characteristics of new versions of CTDM should be
 explored via sensitivity studies to  understand how the model will perform on
 larger scales.

      The question of model sensitivity to the detail of the available input
 data must also  be  addressed.   The  SHIS #1 and #2 data have underscored  the
 need for exceptional resolution in the meteorological data in order to
 understand and  model  the  observed concentrations.  As CTDM becomes  more
 refined,  it will  necessarily become  more dependent on the availability  of
 detailed input  data.  For practical  regulatory application,  parameterized
 "cousins" of CTDM  must be developed  to retain the more important  features of
 CTDM while  reducing  the input data requirements.  This  process will not only
 yield  a version of  CTDM  capable of producing "screening calculations";  it
 will also help  define the absolute minimum information needed  for "refined"
 modeling within the  current regulatory permitting framework.

     6.2.2  The SHIS #2 Data  Base

     Before  significant progress can be made in modeling HBR,  segments  of
 the  SHIS #2 data base must be studied and intercompared.   The work  falls
 into two  principal categories:   refining the measured data and assessing  the
 representativeness of that data for model-ing the tracer gas plumes.

     Four types of noises have  been identified  in the tower meteorological
measurements taken at the Hogback.   Considerable additional work will be
 required  to  filter these noises from the  data and  to  refine the meteorologi-
cal data base.  The so-called 60-HZ and 2-to-4-sec noises have yet to be
 characterized and no methods  have been developed to eliminate them  from the
data.
                                  207

-------
      A necessary initial step in the data refinement  process  is  to  obtain
 more propeller response and F460 cup-and-vane  response characteristics from
 wind tunnel tests,  especially at low-wind speeds.  We  recommend using  the EPA
 FMF wind tunnel to  perform these tests.   Another necessary  step  is  to compare
 the sonic, propeller and cup-and-vane anemometer data.   A careful evaluation
 of the turbulence statistics from the sonic vs. propeller anemometers is
 especially important for modeling.   Once  the instrument  response character-
 istics are better understood,  the data can be  adjusted to the extent pos-
 sible, and the overall accuracy of  the various measurements can be  assessed.

      The issue of data respresentativeness must be addressed if we are to
 construct a reasonable MDA for SHIS  #2.   Many  pieces  of  data are available
 to help us construct  a detailed  spatial and temporal  picture of the varia-
 bility in the  flow.

      Temperature  data from the three main towers and  the  tethersondes must
 be integrated  to  create a  cross  section of the vertical structure of the
 flow upwind and over  HBR,  and  this structure must be  compared with the
 acoustic sounder  records.  This  structure will also be particularly useful
 in modeling the plume rise of  the jet-fogger plume containing the SF6
 tracer gas.

      Wind data from the doppler  sounder well upwind of HBR should be studied
 to characterize the large-scale  incident flow,  and the wind  data  from all  of
 the towers,  the tethersondes,  and the wind directions obtained from images
 of the  smoke plume will be needed to understand the relationship  among the
 vertical temperature  structure,  the  incident flow stucture,  and the  wind
 variability measured just upwind of and over HBR.   It  is  the flow and
 turbulence  in  this zone  that determined the concentration patterns measured
 at  HBR.

     With a refined data base, and a rudimentary understanding of the  forces
 affecting transport  and diffusion of plume material upwind of  HBR, the case-
 study  integration of SHIS #2 data can proceed.   Key elements in successfully
modeling the observed tracer gas concentrations include:

     •    extending the latest version of  CTDM(14083)  to  HBR,

     •    verifying  plume rise and CTZ development algorithms with plume
          photography and lidar data,

     •    creating the HBR MDA, and

     •    summarizing spatial and temporal patterns  in measured
          concentrations, with special emphasis on  the 10-minute data and
          the concentrations measured on Towers B and  C.
                                 208

-------
                              REFERENCES
Burt, E.W. 1977.  Valley Model User's Guide.  EPA-45Q/2-7-7-018.  U.S.
     Environmental Protection Agency, Research Triangle Park, NC.

Cline, A.K. 1974.  NCAR, Comm. ACM _1_7, 4, p221.

Crow, L.W. 1975.  "Meteorological Data Analysis Based on Monitoring
     Stations and Meteorological Data, January - December 1974.  Joint
     Environmental Program,"  No. 153, Loren W. Crow Associates,
     Denver, Colorado.

Environmental Protection Agency 1981.  "Regional Workshop on Air
     Quality Modeling:  A Summary Report,"  Source Receptor Analysis
     Branch, OAQPS, Research Triangle Park, NC.

Fox, D.G. 1981.  Judging Air Quality Model Performance:  A Summary of
     the AMS Workshop on Dispersion Model Performance.  Bull Am.
     Meteorol. Soc. . 62: 599-609.

Gifford, F.A. 1980.  Smoke as a Quantitative Atmospheric Diffusion
     Tracer, Atm. Env. , 14: 1119-1121.

Greene, B.R., and S. Heisler 1982.  EPA CTMD Quality Assurance Project
     Report for SHIS #1.  ERT #P-B348-350.

Halitsky, J. 1961.  Single Camera Measurement of Smoke Plumes.  Int.
     J. Air & Water Poll. ,  4_: Nos. 3/4,  185-198.

Holzworth, G.C. 1980.  The EPA Program for Dispersion Model Development
     for Sources in Complex Terrain.  Second Joint AMS-APCA Conference
     on Applications of Air Pollution Meteorology, New Orleans. LA.

Horst,  T.W. 1973.  Corrections for Response Errors in a Three-
     Component Propeller Anemometer.  JAM, 12: 716-725.

Hovind, E.L.,  M.W. Edelstein, and V.C. Sutherland 1979.  Workshop on
     Atmospheric Dispersion Models in Complex Terrain.   EPA-600/9-79-041.
     U.S. EPA,  Research Triangle  Park, NC.
                                  209

-------
                        REFERENCES (Continued)

Hunt, J.C.R. and R.J. Mulhearn 1973.   Turbulence Dispersion from
     Sources Near Two-Dimensional Obstacles.  Journal of Fluid
     Mechanics 61_: 245-274.

Hunt, J.C.R. and K.J. Richards 1982.   Stratified Shear Flow over Low
     Hills.  To be published.

Lavery, T.F., A. Bass, D.G. Strimaitis, A. Venkatrara, B.R.  Greene,
     P.J. Drivas, and B. Egan 1982.  EPA Complex Terrain Model
     Development Program:  First Milestone Report - 1981.
     EPA-600/3-82-036, U.S. EPA,  Research Triangle Park,  NC,  304p.

Londergan, R.J. 1980.  Validation of Plume Models:  Statistical Methods
     and Criteria.  EA-1673-54.  Electric Power Research Institute,
     Palo Alto, CA.

Moore, G.E., R.G. Ireson, C.S. Liu, R.E. Morris, A.B. Hudischewskyj,
     and T.W. Tesche 1981.  Air Quality and Meteorology of
     Northwestern New Mexico, Draft Final Report No. 81203.  Arizona
     Public Service.

Pearson, H.J., J.S. Puttock and J.C.R. Hunt 1981.  A Statistical
     Model of Fluid Element Motions and Vertical Diffusion  in a
     Homogeneous Stratified Turbulent Flow.  J. Fluid Mech. 129;
     219-249.

Rhoads, R.G. 1982.  "Addendum to Workshop Summary Report,"  OAQPS,
     U.S. EPA, Research Triangle Park, NC.

Sheppard, P.A., 1956.  Airflow over Mountains, Quart. J. Roy. Meteor.
     Soc., 82_: 528-9.

Smith, R.B. 1980.  Linear Theory of Stratified Hydrostatic  Flow Past
      an Isolated Mountain.  Tellus,  32: 348-364.

Snyder, W.H. and J.C.R. Hunt 1983.  Turbulent Diffusion From A Point
     Source in Stratified and Neutral Flows Around A Three-Dimensional
     Hill; Part II, Laboratory Measurements of Surface
     Concentrations.  (Submitted to Atmospheric Environment).

Strimaitis, D.G., A. Venkatrara, B.R.  Greene, S.R. Hanna, S. Heisler,
     T.F. Lavery, A. Bass and B.A. Egan 1983.  EPA Complex  Terrain
     Model Development Program:  Second Milestone Report -  1982.
     EPA-600/3-83-015.  U.S. EPA, Research Triangle Park, NC, 375p.

Venkatram, A. 1982.  A Framework for Evaluating Air Quality Models.
     Boundary-Layer Meteor., 24:  371-385.
                                  210

-------
                        REFERENCES (Continued)

Williamson, H.J. and R.R. Krenraayer 1980.   Analysis of the
     Relationship Between Turner's Stability Classifications  and Wind
     Speed and Direct Measurements of Net  Radiation.   Second  Joint
     Conference on Applications of Air Pollution Meteorology, March
     24-27, AMS, Boston.
                                  211

-------
              APPENDIX A
LABORATORY SIMULATION OF NEUTRAL PLUME
   DISPERSION OVER  CINDER CONE BUTTE
                   212

-------
                   LABORATORY SIMULATION OF NEUTRAL PLUME
                     DISPERSION OVER CINDER CONE BUTTE

                         Comparison with Field  Data
                                    by
                            Roger  S. Thompson,
                            William H. Snyder*
                         and .Robert E. Lawson, Jr.*
                    Meteorology and Assessment Division
                Environmental Sciences Research Laboratory,
                      Environmental Protection Agency
                     Research Triangle Park,  NC  27711
                                April 1983
*0n assignment from the National Oceanic and Atmospheric
 Administration,  U.S. Department of Commerce.
                                     21S-

-------
                               INTRODUCTION
     The purpose of this series of experiments was to duplicate, in the
laboratory, the field experiments performed by Environmental Research &
Technology, Inc. (ERT) at Cinder Cone Butte, Idaho.  In this particular
study, a one-hour period (1700-1800 MST) of Case 202 was modeled.
Meteorological measurements during this period indicated that atmospheric
conditions were neutral.  The laboratory study was performed in the
Meteorological Wind Tunnel, a facility well-suited for modeling neutral
atmospheric flow.  Measurements of vertical profiles of wind speed,
ground-level concentrations, and vertical and lateral profiles of
concentration were made during the wind tunnel experiments.   This report
describes the conduct of the study, makes comparisons of field and
laboratory measurements, and discusses some additional observations of the
laboratory study.
                                  214

-------
                       SUMMARY OF FIELD OBSERVATIONS
     The experimental data collection network used in the field has been
described by Lavery, et al.  1982.  For the one-hour period considered here
(Case 202, October 17, 1980, 1700-1800 MST), the meteorological data were
fairly complete.  The concentration data contained only nine, non-zero,
one-hour average concentrations; however, an additional 14 non-zero,
one-hour values were computed from ten-minute concentration data.  There
were five reported zero one-hour concentrations and an additional ten zero
values computed from the ten-minute data.

Meteorological Data

     Examination of the NOAA Daily Weather Map series for October 17 and
18 showed synoptic conditions at Cinder Cone Butte (CCB) during Case 202
to be dominated by a weak high pressure system centered over southern
Washington.  Skies were clear with surface winds light and variable during
the morning.  Surface temperatures ranged from a low ,.of about 1°C to an
afternoon high of about 12°C.  Upper level winds (500 mb) were from the
north at 10-12 m/s.

     Under these synoptic conditions,  atmospheric stability is dominated
by surface heating and cooling.   Stable conditions prevail during the
early morning hours due to formation of a radiation inversion and, as
surface heating increases during the day,  there is a transition to
near-neutral and finally neutral or slightly unstable conditions.  During
the late afternoon and evening,  the,reverse process takes place as
insolation decreases and surface radiation begins to dominate.  The hour
examined in this case study (1700-1800 MST) falls just prior to sunset
(1800 MST) and, on the basis of these synoptic conditions, would be in a
transition from near-neutral conditions to stable conditions.   Examination
of the detailed meteorological data from CCB shows that this trend is, in
fact, taking place as the stability is initially neutral (D stability) and
becoming more stable (E stability) toward the end of the hour.

     Wind speed and wind direction.were measured and recorded  at five
levels (2, 10,  40, 80 and 150 m) and temperature at eight levels (2,  10,
20,  40,  60,  80,  100,  and 150 m)  ori the tower (Tower A)  located 2 km north
of CCB.   These data are summarized in Table A—1.   Five-minute  averages
were recorded at the time corresponding to the end of  the period.   The
table contains the u (westerly), v (southerly),  and w (upward) components
of the wind as  determined from the u,v,w  anemometers  in m/s.   Computed
from these are the five-minute averages of wind speed  (WS),  wind direction
(WD), along-wind turbulence intensity  (IX),  cross-wind  turbulence

                                 • 215

-------
 1

 o
 o
 CO
              M   S3 in N-tD N —• K> CD CD in CVJ ID
              «   —' N- VD CD IO N- f in CO M" fO O
                                                                                                               tonoj^f. N to to CM <\J CM PJ
 O
 O
 o
 oo
 en
              >-   uj — in i*- m en m.(D TT ori oo —
                                              -
                                                           >-   in in ro ro VD CD —• G) in N-m OD
                                                           —•   CD to iri r^ ^r — i^ CM en PJ — en
                                                                                         —-N-
 ai
 U3
 oa
              X   N-criNf t^M to TT CD ^r inn
              •-•   in co m \n s> *f en \n en N- CD —
                  (MOJ—D — iniDDV—-U3CD
—   VD N. en ED to ID t\HD M in a? ID
                            ' — 6D
                                                                                                               CA—'t^N-IO —
                                                                                                                   '
 CM
 o
 sa
 u
              §
                                                                                                                        — (M — -« -• ID IO
                                                                                                               tototototontoontoto
                                                                                                       rn      -
                                                                                                       3      \r in — to —
I
I
                             . CD ID h^ K-K. K. N-
                   10 O CD D O D CD (D 03 63 O O
                  cniococncncMcnincoiMioco

                  Nntncoco — CDTIO —N-^r

                  tONCVINWIOPJIOIOIOCMN
                   I I  I  I  I  I  I  I  I  I  I  I
                                                                                                               ininiDinioiDinniotDiD
                                                                                                       3      K) CM ID CO CD ID N CD CD CD ID
                                                                                                               CMIM — CM — — IMIM—•-« ~*
                                                                i  i  i  i  i  i  i  i   i  i   i  i
§
O

e
<

la
         E

        OJ

         ti

         M
                  CD ID niO — D ID ID IO (O — tO
                  ea —NIO ^r in n ID (^ CD m ea
                                                                CO — Cn IO tDUDMCTilMMN-CO
                                                      E

                                                      O
uj
C
    co ID n to -- eo co ID in 10 — to
    o — • M to v in in 10 h- CD en to
                                      E

                                      O
     111   _
MHO
                                                                  216

-------
o
o
00
1—I
 I
o
o
             M   CQCnEI — r
             —•   CD CDin \T (M El ED — N-~ ~ in

                  •tf CO O U CO T.   ivi in (vi-— locvicncoinio^rm
             —"   tvj co — co m in m BD — (vi en to
                                                              *-•   —• en —' in f^*in ^r ^3" v UD en N-
                                                                                                          o


                                                                                                          u
                                                                                                               O O CD CO E) CD GD G) E) E) O E)
ca
u
o
             X   tr ro in r^ M — M T m M t^ in
             «—   (nco—-iDCnuMDM — trxri —
                                            '
                                                                                 —•—'—^—«03ln
                                                                                                               -> — en—•
                                                                                                          00   tnco •tf — UD
                                                                                                               —' E) G> ^^ E)
                                                                                                          ISI
                                                                           (M

                                                                           El

                                                                           O
(M

O
co
                                                             Q
                                                                                                               ID en en UD en — — — — ro r^ en
                                                                                                             -  fo CM po M tvi r>j oj (M rM — en co
             s   »
                                                             tn   ^r to to n N- m 10 in N- EI i^ — •
                                                             3   CD o co in o tvj -• — in »D o en
                                                                  co —
                                                                  CDcnoDCOCDCDcricriOcncricn
                                                                                                                        — ro rv in en in   UD
                                                                                                                                          co
8
I
1
-]  3
O  S
CH -H
O  -P
W  p!

W U
        O

        CO
                  in^rrjininnnnnnnn
                  i  i  i  i  t  i  i  i   i  i   i  i
PJ
                  rOUDSDIOUDDIOIOSDtOlDD
             UJ   CD ID niO — D CO ID n 10 — D
             H   BJ — wnv n n 10 K. CD tn D
                                                                  'yncoiMininincococnMco
                                                                  inninninininiDiDiriiniD
                                                                  i  i  i   i  i   i  i  i   i  i   i  i
                                                                  enncnionTcoionco — —
                                                                            — inwioco — in— •
                                                        o
                                                        in
n

N   yj   coionio — coibioinio — o


Q)   r~   N- K- K- f^ N- N- ffc. ^ K_ N- K- CD
CD
CVJ
II

N
E
CD
*
\\
N
-^ ro in in ro in h- r^ cn co — •
co GI c\i o\i GD cvj ^~ ^3" LO r*— r\i
^D LD in in ^D in ^~ "^T to o^ co
_. _


CO G) CO CM — • ^3" CO
rsi — (D ro r^ in ,r\t
\u in ^o ^3" ^" fo r^
— -• — — -* — E3
                                                 CSJ

                                                  II

                                                  N
                                             03
                                             S_
                                             CD
                                             O.
                                                                                                              coujinto
                                                                                                              to
                                                                                                                          -Eicou)inro
                                                                .    217

-------
intensity  (IY),  and vertical  turbulence  intensity  (IZ) in percent.
Table A-lf contains the five-minute average temperatures (°c) recorded for
all eight  levels.

     One-hour average values  of  the meteorological parameters were
computed from the available five-minute  averages.  A one-hour average wind
vector was computed at each of the five  levels  of  the tower as follows:
               U
                                    N
                                              1/2
               6 = arctan
where u^, v^ are the values of the westerly and southerly wind
components for the ifch five-minute period and there are N periods
containing valid data for the hour.  The hour—average values of turbulence
intensity were calculated in the direction of the hourly-mean wind, in the
direction horizontally perpendicular to the mean wind direction, and in
the vertical direction.  The five-minute average turbulence intensities
were projected onto these directions and the variance of the five-minute
mean wind data was added in computing the hourly turbulence intensities.
fi? r<
N i=lU-
IX.WS.COSA9.
100 )
IY.WS.sinA9. „
,(•11 1)2
k 100 '
(WS.cosAG .)2_|- U2
U2
1
2
x 100%
IX.WS.sin,^ .
  11   A9i
    100
                               lY.WS.cos.
                                 i  i   A
                                    100
                                  U2
                                                                    x 100%
"IN IZ.WS ~
» y r L h2
N 4 . ( 100 }
1=1
u2
!_
2
                        x 100%
where the subscript i again refers to the i   five-minute period of the
hour and A9 . = 9 .  - 9 .
           i    i
                                   218

-------
     The one-hourly average values for each level of Tower A are contained
in Table A-2.  Vertical profiles.of the five-minute average wind speeds
are presented in Figure A-l.  The one-hour average wind speed values are
also indicated in Figure A-l and on a log-linear scale in Figure A-2.
From Figure A-2, zo was found to be on the order of 5 to 8 cm.  The
boundary-layer thickness appears to be greater than 150 nu  Vertical
profiles of the five-minute and one-hour average temperature data are
presented in Figure A-3.  A line with the slope of the adiabatic lapse
rate (rd) is included in the figure.   The environmental lapse rate
closely follows the adiabatic lapse rate above about 10 m from the
surface.  This, together with the previously discussed wind data,
indicates neutral stability for this period.
       TABLE  A-2.  ONE-HOUR AVERAGE METEOROLOGICAL DATA FROM TOWER A
z
U)
2
10
20
40
60
80

U
(m/s)
4.39
5.88

7.89

8.79

Q
(°)
312.6
313.7

316.5

308.2
(316.2)*
1
(%)
18.1
13.6

8.1

7.1

2
(%)
14.1
11.7

8.0

6.4

3
(%)
3.7
3.7

3.0

3.2

100
150
9.09
 309.4
(316.4)*
7.7
6.1
3.1
                                                          11.21
                                                          11.53
                                                          11.41
                                                          11.31
                                                          11.21
                                                          11.08

                                                          10.85
                                                          10.40
*The anemometers at z = 80 and 150 m were oriented approximately 8°
 and 7° east of true north,  respectively (Lavery,  et al. ,  1982).
 The values in parentheses are corrected for this  misalignment.


     After the design and data collection of this  wind tunnel experiment
were completed, ERT prepared and released their First Milestone  Report
(Lavery, et al., 1982) on the field project.  The  quality  assurance audit
reported misalignment of the wind systems at the 80 m and  150 m  heights on
Tower A by 8° and 7° east of true north, respectively.  Adjustments for
these have been made in Table A-2, but not in Table A-l.

Concentration Data

     The tracer experiment in the field was designed to examine
ground-level concentrations from a point source upwind of  an isolated
hill.   For Case 202, the source was located 1.0 km from the hill center at
319° from true north.  Sulfur hexafluoride (SFg) was released at a
height  of 50 m above the local ground surface.  Examination of the
topographical maps revealed that the local surface at this location was
8 m below the reference height selected as the base of the hill.
                                   219

-------
     200
CO
oc
LLJ
J—
ULJ
21
     150
     100 -
      SO -
                           	-i	•>-	»	,--....-:	.}	.-,	{
                           average for hour
                                    SPEED    (H/S)
Figure A-l.   Tower A, Case  202,  1700-1800,  5-minute wind speed values and

              hourly average wind speed as  a function of height above ground.
                                     220

-------
     200
Nl
  Figure A-2.    Vertical profile of hourly average wind speed measured at
                Tower A, Case 202,  1700-1800.
                                    221

-------
    200
                       r. - 9.8 °C/1000 m'

                         J..4.4.J...M..AIJ
                                               I  !  19 average for hour
                                           M I  I  I  I  I
                                                                         15
Figure A-3.    Vertical  profiles  of five-minute averages of temperature at
              Tower A,  Case  202,  1700-1800.  The one-hour average values and
              a line with slope  corresponding to the adiabatic lapse rate
              are also  shown.
                                    222

-------
      Time spent in moving the crane supporting the source into this
 position resulted in delay of the source beginning to emit  until  1716,
 sixteen minutes into the hour of interest.   For the remainder of  the
 hour,  the source emission rate was 0.080, ( + . 5.3%)  g/s of  SF6.
 on thpn                      at  a network of  100  sampling stations
 ?able  A  3    TV^'i      i°Cations of  these stations  are  given  in
 Fit    A /    k°th  P°lar  and  rectanS"lar coordinates  as defined in
 Figure A-4.   Sample  ports  were  located on  the model  at corresponding
 positions and assigned identification numbers also listed  in Table A-3.

     A summary of  the field  concentration  data is presented in
 fnt^h    ,    one-hour sample data required an adjustment to correct
 for the  portion of the hour  that the source was not  operating and the
 sample pumps were.  The ten-minute  sample data indicate that the
           tTfrh8 the SamPl6rS ^^ the be§inning  °f the period 1720
           that the concentrations measured during this period were
 roughly  equal to those for the period 1730 to 174?.   Therefore  the
measured "one-hour concentrations" were multiplied by 1.5 (6
-------
TABLE A-3.   LOCATION OF SAMPLERS ON CINDER CONE BUTTE
FMF
• ERT
PORT PORT
1
2
3
4
5
6
7
8
9
10
n
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
37
37
37
37
37
37
37
37
37
37
37
37
37
34
31
31
31
31
31
31
29
28
28
27
26
26
26
63
62
60
60
60
60
60
60
60
60
60
58
58
57
57
57
57
57
56
56
56
55
55
.01
.02
.03
.04
.05
.06
.07
.09
.10
.11
.12
.20
.26
.10
.90
.96
.07
.09
.26
.05
.04
.26
.11
.03
.03
.05
.06
.03
.03
.03
.04
.22
.23
.24
.25
.26
.01
.02
.08
.07
.25
.26
.96
.13
.90
.03
.04
.05
.05
,13
ANGLE
DEG
8
8
8
8
8
8
8
8
8
8
8
8
8
30
52
53
53
53
53
53
68
71
75
83
90
90
90
98
105
120
120
120
120
120
120
120
120
120
135
137
143
143
143
143
147
150
150
150
158
158
•
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
R
m
510
449
391
308
275
250
277
177
154
122
81
193
19
146
146
154
200
154
107
246
187
420
75
260
360
208
142
315
190
362
297
235
164
124
94
43
510
410
195
240
327
140
165
165
162
255
198
86
250
204
X
m
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
' 70.
62.
54.
42.
30.
34.
31.
24.
21.
16.
11.
26.
2.
73.
115.
122.
159.
122.
85.
196.
173.
397.
72.
258.
360.
208.
142.
311.
183.
313.
257.
203.
142.
107.
81.
37.
441.
355.
137.
163.
196.
84.
99.
99.
88.
127.
99.
43.
93.
76.
98
49
42
87
27
79
59
63
43
98
27
86
64
00
05
99
73
99
45
46
38
12
44
06
00
00
00
93
53
50
21
52
03
39
41
24
67
07
88
68
79
25
30
30
23
50
00
00
65
42
505
444
387
305
272
247
224
175
152
120
80
191
18
126
89
92
120
92
64
148
70
136
19
31
0
0
0
-43
-49
-181
-148
-117
-82
-62
-47
-21
-255
-205
-137
-175
-261
-m
-131
-131
-135
-220
-171
-74
-231
-189
Y
m
.04
.63
.19
.00
.32
.57
.79
.28
.50
.81
.21
.12
.82
.44
.89
.68
.36
.68
.39
.05
.05
.74
.41
.69
.00
.00
.00
.84
.18
.00
.50
.50
.00
.00
.00
.50
.00
.00
.89
.52
.16
.81
.78
.78
,87
.84
.47
.48
.80
.14
                         224

-------
TABLE A-3.   LOCATION OF SAMPLERS ON CINDER CONE  BUTTE  (Continued)
FMF
PORT
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
ERT
PORT
55
55
55
54
53
53
39
39
39
39
39
39
39
39
39
39
45
45
45
48
48
48
48
48
48
50
64
64
65
67
67
67
67
67
67
67
67
67
67
69
69
70
70
70
72
72
73
73
74
74
.96
.26
.90
.04
.03
.04
.21
.01
.02
.03
.05
.07
.12
.17
.18
.20
.24
.07
.10
.12
.02
.03
.05
.07
.10
.07
.05
.10
.03
.01
.02
.04
.05
.06
.13
.23
.24
.25
.26
.07
.26
.08
.26
.25
.23
.24
.10
.11
.03
.04
ANGLE
DEC
158
159
159
165
171
173
187
187
187
187
187
187
187
187
187
187
232
232
232
255
255
255
255
255
255
270
277
277
285
300
300
300
300
300
300
300
300
300
300
315
315
322
322
333
338
338
345
345
353
353
•
•
•
•
»
•
a
#
t
%
B
•
*
.
m
•
•
.
.
.
t
^
•
m
.
m
m
m
.
.
•
.
•
.
.
.
•
9
m
^
f
.
•
m
B
9
*
f
t
•
.

0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
R
m
204
165
196
301
139
260
37
510
411
348
291
251
115
180
150
77
107
224
158
63
411
327
258
211
150
238
282
159
252
510
416
343
311
284
230
247
X
m
.0
.0
.0
.0
.0
.0
.0
.0
0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
178.0
130
54
200
325
151
108
275
285
211
127
105
310
197
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
76
59
70
77
21
31
-4
-62
-50
-42
-35
-30
-14
-21
-18
-9
-84
-176
-124
-60
-397
-315
^249
-203
-144
-238
-279
-157
-243
-441
-360
-297
-269
-245
-199
-213
-154
112
-46
-141
-229
-92
-66
-124
-106
-79
-32
-27
-37
-24
.42
.13
.24
.90
.74
.68
.51
.16
.09
.41
.47
.,59
.02
94
.28
.38
.32
.52
.51
.85
.00
.86
.21
.81
.89
.00
.90
.81
.41
.67
.26
.04
.33
.95
.18
.91
.15
.58
.77
.42
.81
.96
.49
.04
.76
.04
.87
.17
.78
.01
Y
m
-189
-154
-182
-290
-137
-258
-36
-506
-407
-345
-288
-249
-114
-178
-148
-76
-65
-137
-97
-16
-106
-84
-66
-54
-38
0
34
19
65
255
208
171
155
142
115
123
89
65
27
141
229
118
85
245
264
195
122
101
307
195
.15
.04
.98
.74
.29
.06
.72
.20
.94
.41
.83
.13
.14
.66
.88
.43
.87
.91
.27
.30
.37
.63
.77
.61
.82
.00
.37
.37
.22
.01
.00
.50
,50
.00
.00
.50
.00
.00
.00
.42
.81
.99
.11
,03'
.25
.64
.67
.42
.69
.53
                               225

-------
Figure A-4.   Map of Cinder Cone Butte showing coordinate systems and
              locations of the 100 ground-level samplers.
                                 226

-------
     TABLE A-4.  GROUND-LEVEL CONCENTRATIONS FROM FIELD EXPERIMENT

 a)   One-hour sample  data
     FMF  Port
      Number
       10
       13
       69
       73
       31
       34
       35
       81
       82
       86
       87
       88
       89
       99
b)  Ten-minute sample data
Measured
Concentration
(ppt SFfi)
189.0
70.0
0.0
0.0
110.0
5.0
61.0
0.0
0.0
0.0
7.0
11.0
26.0
124.0
Adjusted
Concentration
(ppt SFfi)
283.5
105.0
0.0
0.0
165.0
7.5
91.5
0.0
0.0
0.0
10.5
.16.5
39.0
186.0
FM Port
Number
4
5
6
7
9
11
18
19
20
32
52
60
61
64
68
71
74
75
76
79
88
90
95
97
100
Measured Concentration in ppt of SFg
for Ten-Minute Period Ending at
1710
-
-
0
5
-
0
0
0
0
0
0
-
-
'
-
0
-
6
0
0
0
0
-
0
210
1720
0
-
-
-
67
0
0
0
0
0
-
-
-
0
0
0
-
0
0
0
0
0
-
0
194
1730
100
-
265
-
266
119
-
114
101
27
0
-
-
0
0
0
-
-
0
0
. 31
52
.-
0
0
1740
117
87
-
0
352
144
-
178
99
116
0
-
-
0
0
0
-
5
0
0
5
'6
319
- - .
0
. 1750
19 '
-
117
0
116
244
251
294
18
220
0
0
0
0
-
0
0
0
0
0
15
65
490
0
. - .
1800
0
-
0
0
405
-
176
387
8
-
—
0
0
0
-
0
-
-
0
0
47
150
,497
250
35
Average
for
1720-1800
59
87
127
0
285
169
213
243
57
121
0
0
0
0
. 0
0
0
3
0
0
25
68
435
83
— "&
*Data for port 100 were ignored because of high reported
 concentrations for 1700-1720 when source  was not  operating.
                                   227

-------
                           EXPERIMENTAL DETAILS
Model Specifications

     The model was constructed of acrylic plastic by a vacuum-forming
technique.  The nominal scale was 1:640; after fabrication, the actual
model scale was determined to be 1:647 in the horizontal and 1:694 in the
vertical (model height = 14.4 cm).  A survey of boundary-layer generation
systems used  in previous studies in the Meteorological Wind Tunnel
revealed that a boundary layer produced, using 92 cm high vorticity
generators, a 15 cm high fence, and a fine "Sanspray"    surface would
simulate the  full-scale situation.  The model was coated with gravel of
the same size as that used in the Sanspray (average diameter on the order
of 2 mm) and  installed in the wind tunnel, along with the boundary-layer
generation devices (see Figure A-5).  The center of the hill was located
8.0 m downwind of the fence.  The wind tunnel was operated at a freestream
air speed of 4 m/s.

     The source was constructed from a 0.32 cm O.D. brass tube that was
bent over to emit the tracer in the direction of the wind and avoid any
momentum plume rise.  Ethylene (99.8% pure) was released at a rate of
50 cm-Vs, so  that the effluent speed was approximately equal to the
local mean air speed.  Sampling ports were made using 0.24 cm O.D. brass
tubing.  The ports projected approximately 2 mm above the surface
roughness, except for the mast mounted samplers, which were scaled in
height in proportion to their field counterparts.  The sample ports were
connected to a scanning valve which allowed groups of 5 sampling ports to
be connected to the laboratory's system of 5 flame ionization detectors
(FID) for determination of tracer concentration.  The FID's were
calibrated before and checked after each experimental run of 100 samples.
The background level of ethylene in the room was measured before and after
each run and a linear interpolation algorithm was used to subtract the
background from'each sample.   The analog output of the FID's was processed
on the laboratory's minicomputer system using a digitizing rate of 1 Hz
for 120 seconds,  an adequate sampling time to obtain stable averages.

Conversion of Model Concentrations to Field Concentrations

     To facilitate comparison of the field and laboratory data,  the
laboratory data were converted to equivalent full-scale values.   Model
concentration values were recorded in percent by volume of ethylene.   The
minicomputer program for collection and analysis of wind tunnel
                                    228

-------
Figure A-5.   Cinder Cone Butte model in the EPA Meteorological Wind Tunnel.
              Note vortex generators at test section entrance.
                                    229

-------
                                                                    f\
concentration data calculates a nondimensional  concentration x = CUHZ/Q,
where C is the measured concentration,  U is the reference  wind speed, H is
the hill height, and Q is the tracer release rate.   The absolute wind
speed in the wind tunnel is arbitrary.   However,  to convert wind-tunnel
tracer measurements to field values, a  reference  wind  speed, at the  same
locations in the model and field, must  be selected and used in computing
X»  The reference for this study was selected as  the wind  speed measured
on Tower A at z = 40 m.  These nondimensional concentrations were  then
used to convert the model concentrations to field values using equivalence
of the nondimensional concentrations for model  and field  (xm = Xf
- x)«  This results in:
cf -
               xmQf/ufHf
where the subscripts ra and f stand for model and field,  respectively.
Substituting in the field values for the  period  of  this  study gives
        C(ppt)
       X(0.080 gm SF6/s)  (1Q12 ppt)/(6.5xlQ3 gm

                   (7.89  m/s)  (100 m)2
     or C(ppt) = 156x-
                                  230

-------
                  PRESENTATION AND  DISCUSSION OF RESULTS
     For the sake of easy comparison, all wind-tunnel measurements have
been converted to full-scale values according to the scaling concepts '
previously discussed.

Wind Field Measurements

     Verification that the boundary layer developed in the wind tunnel was
representative of that observed in the field was based on comparison of
wind measurments at Tower A in the field with the vertical profile
obtained in the wind tunnel upstream of the model.   As mentioned above,
the absolute wind speed in the wind tunnel is arbitrary.   Adjustment of
the wind-tunnel velocity profile to field values was based on matching the
speeds for the model and field at the 40-m level on Tower A,  the
measurement nearest the height of the source.   The  mean wind  speed
comparison, shown in Figure A-6, is quite good with the exception of the
measurement at z = 80 m on Tower A.  Also presented in Figure A-6 are  the
data for the Tower B site located on top of the south peak of Cinder Cone
Butte.  The mean velocity measurements in the wind  tunnel for this site
are also in good agreement with the field data.

     Vertical profiles of components of turbulence  intensity  were also
computed from the wind tunnel data for the two locations  discussed above.
The longitudinal (direction of the mean wind)  component and lateral
(perpendicular to the mean wind) component of local turbulence  intensity
are shown in Figures A-7 and A-8,  respectively.  The vertical component
was not measured in the wind tunnel.  Overall, the  turbulence levels in
the wind tunnel are somewhat larger than the field  values.  The
longitudinal values are about a factor of two greater and the lateral
values are about one and a half times as large.  Errors introduced by  the
excessive turbulence in the wind-tunnel flow, can be estimated for flow
over flat terrain.   Pasquill (1962) has pointed out that  for  a given plume
cross-section, the maximum ground-level concentration is  determined  by
only the ratio of the vertical spread to the lateral spread of the plume.
Thus, the magnitude of the maximum ground-level concentration should be
correct.  The change in the location of the maximum can be  estimated by
assuming plume spread to be proportional to turbulence intensity.  For
near-neutral conditions and a source height on the  order  of the height for
this study, a 50% excess in turbulence intensity can be expected to  result
in the maximum ground-level concentration occurring 20% nearer to the
source.
                                  231

-------
cuu
600
500
100 •
2:
rvT
30O •

200 •


100 -


0 -
i i i i i i i 1 i i i 1 i i i 1 i i i 1 i py. '
A Approach Flow, Wind Tunnel
A. Tower A, Field m
U Tower B, Wind Tunnel
• Tower B, Field
B
; ^ ;
• AD
A n
A n
A n
A D
A n
A n
A^ n
A n
A n
A n
A 4 n
4*^* M
A A i — i n c~!* — ' BI
— i 	 1_^_: 	 1 	 1 	 1 	 1 	 1 m *=$ 	 1 — S 	 1 	 1 	 Ll-!d — uJL 	 : 	 1 	 1 	 1 	 	
                                             10
12
                     U.  M/S
Figure A-6.    Mean velocity measurements.
                   232

-------
 700
                                   I i i i  i 1
 600
                      A  Approach Flow, Wind Tunnel
                     '4  Tower As Field
                      D  Tower B, Wind Tunnel
                      a  Tower B, Field
500
400
DA


 D  A


  D  A


    D


      n
30O
200 -
100 -
       n  A

         n  A

           n  A

             n  A

             n   A
             n  A
              n   A
               n    A
              • O    A
                n    A
             AD     A
                                    -AA
                    10    1^    .15'   17.5

                     LONG  TURB  INT
                                                   20
25   27.5
         Figure A-7.   Longitudinal turbulence intensity.
                                233

-------
700
     ^gjl
soo -•
                             A  Approach Flow, Wind Tunnel

                             A  Tower A, Field

                             d  Tower B, Wind Tunnel

                             m  Tower B, Field
500 ••
        n A

         n  A

             n A
tiOO --
300 • •
200 - -
100 ••
               n   A


                n   A


                   D    A
                   n
                      n
                       n    A
                       n     A
                        n     A
                        n     A
                       * n
                         D
                         n
                                A
                           n    A
                          n      A
                      06     08     iff     12     1U     16     18


                           LflTERRL TURB  INT  (ft)
                                                                    20
             Figure A-8.    Lateral  turbulence intensity.
                               234

-------
 Ground-level  Concentration Measurements  ...,.,         . „       	    	,

     Data were collected  in  the wind  tunnel  for  three different wind
 direction and stack height combinations.  The first  case was for an
 ambient wind direction of 312° and a  stack height of 50 m  (full'-sc'a'lev
 equivalent) above the wind-tunnel floor.  This wind  direction was selected
 as  an  average of the directions reported  for Tower A before  the  >
 corrections for alignment of the anemometers was later reported."  The
 actual terrain near CCB has  a gradual  slope  away from the  butte and some
 undulations.  For the second run, the wind direction was changed to 316°
 to  correspond more nearly to the direction measured  at the source height.
 Then,  the difference in surface elevation at the source location and the
 field  reference (8 m full scale) was  accounted for by lowering the stack
 to match the height above the field reference.  These situations, referre'd
 to  as  Runs  1 through 3, are  summarized in Table A-5.

     For each of the three runs, groutid-level concentrations were measured
 at  each of  the 100 ports corresponding to field sampler.locations.  These
 data are presented as concentration maps in Figures A-9 through A-ll.   The
 available field data are shown for comparison in Figure A-12.
 Concentration isopleths were hand drawn on these maps to facilitate
 comparison.  Results for each case can be compared with the field data to
 evaluate its success in reproducing.the field results.  In addition,  the
 results from the laboratory  runs can be compared with one  another to
 determine the effects of changing wind direction and"stack height.

                TABLE A-5.  WIND-TUNNEL EXPERIMENT SUMMARY
                          Stack Height
Case
  Wind
Direction

  312°
         316'
 Above Wind-
Tunnel Floor

  7.2 cm
              7.2 cm
Above Field
 Reference

   50 m
                  50 m
         316'
              6.05 cm
                  42 m
    Data Collected

100 surface port
concentrations

100 surface port
concentrations,
1 elevated lateral
profile, and 1 elevated
vertical profile

100 surface port
concentrations
     The concentrations for all three runs were, in general,  a factor of
two lower than the field values.  Runs 1 and 2, with the higher source,
exhibit bands of nearly uniform, maximum concentration that extend from a
point on the upwind surface of the hill to the lee side of the hill.   The
wind direction for Run 1 differs from that of Run 2 by only 4°; however,
the band of maximum concentration for Run 1 wraps around the side of  the
                                  235

-------
Figure A-9.    Concentration measurements for Run 1.
                        236

-------
Figure A-10.   Concentration measurements for Run 2.
                        237

-------
    source
                                                                     MflST R
                                                                          82.
            600 H
               x
SCRLE 1:8000
         200H
            Figure A-ll.   Concentration measurements for Run 3.
                                    238

-------
    source
SCflLE liSOOQ
         200M
     Figure A-12.    Concentration measurements from field experiment.
                                   239

-------
 butte and for Run 2 lays  directly  across the crest.  Run 3, with a lower
 release height,  shows a more localized region of maximum concentration
 occurring on the upwind face of  the butte.  This is in better agreement with
 the field data pattern.

      Scatter diagrams of  the concentrations for each run compared to the
 field values (Figures A-13 through A-15) are another means of evaluating the
 agreement of the laboratory and  field results.  Figure A-13 shows the
 wind-tunnel  concentrations for Run 1 to be much lower,  in general,  than the
 field measurements.  In this type of comparison, Runs 2 and 3 are nearly
 indistinguishable  (Figures A-14 and A-15).   However, there is much better
 agreement than for Run 1  in that most data points fall within an error band
 of  a factor  of 2.   The difference between Run 2 and Run 3 concentrations can
 be  seen in a scatter diagram comparing those two runs (Figure A-16).
 Concentrations,  at  the fixed receptors, from a stack 50 m in height are
 about 85% of those from a stack 42 m high.

      Turner  (1970) presented a solution to  the diffusion equation known as
 the Gaussian plume  formula.  The plume is assumed to have Gaussian
 distributions  in the vertical and lateral directions with standard
 deviations of az and ay, respectively.  The standard deviations  are
 functions of downwind distance and are obtained, according to stability
 classification,  from empirical relationships known as  the Pasquill-Gifford
 curves.

      According to Gaussian plume theory, with complete  reflection at  the
 ground, ground-level concentration is proportional to
 (l/a2ory)exp(-0.5(H/az)2).   If we assume that the plume
 geometry  does  not change (i.e., ay and oz remain the same)  and
 change only  the  stack height from H^ to H2,  the concentration at  the
 fixed point  should change  according to C2/C^ =  exp(-0.5(H2/az)2)/
 exp(-0.5(H^/az)2).  Assuming a value of az  of 48 m (that measured in
 the wind  tunnel over CCB), a change in the  stack height  from 42 m to  50  m
 can be  shown to  result in concentration decreases by 15%,  in general.
 This  is in good agreement  with the comparison in Figure  A-16,  despite the
 presence  of  the hill.

     Another evaluation of the applicability of Gaussian plume theory to
 the prediction of surface  concentrations is  shown in Figure  A-17.
 Ground-level concentrations along the plume  centerline  are  calculated for
 the field source height  of 42 m for downwind distances of 200  to
 2000 meters.  Concentrations from CCB wind-tunnel Run 3  are  plotted for
comparison.  The concentrations on the upwind hill  face  (x < 1000 m)
 follow the Gaussian plume  pattern for D stability.   On  the  lee side of  the
hill, the measured concentrations decrease more rapidly  with downwind
 distance  than  the Gaussian plume predictions and the concentrations
approach values predicted  for  C stability.
                                 240

-------
    1000
     100  -
o.
Q.
e
o
to
O)
o
c
o
u
O)
•-     10  -
c

o:
                                                  100
                                                                       1000
                            Field concentration, ppt
  Figure  A-13.
                Comparison  of concentrations from Run 1 with field data..

                In addition,  of the fifteen field sampler locations for

                which  zero  concentration was recorded, fourteen were

                below  4 ppt in the  wind tunnel.
                                    241

-------
   1000
                            10                  100

                              Field  concentration, ppt
                                                     1000
Figure A-14.
Comparison of concentrations from Run 2 with field data.
In addition, of the fifteen field sampler locations for
which zero concentration was recorded, thirteen were
below 7 ppt in the wind tunnel.
                                 242

-------
    1000
Q.
a.
•£   100  -
£

-------
     1000
 o.
 o.
 c
 o
 •r-
 4->
 CO
 
-------
     Q.
     O.
     c
     o
     s-
     c
     u
     o
     »—
     (U
     o
     3
    o:
                        200             500         1000
                            Distance  from  source,  m
                                                 2000
Figure A-17.
Ground-level concentrations along plume centerline for
Run 3 compared with Gaussian plume predictions for flat
terrain.  Source height is 42 m; curves are for stability
categories C and D.
                                  245'

-------
Elevated Plume Measurements

     The field experiments were limited to direct measurement of
concentration at ground-level only.   A lidar system was used, however,  to
make oil-smoke plume density measurements from which the vertical plume
spread, oz, can be derived.  For this first hour of Case 202, 0Z,
computed as an "hourly value," was found to be 8.2 m at a distance of
200 m from the source and 15.1 m at 520 m from the source.  These
positions are upstream of the hill where the plume is not likely to be
strongly influenced by the presence of the hill.  These data were not
available at the time of the wind-tunnel experiments and corresponding
laboratory measurements were not made.  However, a vertical profile and
elevated lateral profile of concentration in the wind-tunnel plume were
obtained (Figures A-18 and A-19) above the butte during Run 2.  The
vertical profile was based at FMF port number 9 (see Table A-3 for
location).  The lateral profile was obtained at this downwind distance, at
a height of 46 m above the local surface.

     Best-fit Gaussian plume profiles were found for the wind-tunnel data
and are shown in Figures A-18 and A-19.  Complete reflection was assumed
at the surface in calculating the best fit to the vertical profile data.
The dispersion coefficients derived from these best-fits are
o  = 67.3 m and az = 46.6 m.  This ay corresponds to a
Pasquill-Gifford stability category D.  The az is shown with the
available lidar measurements and compared with Pasquill-Gifford values  in
Figure A-20.  The wind-tunnel value falls between Categories C and D,
while the lidar values are between D and E.
                                   246

-------
  mo
  120 -
  100 -
      0      10     20    30     liO     50     60     70     .30     90    100

                           Run 2 Model concentration,  ppt
Figure A-18.   Vertical profile of  concentration.above, -port number.9 of
               Cinder Cone Butte model.in  Meteorological-Wind Tunnel.
                                   24.7

-------
      80
 03
 OJ
 u
 o
 o
O)
•O
O
       -200     -150    -100    -50      0       50      100      150      200


                         Lateral  distance from sampler 9, m
 Figure A-19.
Lateral profile of concentration above Cinder  Cone  Butte
model in Meteorological Wind Tunnel at a height  of  46  m
above the local surface at port number 9.
                                    248

-------
                                                ©  wind tunnel
                                                X  lidar
Figure A-20.    Vertical plume spread coefficients determined from lidar
               measurements in the field and concentration profile in the
               wind tunnel compared with Pasquill-Gifford values.
                                   249

-------
                          SUMMARY AND CONCLUSIONS
     A one-hour period of the CCB field study was modeled in the EPA
Meteorological Wind Tunnel.  Analysis of the available field data
indicated that the atmosphere was neutral for the period.  Three
combinations of wind direction and source height were used in the study.
Concentrations at 100 ground-level sampling positions were obtained for
comparison with available field values.  Also, elevated profiles of
concentration were obtained through the plume over the crest of the hill.

     The conclusions of this study are:
     1)
     2)
     3)
     4)
Modeling neutral flow over this three-dimensional hill in the
wind tunnel. Run 3, reproduced the measured, non-zero, field
concentrations to within a factor of 2 for 43% of the sample
locations; 71% were within a factor of 3.

Changing the approach flow wind direction only 4° resulted in a
significant shift in the concentration pattern on the hill
surface.

Changing only the stack height from 42 m to 50 m resulted in a
15% reduction of ground-level concentrations,  in general.   This
is in agreement with the Gaussian plume theory relationship for
the dependence of ground-level concentrations  on stack height.
Plume spread as determined by computing ay and az from
lateral and vertical tracer concentrations over the hill
                                                                   top
          compared well with Pasquill-Gifford values for flat terrain.

     5)   Concentrations measured on the upwind face of the hill  compared
          closely with Gaussian plume theory predictions.   However,
          concentrations measured on the lee surface of the hill  decreased
          with distance at a faster rate than Gaussian plume theory
          predicts.

The absolute effects of the presence of the hill on the plume and
ground-level concentrations cannot be resolved by this data base.
Additional measurements with the hill and reference measurements  in  the
absence of the hill must be made to evaluate the effects of the hill.  The
wind tunnel data presented here compares reasonably well with the field
data and offers some additional ground-level concentration measurements as
well as measurements within the elevated plume which can be used  to  extend
the field study data base.

                                   250

-------
                                REFERENCES
Lavery, T.F., Bass, A, Strimaitis, D.G., Venkatram, A., Green, B.R.,
     Drivas, P.J. and Egan, B.A., 1982:  EPA Complex Terrain Model
     Development Program; First Milestone Report - 1981, EPA-600/3-82-036,
     U.S. EPA,  Research Triangle Park, NC, 304p.

Pasquill, F., 1962:  Atmospheric Diffusion. D. Van Nostrand Company Ltd.,
     London, England,  297p.

Turner, D.B., 1970:  Workbook of Atmospheric Dispersion Estimates, Office
     of Air Programs,  Pub.  No.  AP-26,  U.S. EPA, Research Triangle Park,  NC,
                                  251

-------
               APPENDIX B
SUPPORTING METEOROLOGICAL MEASUREMENTS -
 ACOUSTIC SOUNDER AND SONIC ANEMOMETERS
                252

-------
   SUPPORTING METEOROLOGICAL MEASUREMENTS -
    ACOUSTIC SOUNDERS AND SONIC ANEMOMETERS
                EPA SHIS NO. 2
                   W.D.  Neff
          Wave Propagation Laboratory
      Environmental Research Laboratories
National Oceanic and Atmospheric Administration
                    253

-------
                              1.  INTRODUCTION
     In addition to the WPL Lidar, the Atmospheric Studies Program Area of
the WPL provided a number of supporting boundary layer measurements for
SHIS #2.  These included two monostatic acoustic sounders, one located just
upwind of the meteorological tower and one adjacent to the Hogback as  shown
in Figure B-l.  A monostatic/ bistatic-Doppler sounder provided wind
profiles and monostatic facsimile records 1.4 km upwind of the tower.   A
portable tethersonde system also provided wind and temperature profiles
first near the Hogback, and during a portion of the experiment,  near the
Doppler site.

     In addition, just prior to the SHIS #2 a preliminary version of a
portable sonic anemometer system was assembled to provide supporting
turbulence measurements.  Two three-axis sonic anemometers were mounted on
Tower A, one at 40 m, a second at 5 m.  Electronics for two fast-response
temperature sensors were also in place.

     In the sections that follow we will describe these measurement systems,
their operation during SHIS #2, current state of data reduction and some
preliminary analysis and interpretation.  In a closing section,  we will
outline some proposed improvements in these systems and their deployment in
future experiments.

Sonic Data

     SHIS #2 provided an opportunity to test a portable sonic anemometer
data system.  At the request of ERT, development of this system began  in
early September, 1982.  Software was developed on an LSI-11,  floppy-disc
based microprocessor system.  Raw data were recorded on a 9-track magnetic
tape.  Software processing in the field provided 20-min averaged data.
While not fully completed and checked prior to SHIS #2, this  system did
provide mean winds, variances, and covariances in a coordinate system
oriented with the sonic device.  Post processing of the raw data tapes  is
being accomplished by the staff of the Boulder Atmospheric Observatory
(BAO).  By adapting this data for input to standard BAD software,  a full
range of analysis products can be made available including turbulence
quantities in standard meteorological coordinates,  in coordinates  rotated
along the mean wind, as well as spectra and cospectra.  Addition of a
hard-disc based, multi-tasking operating system to the field  version of the
LSI-11 system will eventually allow such products to be available in field
settings.

     As operated in this experiment, the sonic anemometer interface
electronics provided both analog and digital outputs.   The fast-response
temperature sensor provides only analog data.  Two versions of the data

                                  254

-------
          u
         i
         a. ^n
                                             (D
                                                CD
                                                •H
                                             (/I  Oi
                                             H
                                             CD  ^
                                            •Xl  O
                                             S  Oj

                                             g-a
                                             w  o
                                             O  0
                                            •H
                                            PH
255

-------
 acquisition program were developed to allow processing of tnese various
 data.  The first version provided acquisition of 16-bit digital data
 directly from the sonic interface.  The second provided 12-bit  conversion  of
 the analog outputs from the sonic interface.   The limitations of a  12-bit
 A/D led to less resolution of both temperature and analog sonic data.
 Evaluation of spectra of temperature will provide some indication of the
 usefulness of these data.  One may expect, however,  that the data resolution
 should be better than that from conventional  instrumentation such as
 propeller anemometers.

      In actual operations, the sonic system performed  reliably  with two
 exceptions.   Following the power outage of 8  October 1982, a number of
 computer system components were destroyed in addition  to the ditigal sonic
 interface bus ICs.   With spare cards from a backup computer system,  the
 sonic system was operational again by 10 October.  Because the  replacement
 ICs for the  sonic digital interface were not  available  until 15 October,  the
 program was  put into a purely analog mode,  bypassing the digital  interface.
 This provided reliable sonic-derived turbulence data for experiments 4
 through 8.   Inspection of w-data in the field and  in the post-processing
 phase show that it  is  free of spurious  noise  except  for glitches introduced
 by the radio transmitter located in the equipment  trailer as shown  in
 Figure B-2.   In general,  these radio-transmission glitches appear only in
 the bottom level of sonic data.   Preliminary  inspection indicates that these
 do not contribute significantly  to  sigraa-w.

      The installation  of fast response  temperature sensors proved more
 difficult.  Early in the  experiment  a platinum wire  temperature sensor was
 installed with the  sonic  on the  lower boom.   This sensor survived the entire
 experiment.   However,  its  absolute calibration is off by a factor of two
 because of the mismatch between  the  differential output  of the sensor and
 the single-sided input  to  the A/D.  The upper sensor presented  additional
 difficulties.   Because  the boom  was  not  retractable, installation was
 accomplished  by  swinging  from a  rope supported at the top of  the tower.
 With the fragility of  the  platinum wire  sensors, repeated breakage occurred
 in this  process.  The decision was then made to record  data from the ERT
 thermistor at  40 m.  A  scaling factor of 8°C/volt combined with the  limited
 resolution of  the 12-bit A/D provided a resolution of 0.04°C, a factor  of 8
 less than that  for the  platinum  wire sensor.

     As  of 15 May 1983, preliminary hourly averaged values of wind and
 vertical velocity variance  had been provided to ERT.   In addition, one
 reformatted,  scaled data tape had been submitted.  Complete preprocessing is
 expected by the  end of May.  A number of BAD analysis routines  have  also
 been applied to  the SHIS #2 sonic data.   Some results from this  preliminary
 analysis are described  in Section 2 of this appendix.

Monostatic Data

     Two monostatic sounders with high-resolution facsimile machines
 (0.30 m/hr; 150 ra to 340 m vertical range displayed over 0.22 m)  were
 operated throughout the experiment.  One close to Tower A provided a
characterization of the boundary layer near the tracer/smoke release point.
 The  second, at the base of the Hogback,  provided a characterization  of the

                                   256

-------
bJ
5>-
UI
X
a.
a:
c£
cs

cc
H-
cr
                                                                                  e
                                                                                  ©
                                                                                     Vl
                                                                                  05
                                                                                     v>
                                                                                     {=>
                                                                                     o
                                                                                     t-H
                                                                                     a:
                                                                                     u
                                                                                  e
IS
v-l
\
                                                                                     ui
                                                                                     H:
                      \f>
                                         Gl
                                                           w


                                                            I
                                                                                                           bO
                                                                                                           fi
                                                                                                           •H

                                                                                                           o
                                                                                                           aJ
                                                                                                           •P
                                                                                                           nj
                                                                                                           •ri
                                                                                                           (D
                                                                                                           •H
                       I

                      I
                     •H
                      0
                                                                                                           TS  -P
                                                                                                           S -H
                                                                                                           aj
                                                                                                           o
                                                                                                           (D

                                                                                                           >
 o
•H
 in
 1/1
• H
 a
                      2  o
                                                                                                           en
                                                                                                           0) •
                                                                                                           fn
                                                  257

-------
 boundary layer flow as  it  approached  the ridge.  Prior to 19 October, the
 ridge-sounder was  located  directly opposite the Tower A access road.  It was
 then moved 180 m south  to  avoid  acoustic signal contamination from the noise
 of the  oil-fog generator.

      Facsimile recordings  from monostatic acoustic sounders show time-height
 cross-sections of  small-scale (0.20 m) temperature fluctuations as they move
 through an antenna beam of some  4° to 8° beamwidth.  Such small-scale
 thermal structure  is formed when turbulent motions mix parcels of different
 potential temperature.  Examples include regions within temperature inver-
 sions where the vector  wind shear is increased sufficiently to reduce the
 gradient Richardson number below a critical value, leading to the production
 of turbulence.  More commonly, the shear layer induced by surface friction
 combined with radiative cooling  of the surface usually produces a well-
 defined echo  layer  on acoustic sounder records.  An example,  showing both
 types of echo layers, appears in Figure B-3.  In this case,  a ground-based
 turbulent  layer extends into the overlying inversion to a depth of about
 40 m while elevated layers occur at heights extending up to 175 m.  In
 general,  the  inversion  is much more variable than shown in this example.
 Often,  the height  of the turbulent layers and the character of the turbu-
 lence change  considerably  over the course of an hour.  Further discussion
 and examples  of the interpretation of acoustic sounder data will be found in
 Section 2.

 Doppler Data

      The WPL  bistatic Doppler sounder was located 1.4 km upwind of the
 experiment site near a power substation.   This system,  providing vertical
 profiles of the wind from  50 m to 250 m,  was intended to characterize the
 approach flow and provide a general meteorology of the  area.   It operated in
 conjunction with a  standard, medium resolution facsimile recorder
 (0.123 m/hr; 820 m maximum range).  Figure  B-4 shows  a  block diagram for
 this  system including the  layout of the bistatic transmitters  and vertical
 receiver.  Conventional wisdom has  always suggested a bistatic  arrangement
 to be preferable because of the increased forward scatter from velocity
 fluctuations that adds to the received signal  strength  and because  the two
 components of  the wind can be computed in a common volume.  The commonly-
noted drawbacks of such a system include  the requirement  for more open space
 to locate  the antenna array and increased emission of noise into the
environment.  While such systems  have  increased signal-to-noise  ratios,
 particularly under adiabatic conditions,  the minimum  range is  somewhat
greater than for monostatic systems.   Also  as  we  shall  show below, with
 strong refraction (due to large wind  shear  or  temperature inversions  near
 the surface),  displacement  of the scattering volume and/or increased ground
 clutter can bias the Doppler shift  from which  the winds are computed.

     From the results of SHIS #2, it would'appear advisable to use  a
monostatic Doppler system for pure  temperature inversion flow  situations  and
where the scale of the flow is greater than the separation of the scattering
volumes  (that results from tilting  two monostatic sounders in orthogonal
directions).
                                  258

-------
Figure B-3.   Monostatic acoustic sounder facsimile record displaying a
              ground-based echo layer as well as elevated echo layers
              (10/29/83, 0200-0300 MDT).
                                  259

-------
     Echosounder
         and
     Doppler Board
Figure B-4.    Block diagram of the WPL  bistatic Doppler sounder  system.
                                   260

-------
Calibration and Data Quality Checks

     Doppler sounders are best checked against fixed tower systems.   In^the
field, basic program operation and parameter settings can be checked  using a
known frequency input.  Calibration against a tethersonde is more diffi-
cult.  Holding the balloon at a fixed level causes, .sidelobe reflections
which introduce a zero Doppler shift, usually at the same height as  the
comparison.  Using a single ascent (descent) of the  balloon introduces the
complications of comparing a 20-min time average with a snapshot wind
profile.  In addition, the motion of the balloon downwind or upwind can
introduce a bias of as much as 1 m/s under stronger  wind conditions.

     To eliminate some of these problems, we sought  comparison cases  where
the balloon ascended and descended in reasonable .proximity to a 20-min
Doppler averaging period.  An example of such a comparison is shown  in
Figures B-5a and B-5b.  In Figure B-5b, the two tethersonde profiles  show a
wind speed difference of about 1.5 m/s above 50 m.  The reason for this  is
that the tetherline has a speed of about 0.5 to 1.0  m/s.  At wind speeds of
more than a few m/s the balloon starts moving downwind.  As the angle to the
surface decreases, the line speed subtracts from the wind speed.  As  the
tethersonde descends, the effect is reversed.  The net difference then
between ascent and descent ranges from 1 to 2 m/s.

     During several cases examined, the presence of  a strong low-level jet
resulted in anomalous Doppler shifts.  Two possible  causes have been
identified.  Both are associated with refraction of  the bistatic acoustic
beam by the wind shear and/or temperature gradient.   This can, for example,
result in a displacement of the scattering volume off the vertical axis  of
the receiving antenna making the Doppler shift reflect the vertical
component of the wind more than the horizontal.  The second possibility  is
that  increased refraction near the ground, results in increased ground
clutter and hence a bias towards zero Doppler shift.  These cases were
evident only with a wind maximum below 40 m and appeared only to affect  the
range gates below 60 m.  Following final quality assurance on these  data,
they  will  be submitted to the CTMD data base.  These results do suggest  the
advisability of using monostatic Doppler geometries  for measuring winds  near
the  surface under shallow drainage conditions.
                                   261

-------
ACOUSTIC  DOPPLER DATA
 STARTING TIME
  10/20   6 0 MDT
                 MIND
INDEX  HEIGHT SPEED DIR
22. . .
21. . .
20. . .
19. . .
IB. . .
17. . .
16. . .
15. . .
14. . .
13. . .
12. . .
11. . .
1O. . .
9. . .
8. . .
7. . .
6. . .
5. . .
4. . .
3. . .
2. . .
1. . .
. . 46O
. . 44O
. . 42O
. . 4OO
. . 380
. . 360
. . 340
. . 32O
. . 30O
. . 28O
. . 26O
. . 24O
. . 220
. . 2OO
. . 180
. . 16O
. . 140
. . 12Q
. . 1OO
. . 80
. . 60
. . 40
0. 0
O. 0
0. 0
O. O
O. O
O. O
0. 0
O. O
O. O
2. 9
0. 0
2. 4
3. O
3. 6
4. 6
5. 7
4. 8
5. 8
6. 0
4. 7
2. 3
2. 8
51
32
41
62
334
51
83
49
87
87
88
86
88
95
93
91
88
81
79
1O2
1O3
9O
ID1
•*••
D
D
D
D
D
D
D
D
D
*
D
*
*
*
*
*
*
*
*
#
*
*
                              DIRECTION (D)

                           27O        O       090      180
                   +#*#*****+#*####*#+*#***#**+**#*****•*•
                                                         *
                                                         *
                                                         *
                                                         *
                                                                      *
                                                                      *
                                                                      *
            S
            S
                                              D

                                              D
                                              D
                                               D
                                               D
                                               D
                                              D
                                              D
                                             D
                                                D
                                                D
                                                                      *
                                                                      *
                                                                      *
                                                                      *
                                                                      *
                                                                      *
                                                                      *
                                                                      #
                                                                      *
                                                                      *
                                                                      *
                                                                      *
                                                                      *
    10/20
6: O
+*##****#*+*#*****#*+#**#*#**#+*#***+
O         5. O      1O. O       15. O   18. O
             WINDSPEED (D)
   Figure B-5a.
    Wind speed and direction profile measured by the  bistatic
    Doppler sounder system (10/20/82, 0600-0620 MDT).
                                  262

-------



















| 	 L-
o in
o w
-a \f)












c
a>
M
^
Q
2
in
in
in
O

1 	 4-
0 If!
O «
•a in
-v- o
1 sji












"*'-1^1>.*-)*j+-
*"l*4-..»-.
* **



_
r-T ,(-„-„., , 1 , , 1 1 , . 1 , „,„ 1 , , . . I , . t
1 1 hi 1 1 r
Kl

.

. o —
f-v O
OJ UI
£4
z
o

_ O 1-
03 CJ
i-l UI
^H
a
a
_ O M
O- 3




. o
o in o sn o in o
in IS C> C-i in fN
>*• ro PS r-j i-i

. ( w ) .1. H a i :;;i H





••



"•
.

"*
/\
>J S.
/ \
-^ \ :

-- -i — i i — i - - i- - i ;
1 i i i i \ \
— in
•r-t


•~ "4
O
lu
s.
- o-5
Ui
Ui
0.
- o en
C3
^-t
3
- M

O
o in o in o in o
in r-. o r-4 in rx
•T* tw*





"






*•





| 	 (. 	 1 	 j 	 1_ 	 ( 	 ^ 	 1 	 .) 	 ,
HJ



. o <-

r-j w
•z.
c
^-i
- O i-
CD tJ
^i Ui
u:
e
C;
* 0 w
l> 3




O
oinoinoinoino
o N in N o w ii"2 r-
•a iiT <• ro M CM »-»

(• w ) .1. |»j .9 ;i; ;«| |-|





-



-
1
O y*.
s / \
^ / M -
i / \
5 / \
0 / \ ..
5 'i
o \

1 ' ij. t ) ' i i i
! , ]-j— -f — j 	 1— 	 1 	 (• 	 \ 	 1
- ll"3
'""'


- PJ
0
U!
V
-0.5
iZi
UI
Ui
ft.
- «o to
w
z
^
- M

O
o in o u~ o in o 'in o
o r-r b~ ix o CM w is

T3 O
pj to
cti cd

CD CM
 — '


*
LO
1
CQ
CD
•H

< W  )    .1. H £) I -il H
( w
                                        263

-------
                           2.   PRELIMINARY ANALYSIS
      SHIS #2 provided a unique set  of  data  from which  to analyze winds and
 turbulence behavior in nocturnal,  complex terrain flows.   In  particular,
 sonic anemometer data provide an accurate measure of turbulence down to very
 small scales.  Monostatic sounders  provide  detailed flow visualizations that
 in general belie any expectation of a  steady,  simply-structured flow.  The
 Doppler sounder together with supporting  tower instrumentation provide an
 appraisal of the variability  in the flow  above the boundary layer.

 Interpretation of Monostatic  Data

      1.    Increasing shear above the surface-based inversion  leads to
 dynamical instability.   Such  instabilities are  easily  identifiable from
 monostatic records and provide a measure  of  the vertical extent of
 large-eddy structure in the inversion.  The growth of  such instability
 regions  seemed to be independent of the proximity to the Hogback.. In fact,
 similar  patterns were observed 1.4  km  upwind at  the Doppler site.  From the
 Doppler  wind data,  increases  in wind speed above the inversion of from 1 to
 2 m/s were observed  during these events.

      As  deduced from second-order closure modeling, turbulence depends
 inversely on the Richardson number.  Since Ri  in  turn depends on the inverse
 square of the vector shear, small changes in the  shear can translate into
 large increases  in turbulence.   From these data,  it appears that mesoscale
 processes above  the  inversion  can strongly affect turbulence within the
 inversion and boundary  layer and hence the dispersion of the plume prior to
 impact on the Hogback.

      2.    The  flow approaching  the  Hogback,  like most nocturnal inversion
 flows, is  characterized by intermittent turbulent layers and patches.   For a
 given turbulence measurement,  say using the 40-m  sonic anemometer,  the
 sounder  record can provide a good indication of the vertical extent over
 which the measurement  is valid  and  the variability of such regions in time.
The data suggest that linear interpolation schemes are not  in general
 reliable (again,  the  sounder record can indicate when such schemes might be
valid).   In addition, the sounder records  show considerable variability on
 time  scales  that are a  significant fraction of a typical averaging period  of
one hour.  In such cases, the sounder record may suggest a  more  appropriate
averaging  period, not necessarily beginning  on an even hour.

     3.   The sounder records, while showing many repetitive features  in the
approach  flow, also revealed many short-term phenomena  such as inversion
height changes due to an apparent local horizontal convergence or divergence
of the flow.  In other circumstances such  patterns have been associated with
10- to 20-rain period internal  waves.

                                  264

-------
 Examples of Data from the Morning of 15 October

      Figures B-6a and B-6b show two sequential half-hour segments  of
 acoustic sounder records between 0500 and 0600 MDT on 15 October.  Prior to
 0500 MDT, the boundary layer,  identified as the lowest echo  layer  on  the
 facsimile recording,  stayed below 50 m.   It would  oscillate  above  this level
 and would at times be perturbed by motions from above.   During  the early
 portion of the record shown in Figure B-6a,  the flow appeared almost  laminar
 at 40 m.  (The absence of an echo on the sounder indicates either  a laminar
 flow or one in which  the potential temperature gradient  is adiabatic.
 Adiabatic regions bounded by temperature inversions are  not  usually clear of
 echo:  in such cases  the decay time for  the  temperature  spectrum is greater
 than the time required to transport some temperature variance (w02) from
 the boundary).  Figure B-7 shows the corresponding time  series  from the
 sonic anemometer at 40 m obtained over a 1-min section of this  record.  In
 this case,  superimposed on a slow background oscillation are fluctuations on
 the order of 0.01 m/s.   During this one  hour period,  the flow aloft
 increased in speed.   This resulted in a  thickening shear layer.  Starting at
 a  height of 70 m it rapidly expanded downward  to meet the boundary layer
 below.   At the same time,  an increase in shear farther aloft expanded to
 fill a  region up to 125  m.   In this case Figure B-8  shows the effect on w
 (and therefore on sigraa-w).

      Evident in the sounder record are descending  echo bands that are most
 noticeable  between the  surface and 80 m.  These have, in previous research
 been identified with  Kelvin-Helmholz instability.   The question then arises
 of the  contributions  of  this large-eddy  structure to sigma-w.  This can be
 estimated from the log-log  plot of frequency versus  the product of frequency
 and power spectrum shown in Figure B-9.  In this case, the product is  almost
 an order of magnitude above  the background spectrum, suggesting a
 significant  contribution, but not  an exclusive one.  (A number of such cases
 will be analyzed with separate  integrations over the wave and turbulence
 portion of  the  spectra).  Almost evident in these records is  the presence  of
 an inertial  subrange.   Within  the  surface layer this tends to occur at
 frequencies  greater than 1 Hz.  In most cases analyzed thus  far, the
 spectrum at  periods greater than  10  s appear pink (freq.  times the spectrum
 is  flat).

     A  feature  noted  during the field exercise was  the occurrence of
 positive heat fluxes  in the 20-min averaged covariances.   The cospectral
 analysis  shown  in Figure B-10 shows that these apparent positive fluxes
 result  from  low frequency motions.  Generally,  we have found  that the  flux
 is  negative  for  frequencies above 0.01 Hz.  In the  particular case  shown the
 flux for frequencies less than 0.01 Hz is positive.   In part  this may  be due
 to  a shallow, fluctuating downslope flow of cold air in the lowest  5 m.
With a  local terrain slope of 0.03 (based on the slope along  the road),' a
 1 m/s drainage flow would produce a 0.03  m/s w.  This negative w would then
produce a positive correlation with the decrease in temperature.

Continuing Analysis

     With acoustic sounder data (monostatic and Doppler),  sonic  anemometer
turbulence information,  and tethersonde data, analysis  such as that
described above will continue.   It is expected  that the analysis of the
primary modeling cases will occur first followed by other cases of interest.

                                    265

-------
50
  0530 MDT
                               Time
                                                                 0500 MDT
    Figure B-6a.   Acoustic sounder record (10/15/82, 0500-0530 MDT)
                                  266

-------
 0600 MDT
                              Time
0530 MDT
Figure B-6b.   Acoustic sounder record f(10/15/82,•0530-0600 MDT)
                              267 -

-------
CM


_J
LU

LU
_I

X
a.
or

CD

cc

a
o
LU
CD
1-4
\
                                       CD
                                                                               U)
                                                                                  
                                                                                  UJ
                                                                              =p
                                                                              C«J
                                                                              G>
                                                         LO
                                                          •

                                                         CD
                                                                          CO
                                                                                                       0)
                                                                                                       e ^-x
                                                                                                       I  H
                                                                                                      O Q
                                                                                                      rt- S

                                                                                                       CD LO
                                                                                                      rC O
                                                                                                      +J LO
                                                                                                         O

                                                                                                       O   •>
                                                                                                       H CM
                                                                                                      <4-l 00

                                                                                                       rt LO
                                                                                                      U) > — '
                                                                                                      o
                                                                                                      •H T3
                                                                                                      f-l  O
                                                                                                      CD .H
                                                                                                      U5  f-l
                                                                                                       1   CD

                                                                                                      i  ^
                                                                                                      H  CD
                                                                                                      4-> -P
                                                                                                          3
                                                                                                      CD -H
                                                                                                      CD  S
                                                                                                      o I  i
                                                                                                      CD
                                                                                                      f-l
                                               268

-------
u
•X
ct:
cc
I—
 t-
                                                                                                             00
                                                                                                               I
                                                                                                             CO
                                                                                                              bo
                                                                                                             •H
                                                    2.69

-------
         821615.  5.48     LEVEL 2
                                       28  MIN
                                              5y n~   (inertial Subrange)'
-4.ee
     -4
00
-3.00
-2.00       -1.00
      LOG  N
0 .00
1 00
 Figure B-9.   Log-log plot of frequency versus  the  product of frequency
               and power spectrum (10/15/82,  0540-0600 MDT).   \
                                  270

-------
           821015.  5.48
LEVEL  1
120 MIN
  -7.15
       -4.
                                         1 88
Figure B-10.   Cospectral analysis of temperature and vertical wind speed
               fluctuations measured at 5 m showing that positive heat fluxes
               result from low frequency motions (10/15/82, 0540-0740 MDT).
                                     271
                                                          *USGPO: 1983 — 759-102/0789

-------

-------