United States
Environmental Protection
Agency
Atmospheric Sciences Research
Laboratory
Research Triangle Park NC 27711
EPA/600/3-84/110
February 1985
Research and Development
EPA Complex
Terrain Model
Development
Fourth Milestone
Report - 1984
-------
EPA/600/3-8V110
February 1985
EPA COMPLEX TERRAIN MODEL DEVELOPMENT
Fourth Milestone Report - 1984
by
David G. Strimaitis
Thomas F. Lavery
Akula Venkatram
Donald C. DiCristofaro
Benjamin R. Greene
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
Atmospheric Sciences Research Laboratory
Research Triangle Park, North Carolina 22771
ATMOSPHERIC SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 22771
-------
NOTICE
The information in this document has been funded
wholly by the United States Environmental Protection
Agency under Contract No. 68-02-3421 to Environ-
mental Research and Technology, Inc. It has been
subject to the Agency's peer and administrative
review, and it has been approved for publication as
an EPA document.
Mention of trade names or cownercial products does
not constitute endorsement or recommendation for use.
11
-------
FOREWORD
The Atmospheric Sciences Research Laboratory (ASRL) conducts
intramural and extramural research programs in the physical sciences
to detect, define, and quantify air pollution and its effects on
urban, regional, 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 relationships between emissions of pollutants
for all types of 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 major
field studies of this six-year program were conducted during 1980 at
Cinder Cone Butte near Boise, Idaho, during 1982 at Hogback Ridge near
Farmington, New Mexico, and during 1983-8A at the Tracy Power Plant
near Reno, Nevada. Data from these field studies along with
measurements of fluid modeling simulations performed in the EPA Fluid
Modeling Facility are being used to quantify the effects of terrain
obstacles on stable plume dispersion. A series of annual milestone
reports has been issued to describe the development of the Complex
Terrain Dispersion Model (CTDM) and to contrast the performance
evaluation of the CTDM against existing complex terrain dispersion
models. This Fourth Milestone Report describes the continuing
development of the CTDM and evaluates the improved model's performance
using measurement data from the field studies.
A. H. Ellison
Director
Atmospheric Sciences Research Laboratory
iii
-------
ABSTRACT
The Complex Terrain Model Development (CTMD) program is being
sponsored by the U.S. Environmental Protection Agency to develop,
evaluate and refine practical models for calculating ground-level air
pollutant concentrations in mountainous terrain. The emphasis of the
program is to develop models with known accuracy and limitations for
simulating 1-hour concentrations in high terrain during stable
conditions. At the time of preparation of this Fourth Milestone
Report, two small hill impaction studies and a feasibility study at a
full scale site, the Tracy Power Plant near Reno, Nevada have been
conducted. This report describes the continuing modeling and analysis
of the Cinder Cone Butte (CCB) and Hogback Ridge (HBR) data bases,
results from the feasibility study, and plans for the Full Scale Plume
Study (FSPS).
Substantial progress has been made in the development of the
Complex Terrain Dispersion Model (CTDM) as a method for simulating
tracer gas concentrations at CCB and as a practical regulatory model.
CTDM includes explicit mathematical expressions that account for the
important phenomena that control dispersion in mountainous terrain.
The model also includes a method to use turbulence intensity
measurements to estimate the vertical growth of a plume with downwind
distance. The current version of CTDM was evaluated by comparing
model calculations to (1) observed SFg concentrations, (2)
concentration estimates based on a flat terrain model, and (3)
concentration estimates based on the COMPLEX I/II plume path
assumptions. The performance statistics show CTDM does better than
the other two approaches in simulating concentrations observed at CCB.
To help understand the phenomena that control dispersion at the
ridge site a simple empirical modeling approach was taken. A model
was constructed by modifying the effective plume height as a function
of Hc. Model simulations were performed using a subset of the
SHIS #2 CF3Br data base. These empirical model calculations were
compared to observed CF3&r concentrations and to calculations made
with a flat terrain model and a model based on the half-height plume
path assumption. The empirical model performed better than the other
two models. Furthermore, for both CCB and HBR the flat terrain model
underestimated concentrations while the COMPLEX (or half-height) model
overestimated concentrations by roughly a factor of two.
The FSPS was conducted at the Tracy Power Plant (TPP) near Reno,
Nevada in August 1984*. The Tracy station was selected based on a
*The FSPS will be discussed in the Fifth Milestone Report (June 1985)
IV
-------
feasibility study conducted at the site in November 1983. The
feasibility experiment resulted in a data base that is useful for
modeling and for showing the important differences between a full
scale site and the two small hill sites. This Fourth Milestone Report
presents a discussion of an initial analysis of the November Tracy
data base and the plans for the FSPS.
This milestone report also includes two appendices prepared by
scientists at the EPA Fluid Modeling Facility. Appendix A contains a
report describing the results of a towing tank study designed to
evaluate the validity of modeling a plume in the flow above Hc as if
both the plume height and the terrain height were reduced by Hc.
Appendix B contains a report describing a wind tunnel study designed
to identify source-terrain relationships for which the effect of an
idealized model of CCB on ground-level concentrations is greatest.
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, 1983 to June 1, 1984.
-------
Foreword i i i
Abut fact iv
Figures viii.
Tab 1 oi3 xv
Symbols and Abbreviations xvii
Acknowledgements xxi i i
1. introduction "I
2. The Complex Terrain Dispersion Model 4
?.] The LIFT Component 5
')..') The WRAP Component 25
2.3 LIFT/WKAP Transition .32
2. i\ Formulation of <37 36
2. '> formulation of Oy 44
2.6 Other Revisions 46
3. Modeling COB 48
,3.1 Data Analysis and Interpretation- the Modeler's
Data Archive 48
3.2 <;ias;-i f irat ion of CCB Kxperiment Hours 62
3.3 Model Performance 63
4. Model ing Hogback Ridge //
4. "I Refinement of Tower Data //
4.2 SHIS it 2 Preliminary Modeler's Data Archive .... 93
4.3 Modeling HBK 99
4.4 Selected Case Study Results 120
5. The Preliminary Tracy Kxperiment 1V3
.'j. 1 Geographic and Meteorological Setting I'j3
5.? Kxperimental Design 1^7
5.3 Preliminary Field Study Results 164
'3.4 Kxample Results from Specific Kxperiments .... 167
5.'J Summary of the Preliminary Kxperiment 213
5.6 Plans for the Full Scale Plume Study 219
6. Summary, Conclusions, and Recommendations for Further
Study 224
References 230
Appendix A: Stable Plume Dispersion Over an
Isolated Hi 1 I 2.33
Appendix B: Dispersion From a Source Upwind of a
Three Dimensional Hill of Moderate Slope. . 269
Appendix C: ALPHA 'i Observations of the Tracy
Oi 1 Fog Plume 287
v i i
-------
F1GUKKS
Number
1 Idealised stratified flow about hills indicating
domain of individual CTDM component algorithms ... 6
2a Illustration of the relationship between the
crosswind average concentration profiles at s
and s0, and the plume from one of many
point- source elements representing the flux of
material across the plane at so ........... 9
?.h Illustration of simulating terrain effects by shifting
the point source elements at so and altering the
diffusivity and flow speed beyond so ........ 10
3a Illustration of the displacement and distortion of
the plume due to the influence of terrain as modeled
in LIFT ...................... 11
3b Illustration of terrain effect on the vertical
distribution of plume materials above Hc as modeled
in LIFT ...................... 12
4 Schematic illustration of terrain effect on the
vertical distribution of plume material when
Hc - 0 ....................... It
5 Depiction of how the terrain factors depend on the
height of the release and the lateral offset of
the trajectory from the crest of the hill ..... 21
Dependence of separation in the lee of a hill on
stratification ................... 23
Definition sketch for deriving the effective
receptor location ................. 26
8 Top view of plume in two-dimensional flow around
a hill ....................... 28
9 Diagram illustrating the geometry for specifying
the dependence of WRAP modeling variables on the
wind direction ................... 31
10 Cross section of plume at so illustrating how
LIFT and WKAP concentration estimates differ at
receptors on either side of Hc ........... 34
VI 1 1
-------
F1GUKES (Continued)
Number-
II Variation of ax/owt with t/TL 41
12 Scatlerplot of
-------
K1GUKKS (ConlInuod)
Number Page
28 Schematic illustrating a Gaussian model that
account;; for wind speed shear 103
29 Variation of az with distance for a typical
case 105
30 Geometry for modeling Hogback Ridge 107
31 Variation of the height of the maximum observed
concentration relative to that estimated by the
flat terrain model with the ratio of the release
height over the dividing streamline height 108
32 Variation of residual with release height relative
to Hc for flat terrain model over the subset of
35 hours 110
33 Variation of residual with release height for
model with half height correction (subset of
35 hours) 112
34 Variation of residual with Cp for flat terrain
model (subset of 35 hours) 113
35 Variation of residual with Cp for flat terrain
with "half-height" correction (subset of 35 hours) . 114
36 Variation of residual with release height relative
to Hc for the empirical HBR model (subset of 35
hours 117
37 Variation of residual with Cp for the empirical
HBR model (subset of 35 hours) 118
38 Time series of 5 min calculated dividing
streamline heights and bulk hill Froude numbers
above Hc (Experiment 11, 10/23/82, 0600 0700 MDT) . 122
39 Time series of 5 min sonic data from Tower A
(Experiment 11, 10/23/82, 0600 0700 MDT) 123
40 Vertical profiles of hourly meteorological data
from Tower A (Experiment 11, 10/23/82, 0600 0700 MUT) 124
41 One hour average observed Cfr^Br concentrations
scaled by emission rate (ys/m^) (Experiment 11,
10/23/82, 0600 0700 MUT) 126
42 One hour average predicted scaled concentrations (vis/m^)
from the HBR (flat) model (Experiment 11, 10/23/82,
0600 0700 MUT) 127
-------
FIGURKS (Continued)
Number
43 One hour average predicted scaled concentrations (p
from the HBK (terrain) model (Experiment 11, 10/23/82,
0600 0/00 MDT) .................. 128
44 Time scrips of 5- min calculated dividing streaml ine
heights and bulk hill Froudc numbers above Hc
(Kxporiment 14, 10/26/82, 0300 0400 MDT) ...... 130
4S Time series of '5- min sonic data from Tower A
(Kxperiment 14, 10/26/82, 0300-0400 MDT) ...... 131
46 Vortical profiles of hourly meteorological data
from Tower A (Kxperiment 14, 10/26/82, 0300 0400 MUT) 132
47 One hour average observed Cb^Br concentrations
scaled by emission rate (jjs/m-*) (Experiment 14,
10/26/82, 0300 0400 MDT) .............. 134
48 One hour average predicted scaled concentrations
from the HBR (flat) model (Kxperiment 14, 10/26/82,
0300 0400 MDT) ................... 13'5
49 One hour average predicted scaled concentrations
(ys/m^) from the HBR (terrain) model (Kxperiment 14,
10/26/82, 0300-0400 MDT) .............. 136
'jO Time series of 5- min calculated dividing stream! ino
heights and bulk hill Froude numbers above Hc
(Kxperiment 6, 10/31/82, 0700 0800 MDT) ...... "137
SI Time series of 5- min sonic data from Tower A
(Kxperiment 6, 10/13/82, 0700 0800 MDT) ...... 139
'32 Vertical profiles of hourly meteorological data
from Tower A (Kxperiment 6, 10/13/82, 0700 0800 MDT) 140
.'j3 One hour average observed Cfr^Br concentrations
scaled by emission rate (ps/m^) {Kxperiment 6,
10/13/82, 0700-0800 MDT) .............. 141
'j4 One hour average predicted scaled concentrations
from the HBK (flat) model (Kxperiment 6, 10/13/82, 0700
0800 MDT) ..................... 143
One hour average predicted scaled concentrations
from the HBK (terrain) model (Kxperiment 6, 10/13/82,
0700 0800 MDT) .................. 144
Time scries of 5-min calculated dividing streamline
heights and bulk hill Froude numbers above Hc
(Kxperiment 8, 10/15/82, 0500 0600 MDT) ...... I4'5
-------
FIGURES (Continued)
Number
5 7 Time sori.cs of 5 min sonic data from Tower A.
(Experiment 8, 10/15/82, 0500 0600 MDT) ...... 146
58 Vertical profiles of hourly meteorological data
from Tower A (Experiment 8, 10/15/82, 0500 0600 MDT 148
59 One hour average observed Cb'gBr concentrations
scaled by emission rate ys/m^ (Experiment 8,
10/15/82, 0500 0600 MUT) ...............
60 One hour average predicted scaled concentrations
(yg/m^) from the HBR (flat) model (Experiment 8,
10/15/82, 0500-0600 MUT) ............... 151
61 One hour average predicted scaled concentrations
(jjg/iiv^) from the HBR (terrain) model (Experiment 8,
10/15/82, 0500 0600 MDT) ............ , . 152
62 The region around the Tracy Power Plant ...... 154
63 91.4 m (300- ft) stack, located near the southwest
corner of the Tracy Power Plant .......... 155
64 Location of the Tracy Power Plant ......... 156
65 Oil Fog and SFg injections were made through
a "door" into the ducting leading to the 300 ft
stack ....................... 160
66 Sampler locations ................. 162
67 Preliminary (November 1983) Tracy Experiment Jayout 165
68 Hourly SFg concentrations, November 12, 1983,
0000 0500 ..................... 171
69 Five minute exposure (Camera 1), November 12, 1983
at 0000 ...................... 1/6
70 Five minute exposure (Camera 1), November 12, 1983
at 0015 ...................... 177
71 Five minute exposure (Camera 1), November 12, 1983
at 0030 ...................... 1/8
72 Five minute exposure (Camera 3), November 12, 1983
at 0030 ...................... 1/9
73 Wind direction measured on 150- ra tower ....... 180
XI 1
-------
KLCUKK'S (Continued)
Number Page
/4 Wind speeds measured on 150 m Lower 181
/'j Meteorological data 150 in level, November 12,
0000 0500 182
76 Duppler sounder wind profiles 183
II Five minute exposure (Camera 1), November 12, 1983
at 0130 18/i
78 Geographic distribution of winds at approximate
plume height 185
79 Kivc minute exposure (Camera 3), November 12, 1983
at 0315 192
80 ALPHA 1 observations of Tracy plume November 12, 1983 193
81 Eagle Picher acoustic sounder records 194
82 Hourly SK^ concentrations, November 15, 1983
0300 0800 195
83 Geographic distribution of 100 m winds,
November 15, 1983, 0300 0800 198
84 Geographic distribution of 150--in winds,
November 15, 1983, 0300 0800 203
85 Meteorological data 150-m level, November 15, 1983 . 208
86 Kive minute exposure (Camera 1), November 15, 1983
at 0315 210
87 Kive minute exposure (Camera 1), November 15, 1983
at 0330 21 1
88 Kive minute exposure (Camera 3), November 15, 1983
at 0330 212
89 Kive minute exposure (Camera 1), November 15, 1983
at 0430 214
90 Photograph (Camera 3), November 15, 1983 at 0615 . . 215
91 Hourly SF6 Concentrations, November 18, 1983 .... 2
xi i i
-------
FLGUKKS (Continued)
Number
92 Five minute exposure (Camera 3), November 18, 1983
at 0030 ...................... 217
93 Geographic distribution of 150 m winds,
November 18, 1983, 0000 0100 ............
KSPS site layout .................. ?22
xiv
-------
TABhKS
Number Page
1 Periods of significant discrepancy between wind
direction estimates made on the basis of Tower A
data and plume trajectories inferred from
photographs .................... b5
7. Periods of significant discrepancy between iz
estimates made on the basis of Tower A data and
iz values inferred from photographs ........ 61
3 Sf'g tracer hours at CCB with Hc < 2 m ....... 64
4 St'g tracer hours at CCB with Hc > zr +• 10 m .... 65
'5 Sb'f, tracer hours at CCB with Hc -- zr + 10 m .... 66
6 Sfr'fc tracer hours at CCB with 2 m < Hc < zr- 10 m . . 67
/ Model performance statistics for 80 hours of St'g
data at CCB .................... /O
8 Kffect of meteorological data resolution on CTDM
model performance for 80 hours of Sfr'g data at CCB . 76
9 Allowable second to second sensor changes used to
filter raw data in processing of preliminary data
base ........................ 9/i
10 Summary of KMF UVW propeller calibrations ..... 95
11 Recommended correction factors for non cosine
response of Climatronics UVW propellers ...... 96
12 Ch'3Br data base summary statistics ......... TOO
"13 Hogback Ridge modeling data ............ 109
l/i Relative performance of models for a subset of 35
hours ....................... 119
"I r> Meteorological instruments at the Preliminary Tracy
Kxpcriment ..................... 166
16 Summary of acquisition of Sb'g concentration data . . 168
xv
-------
TABLKS (Continued)
Number
I/ Data base, preliminary Tracy experiment 169
18 Characterization of the preliminary Tracy
experiments 1/0
19 Five minute meteorological data, November 1.2, 1983
0000 0300 VST 190
?0 KSl'S schedule 220
xvi
-------
1,1ST Ob1 SYMBOLS AND ABBREVIATIONS
SYMBOL
a Ratio of a to u
w *
a(x.) Radius of an ideal hill circular in horizontal
cross section
ot Relaxation scale factor for decrease in terrain effect
away from hi 11
B Boundary layer profile parameter
C Concentration
C Mean concentration from many "filament" plumes
ra
C Observed concentration
o
C Modeled concentration
X Concentration
D Diffusivity
D Mean diffusivity over the interval s s
m o
6 Scale factor for LIFT-WRAP transition zone
Ah Plume height change due to buoyancy of emission
e Krror or residual
V Buoyancy flux
Fr Froude number based on hill height
Fr Froude number based on hill length
LJ
F ,F Correction factors for non cosine response of UVW
u v
propellers
g Acceleration due to gravity
Y Scale factor for the "stable" mixing length
f Scale factor for the "neutral" mixing length
h Terrain elevation above a reference elevation
h Peak height (GC output signal)
II Hill height (crest)
H Critical dividing streamline height
1C Kddy diffusivity for heat
K Eddy diffusivity for momentum
0^ Direction from effective receptor position to source
0 Wind direction
xvi i
-------
0 Potential temperature
0^ Direction of the stagnation streamline
0 Mean wind direction of approach flow
in t
i ,i ,i Turbulence intensities alongwind, crosswind, and
x y z
vertical
k Von Karman constant
I, Mon in Obukhov length
9. Mixing length
8. Neutral mixing length
9. Stable mixing length
s
\ Wavelength of disturbances generated by the hill
-n Arithmetic mean
a
m Geometric mean
K
N Number
N Brunt Vaisala frequency
N Bulk Brunt Vaisala frequency for a layer
p Power for wind speed profile power law
l'(0) Wind direction probability distribution function
t
P(H^) H probability distribution function
P Pressure
4> Wondimensional potential temperature gradient
Q Source strength (mass flow rate)
Q Kinematic heat flux
p Density
r Distance from source to hill center
R* Universal gas constant
s Distance from source to receptor along wind direction
(s,8,,h) General point in a cartisian coordinate system with
s-axis aligned with the mean flow, where s is the
distance from the source
s Arithmetic standard deviation
OL
s^ Distance from the source to the base of the hill
b
s Geometric standard deviation
g
s Mean of s and s
TO O
XVI 1 1
-------
;; Distance along wind direction from source to terrain at
i he elevation of H for the LIFT domain; distance to
c
terrain at the elevation of the receptor for the WKAP
domai n
s _ Distance from source to receptor
o Standard deviation
a Standard deviation of H
lie c
a Standard deviation of alongwind velocity fluctuations
about the mean wind
a Standard deviation of crosswind velocity fluctuations
v
about the mean wind
o a value obtained for a sampling duration T
VT v
a cj value obtained for an infinite sampling duration
v«i V
o Standard deviation of vertical velocity fluctuations
w
a .a Crosswind horizontal and vertical standard deviation of
y z
tracer concentrations
a *.a * a and a for a plume from point source
y z y z
element for the interval s- s
o
a , cj Value of a and a at s - s
yo zo y z o
a ,0 Effective a and a after accounting
ye ze y z
for all terrain effects up to the location of the model
receptor
o Standard deviation of the horizontal meander component
ym
of the wind fluctuations
o _, Total a including terrain effects (a ) on the
yT y ye
"filament" plume and meander
a Total o including terrain effects (0 ) on the
"filament" plume and H variations
a Plume spread above plume center]ine height
a Plume spread below plume center!ine height
T, T Temperature '
t Time
T. Terrain factor for streamline distortion in the
h
vertical direction
T. Terrain factor for streamline distortion in the lateral
direction
XIX
-------
u
'•*
T ,T
ay az
T ,T
Y z
T. ,T.
iy i z
T
T
11
T,
LT
u
r
U
"*
T
U
r
(x.y.z)
x
sep
Torr'ain factor for plume transport speod
Scaling temperature
Terrain factors for diffusivity
Ratio of streamline distortion factor to diffusivity
factor
Kulerian time scale
Terrain factors for turbulence intensity
Lagrangian time scale
Lagrangian time scale of the transverse correlogrrim
Time scale of molecular mixing
Partial height factor (terrain factor)
Time of travel below plume centerline height
Time of travel above plume centerline height
Sampling duration
Generic wind speed
Friction velocity
Wind tunnel transport wind speed
Wind speed at release height
General cartisian coordinates
Distance of flow separation point from hill creut
Tracer release coordinates
Receptor coordinates
Plume release height
Plume centerline height after accounting for wind shear
ABBRKVIATLONS
ARLKKU
ASRI.
ATDI,
inn
G
CF3Br
CCB
CSU
CTDM
Air Resources Laboratory Field Research Division
Atmospheric Sciences Research Laboratory
Atmospheric Turbulence and Diffusion Laboratory
Cent imeters
Centigrade
Freon
Cinder Cone Butte
Colorado State University
Complex Terrain Dispersion Model
xx
-------
CTMD
DKC;
KPA
KPK!
KRT
FMF
ft
FSPS
GC
GLC
HBK
km
Ib
LASL
LMF
MCO
MCP
MDA
MDT
MRi
MSI.
MVJ
3
NAWC
NOAA
•;
ns/m"
ppb
ppt
PDF
PNM
PPC
PST
RTD
Complex Toi-faui Model Development
Digital Equipment Corporation
U.S. KnvironmenLal Protection Agency
Klectric Power Research Institute
Knvironmental Research & Technology, Inc.
fluid Modeling Facility
Feet
Full Scale Plume Study
Gas chroiuatograph
Ground level concentration
Hogback Ridge
Hertz
Kilometers
Pounds
Los Alamos Scientific Laboratory
Linear mass flow meter
Meters
Maximum observed concentration
Maximum predicted concentration
Modelers' Data Archive
Mountain Daylight Time
Meteorology Research, Inc.
Mean Sea Level
Megawatts
Micro seconds per cubic meter
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
Public Service Company of New Mexico
Plume path coefficient
Pacific Standard Time
Resistance Thermometric Device
xxi
-------
K'l'l Research Triangle Institute
s Seconds
Si?' Sulfur hexafluoridc
6
SHIS Small Hill Impact.Ion Study
TP1' Tracy Power Plant
TRC TKC Knvironmental Consultants, Inc
WF'L Wave Propagation Laboratory
WSSI WesiLer'n Scieni. Ific Services, Inc.
xxi i
-------
ACKNOWLKDGKMKNTS
Many people have contributed their talents and energies to
-Maintain continuing progress in the KPA CTMU program. In particular,
we gratefully acknowledge the efforts of the following individuals:
• Bill Snyder and his colleagues at the KPA Fluid Modeling
facility who have performed many towing tank and wind tunnel
simulations in support of the modeling;
• Wynn Kberhard and his colleagues at the Wave Propagation
Laboratory who have supplied the lidar data and many
meteorological measurements;
• Julian Hunt and Rex Britter of Cambridge University for
their ideas and recommendations on modeling stable
conditions in complex terrain and with whom we have had many
useful discussions;
• Steve Andersen and Chris Johnson of KRT in fort Collins who
operated the 150 m meteorological tower and were responsible
for all CTMD logistics at the Tracy Power Plant; and
• Kay Dickson and his staff at NOAA ARLKRU who were
responsible for the flow visualization and tracer
experiments.
We would like to acknowledge the Klectric Power Research
Institute (KPR1) who co sponsored the November 1983 experiment at
Tracy, and the EPR1 contractors (Rockwell International, SRI
"International, TRC Environmental Consultants, and the Research
Triangle Institute). Special thanks go to Mark Cher of Rockwell who
provided the tracer gas data discussed in Section 5.
Finally, we give special thanks to Frank Schiermeier and George
Holzworth, whose enthusiasm, support and encouragement have really
helped us progress.
xxi 1 t
-------
SKCTION 1
INTRODUCTION
The U.S. Environmental Protection Agency (EPA) Complex Terrain
Model Development (CT'MD) project is a muiti year study to develop
improved models for calculating ground level air pollutant
concentrations that result from large emission sources located in
mountainous terrain. At this time, the focus of the project is to
develop models for simulating 1-hour average concentrations during
stable atmospheric conditions.
These models are to be used in a wide variety of applications,
such as the siting of new energy- development facilities and other
sources of air pollution, regulatory decision making, and
environmental planning. Therefore, the models should be easy to
understand, easy to use, and of known accuracy and limitations. The
CTMD project will recommend the types and extent of meteorological
measurements needed to derive input to the models.
The objectives of the program were described by Holzworth (1980)
and generally follow the recommendations of the participants of the
KPA sponsored workshop to consider the issues and problems of
simulating air pollutant dispersion in complex terrain (Hovind et al.
19/9). The program was subsequently designed with a perspective
toward modeling and includes model development efforts based on
physical modeling, field experiments and theoretical work. The
project started with field experiments and physical modeling for
isolated simple terrain features and has progressed to the Full Scale
Plume Study (FSPS), which was conducted at the Tracy Power Plant no a,
Reno, Nevada in August 1984.
The program was begun in June 1980. The first major componcn
was the Small Hill Impact ion Study //I (SHIS #1) , a field experiment*
conducted during the fall of 1980 at Cinder Cone Butte (CCB), Idaho.
The prime contractor for the CTMD is Environmental Research &
Technology, Inc. (ERT). The principal subcontractors for the CCB
experiments were Western Scientific Services, Inc. (WSS1),
responsible for fixed meteorological 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) provided laboratory fluid modeling
support; the NOAA Wave Propagation Laboratory (WPL) supported the
field program with a manned lidar system; and TRC Environmental
Consultants, Inc. provided independent data audits.
-------
CCB is a roughly axisymmetric, isolated 100 meter tall hill Located in
the broad Snake Kiver Basin near Boise, Idaho. The field program
consisted of ten flow visualisation (oil fog) experiments and 18
tnulti hour tracer gas experiments with supporting meteorological,
1idar and photographic measurements.
The second Small Hill Tmpaction Study (SHIS #2)** was conducted
during October 1982 at the Hogback Ridge (HBR) near Farmington, New
Mexico. SHIS #2 included tracer and flow visualization experiments
with concurrent meteorological, lidar and photographic measurements.
The experiment produced approximately 179 tracer hours and extended
the modeling data base to include flow and dispersion around a
two dimensional ridge.
During the course of the CTMD project, three Milestone Reports
(Lavery et al. 1982; Strimaitls et al. 1983; and Lavery et al. 1983)
have been published. These reports, which are available from EPA,
describe the progress in developing and evaluating complex terrain
models using the CCB and HBR data bases. They also describe in detail
the two Small Hill studies and a series of towing tank and wind tunnel
studies performed at the EPA Fluid Modeling Facility (FMF) in support
of the model ing.
This Fourth Milestone Report documents work accomplished from
June 1983 through May 1984. It describes the further development of
the Complex Terrain Dispersion Model (CTDM). In particular, it
provides a detailed mathematical description of CTDM and shows how the
important terrain effects on plume dispersion are specifically handled
in the model. This latest version of CTDM has been tested using the
CCB data base. An 80 hour subset of the data base has been
partitioned into four classes: neutral, weakly stratified,
impingement and very stable. CTDM calculations were compared to
observed SF^ concentrations and to concentrations calculated with
(1) a flat terrain model and (2) a model that uses the COMPLEX I/I I
plume path assumptions. The results show that CTDM performs
significantly better than the other models.
CTDM was also used to test the importance of using high
resolution, onsite meteorological measurements as the basis for input
to the model calculations. CTDM calculations were done with two sets
of input data: (1) derived from the CCB Tower A 10-m and 150-m data
and (2) derived from the 10 m data only. The results confirm the
importance of using onsite turbulence data and measured vertical
profiles of temperature and winds in the model simulations.
** During SHIS #2, KRT 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 Division (ARLFRD) 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 provided lidar, acoustic sounders, a
tethersonde, and optical and sonic anemometers. TRC provided
independent performance audits.
-------
This Fourth Milestone Report also presents the progress in
refining the SHIS #2 tower meteorological data. It discusses new
modeling of the HBR CF3Br data base. To get an understanding of the
phenomena that influence dispersion at a ridge setting, an empirical
modeling approach was taken. The results again confirm the usefulness
of the concept of a dividing-streamline (Hc) in discriminating
between essentially horizontal flow and flows that go over HBR.
The Full Scale Plume Study (FSPS) was conducted at the Tracy
Power Plant (TPP) near Reno, Nevada in August 1984. The Tracy station
was selected based on a feasibility study conducted at the site in
November 1983. The feasibility experiment resulted in a data base
that is useful for modeling and for showing the important differences
between a full scale site and the two Small Hill sites. This
milestone report presents a discussion of an initial analysis of the
November Tracy data base and the plans for the FSPS.
This report consists of a total of six sections and three
appendices. Section 2 provides a detailed mathematical description of
CTDM while Section 3 presents an evaluation of CTDM using the CCB data
base. Section 4 presents the analysis and modeling of the Hogback
data base. The results of the preliminary Tracy experiment and the
plans for the FSPS are given in Section 5. Section 6 presents the
summary, conclusion, and recommendations for future work.
Appendix A contains a report prepared by the EPA FMF describing
the results of a towing tank study designed to evaluate the validity
of modeling a plume in the flow above Hc as if both the plume height
and the terrain height were reduced by Hc. Appendix B contains
another report prepared by the EPA FMF describing a wind tunnel study
designed to identify source-terrain relationships for which the effect
of an idealized model of CCB on ground-level concentrations is
greatest. Appendix C contains an overview of the ALPHA-1 airborne
lidar observations of the oil-fog plume during the November 1983
preliminary experiment at the Tracy Power Plant near Reno, Nevada.
-------
SECTION 2
THE COMPLEX TERRAIN DISPERSION MODEL
The Complex Terrain Dispersion Model (CTDM) is a point source
plume model which 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 important in
controlling the magnitude and distribution of plume concentrations
across CCB and HER. Although the formulation of modeling concepts
contained in CTDM has general inherent applicability, certain details
make explicit use of the geometry of CCB and HBR 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 SHIS #2, the specific formulations will
change as will the range of phenomena contained in the model. In
addition, analysis and modeling of the FSPS data that were obtained in
August 1984 will undoubtedly require a more general treatment of
terrain in the model, and certainly the inclusion of new phenomena.
These changes will be treated as updates to CTDM rather than the basis
for new model nomenclature.
A central feature of CTDM is its use of a critical
dividing-streamline height (Hc) to separate the flow into two
discrete layers. This basic concept was suggested by theoretical
arguments of Drazin (1961) and Sheppard (1956) and was demonstrated
through laboratory experiments by Riley et al. (1976), Brighton
(1978), Hunt and Snyder (1980), Snyder et al. (1980) and Snyder and
Hunt (1983). The flow below Hc is restricted to lie in a nearly
horizontal plane, allowing little motion in the vertical.
Consequently, plume material below Hc travels along and around the
terrain, rather than up and over the terrain. The flow above Hc is
allowed to rise up and over the terrain. Two separate components of
CTDM compute ground-level concentrations resulting from material in
each of these flows. LIFT handles the flow above the Hc, and WRAP
handles the flow below Hc.
The following subsections contain descriptions of the LIFT and
WRAP components, and also descriptions of the plume spread
formulations contained in CTDM, and adjustments made to the plume
height to account for buoyancy effects and wind speed shear effects.
-------
2.1 The LIFT Component
The flow above Hc is considered to be weakly stratified. That
is, the stratification is strong enough to influence the flow pattern
(e.g., lee waves), but not strong enough to inhibit significant
vertical motion. To simplify the modeling task, Hc is assumed to be
a level surface, and the flow above Hc only "sees" that portion of
the hill that lies above Hc.
A fluid modeling study has been performed by Snyder and Lawson at
the EPA FMF to assess the utility of this approximation. The results
of this study, presented in Appendix A, confirm that this
approximation is reasonable with regard to estimating the locations
and values of maximum ground-level concentrations and areas of
coverage on the windward side of the hill. Poorer correspondence is
found in the Lee of the hill for plumes released well above Hc, and
this is apparently due to lee wave effects.
The plume is allowed to develop as if the terrain were perfectly
flat until it reaches the point where the Hc surface intersects the
hill (at x = xr 4- s0, see Figure 1). If Hc is zero, then this
zone extends from the source to the base of the hill, although it
conceptually could extend to any point where the hill is thought to
exert a significant influence on the flow. Beyond so, the plume
material below Hc is disregarded by the LIFT component, and the
evolution of the remaining material is modeled as if the terrain were
flat, and the lower boundary were Hc (with full reflection).
However, the wind speed, plume height above Hc, plume spread, and
lateral position of the plume centerline relative to the receptor are
all modified to reflect the net alteration of these properties between
so and s (the distance from the source to the receptor) induced by
the presence of the hill. The simplicity of the Gaussian plume
solution is retained in this way, while the full dilution of the plume
from the source to the hill (so) as well as the effects of the hill
on both flow and dispersion beyond so are explicitly incorporated.
Aside from the obvious distinction of incorporating the primary
influence of Hc on the plume-terrain interaction, this approach
notably differs from the current regulatory modeling approach (at
least as embodied in COMPLEX I and II) in that the terrain influence
for a receptor on a hill only affects the diffusion of the plume once
it is over the terrain. The mathematical formulation of the "partial
plume height correction" approach of COMPLEX and similar models
actually "lowers" the plume at the source. If this technique were
engineered to produce the "correct" hill-influenced ground-level
concentrations, the terrain correction factor for a particular
receptor would need to be a function of downwind distance, terrain
shape, and distance between source and terrain. As employed in
regulatory modeling, however, the terrain correction factor depends
only on the stability class, so that its use has led to problems of
interpreting "surface reflection" from sloping terrain, as well as to
problems in justifying values chosen for the terrain correction factor.
-------
LIFT
WRAP
'Flat-Terrain'
Domain
LIFT Domain
WRAP Domain
'Flat-Terrain'
Domain
LIFT Domain
WRAP Domain
Figure 1.
Idealized stratified flow about hills indicating domain of
individual CTDM component algorithms. The distance so is the
distance from the source to the intersection of Hc with the hill
surface for receptors above Hc (the LIFT domain), and it is the
distance from the source to the terrain-height contour equal in
height to the receptor elevation for receptors below Hc (the WRAP
domain). Note that in the upper figure the flow is into the page.
-------
Terrain-induced modifications to the plume arise from the
distortion of the flow over the hill. A streamline of the flow will
be deflected to the side (unless it lies in a plane of symmetry) and
its height above the surface will be reduced. Adjacent streamlines
are deflected in much the same way, but are generally displaced by
differing amounts, which in turn changes the spacing between
streamlines and hence, causes changes in the local speed of the flow.
As a result, the actual plume trajectory is curved, the time of travel
does not vary linearly with distance, the plume distorts so that it is
thinner in the vertical direction and wider in the lateral direction,
the turbulence statistics vary, and as shown in Hunt and Mulhearn
(1973), turbulent diffusion across streamlines is enhanced by the
contraction in the distance between streamlines in the vertical
direction, and retarded by the expansion in the distance between
streamlines in the lateral direction.
Hunt and Mulhearn explicitly track these changes through the use
of line integrals along the streamline that coincides with the plume
centerline. LIFT has been designed to take average values of the
changes in flow properties over the interval between so and s.
These average values then guide the distortion applied to the
concentration distribution at the distance so, and the flow speed
used in the subsequent calculations. In essence, LIFT distorts the
plume at so by an amount representative of the average distortion in
the flow between so and s; it removes the hill and displaces the
plume toward the surface by an amount representative of the average
height of plume above the hill surface between s0 and s; and it then
uses a flat terrain computation to estimate the effect of these
distortions on the diffusion of material to the surface over the
interval s-so. Hence, a continuous process is represented by a
two-step process in which the distortion of the flow and the diffusion
of the plume in the distorted flow are treated successively. This
approach, while not as rigorous as the Hunt and Mulhearn approach,
allows us to develop a modeling framework in which the terrain effects
appear as simple factors within the flat terrain solution.
2.1.1 LIFT Structure
The terrain effect as modeled in LIFT includes re-initializing
the flow at a distance so downwind of the release. This
re-initialization can be illustrated first for flat terrain and
uniform flow. The concentration at a receptor downwind of so is
composed of contributions from the entire concentration distribution
above Hc at so. Conceptually, the flux of plume material through
the plane x = xr + so (note that the x-axis lies along the flow
direction, and the plume is released at xr,yr,zr) can be thought
of as a distribution of point sources. If we track the plume material
in terms of the distance downwind of the source, s = x - xr, then
the source strength of one of these point source elements is given by:
dQ(s0,y,z) = C(s0,y,z)u dy dz (1)
-------
Because the flow beyond so is considered to be uniform, the
influence of each of these sources follows the Gaussian plume solution
to the advective diffusion equation:
_0
dQ(s ,y,z) s/fcJLz _0
a *' <•-*€» *' + ew* ^a *' )
, *u y z z
y z
where o»*, oz* denote the plume spread statistics for each
point source element over the interval s - so (see Figure 2a). The
total concentration at a point (s,9,,h) is found by integrating
Equation 2 over all point source elements, so that
,y,z) 5(=)2 _0 5<)2 _0
u e V (e V + 6 V
The plume spread statistics
-------
U
If I / / I I t > I / f t 7 / / / / / t t t I I t / > I I ' I ' ' I ' ' l ' I >
C(s, z; s0, z0)
ft /////// / I "t I / f f ' I / t I ! t I i > > t t I t t / / t / t / f
C(s, z) = C (s, z; s0, z0)dz0
where az
>2 2 2
= fr_^ - a~~
Figure 2a. Illustration of the relationship between the crosswind-average
concentration profiles at so and s, and the plume from one of
many point-source elements representing the flux of material across
the plane at so. The total concentration at a particular point
C(s,z) is constructed by summing the contribution C(s,z; so,zo)
from each point-source element Q(so,zo).
-------
C(s,z) — Flat Terrain
— zr
X r r ffffifff/ff/f/ffff/ ff ft/ i T / / I f / I 7 / t fir*
0
Alter Flow, Source Distribution,
C(s, z; s0, z0)
C(s,z, s0, z0')
/I/////// ///// / / / / / /////////// /// / / / / // / /
0
iC(s,z) — Terrain
'zo
zr
A/////S////S / / ///////////////?
Figure 2b. Illustration of simulating terrain effects by shifting the
point-source elements at s0 and altering the diffusivity
and flow speed beyond so. All source elements are
lowered by the factor Tjj = zo'/zo, and the
diffusivity in the altered flow is changed by the factor
10
-------
VIEW FROM S. LOOKING UPWIND
Yr
(Transformed Terrain)
PLUME SECTION AT SQ
• • • • IN THE ABSENCE OF TERRAIN
PLUME SECTION AT S IN
THE ABSENCE OF TERRAIN
. . PLUME SECTION AT S IN
' •'"• >-•' THE PRESENCE OF TERRAIN
Figure 3a. Illustration of the displacement and distortion of the plume due to
the influence of terrain as modeled in LIFT. In the absence of
terrain, the plume at s remains centered at (yr, zr) as
illustrated by the darkly shaded plume cross-section. In the
presence of terrain, LIFT treats the terrain surface as flat, but
lowers and deflects the centerline of the plume to one side. The
deformed flow results in a decrease in the vertical size of the
plume, and an increase in the horizontal size of the plume, as
illustrated by the lightly-shaded plume cross-section at s. Note
that the centroid height varies with the distance s-s0, and in
this depiction it follows a trajectory that carries the plume nearer
the surface with increasing distance downwind.
11
-------
(Receptor)
PHYSICAL PICTURE: material
above Hc rides up and over hill
in a distorted flow. Material
below Hc passes round the side.
TERRAIN TREATMENT IN LIFE: terrain
factors specify average terrain-induced
flow modifcation from so to s.
zr'-Hc' =Th(zr-Hc)
u ' = Tu u
°zo' = Th °zo
°z = Th °ze
LIFT CALCULATION: net effect of flow
distorition is to increase the
effective rate of plume growth
(aze)over that in the
absence of the hill (crz)
Figure 3b. Illustration of terrain effect on the vertical distribution of
plume material above Hc as modeled in LIFT.
12
-------
y'-y z'-H •
dQ(so.y,z') _Q.5(—B)» o.5(-—H<
y z
where
a*'2 = o'2(s) - o'2(s ) E <*'2 - o' 2 (6a)
z z z o z zo
(6b)
Because the material from elemental point sources below Hc* at s0
are forced around the hill rather than over it, the total
concentration contains contributions only from those source elements
above Hc', so that Equation 3 becomes
00 *• c(s y,z«)
C(s,yR,0;so) . J J
H ' '
C —^o
Z'-H '
_0 5( c_)2
• e 'a*' dv'dz'
z '
where
y'-y' z'-z* z'+z'
o ' 2iru'o' c' yo zo zo
y° Z° (8)
Note that the plume is originally released at the point
(xr,yr,zr). After the deformation at the distance s0, the
plume centerline passes through the point (so, yr, zr).
Primed quantities in Equations 7 and 8 are related to the
unprimed quantities of the undistorted flow by means of terrain
factors (see Figure 3b). These factors are local in the sense that
13
-------
they depend on the position of the receptor on the terrain, even
though they represent the average terrain effect on the flow from so
to s. Tn and TH are factors that specify the amount of
streamline distortion in the vertical and lateral directions; Tu
specifies the resultant change in the flow speed; and Toz and Tay
specify changes in the diffusivity in the vertical and lateral
directions.
The factor T^ should account for the effective contraction of
the distance between streamlines in the vertical. A simple model for
the change in streamline spacing applies a constant depression factor
to all streamlines over a particular location on the terrain. This is
particularly convenient for evaluating Equations 7 and 8. But note
that the perturbation caused by a hill should decrease with height, so
that this simple model must be viewed as an approximation which must
be applied at plume centerline height. Similarly, Tg, is treated
as a constant and is evaluated for a particular plume path. More
details on the specifications of these flow deformation factors are
given in subsection 2.1.2.
The speed factor, Tu, is obtained by conserving mass. The
condition TuTnT^ = 1 must be maintained at every point in the flow.
The factor To (denoting either Taz or T^y) should account for the
effective growth in the plume element in the altered flow between s
and so and can be expressed in terms of the diffusivity. The
diffusivity over the interval s-so can be related to the growth of a
plume element in the following way. For a uniform wind speed but a
variable diffusivity (D) in a flow over level terrain,
*g1 . J DC.)
If a mean diffusivity over the interval s-s0 is denoted by U^,
then Equations 4 and 9 can be combined to give
, 2(s-s )
a*' » o2 - a * = (s-s )~ = - — D (lOa)
o o As u ro
and
a 2(8-8 )
(s-s )fs = - r-2- D' (10b)
o As u m
Therefore Ta, the factor that relates the growth of the
terra in- influenced plume element o*1 to the uninfluenced growth
a*, can be evaluated in terms of the diffusivity
14
-------
and hence, can be evaluated with the aid of Equation 9 as discussed in
subsection 2.1.2.
With these definitions, the explicit relationship between the
initial plume and the distorted plume is given by the following
equations:
z' = T. z ; H1 = T.H ; y* = T.y ;
r h r c her I r
o' = T.
-------
ze zo
The terrain-induced modifications are easier to observe if Hc
is set to zero:
y /T0 - y
- - -
-.
ye ze
Equation 13 differs from the simple flat terrain Gaussian plume
equation only in the appearance of a distortion factor T^ in the
lateral distribution term, and in the appearance of effective plume
spread parameters. If Tz and Ty are not equal to unity, the
effective size of the plume differs from the unmodified plume and the
effective lateral distance from the plume centerline to the receptor
is altered. For illustration, let yr = 0. Then if Tj > 1,
which is generally the case because the streamlines spread out to some
degree as the flow is deflected laterally, the apparent receptor
location (VR/Tg,) lies nearer the plume centerline. Aside from
this lateral shift in the impact region on the terrain, the influence
of the terrain is exhibited only through changes in the rate of
effective plume growth. When Tz is less than unity (again, this is
generally the case) , oze exceeds oz at s and so more plume
material may lie "nearer" the surface. As a consequence, Equation 12
may estimate ground-level concentrations in excess of flat terrain
estimates even when Hc is zero (see Figure 4).
If Hc is non-zero but less than zr, Equation 12 will estimate
even greater ground-level concentrations because the flow in the layer
from the surface to Hc is "removed," allowing the less dilute
portion of the plume to approach nearer the surface (see Figure 3).
In particular, if Hc = zr, then Equation 12 places the centerline
concentration at ground level, producing a centerline "impingement"
result.
In the development of Equation 12 and the terrain factors, the
factors are considered local terrain effects factors for a specific
receptor. The degree of flow deformation is dependent on whether the
flow must go directly over the crest, or pass to one side in
approaching the receptor. Once the local factors are obtained, they
are applied to the entire plume. If there is a large amount of plume
meander over the averaging period of the model computation,,
Equation 12 may not be appropriate.
16
-------
CD
o
** 3
"5 Q.
cc = o
3 -C ;«T
I- CD CD
_i 5 -°
< o -o
2-0 52
w c o
>- co 2
X a.<2
a. 3 -o
1C
3
t"S
ill
— o> •*-
m ^ JP
5 *- =5
>- "S
5?£
< 2 S
oc w S?
oc =5 .5
m « ~
t- .!2 -o
(B—^
''
b
ts
CD
£
C3» *-
0)
CD
CO
(A
O
» £ § « S
c s
z o
o *-
g~
fi§
3 S
^g
O^
CD CO
55
o
tl
C > .E
S 6 -
CO CD
CO
4-1
CO T3
U CD
C CO
o co
U CD
ii
CD ~
CO
® ^
CD $
t I
c 2
3 CD
O C
>- CD
O S>
o
ii
O 1-1
o> rt
In -H
IH f-i
(D -H
rt •(->
O
• H i— t
+-> cd
rt O
B -H
-------
Therefore, consider Equation 12 applicable to a "filament"
plume. The "filament" plume is defined to be a plume described by the
flow field statistics obtained for a sampling period commensurate with
the time of travel from the source to the hill. The mean
concentration at a receptor for an averaging period greater than the
time of travel is a weighted average of many "filament" plumes:
+00
cm = J c(s,er,o;so,ei) P^) de.^ (14)
P(6) is the probability that the wind is from the 9 direction
during the averaging period, and C (s,6r,0;so,6) is the
concentration resulting from a "filament" plume from direction 0,
and 6r is the direction from the apparent or effective receptor
location to the source. Details explaining how 6r is found are
contained in subsection 2.1.3.
For a distribution of wind directions that has a single dominant
mode at the mean wind direction 6m, P(6) may be approximated by
a Gaussian distribution:
(6-6)8
P(e) '
Using a similar arc length notation for lateral distance,
C(s,6r,0;s0,0) may be written as
_ ... (6 -6)s
F (6) r ' a
C(s,e,0;s,6) =
a (16)
where Fz(6) denotes the vertical distribution portion of
Equation 12.
By assuming that the "filament" plume is narrow, set Fz(6) to
Fz(6r), and evaluate Oye f°r a Plume from 6r. This makes explicit
the use of terrain factors that are local to the receptor.
Equation 14 becomes:
„ ,n . +» (6. -6 )s (6,-e )s '
F (6 )s i r x m
C = -7== - J e a e a i (17)
m /Ziro a ye ym
ym ye _
18
-------
The solution of this integral is
(9 -6 )s
„ W -0.5( r m )» (18.)
w — 6 CJ _
» V *T
where
c 2 = o 2 + o 2 (18b)
yT ym ye
Because a,™ is viewed as the statistic for just the meander
component of the wind fluctuations over the averaging period, and
Oyg is viewed as the measure (including the terrain modification)
of the mean "filament" plume spread, Oy? = ove in the absence
of meander, and Equation 12 is obtained once again.
In applying LIFT to the CCB data base, the selection of wind
statistics for the "filament" plume is limited to the 5-min sampling
time, and the mean concentration is generally computed for a 1-hour
interval. Also, when the average concentration is obtained from
simulating each 5-min period separately, cr™ is identically zero,
and Equation 18 is used for both calculations.
Equation 18, in which FZ(0V) represents the portion of
Equation 12 that contains the information on the vertical structure of
the plume, provides the framework for estimating concentrations due to
plume material that travels up and over a hill. As a framework, it
shows how the influence of the terrain, expressed as terrain factors,
affects the magnitude and distribution of plume material along the
surface of the terrain. Consequently, the influence of particular
hills on the flow-field must be prescribed in terms of these mean
terrain factors before concentrations can be estimated by means of
Equation 18. The following two subsections describe a method for
obtaining mean terrain factors, and a method for estimating the
effective position of a receptor after accounting for the lateral
deflection of the plume.
2.1.2 Terrain Factors
Terrain-induced modifications to the flow over a hill are
prescribed in terms of the undisturbed, rectilinear flow and the
terrain factors T^Cx.y.z), Ta(x,y,z), and Tu(x,y,z). The
vertical coordinate (z) denotes the height of the plume in the
approach flow, and the x-y plane is centered on the hill (presumably
at the crest of the hill). The plane z = 0 is the flat surface away
from the hill, generally taken to be the surface elevation beneath the
plume near the source. The x-axis is assumed to be aligned with the
flow direction that would take the plume centerline over the effective
receptor location (see subsection 2.1.3 and Figure 11). If T(x,y,z)
denotes any one of the three factors, then at the position (x,yr)
the distortion factor for the flow between streamlines originally
19
-------
passing through (xr,yr,zr) and (xr,yr,0) is denoted by
T(x,yr,zr) (see Figure 5).
The factors actually applied in LIFT need to be expressed as
average values for the interval s-so. Because the factors are
expressed in the coordinates of the undistorted flow, an average along
the trajectory of the flow is expressed as an integral along "x."
Therefore the mean or effective terrain factor for the flow between
streamlines passing through (xr,yr,zr) and (xr,yr,0) is
given by:
x +s
r
T(s,y ,z ;s ) = J T(x,y ,z ) dx/(s-s ) (19)
X +S
r o
Note that only two of the three factors need be evaluated in this way
because TTT = 1.
In principle, the factors could be computed from a flow model.
However, the approach taken in LIFT involves scaling the magnitude of
the flow modifications on the basis of flow computations rather than
incorporating the results of those flow computations in detail. This
scaling is accomplished by specifying the extreme values of T^,
T£ and Tu. For neutral flow, these values occur at the crest of
the hill. With a certain degree of stratification, these values shift
to the leeward side of the hill. Away from the location of the
extreme, the modification to the flow diminishes, and the terrain
factors approach unity. This relaxation to unity is specified by a
simple analytic function that includes length scales for the lateral
and longitudinal "axes" of the hill. An example of a set of these
extreme values for CCB can be found in subsection 3.3.1.
This method of incorporating the terrain-induced distortion
allows the influence of the terrain to be incorporated by means of a
limited number of parameters. The average terrain modifications
required in the model can be specified analytically without the need
for numerical integration. Consequently, several flow modeling
assumptions can be readily investigated, and the parameters that
specify the flow can also be treated as optimizing parameters so that
the observed distribution of concentrations may be used to infer the
flow modification.
For the case of an axisymmetric hill (an approximation for CCB)
with a length scale equal to its radius at the elevation of the Hc
surface, a(Hc), the factor variation is given the form:
_ a
T(x,y,zr) =1+ (To(zr)-l)e~(a(a(Hc)+xgep))2 ~ ()2 (20)
20
-------
<5y
(xr, o, zr1)~
a<
^ —
f U
* sn •
^
h>
\
\
\
-^^ u ^ -F
V>^
|^-
(xr, o, o) (o, o, o)
<5y <5y
(xr, o, zr2)~|
*
4
te . 1
^ u
>x^
L,
»i U
y>^
5^
^
^
<5y y
n* > »
}
S(
Vr f
-*-U' ^
^ (XR, 0, 0)
5y x
-'
^^-
-i<
«r
" f
— ^- u'
^B- X
VX._^'^=
<5y 'U '
Tj?:u:h
-------
The quantity a is introduced to adjust the rate at which the factors
relax to unity with distance from the location of the extrema.
To(zr) denotes the extreme flow modification factor for the flow
between streamlines passing through (xr,0,zr) and (xr,0,0).
To simulate the effects of weak stratification, the variable
xsep is introduced to shift leeward the location of the nearest
approach of streamlines to the surface. With weak stratification, the
formation of the lee waves influences flow separation in the lee of a
hill. On the basis of laboratory experiments and simple theory (e.g.,
Hunt and Snyder 1980), the separation point resides near the hill
crest when one quarter of the wavelength of the lee wave exceeds the
longitudinal half-length of the hill (measured at 1/2 the hill height)
because the streamlines in the wave pattern exhibit a shallower slope
than the terrain surface. When the wavelength is shorter than this,
the separation is suppressed over the hill until the wave pattern
causes the streamlines to rise again at three quarters of the
wavelength of the wave. This point lies at one half wavelength from
the crest of the hill (see Figure 6).
The wavelength of the lee wave is approximately equal to
2iru/N. Because LIFT addresses the flow above Hc, u/N is taken to
be representative of the flow above Hc as well. The criterion for
where separation is likely to occur can be expressed in terms of the
Froude number based on the horizontal length scale of the hill,
With L defined as in Figure 6,
Consequently, the separation point remains near the crest for
> 2/ir. Therefore,
sep ~ Fr >^ 2/ir
(22)
x = iru/N FrT < 2/w
sep L
We shall assume that the streamlines above Hc come closest to the
hill surface near xsep.
Equations 19 through 22 allow the average factors Tn and Tg,
to be specified for the flow through the source at height zr toward
a particular receptor. Tu is always obtained as the inverse of the
product of Tj and Tjj. Consequently, Tn, T^ and Tu depend on 6r,
the wind direction that brings the centerline of the plume over the
receptor.
The fourth terrain effect factor, Ta (denoting either Toy
or Tcz), depends in part on T^ and Tu. Ta is defined by
Equation 10c, and requires the specification of the mean diffusivity.
22
-------
> 4L : Separation occurs near the crest
A < 4L : Separation occurs near A/2 beyond the crest
(= 27TU/N) = 4L -*-FrL(=u/NL) =
Figure 6. Dependence of separation in the lee of a hill on stratification.
The structure of the lee wave begins to influence the point of flow
separation when the wavelength is less than 4L.
23
-------
In the case of Ttfz, equations 9 and 60 indicate that
D = u izz s ("linear" growth range for oz)
D = u iz 8, ("square root" growth range for oz)
where
1/8. = -~ + --. . (23)
These expressions are derived from the equation for
-------
2.1.3 Effective Receptor Location
The evaluation of Equation 18 and the development of
subsection 2.1.2 require the specification of the effective, or
apparent receptor location. In particular, the wind direction 6r
that transports the plume over the effective receptor must be
obtained. In developing Equation 12, the plume centerline was shifted
by a factor Tj, which represents the average lateral shift along
the trajectory. But a more accurate specification of the shift in the
centerline position requires that the shift be evaluated at the
receptor location without averaging along the path. The effective
receptor location is then related to the actual receptor location by
the terrain factor T^(x,y,zr).
When the coordinate system, with origin at the hill center, is
rotated to align the direction of the x-axis with the direction from
the source to the effective receptor location, (i.e., the flow is due
to wind from 0r)
yr
dy • (28)
That is, the cumulative deflection of the flow from the center of the
hill out to the position (yr) of the undistorted trajectory of the
plume gives the lateral position (yR) of the actual receptor over
which the deflected plume trajectory travels at the downwind position
XR (see Figure 7). Upon substituting for Ta(xR,y,zr),
x_-x
^ , « sep 2
y = y + aa(H )rJL erf(y /aa(H ))(TB (z )-l)e v a(a(H )+x )
R r c2 r cilor c sep ,-«,
(29)
Equation 29 must be solved numerically for yr given yR and
XR. But yR and XR are themselves only known once 6r is
known (due to rotating the coordinate system), which in turn requires
that yr and xr be known. Thus, 6r is obtained from this
implicit set of relationships by iteration.
2.2 The WRAP Component
In CTDH, the flow below Hc is considered to be completely
two-dimensional, allowing no motion in the vertical. Consequently,
the flow must pass to one side or the other of the hill, and the one
streamline that actually touches and passes round both sides of the
hill separates the two flows, and is termed the stagnation streamline
of the flow. The flow on either side of the stagnation streamline
undergoes distortion in much the same way as that discussed in
Section 2.1. A streamline in the flow will be deflected to the side,
but will pass closer to the hill surface than its initial distance
from the stagnation streamline. Adjacent streamlines are displaced by
differing amounts, which in turn changes the horizontal spacing
between streamlines and hence, causes changes in the local speed of
25
-------
Projection of
Actual Plume
Trajectory
onto the Plane
z = 0
Plume Trajectory for 0r Wind
in the Absence of the Hill
Actual Receptor (XR,
(XR, yr) Effective Receptor
Source
Figure 7. Definition sketch for deriving the effective receptor location.
The x-axis is aligned with the flow due to a wind from direction
e,..
26
-------
the flow. The effect of these distortions on ground- level
concentrations is similar to those formulated in LIFT for flow over a
two-dimensional hill, except that the roles of Tj and T^ are
reversed .
The primary difference between WRAP and LIFT formulations arises
from the location of solid boundaries and the relationship between the
position of these boundaries and the wind direction fluctuations. The
terrain effect is modeled in WRAP by re-initializing the flow at the
distance s0 downwind of the source (see Figure 1). Note that
receptors below Hc experience an so different from that
experienced by receptors above Hc. Below Hc, so is the distance
along the stagnation streamline from the source to the terrain contour
equal in elevation to the receptor elevation. The concentration at a
receptor downwind of so is composed of concentrations from that part
of the concentration distribution at s0 that lies below Hc, and
that also lies on the same side of the stagnation streamline as the
receptor (see Figure 8). Reflection of plume material is allowed from
the plane z = 0 over the entire distance s, and reflection is also
allowed from the "stagnation streamline" beyond so. Note that the
stagnation streamline forms the boundary of the hill surface in
horizontal cross section.
The terrain influences are incorporated by deforming the source
distribution at so (Equation 1), and by altering the flow in which
the source elements diffuse. For a receptor located on the hillside
at a distance s (see Figure 8) and a height ZR above the plane
z = 0, the concentration due to one elemental point source located at
(s0,y*,z') in the deformed plume is given by
dQ>
- 2° V (e °r + e «O
y z (30)
Equation 30 assumes that the x-axis of the coordinate system points
along the stagnation streamline, and that the source is located at
(xr,yr,zr). The total concentration at the receptor contains
contributions from those elements below Hc (now represented as
Hcf), and on the same side of the stagnation streamline as yr (now
represented as yr) :
Hc (
-------
O -H
-------
These expressions are analogous to Equations 5 through 8 of the LIFT
component. The integral for dy' has the limits (o) and (<»). This
is meant to denote integrating from 0 to + » if the receptor lies on
the "positive" side of the stagnation streamline, and integrate from
-oo to 0 if the receptor lies on the "negative" side.
Noting that T^ = 1 for this two-dimensional flow, the integrals
in Equation 31 are evaluated and the primed quantities are replaced
according to Equation 11 to obtain
^r y
-------
If the distribution of wind directions over the averaging time is
highly non-Gaussian, then the mean concentration is probably best
estimated by simulating a sequence of "filament" plumes. For the CCB
data base, this would entail averaging the concentrations obtained
from Equation 33 for a sequence of 5-minute average meteorology.
However, for distributions closer in shape to the Gaussian
distribution, an expression of the form of Equation 14 may be used
with the Gaussian distribution specified in Equation 15.
Denote the concentration due to a "filament" plume for wind
direction 6 as
F (e) _ r— 12 y o*/T
C(s,0,zR;so) , -J—e -°'5(o > (1 + sign(yR) erf <£J J ))
ye yo ye (35)
where yr, oye,
-------
0)
u
oo
•H
Pu
31
-------
b.Acf
b.Acj
(38a)
where
A .
(6 -6 )r
yt
-------
2.3.1 Lateral Deflection
LIFT and WRAP produce estimates of ground-level pollutant
concentrations in distinctly different ways. Consider the idealized
"filament" plume shown in cross section in Figure 10, Tha cross
section is taken at the downwind distance s0 for the sake of
illustration. In this example, the centerline of the plume lies
neither along the stagnation streamline nor in the Hc plane; the
relative positions of these quantities are identified in the figure.
for this plume, concentrations at receptors below Hc are derived
'from the cut made by the stagnation streamline. Concentrations at
receptors above Hc are derived from the cut made by the Hc surface.
Adjacent receptors located on either side of Hc illustrate the
effect of these two cuts. Above Hc, the centerline of the plume may
be deflected directly over the receptor (A). Below Hc, the lateral
offset of the plume centerline from the receptor (8) is controlled
entirely by the stagnation streamline. Consequently, distinctly
different concentration estimates say be allocated to receptors
adjacent to one another on either side of Hc.
A remedy for this situation may be constructed by altering the
lateral deflection in LIFT in a zone near Hc. LIFT concentration
estimates may be simply expressed as in Equation 18a
(6 -6 )s
F (6 ) -0.5( — £— - — )a (39)
C .
V
The vertical distribution factor F2 depends on the direction from
the source to the effective receptor position — the position after
lateral streamline deflection has baen included. It is this position
that determines the terrain effect magnitude. We shall leave this
factor unchanged, and replace 6r with 6r' in the lateral
distribution factor to simulate the transition region.
Close to Hc, 0r' should nearly equal QSf the stagnation
wind direction. Beyond the transition region, ©r* should equal
6r. The thickness of the transition region shoul-S depend on the
scale of the hill and the stratification. Hc coaibinea both of these
quantities and is a convenient length scale for the region.
Therefore, introduce the transition zone length ssela &.s iSHC8 and
let 6r" be defined as
6 ' = 0 + (0 ~6 ) e"~(!SlfIle)/'5He <40)
r r s r
where zr is the receptor elevation. Because it is unlikely that the
transition zone could be as large as Hcs & is probably of order
0.1.
33
-------
Concentration Isopleth
in plume
Plume Centerline |
A
A B
Stagnation Streamline
A: Concentration for Receptor Slightly Above Hc (x, y. Hc + c)
B: Concentration for Receptor Slightly Below Hc (x, y, Hc -£)
Figure 10. Cross-section of plume at so illustrationg how LIFT and
WRAP concentration estimates differ at receptors on
either side of Hc. Just downwind of s0, LIFT concentration
estimates are obtained frora fch« cut made by the Hc surface
through the plume, but WRAP concentration estimates are obtained
from the cut made fey the stagnation streamline through the plume
(below Hc).
34
-------
2.3.2 Hc Variability
Variability in the height Hc is specified in terms of a
probability density function P(HC). If we assume a normal
distribution about Hc, the mean value for the averaging period, then
where
Cm
Therefore the effective size of the plume in the vertical is increased
by the He variations. This arises from cutting off the flow below
Hc. As Hc varies, the nearness of approach of the plume center-
line to the surface varies, and this increases the apparent plume size
in the mean.
A similar effect is not seen in the WRAP result. Hc variations
have no effect well below Hc. Near Hc (within oHc of Hc)
the Hc variations modify the amount of plume material contained in
the horizontal flow regime.
35
-------
2.4 Formulation of oz (Venkatram, et al. 1984)
2.4.1 The Theoretical Framework
The crz formulation is based on the behavior described by
o = at ; t « I. (44a)
Z W Lt
a = (2K t)°' 5 ; t » TT (44b)
Z Z L»
In Equation 44a, av is the standard deviation of vertical velocity
fluctuations, t is the travel time from the source, TL is the
dispersion time scale, and Kz is the eddy diffusivity defined by
K = ol , where I = a T (45)
We suggest a plausible formulation for the "mixing length" 8. by
the following argument. In a stably stratified flow, a fluid element
must overcome a stable potential temperature gradient in order to be
displaced vertically. Simple energy arguments suggest that this
gradient imposes a limit, which is defined by a length scale of the
order of ow/N, to the vertical motion. Here N is the
Brunt-Vaisala frequency. Consequently, we assume that the mixing
length I is proportional to this length scale so that
I = y2 «/N (46)
where y is an undetermined constant. Hunt (1982) argues that y
should be a function of the molecular diffusivities of the fluid and
the pollutant in the fluid. The vertical motion of a fluid element (a
control volume larger than the turbulent mlcroscale but much smaller
than the macroscale) beyond ow/N depends on its ability to exchange
its density with its surroundings. This exchange is governed by the
molecular diffusivity of the fluid. The rate at which the pollutant
escapes the fluid element is determined by the diffusivity of the
pollutant in the fluid. Hunt suggests that the time scale of
molecular mixing Tm can be long enough to inhibit the growth of
az beyond ow/N for long travel times.. We will assume that Tm is
always much smaller than the travel time t so that y is independent
of molecular diffusion processes. We will allow the comparison of our
predictions with observations (presented later) to justify this
assumption.
Surface-layer relationships (Businger 1973) are employed to
estimate y. When the nondimensional potential temperature gradient
<}>H is independent of z, the kinematic heat flux Q0 can be
written as
Q = -T*u* -- Lu* (47)
36
-------
Where L is the Monin-Obukhov length given by
Notice from Equation 47 that the eddy diffusivity for heat KH is
KU = « u* L (49)
n o *
Substituting Equation 48 into Equation 47 we get
L - ^ ^ (50)
where N is given by
•-J-S"
If we now substitute Equation 50 into Equation 49, and recast Kjf
using the relationship ow =» au*
<52)
Assuming that KH = K2, we find that y2
2 1
(53)
With fl » 4.7 and a = 1.3, y equals 0.52. The expression for KH
was derived by Hunt (1982), but on the basis of arguments presented
earlier, he maintains that KH is not generally equal to Kz. He
suggests that KZ/KH ranges from 0.5 in the surface layer to 0.1 at
higher elevations. Our observations of oz, to be described in
later sections, do not support this contention. In our analysis we
will take the stable mixing length 8-s to be given by
ls » Y2 <»W/N where y - 0.52 (54)
When N is small, ls can become very large, and it becomes
necessary to consider the effect of the ground on limiting the length
scale. In the absence of stratification, one expects the mixing
length to scale with z,
ft = Fz (55)
37
-------
Where the subscript n is used to distinguish the neutral length scale
9,n from the stable scale 8,s. To estimate T we notice that
KH can be written as
KJJ = cwkz/(aH(0)) (56)
Equation 56 indicates that
F = k/a$H(0) (57)
If we take the Kansas values (Businger 1973) k = 0.35 and
4»H«» = 0.74, F equals 0.36.
We have chosen to interpolate between S,n and 8,s with the
following formulation for 8.:
1/8. = 1/8. + 1/8. (58)
s n
Equation 58 has been used by other investigators (see Hunt et al.,
1983 for example) . Then the dispersion time scale is given by
(59)
The oz formulation which interpolates between the linear and
square-root growth rates is one used by other authors (Deardorff and
Willis, 1975),
a~ • O'd + t/2T.)°-5 (60)
Z wf Li
Equation 60 was used to analyze the observations described in the next
section.
2.4.2 Estimating az from Observations
Data collected during SHIS #1 conducted in the fall of 1980 at
CCB have been used to evaluate Equation 60. Estimates of c»z are
made from individual 5- or 10-minute long photographic time exposures,
and from individual lidar scans of the oil-fog plume.
Four of the 18 CCB experiments have been analyzed for plume size
in the vertical: Experiments 201, 206, 210 and 218. Lidar data have
been reduced for two hours of Experiment 206, and two hours of 210.
Consequently, the lidar data have been used to help determine a link
between e»z and the apparent vertical extent of the plume.
The lidar data were obtained and reduced by the NOAA Wave
Propagation Laboratory (WPL) . WPL furnished the laser backscatter
intensities for each plume section, and also the second moments of the
distribution of these data. We have computed the az of the
38
-------
backscatter intensities by integrating across the scan section along
the lidar line-of-sight (nearly perpendicular to the transport
direction), and fitting a reflected Gaussian envelope by the method of
least squares. The location of the centerline is assumed to be
equivalent to the centroid of the backscatter distribution, and the
reflecting surface is set equal to the terrain elevation beneath the
centroid. This fitting procedure provides both the centerline
intensity and the best-fit cz for the cross-wind-integrated
distribution of plume material. The estimates of cz from this
procedure are nearly equal to the computed second moments of the
distribution for most of the scans, and are used in preference to the
calculated second moments only because the fitting procedure reduces
the influence of "shadow zones" on oz.
Photographs of the oil-fog plume taken from positions off to the
side of the plume trajectory have been analyzed for plume spread in
the vertical at several distances along the plume trajectory. Daytime
photographs were nearly instantaneous pictures of the plume, while
nighttime photographs were time exposures over nominal 5-minute
periods, although some photographs extended over 10-minute periods.
Plume illumination at night was provided by moonlight or by a carbon
arc lamp. When moonlight dominated, the edges of the plume in the
photographs are generally diffuse owing to the blurring effect of
passing puffs. When the arc lamp dominated, the plume image is made
up of approximately ten distinct instantaneous plume images produced
by the rotating lamp beam. These differences appear to have only a
minor impact on estimates of oz.
Gifford's (1980) technique for deriving estimates of az from
photographs cannot, in general, be applied to the CCB photographs.
This technique requires not only a reasonably uniform background and
uniform illumination of the plume, but also that the plume be
dispersed sufficiently within the field of view that the distance
between the edges of the plume, defined as a particulgfc light
intensity value, can be seen to reach a maximum value. The latter two
requirements are generally not satisfied in the CCB photographs. As a
result, a comparison of plume thickness measurements with the lidar
sections provides the best guidance in estimating cz from the
plume photographs.
The lidar data and plume photographs from Experiment 210, hour 3
are the most detailed of the four experiments considered here. The
plume image is well-defined in the 5-minute exposures, and one to two
lidar scans upwind of the hill are available during nearly each
5-minute period of the hour. Pairing only those lidar and photo data
from the same 5-minute time period, and within 50 m of each other in
distance from the release point, we find that the six resulting data
pairs produce an average ratio of plume depth to lidar-az equal to
3.69 with a variance of 0.24. If we had assumed a top-hat
distribution of plume material, this ratio would be expected to equal
3.46. If we had assumed a semi-circular distribution, it would be
expected to equal 4.47. The average of 3.7 therefore appears
consistent with the highly coherent nature of the plume during this
39
-------
hour. Similar comparisons for the other three hours with lidar data
are not as conclusive, but the result of each is not inconsistent with
a ratio of about 3.7. Therefore, all estimates of the observed az
values used in the following analyses have assumed a factor of 3.7 in
converting plume depths estimated from photographs to oz.
2.4.3 Comparison of Model Predictions with Observations
Figure 11 shows a plot of oz/tfwt versus t/TL, where the
dispersion time scale is
w
where Y = 0.52, r = 0.36, and zr is the height of release. Both
ow and N are evaluated at zr. The travel time t is taken to be
x/ur where ur is the wind speed at zr. Using zr as the
reference height implies that the scale of vertical variation of
ow, N and u is larger than az.
The solid line in the figure corresponds to Equation 60. We
conclude that this theoretical curve explains the observations
reasonably well. Note that the observations span the linear to the
square root regions of oz growth. This supports the appropriateness
of the TL formulation. The circled points in the figure refer to a
few 5-minute time periods in which the measured ow were small when
the photographs clearly indicated large sigmas. We believe that the
relatively large deviation of these points from the other data
reflects the inevitable uncertainties associated with the measurement
and interpolation of the micrometeorological variables. However, the
overall agreement between theory and experiment is good. Note that
over the range of t/TL considered, the form of the interpolation
contained in Equation 60 produces oz values that differ little
from those produced by the oz expression based on an exponential
correlogram for the vertical velocities.
The performance of Equation 60 is further illustrated in the
scatter?lot of Figure 12. The model for crz does seem to
underpredict slightly at large oz. Note, however, that 95% of the
680 observations used in the analysis are within a factor of two of
the az predictions.
The importance of including SLn (neutral length scale) in
determining TL is illustrated through Figure 13.' Here oz is
computed using only 18. Notice that most of the points are below
the horizontal "perfect fit" line. For an adequate model„ the points
should be symmetrically distributed about unity if we make the
assumption that the ratio az ( observed )/crz (predicted) is lognormally
distributed. Note that ln becomes important when ls is comparable to
S,n. If ln is not accounted for, the overall length scale I, and
hence TL is larger than it should be. This would lead to an
overprediction of 0.5 (points for which 18 > ln are plotted at ls/8.n = 1) •
40
-------
4*
to
cd
41
-------
•o
3
O
+> e
(0 -H
CO «
to*
rt u
§2
u m
«
0) (0
^ 0)
O -H
^ u
CO
N-O
*> CM
O
r-l IM
g.0
4) U
*> O
4^ *>
CO O
O (0
M Cb
CM
(ui) (peAjesqo)
42
-------
•f + *• +
+ «
•»•+*•
0)
• 00
(O
C
» x
ro
CM
'e
•o
« 40
4J a
o •*
•H 4J
•a at
a> H
a* u
«•- r-l
N e<
co
60
CO CM
CO
43
-------
Figure 14 shows the effect of including 8,n in the formulation
for TL- We see that there is an improvement in the distribution of
points for ls/8.n > 0.5.
2.5 Formulation of cfy
Past versions of CTOH have estimated ay by assuming that the
transverse spread of the plume grows linearly with time. This implies
that the Lagrangian time scale for the transverse spectrum is long
compared to the time of travel to the hill. While this may be the case
for many of the CCB experiments, there are some experiment-hours in
which the linear growth law appears to overestimate the dilution of
the plume. This is deduced from comparisons of observed ground-level
SFg concentrations with estimates of the plume "centerline"
concentrations .
The Lagrangian time scale can be incorporated in the expression
for tjy by means of the interpolation contained in the az
formulation:
a = i s/(l + ~ -— )o.s (62)
y y 2uTLT
The Lagrangian time scale for the transverse spectrum, TLT cannot be
estimated from the flow properties, but it may be estimated from the
turbulence measurements. Pasquill and Smith (1983) point out that if
the turbulence is assumed to be isotropic, and if the longitudinal
correlogram is modeled by an exponential with an Eulerian time scale
of Tg, then
-------
»+•»*• -H
•***+
0>
* \
(A
CO
2
.5
«
3
g
fr
8
2 ON
Li O
•H O
I>
09
2
S
45
-------
2.6 Other Revisions
2.6.1 Density of the Tracer Gas
SF$ has a molecular weight of 146, and CF3Br has a molecular
weight of 149. Dry air has a molecular weight of only 29, so that
both tracers are approximately five times as dense as air. Because
the tracers were not mixed into the fogger jet at CCB, they
undoubtedly experienced some fall in height beyond the release point.
The same is true for the CF3Br released at HBR because it was
released at a height different from the oil-fog generator. The SF&
was completely mixed into the jet fogger at HBR.
This fall is estimated by means of the Briggs (1975) stable and
neutral plume rise formulas:
Ahn » 1.6 (F s2)0-"'/u (neutral)
(66)
Ahs - 2.03 (F/uH'2)0-" (l-cosdTs/u)0-" (stable)
where N'2 = N/M » N2/2.25
Note that M is the flux of vertical momentum within a bent-over plume,
and Meff if the effective flux. For the stable rise formula,
Ah8 is constant beyond s=1.5wu/lT. In practice, the minimum of
the two height change formulas is taken.
The buoyancy flux F is the product of the buoyancy and the volume
flux of the tracer gas. For
(68)
P
F a ((PSF6 ~
The densities of air and, in this example, SFg are given by
'air
PSF6 ~ R*T
where R* » 8.3144 J/mole/°K. For T » 0*C and P = 840mb, the buoyancy
flux of SFg is related to the emission rate (g/sec) by
F * 0.00733 Q (mVgs2) (69)
Equation 66 will probably overestimate the height change
resulting from the weight of the tracer. These relationships do not
account for the turbulent mixing caused by the eddies in the wake of
the release crane. Also, when the tracer was released near the
oil-fog generator, it may at times have been mixed by the turbulent
eddies induced by the fogger jet.
46
-------
2.6.2 Wind Speed Shear
The increase in wind speed with height is significant in many of
the CCB experiments. When Kz is considered constant with height,
the effect of this shear on the vertical distribution of plume
material can be estimated.
Hunt (1981) quotes a result obtained by Lauwerier (1954) for the
plume concentration profile in the vertical for Kz constant. This
result is approximate, and is valid near the centerline of the plume
(say, within oz/2 of the plume centerline). When a new Gaussian
distribution is fit to this vertical distribution, the effective plume
height is found to be lower than the original height, and it is nearly
equal to the height of maximum concentration in Lauwerier*s solution:
a du/dz
(1- <-*— )2)0-5] (70)
Consequently, the shear-altered vertical distribution of plume
material is approximated in CTDM by a reduction of the plume height as
estimated above. The point chosen for evaluation in the LIFT module
4 a "a "
IS SQ.
47
-------
SECTION 3
MODELING CCB
3.1 Data Analysis and Interpretation—the Modelers' Data Archive (MDA)
Meteorological data obtained at the 150-m tower (Tower A) and
inferred from lidar data and the available photographs of the oil-fog
plume have been compared for all CCB experiments. The comparisons
include estimates of the wind direction and the vertical intensity of
turbulence at the height of the oil-fog plume.
3.1.1 Wind Direction
Photographs taken from behind the release crane looking along the
plume trajectory and those taken from atop CCB looking back toward the
release crane are the key views of the oil-fog plume that are most
important in estimating wind directions. For those views from behind
the crane, wind directions are most easily estimated when the plume
passes over some recognizable portion of the hill. Any horizontal
displacement of streamlines that may occur when the plume is close to
the hill is subjectively taken into account. For views from the
hilltop, wind directions are most easily estimated when the plume
passes over one of the camera positions on either peak, or when the
plume obscures a known landmark on the ground away from the hill (such
as a turn in a road). The wind direction estimates from the
photographs are least accurate when the plume misses the hill to one
side or the other. The most useful information that can be obtained
during such periods when the plume is off to the side of the hill is
that the plume is not currently contributing to one-hour-average
tracer concentrations measured in the sampler array.
The comparisons between the photo estimates of wind direction and
those interpolated to the release height from the Tower A wind data
are shown in Figure 15 (a-e). In these figures, the direction from
the origin of the polar grid coordinate system (at the "center" of
CCB) to the source is indicated by the solid horizontal line. The MDA
wind directions at plume height are indicated by a dashed line.
Estimates of wind directions taken directly from the photographs are
indicated by a solid line labeled with a "p." When the photographs
were ambiguous, the photos in combination with other sources of
information were used to estimate the wind directions. For these
periods the solid line is labeled with a "x."
48
-------
I
u> «
tn m
as
K
R
c
x
UJ
N.
Figure 15a, Time series of MDA wind directions at oil-fog height ( ) and
estimates made from photographs (-p-). Estimates derived from
the photos and other sources of information are denoted by (-x-)
49
-------
s
I
i
I
_r
"5=--
—r "> T
o
X
I I £ ! g § S §
c
«*
i-
X
s
m
o>
-------
8 3 ft
sill I I S 2 5 I i § * * *
•i
8
8
00
i
Figure 15c. Time series of MDA wind directions at oil-fog height ( ) and
estimates made from photographs (-p-). Estimates derived from
the photos and other sources of information are denoted by (-x-)
51
-------
(0
5
(M
8 9 ft ft
m
(M
Figure 15d.
Time series of MDA wind directions at oil-fog height ( ) and
estimates made from photographs (-p-). .Estimates derived from
the photos and other sources of information are denoted by (-x-)
52
-------
388
a o IB
TiM (Hour)
Figure 15e.
Tiro* series of MDA wind directions at oil-fog height ( --- ) and
estimates made from photographs (-p-). Estimates derived from
the photos and other sources of information are denoted by (,-x-)
53
-------
In general, the oil-fog plume trajectory agrees with the wind
direction data from Tower A interpolated to release height.
Significant deviations from the photographic evidence do occur and
these are summarized in Table 1. Many of the discrepancies may be
readily attributed to errors in estimating trajectory directions for
plumes traveling to the side of the hill. In other cases, the
discrepancies appear to arise from insufficient vertical resolution in
measuring the wind field. The 40-m level wind directions are not
available for four experiments.
3.1.2 Vertical Intensity of Turbulence
Estimates of az are made from individual instantaneous
photographs, individual 5- or 10-minute-long photographic time
exposures, and from individual lidar scans of the oil-fog plume.
Photographs suitable for inferring values of az are available from
all CCB experiments except Experiment 216, and lidar data are
available from 11 experiments. The'method used to calculate az
from the lidar data and the photographs is described in Section 2.4.
The estimates of az obtained from the lidar data and the
photographs, the inferred distance from the source to the observation
(s), and the MDA values of the wind speed (u) and the Brunt-Vaisala
frequency (N) for each 5- or 10-minute period are used to find an
optimum value of iz. The following formulation for az is used
(71)
iz8
(1 +
i s
z
2
t N H
2,
Y i u
' z
O .5
'^
where y2 = 0.27, f = 0.36 and zr is the release height.
As shown in Figure 16 (a-e), the lidar and*photo estimates of
iz are in general agreement with the MDA values of iz. Signifi-
cant deviations (differing by more than a factor of two) from the
photographic and lidar values do occur and these are summarized in
Table 2. The values of iz interpolated to the release height from
Tower A data are generally appropriate, although the lower values of
iz which approach 0% appear consistently too small. Whether or not
the response of the instruments can detect values that approach 0% is
not known. The lowest photo-inferred value of iz is 1%.
Wider variability in the lidar estimates are to be expected
because the lidar estimates for each 5-minute period are obtained from
one to two instantaneous scans of the plume. The lidar estimates tend
to be biased toward under-representing the magnitude of the 5-minute
iz data. The lidar estimates corroborate the large photo-inferred
values of iz (>10%) for Experiments 204 and 209. The estimates of
iz differ by less than a factor of two. Conversely, the lidar
estimates contradict the large photo-inferred values of iz observed
during Experiment 213. The photo observations are more than a factor
of three times the lidar estimates of i~.
54
-------
rH U
rH 0)
O
JJ
£ ti
0) rH
I
J3
JJ
•H *rl C 0 *H
IB O £ > J
$i
O P _
8 I s I
S* H
-;• ft
—« a
O -H §
* > 0
O rH JJ
"3 1
0 t4 B
ffg 5
I
5
o
0
I
to
s
I
I & I?
JS JJ JJ
I O •
a*«v
•H o
(8 r-l
15
:s
u
j
0»-«
S5
H JJ
o
en
o
in
o a>
to m
O O
en en
O
en
O to 00 00
•
-------
Si
i
M
N*
N
Kl
c~
- - -'*' *•
5
*..
« R ?
CD
I
-1
"«
% I
8
a
8
I
«*
1
M
-t *
\
I
» i
Figure 16a. Time series of HDA vertical turbulent intensities at oil-fog
height ( ), photo estimates (-p-), and lidar estimates (L).
56
-------
^
i sr
I
y
\
—rll
J'
' (O
u»
<.__
M(0
• « 10
(M
I
M
OB (fl
N
K 7
Figure 16b. Time series of MDA vertical turbulent intensities at oil-fog
height ( ), photo estimates (-p-), and lidar estimates (L).
57
-------
•*4
I
„_":::: i :===•»
HIM
I
j
U)
*.-
•~—•-. _—
•>. -i
7
8 a
(0
U> •
."S-
.
»*L -J
u>
u>
-------
33 9 9 8
(A
_^ I
NO)
' CO
S K 8
s ~
U)
N
$" ft.
N «
a>
Figure 16d. Time series of MDA vertical turbulent intensities at oil-fog
height ( ), photo estimates (-p-), and lidar estimates (L)
59
-------
Is (X)
Sit
e te
lM (Hour)
Figure 16e. Time series of MDA vertical turbulent intensities at oil-fog
height (—), photo estimates <-p-), and lidar estimates (L).
60
-------
TABLE 2. PERIODS OF SIGNIFICANT DISCREPANCY BETWEEN
iz ESTIMATES MADE ON THE BASIS OF TOWER A
DATA AND
VALUES INFERRED FROM PHOTOGRAPHS
Exp.
201
202
203
206*
207
209
210*
211*
213
214
215
Ri
Time 1
2100-2200
1700-1800
2200-2300
500-600
500-600
350-400
1755,1810,1900
1905-2000
2400-2500
100-200
410-415
455
125
300-400
305-320
335, 345
100-200
alease
It (m) R
30
50
40
30
35
30
40
40
.30
57
58
20
40
50/30/40
47
47
57
eleai
NW
NW
NW
SB
SE
SE
NW
NW
SE
SE
SE
SE
NW
SE
SE
SE
SE
Probable Cause of Discrepancy
Time Shift/interpolation
\
Instantaneous photos
MDA iz approaches 0%
(Photo iz agrees with 10-m Tourer A
iz with a time shift)
Horizontal plume meander/faint plume
image
Missing 40-m data
Unknown
MDA iz approaches 0%
Unknown
Interpolation (iz at 10m is 2 to 5
times
larger than iz at 40m)
Missing 40-m data
MDA iz approaches 0%
(Ph(*to iz agrees with 2-m and 30-m
Towe'r B
iz and 10-m Tower C iz)
MDA iz approaches 0%
(Photo iz agrees with 2-m Tower D
MDA iz approaches 0%
Interpolation
Tower A not representative
(Photo iz agrees with 10-m Tower C,
D, and F values of iz)
MDA iz approaches 0%
(Photo iz agrees with 10m Tower C
Tower A not representative
(Photo iz agrees with 2-m and 10-m
Tower C iz)
*40-m wind speed not available.
61
-------
Other possible sources of discrepancy are listed below. The
larger values of iz estimated from the photographs often occur
during periods of significant horizontal plume meander and this
horizontal "smearing" could appear as an increase in the vertical
dimension of the plume in the photographs. The larger photo-inferred
values of iz do not agree with the MDA values but are often
corroborated by values of iz measured at the 10-m and 30-m towers on
CCB. Some of the photo-inferred values of iz were obtained from
instantaneous photographs and 10-minute long photographic time
exposures; whereas the MDA values were obtained from 5-minute averaged
Tower A data.
\
3.2 Classification of CCB Experiment Hours
The CCB MDA for SF$ has been partitioned according to the
relationship between the tracer release height above ground and the
one-hour average value of Hc. A "neutral" class contains all hours
in which He is less than or equal to 2 m (the lowest instrumented
level at the 150-m tower). A "very stable" class contains all hours
in which He is considerably greater than the release height. For
this classification, Hc values nearly equal to or greater than
zr + 10 m are included. The 10-m value is chosen so that the bulk
of a typical stable plume would be below Hc. A third class contains
all hours in which Hc is nearly equal to zr, i.e., Hc
approximately equals zr + 10 m. This class includes the hour in
which the greatest scaled concentration (x/Q) was observed. The
fourth class contains all remaining hours.
Of the 111 hours in the SFg MDA, 22 are in the class
Hc < 2 m; 29 are in the class Hc £ zr + 10 m; 26 are in the
class Hc » zr + 10 m; and 34 are in the remaining class. In the
modeling presented in the following sections, we are primarily
interested in evaluating model performance for those hours in which a
reasonably good concentration pattern was observed. Hours with near
zero concentrations at all but a few samplers riear the base of the
hill (to one side or other with respect to the release position) may
be "modelable," but are not as important from the point of view of
developing and testing a model for stable plume impingement
conditions. When these hours are removed, 21 hours remain in the
class Hc < 2 m; 20 remain in the class Hc > zr + 10 m; 20 remain
in the class Hc =» zr ± 10; and 26 remain in the fourth class.
Each of these hours has been reviewed to see to what extent the
MDA data reflect the wind directions (WD) and intensities of vertical
turbulence (iz) inferred from smoke photographs and lidar data, as
well as the temporal and spatial concentration patterns revealed by
the 10-minute average concentration data. The MDA was changed
Whenever possible when significant discrepancies were found. In some
cases, an hour had to be removed because the directional wind shear
was too great between measurement heights on the tower to adequately
reconcile the MDA and the smoke plume appearance. Furthermore, a few
additional hours were removed because the Tower A data were thought to
62
-------
be unrepresentative of the release location, or because the observed
concentrations appeared to make no "sense." Tables 3 through 6
summarize several meteorological variables of each hour remaining, and
indicate which hours were modeled with the modified MOA.
The class boundaries prescribed above were obtained somewhat
arbitrarily. In particular, the choice of a 10-m zone above and below
the release height in defining what is meant by "zr approximately
equal to Hc" was guided mainly by the vertical size of a
representative stable plume. However, for the range of release
heights in the CCB data base, this choice of class boundaries is
nearly equivalent to the statements:
"Very Stable": Hc/zr > 1.25
"zr nearly equals Hc": 1.25 > Hc/zr > 1/1.25
"Moderately Stable": 1/1.25 > Hc/zr > 0
"Weakly Stratified": Hc = 0
where Hc = 2 m is considered to be equivalent to Hc =» 0. This
form of classification could provide a better indication of what
streamline patterns may be expected in each class.
3.3 Model Performance
The performance of CTDM is prescribed in an absolute sense by
presenting the statistics of the residuals of the observed and modeled
concentrations. These statistics are compiled for the concentration
fields grouped by experiment hour, grouped by the four classes
discussed above, and grouped together for the entire 80-hour SFg
data base selected for this evaluation. The specific statistical
measures are discussed in subsequent subsections.
The performance of CTDM is prescribed in a relative sense by
comparing its residual statistics with those from a "flat terrain"
version of the model, and a version that incorporates the plume path
assumptions of COMPLEX I/II. This type of comparison places the
performance of CTDM in perspective by contrasting it with that of two
well-known modeling approaches. The relative performance of these
three models in each of the four classes will indicate when terrain
effects are most important, and when the simpler modeling algorithms
are just as good or better than the more complex algorithms of CTDM.
The performance of CTDM is also evaluated for three alternative
sets of meteorological data which reflect varying quantities of
on-site meteorological information.
3.3.1 Absolute and Relative Performance
CTDM
Several parameters need to be specified before CTDM can produce
concentration estimates. These are
63
-------
TABLE 3. SF, TRACER HOURS AT COB WITH H < 2 m
6 C ~
Exp-hr
201-1
202-1
202-3
202-4
208-3
208-4
214-8*
217-1
217-3
217-5*
217-6*
217-7*
217-8*
218-2
218-3
218-4
218-5*
218-6*
218-7*
218-8*
Time Ending
18
18
20
21
20
21
10
3
5
7
8
9
10
4
5
6
7
8
9
10
u (m/s)
6.8
10.3
7.7
7.0
9.1
8.4
2.4
4.8
5.4
7.2
7.2
6.7
5.8
6.7
8.7
8.1
7.2
7.9
6.9
8.5
2r (m)
30
50
20
20
30
30
24
30
25
40
40
40
40
30
30
30
30
30
15
15
Hc (i
0
0
0
1
0
0
0
2
2
0
0
0
0
0
0
0
0
0
0
* 0
Total Number of Hours: 20
m) MDA Modifications
z
WD
WO
*0il-fog and SF$ were not released at the same height for all or part
of the hour.
64
-------
TABLE 4. SF, TRACER HOURS AT CCB WITH H > z +10 m
6 c — r
Exp-hr Time Ending u Cm/a) zr (m) Hc (m) MPA Modifications
204-1
204-2
204-5
204-6
204-7
204-8
205-7
213-5*
213-6*
213-7*
213-8*
214-2*
214-3*
215-4
215-6*
1
2
5
6
7
8
7
5
6
7
8
4
5
4
6
2.2
1.1
1.2
1.7
1.0
0.5
1.7
1.2
0.7
1.1
1.0
1.2
1.6
2.3
1.4
30
30
30
30
30
30
30
30
30
30
30
17
24
30
30
61
81
44
39**
45
52
41
42
49
44
40
31
43
37**
42
WD
Total Number of Hours: 15
*0il-fog and SFg were not released at the same height for all or part
of the hour.
**Two hours are included in this class even though they do not meet the
criterion Hc ^ zr + 10 m.
These are included because: (1) This class has the fewest members.
(2) Hc lies substantially above the plume
elevation.
65
-------
TABLE 5. SF. TRACER HOURS AT COB WITH H - z i 10 m
D c r
Exp-hr Time Ending u (m/s) Zr> (m) H/. (m) MDA Modifications
u (m/s)
3.8
3.7
3.2
3.0
2.5
3.4
2.8
1.8
2.4
1.8
2.2
2.3
3.0
2.5
2.3
3.4
2.8
3.2
2.8
3.5
Sr-iffil
30
30
30
30
30
50
30
35
35
35
40
40
30
30
20
20
20
24
30
30
Hc-fc
23
26
31
33
31
40
30
37
31
37
37
45
27
33
25
20
21
23
23
*26
201-4 21 3.8 30 23 WD,
201-5 22 3.7 30 26 WD,
201-6 23
203-2 2 3.0 30 33 WD
203-3 3 2.5 30 31 WD
203-8 8
205-8 8 2.8 30 30 WD
206-6 6 1.8 35 37 WD, i.
206-7 7
206-8 8 1.8 35 37 WD
209-3 20
209-8* 25
211-2 2
211-3 3
211-5* 5
211-6* 6 3.4 20 20 WD
211-7* 7
214-4* 6
215-1* 1
215-5* 5
Total Number of Hours: 20
*0il-fog and SFg were not released at the same height for all or part
of the hour.
-------
TABLE 6. SF, TRACER HOURS AT CCB WITH 2 m < H < z -10 m
o c r
Exp-hr Time Ending u (m/s)
201-2 19 5.4
202-5 22 7.6
202-6 23 9.3
205-3 3 7.0
205-4 4 6.2
205-6 6 6.2
206-3 3 6.7
206-4 4 4.5
206-5 5 4.1
207-2 2 3.6
207-4 4 3.0
209-1 18 4.4
209-2 19 3.1
210-2 2 4.8
210-3 3 6.4
210-4 4 6.8
210-6 6 7.2
210-7 7 7.4
214-5* 7 4.1
214-7* 9 2.7
216-1* 1 3.9
216-2 2 2.7
216-3* 3 3.1
216-4* 4 4.0
217-2 4 5.1
Total Number of Hours: 25
. (a)
i ™«^""""»
30
30
40
40
40
30
46
35
35
30
30
40
40
57
57
57
58
58
24
24
30
30
30
30
30
Sc-Iffil
8
7
7
8
15
12
16
20
21
13
13
3
17
35
26
22
9
15
12
* 11
5
11
15
8
4
MDA Modifications
WD, i
z
i
z
<
z
-
-
-
WD, i
z
WD
-
-
-
WD, i
z
WD
WD, i
Z
WD, i
z
-
WD
WD
-
-
-
-
-
-
*Oil-fog and SFg were not released at the same height for all or part
of the hour.
67
-------
1. turbulence modification factors:
2. extreme flow modification factors: T^o(zr), T|,o(zr),
Tuo(zr)
3. flow relaxation length scale factor: a
4. LIFT-WRAP transition length scale factor: 6
Guidance for specifying each of these may come from both laboratory
and theoretical studies. They may also be inferred from field studies
such as SHIS #1 and #2 and the FSPS. In this evaluation of CTDM, only
the flow modification factors are taken from theory. The turbulence
modification factors are assumed to be unity because the data obtained
at the 10-m and 30-m towers on CCB do not show substantial variations
in the turbulence intensity compared to that measured on Tower A. An
increase in Oy is observed, but this is generally accompanied by
an increase in the flow speed as well, so that ic remains about the
same. Note that these measurements were made on towers at terrain
elevations greater than 70 m, so that the structure of turbulence
closer to the base of the hill is not documented.
The flow relaxation length scale factor, a, is assumed to be
1/1.5. This factor assures that the flow modification at the base of
the hill is about one tenth of its extreme value at the crest. The
LIFT-WRAP transition length scale factor, 6, is assumed to be 0.1,
so that the e-folding scale of the transition region is 0.1 Hc.
Flow modification factors are obtained from flow deformations
estimated from potential flow theory. In the case of the LIFT
component, CCB is assumed to be an ellipsoid of revolution about the
vertical axis, with a half-length equal to 2.7 times its height. The
streamline height modification factor T^Q evaluated at the crest for
plume heights equal to .2H, .4H, and .6H equal .47, .51, and .54,
respectively. The corresponding factors T§,0 and Tuo are 1.73,
1.61, 1.54 and 1.23, 1.22, 1.20, respectively. Because the majority
of the plume heights at CCB are between ,3H andt.6H, representative
values are used for all LIFT calculations; T^o » .5, TIO = 1.6,
Tuo » 1.25. In the case of the WRAP component, Tno is unity.
Because the flow is considered to be two-dimensional in the x-y plane,
Tlo • <5i and Tuo * 2.0. These values are consistent with flow
about a two-dimensional circular cylinder.
FLAT
Flat-terrain concentration estimates are obtained by replacing
the terrain-modified portions of the CTDM code with the
straightforward Gaussian plume solution. Therefore, calculations of
Oy and oz are done the same way as in LIFT and WRAP, as are
calculations of all other quantities that are not influenced by the
terrain. Ground-level concentrations are then estimated 'by
nt\
(.'2)
68
-------
where Oy^ is the horizontal spread of the plume due to both the
'filament' and 'meander' portions of the crosswind turbulence
statistics, as done in CTDM. zr is the plume centerline height, and
yr is the lateral position of the release point in a coordinate
system with origin at the hill center, and x-axis aligned in the
direction of the mean flow. The quantity yg is the lateral position
of the receptor in the same coordinate system.
CMPLX
Plume path modifications contained in the COMPLEX I/II codes are
tested within the context of CTDM by introducing the partial height
factor Tp into the vertical distribution factor of Equation 72 so
that
-0.5 zr
(74)
Tp = 1 - (1 - PPC) ZR/zr ZR < zr
In COMPLEX I/II, the recommended value of PPC varies by stability
class. PPC equals 0.5 ("half-height") for stability classes A, B, C,
and D; and PPC equals 0.0 for classes E and F. However, the product
TpZr is allowed to be no smaller than 10 m , the minimum "stand
off" distance.
Performance Results *
The performance of each model is summarized in Table 7 .
Geometric means and standard deviations are computed so that the
degree of model over- or underestimation is readily apparent.
Specifically, the geometric means and standard deviations are computed
using N-weighting as
m = exp (In C /C )
8 ° P (75)
2
O .5
s - exp ((In C /C )2 - In C /C )
g op op
A problem with statistics generated from the ratio Co/Cp is
that zero values of C0 or Cp must be excluded. This problem is
approached in two ways. First, the statistics are computed only for
69
-------
TABLE 7. MODEL PERFORMANCE STATISTICS FOR 80 HOURS OF SF, DATA AT CCB
Top 5
(Time-and-space-
paired)
Heut:
Weak:
Impg:
Stab:
Comb:
avg.
all
avg.
all
avg.
all
avg,
all
avg.
all
o
1.46
1.47
1.38
1.32
0.53
0.52
0.40
0.40
0.85
0.84
e
2.0
4.5
3.5
10.0
2.8
7.6
1.9
9.1
3.0
8.2
0
0
6
6
1
1
0
0
7
7
e
2.31
2.48
3.62
3.22
9.61
9.99
3.06
2.71
3.47
3.53
o
2.8
8.3
4.1
23.6
10.7
69.5
11.7
607
6.4
63.5
0
0
15
15
11
11
3
3
29
29
e
0.73
0.75
0.92
0.85
2.40
2.10
0.42
0.35
0.95
0.90
O
2.2
8.2
7.0
55.8
34.5
26271
5.1
61.1
10.1
426
0
0
0
0
0
0
0
0
0
0
Top 5 (Time-paired)
Neut:
Weak:
Impg:
Stab:
Comb:
Notes:
peak
avg.
all
peak
avg.
all
peak
avg.
all
peak
avg.
all
peak
avg.
all
1.57
1.45
1.45
1.37
1.20
1.20
0.63
0.53
0.53
0.67
0.46
0.46
1.00
0.84
0.84
2.0
1.9
2.0
2.5
3.1
3.2
2.5
2.4
2.5
1.8
1.8
1.9
2.5
2.7
2.8
0
0
0
5
5
5
0
0
0
0
0
0
5
5
5
2.06
1.80
1.80
3.21
2.56
2.56
5.91
5.18
5.18
1.71
1.25
1.25
2.64
2.18
2.18
2.2
2.3
2.3
2.1
2.7
2.8
1.7
2.0
*2.1
2.4
2.2
2.4
2.4
2.6
2.7
0
0
0
14
14
14
10
10
10
3
3
3
27
27
27
0.71
0.63
0.63
0.48
0.40
0.40
0.55
0.43
0.43
0.31
0.24
0.24
0.51
0.42
0.42
1.7
1.8
1.9
2.6
3.0
3.1
2.6
2.6
2.6
2.3
2.5
2.7
2.4
2.7
2.8
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
# denotes number of hours excluded because the mean of the five largest modeled
concentrations was less than
70
-------
the top "N" observed or modeled concentrations to quantify model
performance for the region(s) of greatest observed or modeled impact.
And second, hours in which the mean of the top "N" modeled
concentrations is less than 1 ysec/m^ are removed. The 1
psec/m3 threshold is chosen because it provides a convenient flag
for cases in Which the modeled plume impact is virtually
insignificant. In Table 7, the number of hours that fall into this
category is indicated in the data column headed by the # symbol.
Statistics are computed for the following classes:
Neut: Hc < 2 ro ("neutral")
Weak: 2 ra < Hc < zr - 10 m ("weakly stratified")
Impg: zr - 10 m < Hc < zr + 10 m ("impingement")
Stab: zr + 10 m < Hc ("very stable")
They are also computed for two pairing possibilities. The top
"N" time-paired statistics are computed from N pairs which represent
the N largest C0 values, and the N largest Cp values. For
example, the largest Co value for the hour is paired with the
largest Cp value for the hour; second largest C0 with second
largest Cp; etc., to the Nth largest values. These pairs are not
required to be paired in space, but are paired in time. The top "N"
time-and-space-paired statistics are computed from as many as 2N pairs
which represent the N largest Co values and their corresponding Cp
values, plus the N largest Cp values and their corresponding Co
values. Each pair is included only once. Thus the total number of
pairs may be less than 2N, but not less than N. If the observed and
modeled concentration patterns overlap well, then the number of pairs
should be nearer N than 2N. Consequently, the time-paired statistics
characterize how well the larger observed and modeled concentrations
agree without regard for where these values occur, while the
time-and-space-paired statistics characterize how well the region of
greatest impact is modeled. k
Within the time-and-space-paired subset, residual statistics are
based on all of those individual concentration pairs, and statistics
are also based on the mean C0 and Cp from these pairs for each of
the 80 experiment-hours. The "avg. label denotes those statistics
for the 80 (less the number of hours in the # column) hourly geometric
mean residuals, and the "all" label denotes those statistics for all
individual pairs in this subset. Within the time-paired subset, the
"avg." and "all" labels have the same meaning, and a third label
(peak) is added to denote the statistics of the residuals of the
hourly peak observed and modeled concentrations.
The number "N" chosen to evaluate model performance for these 80
hourfe is 5. Although arbitrary, this selection was guided by the need
to avoid averaging over a region large compared to the peak impact
region, as well as the need to avoid limiting the pairs to the extreme
values. With a total sampler number of about 100, the paired
statistics cover as many as 10% of the total sampler positions.
71
-------
Because the plume impact could at times be limited to roughly 20 to 25
of the samplers, the value of 5 is thought to be a reasonable choice
for "N." The sensitivity of the comparison statistics to "M" has not
been tested.
CTDM generally performs better than the other two "models." In
the case of the time-paired statistics, FLAT underestimates observed
concentrations in the mean, While CTDM and CMPLX tend to overestimate
these concentrations. The noise (sg) is similar for each model, but
note that 5 hours were "missed" by CTDM, and 27 hours were "missed" by
FLAT. These hours are not included in the statistics. When broken
out by meteorology, CTDM tends to overestimate concentrations for the
"impingement" and "very stable" classes (Hc >. zr - 10 m). On the
other hand, FLAT tends to underestimate in all classes, and CMPLX
overestimates in all classes.
The time-and-space paired statistics generally indicate a greater
degree of noise for each of the models. The increase in the noise is
indicative of differences in the spatial distribution of the
concentrations. It is evident that CTDM fares far better in the s~
statistics for the concentrations paired in time and space than does
either FLAT of CMPLX. Hence, we would expect the CTDM estimate of the
distribution of concentrations to be a better representation of the
observed distribution then either the FLAT or CMPLX distributions.
Both the time-and-space-paired and time-paired statistics
indicate that CTDM performs the worst when 2 m < Hc <. zr - 10 m,
what is termed the weakly stratified class. The number of "misses" in
this class indicates that too little plume material reaches the
surface in several of the hours in this class. Unlike the more
neutral class, plumes for many of these hours may be quite narrow in
the vertical. Hence, the concentrations may be particularly sensitive
to estimates of Hc, iz, and possibly the specification of the
influence of the hill on the turbulence in the vicinity of the hill.
The CMPLX results for this class include no "milses", while the FLAT
results show many more. It may be possible to model this class better
with alternate choices of several of the "free" parameters in CTDM.
The FLAT results are of interest in themselves. In answer to the
query: when is a flat-terrain calculation at CCB as good as CTDM?, we
can answer: never. However, among the FLAT simulations the "neutral"
class is clearly modeled best. This is reasonable in that one would
expect the hill effect to be weakest when Hc is virtually zero, and
(presumably) when the wind speed is greatest so that the boundary
layer shear zone encompasses both the hill height and the plume. FLAT
tends to underestimate concentrations to a greater extent As the flow
becomes more stably stratified. But When the most stable class is
reached, FLAT tends to do better again in estimating the magnitude of
the impact, although the spatial distribution is not reproduced well.
The success of FLAT in |^f time-paired results for the "stable" class
would appear to result from deflecting the plume too much in the
vertical, but not enough in the horizontal. That is, FLAT seems to
benefit from compensating errors.
72
-------
With a 10-m minimum "standoff distance" for the most stable
flows, CMPLX does not deflect the plume "too much" in the vertical,
and it does not deflect it at all in the horizontal. As a result, the
larger concentrations are overestimated to the greatest degree in the
"stable" class.
Taken together, these results indicate that the structure of CTOM
is necessary to overcome deficiencies in both the flat-terrain
algorithm and the terrain algorithm contained in COMPLEX I/II. The
importance of both the critical dividing-streamline height and the
stagnation streamline can be seen in the modeling results.
Furthermore, the present choices of the "free" parameters in CTDM
appear to be reasonable, but more work is needed to understand the
reasons for the poorer model performance in the "weakly stratified"
class. Also, the trend towards underestimating the "neutral" class
hours by about 50%, while overestimating the "impingement" and
"stable" class hours by a factor of two should be explored. Changes
to the turbulence over the hill and possibly just upwind of the hill
may account for some of the discrepancy. Also, the variation of wind
speed, direction, and turbulence with height may be an important
factor during some of these hours.
3.3.2 Model Performance for Alternate Meteorological Data
The sensitivity of CTDM performance measures to the input
meteorological data has been tested by forming 1-hour concentration
fields from simulations with the 5-minute sequence of meteorological
data in the MDA, by constructing the 1-hour wind and temperature
structure from data at 10 m and 150 m, and by constructing the data
from 10-m measurements alone. In each case, however, the wind
direction is taken from the MDA—we already know how sensitive the
concentration pattern is to the wind direction.
The 10 m - 150 m construction is analogous^ to assuming that a
site has a tower with wind and temperature measurements at two
elevations. No turbulence data is presumed to be measured. A
stability class is prescribed on the basis of the 5-minute sequence of
10-m wind speeds, and net radiation data (the stability class is
already contained in the MDA, and its computation is described in the
CTDM Third Milestone Report). The Froude number between 2 m and 150 m
is contained in the MDA, and this is used to obtain Hc by the
relation:
H = (1 - Fr) H (Fr < 1.0) (76)
c —
H « 0 (Fr > 1.0)
c
where H is the height of the hill. For this evaluation, we have also
assumed the MDA wind speed at release height is similar to what would
have been interpolated between 10 m and 150 m, so that N is obtained
from the relation:
<77)
73
-------
The bulk Froude number for the layer above Hc is taken to be the
greater of 1.0 or Fr.
Turbulence intensities are inferred on the basis of the stability
class for each five minutes. The oy and oz relations proposed
by Briggs (1973) contain turbulence intensities iy and iz that are
consistent with the PG sigma curves:
iy = .22 iz = .20 (Class A)
iy = .16 iz = .12 (Class B)
iy = .11 iz = .08 (Class C)
iy = .08 iz = .06 (Class D)
iy = .06 iz = .03 (Class E)
iy = .04 iz = .016 (Class F)
These intensities are used in place of the intensities contained in
the MDA.
The data set based on measurements at 10 m contains "default"
wind and temperature profiles. As is the practice in many regulatory
models, the temperature lapse rate values depend on the stability
class, and the change in wind speed with height follows a power law
profile, with the power exponent tied to the stability class.
The wind speed at release height is estimated as
u = u10 (zr/10)P (78)
Where the exponent depends on the stability class in the following way:
p 3 .10 (Class A)
p = .15 (Class B)
p = .20 (Class C)
p = .25 (Class D)
p = .30 (Class E)
p = .30 (Class F)
The wind direction is taken directly from the MDA. The potential
temperature gradient also varies by stability class:
de/dz = 0 °K/m (Classes A-D)
de/dz =0.02 °K/m (Class E)
de/dz = 0.035 °K/m (Class F)
With these values of the potential temperature gradient, the
Brunt-Vaisala frequency is computed from
N = ((g/T) de/dz)0-5 (79)
-------
where g is the acceleration due to gravity, and T is the temperature
interpolated to release height in the MDA (°K). For consistency, T
should have been the temperature at 10 m, but this value is not
contained in the MDA.
The bulk Froude number is obtained from this estimate of N, the
hill height H, and the wind speed at one half the hill height
(~ 50 m)
Fr = u50/(N H) (80)
where U$Q is obtained from the power law profile. Now the critical
dividing-streamline height is again given by Equation 76, and the bulk
Froude number above Hc is the greater of 1.0 and Fr. The turbulence
intensities are obtained from the Briggs <*» and az relations,
as described above.
Table 8 contains the performance statistics for CTDM as driven by
each of the three meteorological data sets just described.
Modeling performed with the sequence of 5-minute meteorological
data has led to an improvement in model performance in none of the
classes. The number of "misses" in the "weakly stratified" class
drops, but additional hours are now "missed" in the "stable" class.
Furthermore, although the overall bias lies closer to unity, the noise
has increased slightly. Therefore, although the 5-minute modeling has
helped a few hours, it tends to degrade model performance in general.
Modeling performed with the 10 m - 150 m "tower" data is
substantially inferior. A total of 39 out of 80 hours is "missed."
For the hours remaining, the modeled spatial distribution of
concentrations is generally quite poor. The best performance is seen
in the "neutral" class, and the worst is seen in the "weakly
stratified" class on the basis of the time-paired statistics.
Apparently, the use of the bulk Froude number between 2 m and 150 m
underestimates the stability of the flow, and therefore Hc as well.
Modeling performed with the 10-m data alone is better in that far
fewer (24) hours are "missed." But the statistics for the remaining
hours are still quite poor. The time-and-space-paired statistics
indicate that modeled and observed concentration patterns are
generally dissimilar. Once again, however, model performance is best
for the "neutral" class.
These results show that the resolution of the meteorological data
in the vertical is a key factor in modeling COB successfully. This is
probably due to the relatively narrow plumes at CCB in combination
with the complex vertical structure of the lower atmosphere. The
model must know what is going on at plume height.
75
-------
TABLE 8. EFFECT OF METEOROLOGICAL DATA RESOLUTION ON CTDM MODEL
PERFORMANCE FOR 80 HOURS OF SF, DATA AT CCB
D
Top 5
(Time-and-space-
paired)
5-minute Met.
10 m - 150 m Met.
Neut:
Weak:
Impg:
Stab:
Comb:
avg.
all
avg.
all
avg.
all
avg.
all
avg.
all
e>
1.62
1.71
2.39
2.09
0.54
0.56
0.51
0.50
1.12
1.08
e
2.0
5.0
7.4
31.1
2.5
8.5
1.9
6.1
4.2
12.4
0
0
4
4
1
1
3
3
8
8
o
1.70
1.67
4.82
5.19
89.2
85.1
16274
17341
9.97
11.70
e
2.2
4.4
2.4
20.1
834
874011
4.5
*.*
67.0
8270
0
0
19
19
12
12
10
10
41
41
D
1.54
1.57
374
777
312
353
99542
108302
75.5
110.2
o
1.9
4.3
36621
*.*
336
*.*
7.0
*.*
1640
786052
2
2
8
8
5
5
11
11
26
26
Top 5 (Time-paired)
Neut:
peak
avg.
all
1.74
1.68
1.68
2.3
2.2
2.3
0
0
0
Weak:
peak
avg.
all
2.98
3.49
3.49
4.4
6.5
7.1
17
17
17
Impg:
peak
avg.
all
0.32
0.38
0.38
1.6 12
1.9 12
2.2 12
Stab:
peak
avg.
all
0.39
0.78
0.78
2.4 10
4.7 10
5.2 10
Comb:
paak
avg.
all
22
35
1.35
3.5 39
3.9 39
4.2 39
Notes: # denotes number of hours excluded because the mean of the five largest modele
concentrations was less than 1 ps/m-*.
*.* denotes a number greater than 999,999.
76
-------
SECTION 4
MODELING HOGBACK RIDGE
4.1 Refinement of Tower Data
Four meteorological towers were instrumented for SHIS #2 at the
Hogback Ridge site. The 150-m tower (Tower A) was erected about 800 in
east of the crest of the ridge and instrumented at 10 levels; the 30-m
tower (Tower B) was among the hillocks at the base of the ridge on the
east and was instrumented at five levels; a 10-m tower (Tower C) was
located on the crest of the ridge and was instrumented at three
levels; and an existing 60-m tower (Tower P) approximately 4 km east
of the ridge was instrumented at two levels. The signal conditioning
electronics for the 150-m and 60-m towers were enclosed in
aluminum-clad, environmentally controlled, insulated shelters; those
for the 10- and 30-m towers were in naturally ventilated steel
enclosures.
The instruments on all the towers were scanned once per second by
the data acquisition system. Data from the 150-m tower's shelter were
communicated over shielded signal cables to the bus containing the
data acquisition computer, a distance of about 50 m. Radio
communications links telemetered the data from the other three
towers. Further details regarding this data system are reported in
the Third Milestone Report.
At least three different sorts of noise were observed in the
1-sec data during the SHIS #2 experiments—large "hits," which drove
the instrument output voltages outside their O-fbo-5 VDC range;
"channel-skipping," in which the data from one 'input channel was
skipped and replaced by the data from the next sequentially polled
channel, with the shift of data continuing to the end of the
16-channel multiplexor; and "high-frequency" noise bursts that caused
a few seconds of data to oscillate unrealistically at consistent
periods within each 5-minute averaging period. These types of noise
are described more fully in the Third Milestone Report.
The large hits are generally easy to identify and remove from the
data; the other two types of noise are less so. ARLFRD developed a
"filtering" routine that examined the second-to-second changes in
instrument output and replaced values that exceeded what they regarded
as reasonable limits for such changes; these limits are shown in
Table 9. Data removed from the time series by this filtering
procedure are replaced by linear interpolation in time between the
last good value and the next good value.
An interim tower data base was produced by ARLFRD in September
1983 in order that the modelers could begin analyzing the Hogback
77
-------
data. The 5-min and 1-hr values were produced from 1-sec data edited
by the filtering and "filling" routine described above. The
anemometer data were also treated by calibrations specific to each
instrument that had been made by ARLFRD in their wind tunnel prior to
installation at the Hogback. The UVW propeller anemometer data were
modified at the 1-sec level by cosine-response corrections developed
by John Clarke of EPA from wind tunnel experiments on propeller
systems manufactured by R.M. Young Co.
Examination of the data in the interim data base suggests there
are consistent errors remaining that might be removed by further or
different processing. Examples are shown in Figures 17 to 22 for data
from the 40-m level of Tower A.
Figure 17 shows the difference between the scalar mean wind
direction from the F460 vane (DX) and that from the UVW propeller
anemometers (D) as a function of vane direction for all 5-min average
data from SHIS #2. Although the data are somewhat scattered,
particularly for light winds, clear trends are evident, and the
differences are often larger than those observed in the SHIS #1 data
before they were corrected, exceeding 25° with winds from the south.
Figure 18 is a plot of the differences between cup-derived wind
speeds (SX) and prop-derived wind speeds (S) expressed as a fraction
of the cup speed and plotted as a function of vane direction. Again,
some trends can be seen through the scatter, and the magnitude of the
differences is like that of the uncorrected SHIS #1 data.
The 5-min
-------
+ fc
B +
« *+ *
I
2
2,
•rl
fe
Xa
79
-------
(0
w
0)
a
w
1
•a
4)
4)
•o
o.
3
§
4)
-p
4) •
Xi w
•o
4) 4)
U 4)
SO.
M
4) -O
•O
•O
4) 4)
Li
>M 4)
O "O
•u a.
o o
^H Ll
d. CU
00
l-l
4)
3
xs/(s - xs)
80
-------
I
•
(S/UI)
81
-------
o
0)
p.
2 .
O, -
S g
0.5
•H U
c >A
O v^
0) Ul
-U U
3 0)
C -t>
•i-l Q)
4) 3
> 4i
•H C
fa td
a
•I-t
82
-------
•8
o
LI
4)
*4.
•o e
O lj
•H (1)
VI W
0) 00
.c e
-P -H
Ll
e 3
O T>
Li
01 I
O ^
(1) tfl
9
-------
0
p
CS1
CM
(8/UI)
-------
some of it may not be recognizable, further attempts will be made at
signal-processing to improve data quality. A first step might be to
reduce the allowable second-to-second changes (See Table 9, p. 94) to
capture more noisy values and cases of channel-skipping. The noise
remaining in the present interim data probably affects variances and
covariances more than the means, but removing some of it will also
reduce the scatter in Figures 17 and 18.
4.1.1 Wind Tunnel Studies
The Third Milestone Report recommended that further wind tunnel
studies be performed at the EPA Fluid Modeling Facility (FMF) to
characterize the response of the Climatronics UVW propeller
anemometers more completely than had earlier experiments done by ERT
in the wind tunnel at Colorado State University (CSU) in 1981, from
which the corrections for the SHIS #1 data were largely derived. The
experiments were performed by ERT with the cooperation of FMF staff in
June 1983. Primary objectives of the experiments were to determine
not only the non-cosine response corrections but also the calibration
lines for 0° angle of attack.
Calibrations of UVW Propeller Anemometers
The calibrations of the propeller anemometers done by ARLFRD in
summer 1982 showed a nonlinearity of response at low wind speeds when
plotted against the estimated tunnel speed but not when plotted
against the "counts" data from the output of their standard cup
anemometer. ERT's calibrations of the propeller sets in the CSU
tunnel did not have this characteristic, but they had not been well
defined at light winds. Since the SHIS #2 data have been processed
with ARLFRD's calibrations, and since most of the SHIS #2 data
indicate light winds, it was important to try to verify the
nonlinearity in another facility.
»
The six component arms of two UVW sets were calibrated in the
large wind tunnel at the FMF, the test section of which is 2.1 m high
and 3.7m wide. Air speed during the calibrations was varied from
approximately 0.5 to 8.0 m/s, the maximum speed possible in the
tunnel. Fan tachometer settings were used to determine air speed
after linear regression relationships were derived between tachometer
settings and pitot tube readings for speeds over about 1.7 m/s and
from smoke puff tracking experiments for lower speeds. The
calibration of tunnel air speed vs. fan speed is based on the
following premises:
• Air speed for a given fan speed is independent of air
density.
• Pitot tube readings, when corrected for pressure and
temperature (i.e., density), give adequate air speeds above
about 2 m/s.
• Timing the passage of smoke puffs yields adequate air speeds
at low fan speeds.
85
-------
Separate calibration lines were derived for fan tachometer
settings between 59 and 190 and for settings in excess of 190. The
lower speed calibration was derived from data on passage of smoke
puffs and the higher speed calibration from pitot tube pressure
differences corrected for ambient temperature and pressure. Although
the smoke puff experiments included data from tachometer settings of
20 to 40, these data were not used because no evaluations of the
propeller anemometers were done at such low tunnel settings. The
results of the linear regressions of air speed on tachometer settings
are as follows:
UT (smoke) = .009577 * tach - .13129 for tach < 190
Coefficient of determination R2 = 0.99963, N=5
(Each of 5 data points is the average from 5 or 6
smoke puffs)
UT (AP) = .009854 * tach - .12607 for tach > 190
Coefficient of determination R2 = 0.99990, N=73.
At tach = 190, the smoke derived calibration gives UT = 1.688 m/s
and the AP derived calibration gives Uf = 1.746 m/s. The
difference is consistent with the general observation that FMF tunnel
speeds determined from pitot tube measurements are consistently higher
(by about 4%) than those derived from smoke puffs. The precision of
the smoke puff measurements at 1 m/s is about 1.2 percent, so it would
appear that the pitot tube may overestimate the velocity at the center
of the tunnel. This possibility is consistent with the fact that
lower fractions of estimated tunnel speed were indicated by the
propellers when facing directly into the wind in the FMF tunnel than
in the CSU tunnel, but this evidence is only suggestive.
An analysis was made of the "group" performance of the six
propeller anemometers calibrated in the EPA/FMF. Linear regressions
of best estimates of tunnel speed vs. instrument output voltage were
very similar for the six transmitters, as shown.in Table 10 (p. 95).
When the slopes of the "W'-props* regression lines are multiplied by
2.5 to make them equivalent in voltage vs. speed to the "U" and "V"
props, the range of the slopes is only 10.337 to 10.524 or 1.8%.
Inclusion of the zero-speed points, which were available for the first
three sensors, didn't cause the slopes to change appreciably and in
fact slightly improved the coefficients of determination R2, all of
which lie between 0.99996 and 0.99998. This result implies that there
is very little fall-off from linear response as the ambient speed goes
to zero. There are some minor departures of response from linearity
at low speeds (<1.5 m/s), but some part of this uncertainly is
probably due as much to errors in estimating tunnel speed as to
imprecise instrument response. In any event, the nonlinearities
observed in the ARLFRD tunnel are not apparent in the FMF tunnel.
Refinement of the SHIS #2 propeller data would be much simplified
if the same calibration curve could be used for all the sensors. The
apparent group precision of these six transmitters (and the similar
results for the experiments with propeller systems at Colorado State)
suggests that this should be possible without substantial increase in
uncertainty.
86
-------
Non-Cosine Response Corrections for Propeller Systems
Analysis of the cosine response data obtained in the FMF wind
tunnel with the project's Climatronics UVW propeller anemometers
indicates the data are generally consistent with both the results of
the experiments done for the CTMD program in the Colorado State
University wind tunnel and the results of the tests of R.M. Young UVW
systems done by EPA's John Clarke. The experiments in the FMF were
somewhat more detailed than the CSU experiments in an attempt to
define the local maximum of response at 60° angles of attack and the
response near the stall regions around 90° angle of attack.
Two sets of instruments were studied, but time allowed for
experimentation with the response of only one W-propeller. Responses
were recorded at three tunnel speeds---approximately 0.5, 1.9, and
4.9 m/s. Low speeds were emphasized because most of the CTMD
experiments have been done in fairly light winds.
The results of the FMF experiments are plotted in Figures 23, 24,
and 25, with each plot displaying data for both sets of horizontal
propellers at one wind speed. The "angle of wind" Q in the plots is
oriented for the way the U-props are mounted in the field so that 90°
is a wind blowing directly into the outward face of the propeller.
The V-prop output is plotted vs. 90°-6 to show the general (and
expected) symmetry of the UVW system about the 225°-45<> line. The
principal departure from symmetry is at wind angle 270° (V-prop angle
180°), where the V-prop response is reduced below that of the U-prop
because of the wake of the signal cable connector, which extends to
the west (270°) from the connector block to which all three component
arms join.
The fractions of cosine response shown in the analysis are
determined with the instruments' output when facing "upwind" (wind
angle 90°) taken as the normalizing factor so that the fraction of
response is 1.00 for all component arms when in this orientation.
The results shown in the accompanying figures support the
following generalizations:
• The local increase in response at 60° angles of attack is
consistent from prop to prop.
• The responsiveness of the instruments is nearly independent
of wind speed outside the stall regions.
• The angular width of the stall region is inversely related
to wind speed, as expected.
• The U-prop of the second set of instruments is not like the
other three horizontal props, displaying large
over-responses on either side of the broad stall region.
This characteristic was not recognized during the
87
-------
3SNOdS3a 3NISOO dO NOIlGUdd
88
-------
0) -U
O 01
.H
C -H
O >
o -o
.
(0
a. at
o -o
o. a)
I 01
< o in
PL, XJ 4)
in
01 c -H
P. U
in a)
01 J->
P 0)
a) o
•rl 4)
n c
o a
0
o
^-<
to
k, D
4)
tj
HO
•H
P.
O
O b
P.
•O I •
C > >»
•H Vl
S jJ
• 0)
O *J 1
w >.
-------
01 -t>
o cd
•H
I 01
1u
4)
C *H 00 -P
O 4) C 4J
fcl C -H O
-p C O rH
cd 3 cd o.
O C P. cd
•H O 13
x i u
.0 P. 0)
I
o
VI
C
o
01
0)
u
01 CU M-i
Cd
0) 01
01 C -H (X
C .H
0. b
(D
•a
*J C
(X
C
•H
01 C
o
o
-p
4)
14-t
O -U g
01 X
a) cd 01
^ 4)
60 %
4) C 4) O
r-i «n f:
o o
-p
01 a)
C O. •
O O 01
•ri U ^
4J a g 01 CD
O -O I
ed 3 in S-i o
U > • cd o
tt) o ^5 s CT*
IT)
CM
01
(j
3SNOdS3d 3NIS03 dO NOIlDUdd
90
-------
experiments, and consequently neither the propeller nor the
precision of its mounting on the transmitter shaft was
examined for causes of anomalous behavior.
The responses of the Climatronics instruments are generally
a few percent lower than those derived by Clarke for the
R.M. Young anemometers. The stall regions of the
Climatronics instruments may also be wider; and the
over-response for non-stalled props between 92° and 95° and
between 265° and 268° for the V-prop (similar regions for
the U-prop) seems not to have occurred with the R.M. Young
devices.
The reason for this latter effect may be the propeller shaft
extensions on the Young instruments. Another possible contributor,
however, is a general divergence of flow around the Climatronics UVW
system. A smoke streak in the FMF tunnel aimed into the center of the
vertical W-arm rose quite noticeably as it approached the equipment;
this accounts for the fact that the W-props frequently turned in the
positive (upward) sense when the horizontal arms were being tested for
response. It may also account for the small positive mean vertical
wind speeds often measured by the props on the 150-m tower at Cinder
Cone Butte.
Listed in Table 11 (p. 96) are the correction factors ERT
recommends for the Climatronics UVW data from SHIS #2. Because of the
variability and rapid changes in response shown by the instruments
near the stall regions, the corrections in these areas are
ill-defined. It is not clear how best to refine the data from times
when one propeller is near or in its stall region. The wind component
speed will generally be quite low when this occurs, so that the effect
on resultant wind speed may be quite small, but the effect on
resultant direction may be several degrees. Over a five-minute
averaging period, unless the winds are very steady, however, even the
error on average direction may be less than three to five degrees.
Measures of lateral wind variability may be the most adversely
affected.
The response of the W-prop shows it to be similar to the
horizontal props but with a narrower stall region at low speeds, as
might be expected because of the reduced inertia due to gravitational
forces on any propeller imbalance. Otherwise it seems to fall into
the general range of response exhibited by the U and V props. The
differences between responses with effective wind direction at 90° or
270° are also not great. In short, the tests in the FMF wind tunnel
do not provide evidence that the W-prop's response is substantially
different from the U and V. We are aware of no other studies that
corroborate or contradict this premise.
The non-cosine response corrections that were applied to the
SHIS #1 data were combinations of the results of the CSU experiments
and factors determined from comparison of data from nearly collocated
propeller and cup-and-vane systems at Cinder Cone Butte. The values
in the COSCO (COSine correction) algorithm used by ARLFRD for the
interim SHIS #2 data are derived from experiments with modified
R.M. Young instruments by EPA's John Clarke.
91
-------
The principal differences among the three sets of results are:
• Clarke's (and COSCO's) corrections are the same for the U-,
V-, and W-props, and the corrections in the first 180° of
wind direction are assumed to be mirrored exactly in the
second 180°. This assumption is probably more accurate with
Clarke's R.M. Young instruments, which had the U- and V-arms
offset vertically to reduce the effects of one arm's wake on
the other. In the Climatronics results, the effects of
these wakes can be seen in the reduced response between 280°
and 340° in comparison to that between 200° and 260°.
• Clarke's propellers were 23-cm polypropylene and the CTMD's
are 19-cm polystyrene. The smaller, lighter props were
selected for CTMD because of their increased responsiveness,
but the polypropylene props are smoother and harder.
Clarke's instruments also had extensions of the propeller
shafts outwards from the hub. Both of these differences may
improve the response of propeller anemometers near the stall
regions around 90° angles of attack. The roughness of the
polystyrene props is not uniform over their surfaces, which
may contribute to the large variability of response among
units at low component wind speeds. The softness of the
material and the fact that they have been put on and taken
off the shafts several times makes their hub holes
vulnerable to enlargement and distortion, and consequently
their mounting may be imbalanced or skewed. The effect of
imbalance was evident during the experiments; on several
occasions a prop would oscillate several degrees but not
turn. This behavior results in apparently unpredictable
results that depend on whether the motion causes light
signals in the three-hole chopper wheel. The instrument
responses in the present analysis have been set to zero
speed when visual observations indicated the prop was not
turning.
• ERT's experiments in the CSV wind tunnel differed from the
FMF experiments in several ways. First, the FMF tunnel
maintained steady low speeds better than CSU's so that we
now have valid responses for a speed of approximately 0.5
m/s. Second, the instrument response from the CSU tunnel
was mechanically determined by counting the number of pulses
from each transmitter in a 30-second sample, whereas the
instrument response during the FMF experiments was the
voltage produced by the translator card. This output was
measured both by the FMF's data acquisition system (DAS)
sampling at lOHz for 20 seconds and by ERT's digital volt
meter (DVM). Also, the tunnel speeds at CSU were estimated
from the output of ERT's "transfer standard" cup anemometer,
which had been calibrated in the National Bureau of
Standards wind tunnel in Gaithersburg, MD. The tunnel
speeds at the FMF were estimated from the fan tachometer
settings as described above.
92
-------
In summary, the interim HBR data base still contains noise,
especially in the data from the early SHIS #2 experiments. We plan a
reanalysis of this data base to remove some of the residual noise.
The first step will be additional signal processing to reduce the
allowable second-to-second variations listed in Table 9. The second
step will be an analysis and subsequent removal of the high frequency
noise apparent in the data. For the propeller anemometer data we will
apply the calibration factors listed in Table 10 and the correction
factors for non-cosine response given in Table 11 after filtering the
noise. The fourth step will involve correction for physical
misalignments of instruments to true north. Finally, the 5-minute and
1-hour averages of the measurements and derived measures will be
recalculated.
4.2 SHIS #2 Preliminary Modeler's Data Archive
The preliminary MDA developed from the SHIS #2 data archive
contains 1-hour average tracer gas concentration data for each of the
hours during SHIS #2 in which either SFg or CF3Br tracer gas was
released. The MDA excludes the 19 hours which were withheld by the
EPA Project Officer for independent model validation purposes (p. 184
Third Milestone Report) . The tracer release and meteorological data
portion of the MDA has been compiled only for those hours in which
CF3Br tracer gas was released because the analysis of plume rise for
the coincidental oil-fog and SFg tracer gas plume is not yet
complete.
The SFg data base, along with a revised meteorological-
data base, will be compiled once final refinements have been made to
the tower meteorological data. Refined estimates of wind directions
and <3Z values as derived from photographs and lidar will be
included in the MDA in the future.
4.2.1 Tracer Concentration Data k
Tracer gas data that were selected for the MDA are included as
1-hour average values of the concentrations scaled by the emission
rate. These averages include concentrations from all 1-hour sampling
bags, the largest of the two samples collected at collocated sampler
sites, and the average of 10-minute concentrations at standard
10-rainute sampler sites. For the 10-minute samples, 1-hour averages
are included only if no more than one 10-minute period is missing.
All concentrations are given in units of nanoseconds /m^.
The tracer gas archive also includes the position of each
sampler. A Cartesian coordinate system is used with the origin at
Tower A. The x-axis is oriented toward true east and the y- axis is
oriented toward true north. The sampler elevation is reported as the
height in meters above 1600 m MSL. The sampler position
identification code and a data quality flag are also included for each
concentration. The intake port for the sampler is approximately 0.5m
above the ground.
A map of HBR with each sampler and release position is presented
in Figure 26. The solid lines are terrain contours and the dashed
93
-------
TABLE 9. ALLOWABLE SECOND-TO-SECOND SENSOR CHANGES USED TO
FILTER RAW DATA IN PROCESSING OF PRELIMINARY DATA BASE
Sensor
Max voltage Increment
Approx. change in units
u, v prop
w prop
cup anemometer
wind vane
temperature
delta temperature
fast temperature
0.200
0.200
0.200
0.150
0.100
0.100
0.100
2.0 m/sec
0.8 m/sec
2.0 m/sec
16.0 degrees
1.6 degrees C
0.4 degrees C
1.6 degrees C
94
-------
*
*
CO
CO
•>
o
o
00
.
CM
^
e
in
•^f
r-l
*
CO
r«. \o
in -sr
i-l O
• .
CO CO
r*»
r-l
p-4
*
CO
O
r»*.
0
•
CO
f>*h
o^
0
•
CO
O in co
O O •»
o r-i o
CO
z;
0
>-i
H
5!
c
0)
r-l
CO
'3
a
Ed
->
O
O
\o
CM
^
O
in
.
CM
CO
e
00
^
o
,_!
(0
^
g
^
CM
m
CM
in
o
,_,
\o
CM
in
o oo
vO l-l
o* o
O i-l
00 CO
CO CT>
•«T -sj-
CO
o
o
i-i
vD
00
• CM
o co
o o ->r
CD
0)
03
W
00 00
CM
CTi
en
cr>
3;
*>
D
^
fe
fc
O
>i
pj
^
s
CO
•
o
iH
Lx3
j
P3
^4
H
U
-p
eu
o
!Z
0)
c
•H
rJ
C
o
CO
w
CO
00
^
.
0
I-l
H
3
in
i
H^
*
o
o
in
CM
1
CO
^J
I—I
o
>
^>
CM
in
o
i-i
n
3
vO
J
^^
CO
o
in
CM
I
CO
r-l
O
>
•^
CM
i-l
•«r
n
3
CO
^
j^n
CO
o
in
CM
I
(0
t-^
o
>
^
o*
^
•
o
I-l
II
3
-j
^f
CO
o
m
CM
i
w
,— I
o
>
r^
CO
CO
O
r-l
II
3
CO
T~4
.q.
0
m
CM
i
CO
t—l
o
>
\o
in
00
r-l
•*
II
3
3 > D 3
I— 1
CO •
.r4 O
Ll Si
«•
H H
CO
in
^
in
CO
»
u
0)
Jj
3
-P
O
co
3
C
co
X
XN
O
0
in
CM
I
W
-p
r-l
O
vx
o
o
o
o
II
3
C
o
•H
^)
CO
Ll
.H
CO
0
*
r~\
M
CO
•H
tQ
v^
Jj
o
LI
u
OJ
OJ
60
CO
Li
4)
j>
<
/f^
o
.H
U
•r-l
O
0)
Ll
p.
\*s
{^
O
Li
Li
*4-l
o
C
0
.r4
-P
CO
•H
^
a)
•o
*
T>
W
x™\
jj
C
o
Li
0)
(X
0}
(0
v^
(-1
o
Li
Li
4J
o
> 0
,0 to
•t-i
-0 ti
-------
TABLE 11. RECOMMENDED CORRECTION FACTORS FOR NON-COSINE
RESPONSE OF CLIMATRONICS UVW PROPELLERS
Angle
0°
1
2
3
4
5
7
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
88
90
92
95
100
105
110
115
F
u
(0.50)
(0.30)
(0.35)
(0.38)
.40
.48
.58
.62
.68
.72
.75
.79
.77
.78
.82
.86
.91
.94
.96
.97
.98
.99
1.00
1.00
1.00
1.00
.995
.99
.98
.97
.96
F
V
1.00
1.00
1.00
1.00
1.00
1.00
1.00
.995
.99
.97
.96
.94
.91
.87
.83
.79
.78
.80
.76
.73
.69
.64
.57
(0.35)
(.50)
(1.60)
1.10
.84
.81
.80
.81
Angle
120
125
130
135
140
145
150
155
160
165
170
173
175
176
177
178
179
180
181
182
183
184
185
190
195
200
205
210
215
220
F
u
.94
.91
.88
.83
.78
.78
.80
.75
.73
.70
.66
.60
.48
.35
(.35)
(.35)
(.30)
(.50)
(1.80)
(1.60)
(1.40)
(1.20)
.90
.87
.84
.82
.80
.80
.81
.84
F
_v
.81
.79
.77
.77
.78
.83
.86
.87
.87
.88
.88
.85
.82
.81
.80
.78
.78
.79
.79
.81
.83
.84
.86
.94
.97
.97,
.96 '
.94
.92
.89
Angle
225
230
235
240
245
250
255
260
265
270
273
275
280
285
290
295
300
305
310
315
320
325
330
335
340
345
350
355
357
359
360
F
u
.86
.88
.91
.94
.96
.97
.98
.97
.95
.90
.86
.86
.85
.87
.88
.88
.87
.85
.81
.78
.79
.80
.81
.81
.83
.84
.89
(1.10)
(1.40)
(1.80)
(0.50)
F
V
.85
.83
.81
.80
.80
.81
.83
.83
1.05
(0.50)
(0.40)
.57
.65
.68
.72
.75
.80
.78
.78
.83
.87
.91
.94
.96
.97
.98
.99
.995
1.00
1.00
1.00
Fu and Fv are the fractions of correct cosine response shown by the
Climatronics U and V props in the tunnel with the wind from direction
"angle." F^1 and F^1 are the corresponding correction factors.
Values in ( ) are not firmly founded from experiments but are estimates
to be used for completeness and program coding.
96
-------
10
o
•rl
-P
to
O
O
0)
CD
•a
CO
01
ta
cd
a)
t-i
0)
V)
vD
CM
97
-------
Lines are roads. Elevations (MSL) for each contour and release
location are presented in meters. The towers and collocated sampling
positions are not shown.
4.2.2 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
ground beneath the source, 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 (g/s) is an average mass release rate from the
time at which the release valve was opened to the time at which it was
closed. In some cases, this period was less than 1 hour, but in most
cases it was several hours. The start and stop times for the release
are referenced to the beginning and ending times of each experiment
hour, respectively. 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 Tower A. The zero height
contour in this system corresponds to the 1600-m elevation MSL.
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 1600 m MSL. A topographic map of HER can be
found in the Third Milestone Report.
4.2.3 Meteorological Data
Meteorological data contained in the MDA differ from those
contained in the SHIS #2 data base in that: all quantities apply to
the release height of the tracer gas rather than to the height of the
fixed instrument levels; derived parameters computed from the MDA are
included; and 1-hour averages are ultimately utilized in the modeling.
First, the sequence of 5-minute data in the MDA was constructed
as follows:
• The NOAA/WPL sonic anemometers on Tower A at 5 m and 40 m
acquire mean vertical wind speed and direction and vector
mean horizontal wind speed and direction, and
the turbulence velocity scales (cru, crv, aw) . These sonic
data are linearly interpolated to the plume height. All the
CF3Br tracer release heights are between 5 m and 40 m,
except one hour when CF3Br was released at 50 m.
• The dividing-streamline height, Hc, is obtained from
splined (splined under tension) profiles of temperature and
wind speed by means of the integral formula presented in the
First Milestone Report. The 5-minute temperature and
propeller wind speed profiles (as received from ARLFRD in
October 1983) from Tower A are used. The effective height
of HER is taken to be 85 m.
98
-------
• A bulk hill Froude Number (Fr) is 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 85 m and the height
of the bottom of the layer (either Hc or 2 m) .
• The local Brunt-Vaisala frequency (N) is estimated at source
height by evaluating the temperature change along the
splined Tower A temperature profile in the immediate
vicinity (release height + 0.1 m) of the source height to
obtain the local temperature gradient. The bulk
Brunt- Vaisala frequency (N^) is estimated at source height
by evaluating the temperature change between 2 m and the
release height.
Then, the 1-hour average data in the MDA are constructed as
follows:
• The 1-hour averages of speed and direction reported by the
NOAA/WPL system are vector averages. These speeds and
directions as well as au, av, and av are linearly
interpolated to release height. The 1-hour "scalar" wind
speed and wind direction averages are computed from the
5-minute vector average data interpolated to release
height. The 5-minute average wind data interpolated to
release height are also used to compute a vector average of
unit vectors along each 5-minute wind direction.
• The parameters calculated from the splined profiles of the
5-minute temperature and wind data from Tower A (Hc, Fr,
N, Nfc) are averaged to provide 1-hour average values.
The tracer release and meteorological data portion of the MDA
have been compiled only for the CF3Br tracer release hours. There
were 90 hours of CFgBr tracer releases, data from 11 of which were
withheld by the EPA Project Officer for independent model validation
purposes, leaving a total of 79 CF3Br MDA hours (Table 12). Sonic
anemometer data are not available for 12 hours. Of the remaining 67
hours, nine hours were judged inadequate for modeling because there
were few non- zero concentrations observed, or the plume missed the
sampler array, or the release was from Tower C on top of HBR. Thus, a
total of 58 hours are available for modeling.
4.3 Modeling HBR
The development of CTDM has been largely guided by a theoretical
understanding of dispersion and flow about simple, three-dimensional
hills. A different approach has been taken in the initial stages of
developing a model for HBR. Although many of the theoretical concepts
contained in CTDM are applicable at HBR, the nature of the flow below
a "critical dividing- streamline" at HBR appears to differ from that at
CCB. The flow below Hc at HBR was frequently more unsteady than the
flow above Hc. The plume tended to travel along the ridge, either
toward the south or the north, before reaching the surface.
99
-------
TABLE 12. CF3Br DATA BASE SUMMARY STATISTICS
(EXCLUDING WITHHELD HOURS)
1. Number by release location
216/215 203 111 A _C_
15 27 3 32 2
II. Number by height of source above the ground
<10m 11-20m 21-30m 31-40m >40tn
12 30 21 15 1
111. Number by wind direction*
N E S W
(316°-45°) (46°-120°) (121°-225°) (226°-315°)
19 26 18 4
*Sonic anemometer data not available for 12 hours.
IV. Number by release height (z ) vs. H
T^ *^ " c* ^* T" ~ *Af* Zv* ^ rl rv
2lT 10~~ 48~
100
-------
Fluctuations in this "blocked" flow and the two dimensionality of the
ridge make the use of a single stagnation streamline concept
inappropriate. A dispersion model for HBR must include a description
of the nature of this blocked flow and its effects on dispersion.
Because the analyses involved in specifying Hc for the unsteady flow
in this blocked region have just begun, we have developed an empirical
model for HBR as the first step towards structuring a theoretically
based model. Once we have an adequate empirical understanding of the
relationship between concentrations and observed meteorological
conditions, we can attempt to develop a better theoretical
understanding of the processes.
4.3.1 Importance of Hc at HBR
Analysis of data collected during the Cinder Cone Butte (CCB)
experiment shows that the concept of the dividing-streamline height
(Hc) plays a critical role in the description of dispersion in
complex terrain. A plume below Hc tends to remain horizontal as it
flows around CCB, while one above Hc has enough energy to climb over
the hill.
A plume released near Hc can, under certain circumstances,
"impinge" on the hill and cause concentrations that can be as large as
the centerline concentration. This was the case at CCB, where the
highest concentrations occurred when zr was approximately equal to
Hc. At HBR, the highest CF3Br concentrations were observed when
zr was less than Hc, i.e., in the blocked flow upwind of the
ridge. This is one of the major differences in dispersion between the
two sites. Figure 27 shows that the averages of the top five
concentrations observed during each hour at HBR take on their largest
values when the release height zr is below Hc. Because a plume
released below Hc has a tendency to remain horizontal, one would
expect the position of the maximum concentration to be lower on the
hill than it would have been had the plume been^ lifted as it traveled
toward the ridge. This hypothesis can be examined by comparing the
locations of the observed maximum concentrations to those estimated by
a flat terrain model formulated in terms of a terrain-following
coordinate system (the height of the plume centerline above the
terrain remains constant).
The flat terrain model used for this analysis includes a method
to account for the large wind shear observed during stable
conditions. This method is based on a model to simulate the
dispersion of plumes released into a sheared boundary layer (Venkatram
and Paine, 1984). The performance of this flat terrain model was
improved significantly by incorporating the variation of wind and
turbulent velocities into the sigma-z formulation. Figure 28 shows a
schematic of plume behavior and the assumed speed profile. The upper
and lower parts of the plume grow at different rates because of the
speed change with height. It is assumed that azu and
can be written as:
101
-------
u
X
,£2
& ^
00 -P
-i-i
o
P
cd
o>
o -P 4->
o
CO
.1-1
u
co
r-
CM
Ll
00
•H
fa
oo
•H
O
%
Ll
oo
C
8
S
(B
QJ
S
5
102
-------
Figure 28. Schematic illustrating a Gaussian model that accounts for
wind speed shear.
103
-------
(1 + t /2TT)
u L
(82)
where the times of travel to, and ty are implicitly given by
tn
s = / u dt (83a)
0 e!l
and
t
s = J u u dt (83b)
0 eu
where s is the along -wind distance to the receptor, and
u
u 0 = — (z - o n/2); <* n < z (84a)
eil z r zl zl — r
r
u
= ;r <* „ ; ° (i > z
2z zS. ' zl r
v
and
u
u = — (z + o /2) fc (85)
eu z r zu ?
r
The effective transport velocities ueu and uep, correspond to the
"midpoints" of the upper and lower plumes. The sonic observations of
zr, the effective
wind speed associated with the travel time of 0zp, starts
increasing and the growth rate of az^ becomes comparable to that
associated with ozu. The figure shows that (ozu + ozo.)/2
does not differ very much from oz obtained by setting t = s/ur.
104
-------
bN\
0>
u>
I
M
cd
o
cd
o
•H
to
•R "-"
I ^
CD -l->
y .H
S *M
W N
(4-1
o
o
•H
-p
CO
.1-1
t-l
cd
CM
0)
ao
105
-------
Therefore, the new az formulation does not alter the centerline
concentration. Its main effect is to decrease the distance at which
the maximum ground-level concentration occurs.
Ground-level concentrations are expressed in terms of the two
sigma-z values by
C(s,yR,0) = x+q—— exp —^-j exp v ——|' (86)
^r °zl +0zu °y 2ozjl 2oy
where yp is the crosswind coordinate of the receptor (see
figure 30). The crosswind spread oy is taken to be iyS, where s
is measured along the profile of the hill.
Equation 86 was used to calculate concentrations at the receptors
on HBR. Figure 31 shows the ratio of the height of the receptor at
which the predicted maximum occurred to that at which the observed
maximum occurred as a function of zr/Hc. As the release height
moves below Hc, the positions of the largest concentrations produced
by the flat terrain model become higher on the hill than vj«re
observed. This analysis suggests the importance of including Hc in
the modeling of HBR.
4.3.2 Performance of the Flat Terrain Model
B'igure 30 shows the idealized geometry used to model Hogback
ridge. We assume that the ridge is two-dimensional and the stagnation
streamline can be defined by a line through the source perpendicular
to the ridge. The travel time to the receptor is calculated using the
distance, (s2 + (yR-yr)2)1/2. The model was applied to 58
case hours listed in Table 13. Because the model assumes steady state
conditions, and does not account for "sloshing" along the ridge,
we chose to concentrate on a subset of hours for which ur > 0.5 m/s
and for which the wind direction falls in the range 55° < Gm < 180°.
With Os = 117°, this choice of wind directions selects those cases
in which the mean wind was directed towards the hill. This subset
consists of 35 case hours which are indicated by asterisks in
Table 13. All models discussed in this section are evaluated with
this subset of 35 hours.
In this preliminary analysis we have chosen to quantify the
performance of the models with the ratio of the average of the top two
observations with that of the model predictions (see also Section
4.3.5). Figure 32 shows the variation of this ratio for the flat
terrain model as a function of (zr-Hc)/Hc. Note that, the
performance of the model deteriorates as the release height b*:c
-------
Mean Wind Direction
Figure 30. Geometry for modeling Hogback Ridge.
107
-------
0
X
0)
>
4-> O
CO
r-t 4->
144 fi
00
•o a) .H
a) ^ 4-> .e
LI
o> >> o>
01 jo in
fl CO
o -o a>
0) .-4.
g 4->
g a)
f CO
^5
O 4J
O 00
<44 4J 0)
O O
.C 0)
00 >
•H -H
• H
3
II
tn
.—I ,
O) 00
•8 C
O -H
e -o
.H
c >
•r< -i-l
co -a
u
(j 01
a> .c
8
108
-------
TABLE 13. HOGBACK RIDGE MODELING DATA
tXD
4
4
4
4
*4
*5
5
5
*5
5
*5
*6
6
*6
o
*6
*8
8
8
8
8
8
*8
*8
*8
9
9
9
*9
*10
*10
10
MO
*10
*10
*11
*11
11
11
*11
1 1
*1 1
*12
*12
*ia
*14
14
*ia
*1<4
14
*14
*15
*15
*1 5
*15
*15
*15
*15
EXD-
Hour
2
3
5
6
7
1
a
3
<4
5
6
1
3
6
*J
9
1
a
3
4
5
o
7
H
9
1
2
5
•4
S
4
5
b
7
8
I
3
4
7
8
9
1 0
2
3
5
b
7
H
10
1 1
12
2
•'»
5
7
8
9
1 1
zr
(ti)
10.0
20.0
10.0
20.0
20.0
25.0
15.0
20.0
20.0
15.0
15.0
30.0
20.0
30.0
25.0
20.0
20.0
10.0
5.0
5.0
30.0
25.0
25.0
30.0
15.0
20.0
20.0
20.0
20.0
30.0
30.0
30.0
30.0
30.0
30.0
20.0
20.0
20.0
25.0
25.0
10.0
10.0
40.0
50.0
20.0
20.0
20.0
20.0
35.0
35.0
35.0
40.0
40.0
40.0
4 0 . 0
40.0
40.0
40.0
nc
(m)
30.0
26.0
31.0
47.0
29.0
18.0
31.0
29.0
30.0
49.0
48.0
25.0
22.0
31.0
29.0
35.0
61.0
33.0
48. o
37.0
S6.0
41.0
35.0
39.0
25.0
20.0
13.0
19.0
22.0
37.0
50.0
59.0
33.0
34.0
38.0
59.0
42.0
54.0
27.0
21.0
37.0
21.0
46.9
31.0
31.0
16.0
12.0
22.0
26.0
19.0
11.0
41.0
'» 5 . 0
44.0
31.0
33.0
41.0
1.0
u
(m/s)
.9
.7
.2
.3
.5
2.4
.8
1.1
.4
.5
.7
1.3
2.0
1.5
1.3
1.0
.7
.6
1.7
1.5
.8
.9
1.2
1.1
.8
1.2
2.1
2.6
1.5
.9
.5
.4
1.3
1.0
.9
.8
.8
.9
1.5
1.6
.fa
.8
1.5
2.4
.7
1.2
1.6
.9
.8
.6
1.2
.9
1.4
1.0
1.5
1.4
.9
4.6
thta
(rteg)
10.5
30. V
187.4
261.7
119.7
55.7
26.9
28.3
170.7
186.4
178.2
131.3
33.7
64.0
45.3
67.1
99.3
322.3
17.0
15.9
276.0
53.0
85.7
141.8
81.8
38.7
24.1
29.8
65.2
120.7
139. b
247.0
75.6
65.4
58.8
59.3
io9.a
298.2
51.3
93.9
2^.7
67.3
167.1
13ft. 4
55.9
59.3
41.0
137.2
132.2
48.9
102.8
63.0
151.3
14o.l
97.5
106.9
144. b
114.4
s 1 g m a - v
(m/s)
.484
.512
.547
.611
.449
.464
.547
.386
.376
,5B5
.770
.366
.491
.792
.534
.495
.629
1.117
.262
.338
.146
.436
.697
.788
.614
.379
.521
,b94
.539
.295
.328
.39b
.362
,48«
\522
.475
.506
.245
.543
,4V1
,451
1.025
.875
.794
.691)
.787
.682
.743
.872
.484
.744
.461
.388
.324
.597
.508
.501
.6b7
si gma-w
(m/s)
.136
.157
.115
.122
.165
.139
.151
.136
.083
.112
.095
.118
.163
.201
.167
.096
.096
.118
.143
.128
.088
.092
.207
.198
.150
.125
.217
.265
.283
.116
.136
.138
.137
.196
.235
.111
.233
.107
.241
,23b
. 1 8«
.223
.115
. 485
,iU£
.218
.2<»«
.295
.253
.242
.294
.159
.151
.175
.296
.206
.212
-33b
N
(1/s)
.0326
.0185
.0540
.0321
.0501
.0519
. 0246
.0274
.0306
.04ul
.0415
.0215
.0427
.029?
.0226
.0225
.0243
.Ubo3
.1031
.0659
.0340
.0356
,03b3
.0276
.0365
.0233
.0243
.0393
.0254
.0181
.0198
.0316
.1)337
.0281
.0355
.0302
. (J 4 b 1
.0433
.0258
.0237
.0391
.U287
.0436
.0376
.0275
.0204
.0115
.0133
. 0 0 7 3
.0037
. U 0 i| 1
.0276
. u 4 0 5
.0293
. 0 3 U H
. 0296
. 0356
. (!OS^
*Mean wind was directed toward the ridge.
109
-------
K +
t
s
8 S S 8
i
N"
«
o «j
*» 3
•H 10
CO
O O
+>
-P 0)
£ 0)
•H 3
0) 0)
«
a) f
o) -P
CD
CD u
.H 0
01 >
f! -H
4-> 0)
-H -O
*g
"cd C
3 -H
•O CO
•H U
en u
(D (1)
b -p
O CO
C« ^"^
.^ fl)
CO -P
•H
U. ti
CO O
CM
CO
a
•H
dO Z d°l eBBJ8AV/°O 2
110
-------
It is informative to examine the performance of the model when
the traditional "half-height" plume path modification is incorporated
into it. The version of the modification that was used is given by
1 ZR
T = 1 - - — - ; z0 < z (87a)
p 2 z R ~ r
r
(87b)
and is equivalent to that defined in Equation 74. Here zg is the
receptor height. It should be noted that the "half-height"
modification has been widely used to simulate non-stable or slightly
stable conditions in complex terrain. It has only been occasionally
used to simulate very stable conditions. Nevertheless, the
half- height modification has been applied to all 35 HBR hours.
Figure 33 shows a plot of the residuals with release height. It is
seen that the half-height modification moves most of the points down
into the "overprediction" region for releases below Hc. Performance
statistics for both tests are compared in subsection 4.3.5.
4.3.3 Development of the Empirical Complex Terrain Model for HBR
In developing the empirical HBR model, we assume that the effects
of complex terrain on dispersion occur largely as a result of the
influence of Hc and can be accounted for empirically by reducing the
effective height of the plume centerline over the terrain. Future
efforts will attempt to develop a theoretical description of actual
plume behavior.
Before introducing the height correction, it is useful to
re-examine the residual plots presented earlier. In Figures 34 and 35
the residuals have been replotted as a functiort of the predicted
concentration. For an "adequate" model, we know that the residuals
should not exhibit any trend with Cp (see First Milestone Report) .
It is seen from the figures that residuals from both the flat- terrain
and the half -height model do not follow this behavior; Co/Cp
decreases with Cp. The flat-terrain model tends to underestimate as
Cp decreases while the half -height model tends to overestimate as
Cp increases. While the flat terrain model produces underestimates
on an average, the half-height model provides estimates that are
larger on an average than the observed concentrations.
The results from towing tank experiments (Appendix A) show that,
at least for a three-dimensional hill, the Hc surface can be
approximated by a ground plane in simulating the flow above Hc.
This would suggest the following form for Tp:
Tp = (zr ~ Hc)/zr d zr * Hc (88)
T = 0 i z < H
p re
111
-------
u
z .
0)
o u
33 3
O
O £.
J->
IO
0) CO
>
•H U-i
*> o
eo
.•-i
0 -P
M O
-------
+
+
+
•f
1 7
«
-f
+
•f
,
*
+
+
"^* +
+
^ +
f +
.
4-
4-
1
•
+
*
w
*
1
CD
-g
0)
•o
o
6
.s
- « cd
3 u
t*
o
• rt
"9 *"*
A H-l
"E o
""V **^
^ 0.
O fi
a. S
0
« a, cd ^.
~ ® ^ ^ £
5 .H 3
§ M O
< 2 ^
a)
cd 01
•rH ^)
U 3
cd 01
!_ o > ^
CO
CO
0)
3
00
-------
1
4-
4-
4-
4-
+
4-
+ *
4-
4-
.
4- *
4-
4-
4-
4-
1
4*
4-
4-
4-
4-
•f
4-
4-
4.
t-
4-
k
|
I
O)
"i 5
.H
*
C •
•H «-s
to co
u u
01 U 3
- jz
-P in
cd co
u-i u-(
CD O
- a> ^T- ^
* 1. O -4J
g IM Q) O
* u .
O f.
oo
C -H
O 0)
•H .C
. « « ui,
Ul .H rH
U CO
to £
> :
m co
0)
u
3
&0
fa
s
_
8 8 8 8 8 S K 2 5
8 2 * ~
doj. e6B4eAv/°O 2
114
-------
The main problem with this formulation is that it is equivalent
to assuming that the plume height is a constant (zr - Hc) from the
source to the receptor. This in turn leads to problems in properly
accounting for the dispersion of plume material over terrain well
upwind of the hill. To get around this problem we assume that Tp
varies linearly from unity at the source to its minimum value given by
Equation 88 at so where Hc intersects the hill surface. Beyond
so it remains at this minimum value (Tpm). Then the variation of
T is given by
s < s
a - o
(89a)
T : s > s
pm a o
(89b)
where the along-wind distance sa is measured along the profile of
the hill so that
2 a o. 5
s = s + ((s-s ) 4- (z -z ) ) for s > s
a o o R r o
(90)
sa = (s2 + (ZR - zr)2]
for s < s0
and Tpm is the minimum value given by Equation 88. The terrain
correction factor Tp used in the Gaussian dispersion equation is
then the average of Tp over the distance sa,
, s
T = J -* + T
p 2s pm
* a *
s > s
a - o
(91a)
and
T = 1 -
P
(1-T ) s
pm a
s < s
o.
(91b)
For completeness we rewrite Equation 88
T = (z - H )/z ; z > H
pm r c r r - c
= 0
z < H
r c
(92a)
(92b)
4.3.5 Summary of Model Performance
This section compares the performance of the empirical HER model,
the half-height model, and the flat terrain model. We have chosen to
quantify the performance of the models with the residual c defined
by
115
-------
Average of top N observed
Average of top N predicted
where N is taken to be 1, 2 or 5 and the concentrations are paired in
time but not in space. If we assume that e is lognormally
distributed, an adequate model (see Section 5.3 of the First Milestone
Report) should yield an c that is randomly distributed about unity
when it is plotted on a logarithmic scale. For an ideal model, nu
E In e should be unity and the logarithmic standard deviation
should be small. Figures 32 and 33, presented earlier, and Figure 36
show the variation of e(N=2) for the flat terrain model, half-height
model and the empirical HBR model as a function of (zr-Hc)/Hc.
As discussed earlier, the flat terrain model performs poorly and
underpredicts the concentrations by large margins. There is some
indication that the performance is better when zr becomes much
greater than Hc.
The half-height model overestimates the observed concentrations
with the tendency to overpredict increasing as zr becomes less than
Hc. The residual pattern for the empirical HBR model is similar to
that for the half-height model except that the tendency toward
overestimation for zr less than Hc has been reduced. This
reduction in the tendency to overestimate peak observed concentrations
for zr less than Hc also reduces the degree of overestimation as
Cp increases. A plot of the residual versus Cp presented in
Figure 37 shows that the empirical HBR model has improved upon the
performance of the half-height model (Figure 35) in this aspect of
model performance.
The relative performance of the models is better illustrated
through the statistics presented in Table 14. Because the sample size
was relatively small (=35) it was necessary to calculate m» and Sg
using a method that weighted the outliers less than the residuals near
the middle of the distribution. We assumed th^t an appropriate
estimate of itu was the median of the 35 ratios 'of the observed to
the predicted concentrations. To estimate sg we assumed that c
was lognormally distributed. Then In Sg corresponded to the
standard deviation of a normal distribution with the observed fraction
of residuals within a factor of two of unity.
We see from Table 14 that the flat terrain model underestimates
the concentrations by almost a factor of two. (An nu greater than
one indicates an underestimation). The standard deviation Sg is
relatively large. The model with the half-height terrain correction
factor overestimates the observed concentration by a factor of 1.5 for
all the chosen values of N. The sg, while smaller than that of the
flat terrain model, is still large.
The empirical HBR has an m~ which is close to the ideal value
of unity. Notice that the best combination of the mg and Sg
statistics correspond to the largest observed concentrations (N - 1).
The empirical HBR model outperforms the other two models for all three
values of N.
116
-------
•fr
1
*
•h
*
+• + + 4
,
1
+
*
, + +
A
•
•f
|
a>
/ — \
o> 3
> 0
- ^ rt LO
0)
o
+J
xl ^
bO (D
•t-f tn
(D '-/I
- . 03 — i
~ ^
- • t>0
e»
i 8 8 8 S K
O Z
117
-------
+
1
*
+
+
* •»•
*.**
+
+
•»•
4-
*
•f
4-
I
+ + +
*
k
*
*
1
i
CC
® CQ
•• *
CO
o
•H
.r-l
A Oj
-8 jg
4J
u
o
" * W
S *"~* *j
— S3
«M 3m
0- "5 °°
£ w «w
•^ 0) 0
- * 1 ^ •£
Oft w M-i O
- £ o o>
> ^3
< C 3
o w
CO -H
•H Q)
u -o
«» CO O
• ui > e
n
" m *->
3
oo
fc.
ui as -
» 5 CD S in N — s>
• W * (V —
Z dOJ. 9BBJ8AV/°O Z d°l 8BBJ8AV
118
-------
TABLE 14. RELATIVE PERFORMANCE OF MODELS
35 Hours
N=l N=2 N=5
z > H (10 Hours)
r c
z < H (25 Hours)
r c
Model m S m S^ m £
Flat
Half-Height
HBR Model
1.98
0.67
0.99
3.70
2.90
2.28
1.91
0.66
0.95
3.37
2.90
2.40
1.62
0.64
0.87
3.37
2.55
2.70
N=l N=2 N=5
Model
Flat
Ha If -Height
HBR
m
2
0
1
g
.11
.84
.12
3
1
1
S
g
.74
.72
.95
m
2.02
0.
1.
87
12
S
3.74
1.
1.
72
95
1
0
1
m
~g
.75
.81
.06
S
~g
2. 79
1.72
1.95
N=l N=2 N=5
Model m S m ^S m
~g ~g ~g ~g ~g
Flat 1.66 3.74 1.60 3.27 1.21 3.14
Half-Height 0.42 4.37 0.42 4.37 0.46 3.74
HBR 0.64 2.45 0.63 2.67 0.58 3.28
119
-------
Table 14 also includes the statistics for the three models when
the hours are separated into two classes: (1) zr greater than Hc
and (2) zr less than Hc. Although the number of hours in each
class is considerably smaller than the total number of hours, the
statistics are still instructive. For zr greater than Hc the
half-height model and the empirical HBR model have virtually the same
Sg, and the half-height model overestimates by approximately the
same amount that the empirical HBR model underestimates. For zr
less than Hc, the empirical HBR model exhibits a smaller tendency
toward overestimating the observed peak concentrations than the
half-height model. The HBR model also performs a little better in
reproducing the peak concentrations (N = 1 and 2) for releases below
Hc than the two other models. However, for N = 5 the flat terrain
model performs as well as the empirical model.
4.4 Selected Case-Study Results
A total of 58 hours have been used in the modeling analysis. Of
these 58 hours, 36 hours are classified as zr < Hc, 12 as zr > Hc,
and 10 as zr * Hc.
To illustrate plume behavior at HBR, detailed analyses of four of
the hours modeled are presented in this section. These hours include:
• Experiment 11, 0600-0700 MDT, zr > Hc
• Experiment 14, 0300-0400 MDT, zr * Hc
• Experiment 6, 0700-0800 MDT, zr < Hc
• Experiment 8, 0500-0600 MDT, zr < Hc
These four case-studies include those hours in which the highest
concentrations were observed for release heights greater than, equal
to, and less than Hc. Also, these hours were selected because the
MDA wind directions are consistent with the observed concentration
patterns. Each case-study analysis includes a description of the
release information, meteorological data, and observed CF3Br
concentrations. Also, the performance of the flat-terrain model and
the empirical HBR model discussed in subsection 4.3.3 are contrasted
for each case-study hour. These two models are referred to as HBR
(Flat) and HBR (Terrain), respectively.
4.4.1 Experiment 11, Experiment-Hour 8 (0600-0700 MDT)
Release Description
The CF3Br tracer gas was released 25 m above the ground from
Tower A for the entire hour. The release was continuous from the
previous two hours; therefore, the tracer plume was well established
by the beginning of experiment-hour 8.
Local terrain elevations near the release point are estimated to
be 3.7 m above the base elevation of the hill coordinate system, so
the net release height corresponds to the 28.7 m height level on the
ridge. The CF3Br release rate is calculated to be 1.31 g/s.
120
-------
Meteorological Information
The 5-minute temperature and propeller wind data measured during
the hour at the 10 instrument levels of Tower A (2, 5, 10, 20, 30, 40,
60, 80, 100, 150 m) are used to characterize the flow in terms of the
dividing-streamline height (Hc) and the bulk hill Froude number
above Hc (Fr). Time series plots of the calculated 5-minute Hc
and Fr values for this hour are presented in Figure 38. The dashed
line shown in the Hc time series plot represents the tracer gas
release height. Hc drops from a high of 30 m at the beginning of
the hour to a low of 14 m, then rises to approximately 22 m for the
last 25 minutes of the hour. The average of the 5-minute values over
the hour is 21 m. Because Hc is greater than the release height for
only one 5-minute period, this hour is representative of flow above
the dividing-streamline height. Fr varies from a low of 0.7 during
the first half of the hour to a high of 1.3 during the last half of
the hour. The one-hour average value of Fr above 21 m is 1.0, which
indicates that stratification has a significant influence on the flow
over the top of the ridge.
Time series plots of the 5-minute sonic anemometer data from
Tower A are presented in Figure 39. The dashed line represents data
from the 5-m level of Tower A and the dotted line represents data from
the 40-m level. The 5-m and 40-m data are linearly interpolated to
the tracer gas release height and these release height values are
represented by the solid line.
The 5 m winds are light and variable during the hour with speeds
ranging from 0.3 to 1.2 m/s and wind directions ranging from 350 to
140°. The 40-m winds are steadier with speeds ranging from 2.2 to 3.0
m/s and wind directions ranging from 90 to 115° during the hour. The
hourly averaged vector wind direction and wind speed at the tracer
release height are estimated to be 93.9° and 1.6 m/s, respectively.
»
The vertical turbulence intensity (iz) values at the tracer
release height vary from 8 to 17% and the ow values at the release
height vary from 0.15 to 0.32 m/s. The 1-hour value for aw
estimated at the tracer release height is 0.24 m/s (iz - 15%).
An hourly average of the 5-minute temperature and propeller wind
data measured during this hour at the 10-instrument levels of Tower A
are used to construct vertical profiles. A "spline under tension"
method is used to interpolate the meteorological variables for every
5 m between instrument levels on Tower A. Figure 40 shows the
vertical profiles of hourly averaged wind direction, wind speed, and
temperature. The wind data from the propeller anemometer at 5 m and
40 m compare favorably with the sonic anemometer wind data at the same
levels. Also, the vertical profiles between 5 m and 40 m are
approximately linear, so the linear interpolation used to estimate MDA
values for wind speed and direction at the release height should be
acceptable. The interpolated profile values for the wind direction
and speed at the tracer gas release height are 98° and 2.0 m/s,
respectively. These values do indeed indicate that the linear
121
-------
SB
79
ea-
se-
481
38
Release Height
Time (Hour)
2.0
1.5-
1.0-
Time (Hour)
Figure 38. Time series of 5-minute calculated dividing-streamline
heights (Hc) and bulk hill Froude numbers above Hc
(Fr(Hc)) (Experiment 11, 10/23/82, 0600-0700 MDT).
122
-------
« M
(«/iu) poods puuw
3
o
—I 1 1 1—
? S ft •
-U
CO
01
LJ cd u-i
* -o
o a)
H • -P •
'"^ Cd E
e H -<
O Q O ^
& :
0) •
cd
U-i
o
o
o
-fl-
ed o
•O I
O 0)
O O Li 13
•i-l VD Cd P
CO cd
o
01
CM
(!) 00
D f)
C CN
.1-1 ^.
e o
i
•4-1
o
r-l -H in
0>
• H
H
• rl L, 0)
t-l Ll
0) Li 3
p. OQ 01
' n cd
DLJ n)
o I
00
(60Q) UOIJ09JIQ pUMA
123
-------
o
£
i
I
u
I
o
H
3 ? 8 S
g 3
to
-4J
•o
S
a
1
a>
M —
3 ?
S 2
S S
S 3
-8
0
I
CO
o
o o
u o
O r-
0) O
QJ O
e o
vO
!>> O
3
a. c
CO -i-l
O U
-H Q)
t: s-
OJ S3
O
•5T
10
124
-------
interpolation used to construct the MDA apparently provides a
reasonable estimate of the meteorology at release height.
In summary, the plume was released above Hc into a flow
characterized by some plume meander, 1.6 to 2.2 m/s wind speeds, and a
hill Froude number equal to unity. The largest observed
concentrations are expected to be found near or above Hc.
C F 3 B r Concentrations
The distribution of the observed hourly averaged CF3Br
concentrations scaled by the emission rate (ps/m3) is shown in
Figure 41. The concentrations from samplers suspended from Towers B
(sampler id: 702 and 703) and C (701) are listed above the release
information. In this figure, the 5-minute average wind flow vectors
estimated at the tracer gas release height are drawn at the release
position. The length of each flow vector is proportional to the
5-minute vector wind speed. The 1-hour average flow vector, derived
from a vector average of the 12 5-minute vector wind directions, is
depicted by the long dotted line emanating from the release position.
The largest concentrations are found towards the middle of the
sampler array near or above the hourly average Hc value (21 m above
the release base). The maximum observed concentration (52 ps/m^)
is found near the estimated hourly-averaged plume centerline at
sampler 112.
Model Performance
The hourly-averaged scaled concentrations estimated from the HBR
(Flat) model are presented in Figure 42. The summary table, shown in
the lower right-hand corner of the figure, includes statistics for all
one-hour average concentrations except those from samplers suspended
from Towers B and C. MCO/MCP is the ratio of the maximum observed to
the maximum predicted concentrations, unpaired in space. The
estimated concentration pattern compares favorably with the observed
(r^ = 0.61), but peak concentrations are underestimated (MCO/MCP =
1.62).
The hourly-averaged scaled concentrations estimated from the HBR
(Terrain) model (see subsection 4.3.3) are presented in Figure 43.
The maximum predicted concentration has increased from 32
from the flat-terrain version to 47 ys/m3. This still
underestimates the maximum observed concentration of 52
Overall, the estimated concentration pattern compares well with the
observed (r^ = 0.61) with the ratio of the mean observed to the mean
predicted equal to 0.89.
4.4.2 Experiment 14, Experiment-Hour 6 (0300-0400)
Release Description
The CF3Br tracer gas was released from position 203 at 20 m
above the ground for the entire hour. The release was continuous from
the previous hour at a rate of 0.93 g/s.
125
-------
ro I OQ X
wxn
X X
L3X
X
O 00 *~* *~*
ui i • ro en
-" in
Cf — 01 — '
UJ ••
01Q. II II II
CD xo: >
o iu NI ca tn
126
-------
» » i • •
o o. Q. —
u u u a.
— — i u v
• B O I O
U O U
-OX
127
-------
"I« ".". -
« «i r» 10 •
k k O I o
ess
k -
rv w r«. o> o> » in
jr w - T * to *
o a. a. — a. — 01
o r^ — ru •
•• <>• ii u ii
ceui
10:310
3
ta :c m x in
to to
\ x
ox
CO -•-«
i -men
--in • ^r
OCL II II II
t-XDC >•
O LU M C3 (rt
en
^
-------
Local terrain elevations near the release point are estimated to
be 17.4 m above the base elevation of the hill coordinate system, so
the net release height corresponds to the 37.4 m height level on the
ridge.
Meteorological Information
Figure 44 contains plots of the 5-minute Hc and Fr values for
this hour. Hc decreases from 23 m in the beginning of the hour to
5 TO half-way through the hour, then rises to 34 m during the latter
part of the hour before decreasing to 10 m at the end of the hour.
The average of the 5-minute values over the hour is 18 m. Fr
gradually rises from 1.3 to 1.8 during the hour. The one hour average
value of Fr above Hc is 1.5, which is an indication that
stratification has only a weak effect on the flow over the top of the
hill.
The hourly averaged vector wind direction and wind speed at the
tracer release height are estimated to be 59.3° and 1.2 m/s,
respectively. Figure 45 shows the trend in wind speeds and directions
between 5 m and 40 m during the hour. There is little directional
wind shear for approximately two-thirds of the hour, but there is up
to 100° of directional wind shear during the remaining third of the
hour. Large directional wind shear between 5m and 40 m occurs in
combination with increasing Hc above 5 m. The flow direction is
more variable well below Hc than above and this is apparently due to
the blocking effect of the nearly two-dimensional ridge. The wind
speed time series varies inversely with the Hc time series. As the
wind speed increases during the first half of the hour, Hc
decreases; conversely, as the wind speed decreases during the second
half of the hour, Hc increases.
The vertical turbulent intensity values estimated at the tracer
release height vary from 8 to 27% and the av values estimated at
the tracer release height vary from 0.18 to 0.29 m/s during the hour.
The one hour value for aw estimated at the release height is
0.22 m/s (iz = 18%).
The vertical profiles of the hourly averaged wind direction, wind
speed, and temperature are shown in Figure 46. For the layer 10 m
above and below the release height, the wind speed shear is 0.5 m/s
and the directional shear is 32°. The wind data from the propeller
anemometers at 5 m and 40 m compare favorably with the sonic
anemometer wind data at the same levels. The wind direction profile
between 5 m and 40 m is approximately linear, so the MDA value for the
wind direction should be reasonable. The hourly average of the
5-minute propeller wind direction and wind speed values measured at
the tracer gas release height are 50° and 1.7 m/s which corroborate
the one-hour average MDA values.
In summary, the plume was released above Hc during the first
half of the hour and below Hc during the second half of the hour.
The hourly average value for Hc is 18 m which is close to the
release height. The flow is characterized by substantial wind
129
-------
u
Release Height
Time (Hour)
2.0T-
1.5-
o
5- i .0-
(Experiment 14, 10/26/82, 0300-0400 MDT).
130
-------
m IM
(s/ui) peodg PUIAA
3
o
I
o
0)
tn
*£ 3
r-f
lj (d
jf
O
fl •
<*-s
e H
0 Q
£ s
O
(D O
4J *^
n) o
•O 1
o
o o
••-I CO
G O
o
w -
CM
Oi 00
-*J ^.
D \o
C CM
•H \
e o
1 H
in
e
o
Ll
14J
•a
H ^ o e
60
IM IM
-------
t- :
O.
o
H
O
U
8 8
8 ? 8
8 8 S
S
JR R
S 8
CO
•o
o
•a
to
•o
a
o
00
o
— o
M (1)
\ ^
1 I
o
o
O
o
CO
O
•_ .
3 CM
O 00
(4-1 CM
O >»
O
14-1
o
- §
« -rM
O U
• r-t 0)
*•» ^
vo
-5T
Q
t-i
00
132
-------
meander, wind speeds varying from 0.7 to 2.2 m/s, and a hill Froude
number of 1.5.
Cjf.3§r Concentrations
The distribution of the observed hourly averaged CP^Br
concentrations over the surface of the ridge is shown in Figure 47.
The concentration pattern is consistent with the variable wind
directions observed during the hour. The largest concentrations are
found near the bottom-half of the sampler array at an elevation that
is less than Hc. The maximum observed concentration (117 ys/m^)
is found near the estimated hourly-averaged plume centerline. There
is a sharp decrease of observed tracer concentrations above the Hc
surface (1635 m).
Model Performance
Results from the HER (Flat) model are displayed in Figure 48.
Overall, the mean of the concentration estimates is in close agreement
with the mean of the observed concentrations (Co/Cp - 1.04,) but
the spatial correlation is poor (r^ = 0.31). The maximum observed
concentration (117 ps/nP) is more than a factor of two greater
than the maximum modeled concentration (52 ps/m3).
Estimates from the HBR (Terrain) model, shown in Figure 49, are
somewhat larger than the flat-terrain estimates. The maximum modeled
concentration (79 v>s/m3) is less than the maximum observed
concentration (117 ys/rn-*), although on average, the model is over
estimating (cfo/cl = 0.74). Because the distribution of the peak
observed concentrations in space is so unlike that estimated by the
model using one-hour average data, the temporal variations in the
meteorology (particularly iz) may be responsible for the poor model
performance in estimating the peak concentrations.
»
4.4.3 Experiment 6, Experiment-Hour 9 (0700-0800)
Release Description
The CF3Br tracer gas was released from position 203 for the
entire hour at a release height 20 m above the ground. The release
was continuous from the previous three hours. The CF3Br release
rate is calculated to be 1.30 g/s.
Local terrain elevations near the release point are estimated to
be 17.4 m above the base elevation of the hill coordinate system, so
the net release height corresponds to the 37.4 m height level on the
ridge.
4
Meteorological Information
The plots of the 5-minute Hc and Fr values for this hour are
displayed in Figure 50. The 5-minute values of Hc are all greater
than the release height, except for the second 5-minute period when
Hc drops to 14 m. This sudden drop in Hc corresponds to a 1.0 m/s
133
-------
u>
I
I E
co
CO
:D
10 T
-. r- in
X
•
O uj M C3 CN
u o
-4-1 i-H
0) -
O -3-
C rH
o
o -P
ti a)
CD e
CO-H
ii U
u a)
CJ *~*
mm
oo -
cd
H
O
(U
C
O
t>0
O
C O
o <•
.rH O
CO I
en o
134
-------
135
-------
1
I o
I <-<
" <•> in c; f-
«) w « sr •
(VI (V (U — -•
a a. a. — o
o u o a. x
*» *» o i a
o o o
- D JC
7 oo <>• 19 ar
ULU N O t/1
CO
e -
(N
tn ^
c o
O iH
CO
c e
O -H
o u
(U
*o o*
3 J=
O *1
0) O O
C [j co
O "4-1 O
ao
bu
136
-------
o
Z
Release Height
Time (Hour)
2.8-
1.5-1
.0-
Time (Hour)
E'igure 50. Time series of 5-minute calculated dividing-streamline heights
(Hc) and bulk hill Froude numbers above Hc (Fr(Hc))
(Experiment 6, 10/31/82, 0700-0800 MDT).
137
-------
increase in Tower-A propeller wind speeds. This increase in speed is
not corroborated by the sonic anemometer wind data. The average of
the 5 minute Hc values over the hour is 35 m. Because Hc is less
than the release height for only one 5-minute period and this value is
suspect, this hour is considered representative of flow below Hc.
Fr remains steady during the hour varying from 1.0 to 1.3, with the
one hour value of Fr above Hc being 1.2.
The hourlr averaged wind speed and direction estimated at the
tracer release height are 1.0 m/s and 67.1°, respectively. Figure 51
shows the trend in wind speeds and directions between 5 m and 40 m
during the hour. The linearly interpolated values estimated at the
tracer release height are also shown. The 5-m wind directions vary
from 1 to 51° during the hour and the 40-m wind directions vary from
72 to 164°. The wind speed measured at 5 m oscillates from a high of
1.5 m/s during the beginning of the hour to a low of 0.6 m/s in the
middle of the hour, then increases to 1.2 m/s towards the end of the
hour. The wind speed measured at 40 m steadily decreases from 1.6 to
0.6 m/s during the hour, with the exception of a 1.7 m/s peak measured
during the seventh 5-minute period.
The trend in iz and aw values during the hour are shown in
Figure 51. The values of iz and aw estimated at the tracer
release height vary from 7 to 10% and 0.06 to 0.13 m/s, respectively.
The one hour value for
-------
(M
(s/iu) peeds
O
(9
(O
a>
U)
—r—
s>
m
CB
(M
IZI
o
o
H
(B
I
CD
W
3
r-l
CQ
o
H
e
o
u
y
•r-l
c
o
w
o
o
oo
o
o
o
CO
r-l
o
o
i m
co
At
u
(Baa)
111
(s/iu) M/>
139
-------
o
I
I
0)
o
H
O
8 5 S 8
co
•o
8 5 S
I S 8
3 3
a
O
CJ i
g o
,-t O
u
O CM
.c oo
in n
O r-<
-V
w o
P. C
-1
CO -H
o
.H
o
s-
CM
LO
ao
140
-------
141
-------
release height. This is the largest observed concentration found in
the Ct^Br modeling data set. Also, this sampler is collocated with
sampler 809 where the observed concentration is 425 ys/m-*. A
sharp decrease of observed tracer concentrations is found above the
mean Hc surface (1652 m).
Model Performance
The hourly averaged scaled concentrations estimated from the HER
(Flat) model are shown in Figure 54. The model grossly underestimates
the mean of the observed concentrations by more than a factor of 4
with~Co/Cp - 4.71 and r^ = .004. The maximum observed
concentration (445 ps/m^) is found towards the bottom-half of the
ridge at an elevation of 26 m; whereas, the maximum estimated
concentration (24 ys/m-*) is found towards the top of the ridge at
75 m.
Estimates from the HBR (Terrain) model, shown in Figure 55, are
considerably larger than the flat-terrain estimates with Co/cT =
1.02. However, the maximum estimated concentration (105 ys/m^)
found towards the center of the sampler array at 56 m is still much
less than the observed maximum concentration, and occurs at a location
too far up on the hill. The model performance apparently suffers
during this hour because the plume seems to have "dropped" in height
over a short distance from the release and HBR (Terrain) does not
account for this.
4.4.4 Experiment 8, Experiment-Hour 7 (0500 - 0600 MDT)
Release Description
The CF3Br tracer gas was released from position 215 at 25 m
above the ground for the entire hour. The CF3Br release rate is
computed to be 0.97 g/s. >
Local terrain elevations near the release point are estimated to
be 12.9 m above the base elevation of the hill coordinate system, so
the net release height corresponds to the 37.9 m height level on the
ridge.
Meteorological Information
Figure 56 contains plots of the 5-minute Hc and Fr values for
this hour. The tracer gas release height is less than Hc for the
first eight 5-minute periods of the hour. The average of the 5-minute
values over the hour is 35 m. Fr is steady during the first eight
5-minute periods of the hour ranging from 0.8 to 1.0. For the last
third of the hour, Fr rises to 1.5. The one hour average value of Fr
above Hc is 1.0.
Time series plots of the sonic anemometer data from Tower A for
this experiment-hour are presented in Figure 57. Large directional
wind shear is found between 5 m and 40 m until the tenth 5-minute
period when the 5-m wind direction shifts from 56 to 152°. There is
142
-------
en
^v
r/l
3.
CM
00
IJ
£ -u
O C
O
•O
4) ,-H
4-> 0)
O 73
.* O
*o c
0)
a -P
a)
00 b
0)
>
ca
u
o
£
o
01
Jj
3
00
Dd Q
m Z
o
-> o
I
e o
o o
143
-------
W CM
a oo
in r-t
C x
O O
•H rH
-P
n) -
c *>
a) c
o a)
c e
o -^
o u
01
*O &•
a) 5<
i-t W
B) ^
O
01 ^1
•a -a
<]) O
+> e
o
•a c
o 4-1 o
e o
o o
&o
•H
fc.
144
-------
1.5-
U
I
Release Height
Time (Hour)
~1
Time (Hour)
Figure 56. Time series of 5-minute calculated dividing—streamline heights
(Hc) and bulk hill Froude numbers above Hc (Fr(H))
(Experiment 8, 10/15/82, 0500-0600 MDT).
145
-------
(s/iu) peedg PUJM
3
O
(%) Zl
3
O
\
u
0)
o
H
o
Ll
14-1
o
cn
cu
-P
C
• H
e
H
Q
£
O
O
O
I
o
o
0)
-P •
CO 6
*—1
O /-N
& :
-------
little wind speed shear for the first eight 5- minute periods of the
hour. During the final third of the hour, the wind speeds between 5 m
and 40 m differ by as much as 2.0 m/s. The hourly averaged vector
wind direction and wind speed estimated at the tracer release height
are 83.7° and 1.2 m/s, respectively.
The iz and aw values are fairly steady and low for the
first half of the hour with iz values less than 10% and crw
values less than 0.1 m/s. During the second half of the hour, there
is a sharp increase in measured turbulence. The estimated value of
aw at the release height peaks at 0.4 m/s and iz peaks at 30%.
The one hour value of ow estimated at the tracer release height is
0.21 m/s (iz - 17.5%).
The vertical profiles of the hourly averaged wind direction, wind
speed, and temperature are shown in Figure 58. There is approximately
60° of directional shear in the vicinity (+10 m) of the tracer release
height. The wind speed decreases for the 10-m layer below the release
height. The wind data from the propeller anemometers at 5 m and 40 m
are in fair agreement with the sonic anemometer data at the same
levels. However, the vertical profiles of the propeller wind data
between 5 m and 40 m are not linear. The hourly average of the
5-minute propeller wind direction and wind speed values interpolated
to the tracer gas release height are 130° and 1.3 m/s, respectively.
The wind speed value is in close agreement with the hourly averaged
MDA value. However, the wind directions differ by 46° which indicates
that the linear interpolation is questionable. In this case, the 40 m
sonic wind direction (128°) is better than the interpolated value.
In summary, the plume was released on average below Hc into a
flow characterized by significant directional wind shear, 0.9 to
2.2 m/s wind speeds, and a hill Froude number above Hc equal to
unity.
»
Concentrations
The distribution of the observed hourly averaged CF3Br
concentrations over the surface of the ridge is shown in Figure 59.
The largest concentrations are found towards the middle of the sampler
array at an elevation that is near or less than the hourly averaged
value for Hc. The maximum observed concentration (72 ps/m^) is
found away from the estimated hourly-averaged plume centerline at
sampler 205. This large concentration may be associated with the last
third of the hour when Hc was near the tracer release height and the
wind direction was towards this sampler. The second highest observed
concentration (69 vis/m3) is located near the estimated hourly
averaged plume centerline at an elevation that is less than Hc. It
is most likely that the estimated hourly-averaged plume centerline is
incorrect. The concentration pattern is more representative of flow
from 130°, as indicated by the propeller anemometer data.
147
-------
IB
tn
CD
(M
S>
CT
(LU)
o
¥
8 <
CD
Q. P
I s
* £
e
o
IB
ID
(M
(B
a
IT
CO
•a
CO
O
f
TJ
Q.
(A
TJ
00 H
O Q
-< s
O
u o
o o
0) vO
-P O
Q) I
e§
X m
r-t O
O CSJ
^r oo
tn
(B
(9
c
O
O
£
a
•o
c
e-
w
00
in
!iO
148
-------
149
-------
Model Performance
Scaled concentrations from the HBR (Flat) model, shown in
Figure 60, are much less than the observed concentrations with
CQ/CP = 2.17 and r2 = 0.05. The largest estimated
concentrations (20 ys/m^) are found near the crest of the ridge at
a total of nine sampler locations.
Scaled concentrations estimated from the HBR (Terrain) model are
shown in Figure 61. On average, the model does well with Tfg/Cp =
0.88 and MCO/MCP = 1.29, although the correlation is low (r" =
.08). The observed concentration pattern exhibits a bi-modal tendency
which mirrors the wind direction pattern. This appears to contribute
to the low correlation with the concentration pattern estimated with a
Gaussian distribution about the mean wind direction.
4.4.5 Summary
The flat terrain model and the empirical HBR model have been
tested against CF3Br observations for four case study hours. These
case study hours represent three classes of meteorology (zr < Hc,
zr ~ Hc, zr > Hc) and include the highest concentrations
observed within the*three classes. Also, the MDA wind directions are,
in general, consistent with the observed concentration patterns during
these hours.
Comparisons were made between the sonic and propeller anemometer
data to evaluate the method used to construct the preliminary MDA for
the release height. In general, data from the sonic anemometers
compare favorably with data from the propeller anemometer for these
case study hours. However, there are many hours in the preliminary
CE^Br MDA that contain uncertain meteorological data, probably
because the linear interpolation between 5 m and 40 m is
inappropriate. The greater vertical resolution provided by the Tower
A propeller anemometer data will most likely provide more
representative meteorological data which should improve the modeling
results.
For each case study hour, the ratio of the maximum observed to
the maximum predicted concentration (unpaired in space) is closer to
unity for the empirical HBR model than the flat-terrain model.
Significant discrepancies between observed and modeled concentrations
(from the empirical HBR model) appear to result in part from
variations in the meteorology during the hour that are not adequately
represented in the "hourly" Gaussian plume formulation.
150
-------
O 0. 0. — I)
o u u a
~ ~ i u N
e b o i o
<-> a u
- o r
151
-------
-• OD r* ai
• • • "
U) (D O) ~'
If II M H It
o a. a. — u
u u o a. z
— ~ I u> x
» k o i o
u a u
» —
t
-« 0
• • • * OB CD r**
u> m 10 • «»
(U Al I -
n ii i i it H •
o a. a. ~ a. ai z
u u o a. u w
i u N w
o i o or
u a o
u
f)
)K
m
CO
CO
Of)
:co
x
a: •>:
CD oj rv
ro (s -. • o
U. (3 CD IT) (VI
o co —• ro •
•• •• u u ii
(K UJ
co :DCO
CJ LU M C5 r/i
'f, CM
3. 00
v-^ V.
LO
01 r-(
C ^
O O
•H H
-P
Cd -
U 03
C
0)
O
CD
O -H
O Ui
CD
•a a,
4) X
•—I [T1
cd ^
CJ
m ^^
0)
•o -a
6
o
•O "c
CD -H
'M Cd
O. !J
CD CD
OB 4-" •
'<-• H
CD « Q
> 33 2
Ct) X
3 .C
-------
SUCTION 5
THE PRELIMINARY TRACY EXPERIMENT
5.1 Geographic and Meteorological Setting
The Tracy Power Plant (TPP) near Reno, Nevada has been selected
as the site for the Full Scale Plume Study (FSPS), the third field
experiment in the CTMD project. The Tracy plant was tentatively
selected in May 1983 after (1) a review of operating power plants in
the western United States and of available topographic, meteorological
and air quality data, and (2) site visits to three power plants. The
final selection was made after a preliminary experiment that was
conducted at Tracy in November 1983. This section summarizes the
results from the preliminary experiment and discusses the plans for
the FSPS.
The Tracy station is operated by Sierra Pacific Power Company.
The Sierra Pacific personnel had agreed to participate in the program
and were very cooperative in the design of the FSPS. The power plant
is located about 27 km (17 mi) east of Reno, Nevada in the Truckee
River Valley. It has three units--53 MW, 80 MW and 120 MW, although
only the latter has been used recently. The 120-MW unit is serviced
by a 91.4-m (300-ft) stack. This stack was used to release the
oil-fog and SF$ during the preliminary experiment.
Figure 62 shows the location of the Tracy station on a 1:250,000
scale topographic map. The plant is located east, of the Reno-Sparks
metropolitan area and about 40 km (25 mi) east of* the Sierra Nevada
Mountains, the source of the Truckee River. The river runs eastward
near the plant and eventually drains into Pyramid Lake.
The 91.4-m stack is located near the southwest corner of the
plant (Figure 63) south of the Truckee River. The plant site is in a
relatively narrow valley with mountains surrounding the plant on all
sides. Figure 64 shows the location of the plant on a 1:62,500 scale
map. Mountain peaks rise to elevations of 900 m above the stack base
elevation within 6.5 km of the plant. The area is characterized by a
sparse vegetative cover of shrubs and grasses. Much of the
uncultivated area in the immediate plant environs is covered by small
boulders.
The Truckee River enters the valley through a narrow opening near
Patrick. It flows eastward just north of the plant and then takes an
abrupt turn to the north about 4 km east of the plant. The river
flows between two mountains at its northward bend. These two
mountains are the primary "target" areas for the dispersion
experiments.
153
-------
'
n.
Ui
o
&>
o
(0
0)
5
•o
I
tJ
CO
c
o
•H
00
0)
CM
\D
O
u
00
154
-------
155
-------
c
1-1
a.
o
Du
U
-------
In addition to the Tracy station, there is a diatomaceous earth
plant run by Eagle Picher Industries Inc. in the valley near Clark.
There are also a few ranches in the valley. Interstate 80 runs north
of the plant along the river.
No historical meteorological record is available from anywhere
in the valley. However, a previous field program (Kapsha et al.
1976), which included aircraft and mobile van measurements of SO2 as
well as pilot balloon data, suggested plume transport from the Tracy
stack to locations that would produce ground- level concentrations on
mountains 5229 (called beacon hill) and 5764 (called target mountain)
east of the plant and around the mountain 5610 complex northwest of
the plant. This field program was conducted in December 1975 and
experienced easterly wind flows* associated with transistory
anticyclones, as well as the more usual stable drainage winds. In
July 1983, as part of their initial feasibility testing, ARLF'RD
released oil-fog from the Tracy stack, and the smoke plume was
transported to and interacted with target mountain and beacon hill.
Although there is currently no corroborative data base, it is expected
that during the late summer, nighttime winds in the valley will be
dominated by terrain effects--probably producing a prevailing westerly
wind down the Truckee River with superimposed local katabatic effects.
5.2 Experimental Design
The preliminary flow visualization and tracer experiment that was
conducted during the period November 7-20, 1983 was co-sponsored by
EPA and the Electric Power Research Institute (EPRI). The EPR1
participation was in anticipation of a complex terrain field
experiment as part of their Plume Model Validation and Development
project. The results of the preliminary Tracy experiment will be used
to guide the design of the next PMV&D field experiment and to provide
the PMV&D modelers preliminary information on the relationship among
emissions, meteorological conditions, and observed concentrations in a
complex terrain setting.
The EPA objectives of the November experiment were (1) to assess
the feasibility of the Tracy site for the FSPS and (2) to obtain
sufficient information to design and plan the full scale experiment.
These objectives were satisfied and the site was selected for the
FSPS.**
The experimental methods were similar to those used and tested at
CCB and HBR and at the two previous EPRI field sites. The experiment
included:
• Releases of SV& and oil-fog from the 91.4-m stack;f
*Easterly winds were also experienced during the 1983 Tracy
experiment. See Section 5.5
**A detailed work plan for the FSPS was prepared. See also
Section 5.6.
t In the two small hill experiments, releases were made from mobile
cranes.
157
-------
• Ground- level SF^ concentration measurements at up to 53
sites;
• Fixed meteorological measurements:
a 150-m tower instrumented at four levels (5, 10, 100
and 150 m),
two 10-m towers instrumented at one level,
two monostatic acoustic sounders,
a doppler acoustic sounder, and
two optical crosswind anemometers;
• Two tethersondes;
• T-sonde releases with double theodolite tracking at two
locations;
• Two solar-powered electronic weather stations;
• Airborne lidar; and
• Photographs and videotapes.
The participants and their principal responsibilities were:
• ERT (EPA prime contractor)
(1) directed experiment operations in consultation with
other participants;
(2) operated the command post (Sierra Pacific Power
provided office space at the Tracy station);
(3) provided smoke candles, two carbon arc lamps, cameras
and personnel for flow visualization experiments and
scientific observations;
(4) provided, installed and operated two electronic weather
stations to measure winds and temperature at two
locations;
(5) provided, installed, and operated a 150-m tower
instrumented at four levels to measure winds,
turbulence, and temperature;
(6) provided for other site logistics as needed; and
(7) produced the master data archive ayd disseminated it to
participants.
• NOAA ARLFRD (via an interagency agreement with EPA)
(1) provided, installed, and operated two 10-m towers with
wind and temperature instruments telemetering to
command post for display and storage;
(2) provided data logging equipment; archived, reduced and
disseminated the meteorological data;
(3) adapted fogger to Tracy flue, provided oil and operated
fogger; provided SFg and injected into flue (with oil
fog);
(4) provided one tethersonde and operator;
(5) provided radios and repeater station;
(6) provided for photography contract (two photographers,
about four time-exposures/hr and one video during
daylight).
• NOAA WPL (via an interagency agreement with EPA)
(1) provided one tethersonde and operator,
(2) provided Doppler acoustic sounder and operator,
158
-------
(3) provided two monostatic acoustic sounders,
(4) provided two optical crosswind anemometers.
• SKI International (KVRl contractor)
(1) provided and operated airborne lidar (approximately two
3 hour missions per experiment).
• Rockwell. International (KPR1 contractor)
(1) provided 1-hour sequential (9 hrs) syringe samplers
(including spares) to operate on consecutive nights at
53 locations, and provided necessary deployment crews
and vehicles (including helicopter);
(2) determined SFg concentrations (by GO in all syringes
and bags with a turnaround time of about 24 hours;
(3) provided equipment and took T-sondes to 2.5-3 km above
ground at 1-hr intervals at two locations during tracer
releases; and
(4) provided survey/identification of all SFg samplers,
meteorological instruments, and fixed photography
locations.
• Research Triangle Institute (EPRI "External Audits"
contractor)
(1) provided an independent review of QC plans of other
EPRI contractors; conducted onsite systems audits of
the field measurements and the data handling activities;
(2) provided independent checks of the precision and
accuracy of the field measurements and data handling
results. Onsite performance audits were performed at
the Tracy Power Plant for the following measurements
systems: tracer, T-sonde, tethersonde, and two 10-tn
tower? with wind speed, direction, turbulence, and
temperature measurement systems; and
(3) provided reports to EPRI, through TRC, on the results
of systems and performance audit results.
• TRC Environmental Consultants (EPRI technical management
contractor)
(1) represented EPRI in the field and coordinated Kf'Rl
contractors, and
(2) undertook scientific observations and analyses as
appropriate.
5.2.1 Oil-fog and Tracer Gas Release System
ARLFRD provided an oil fog generator and a SFg release system
to inject oil-fog and SFg directly into the 91.4 m stack flue at the
Tracy station. The injections were made through a "door" into the
ducting leading to the 91.4-m stack (Figure 65). The SFg tracer gas
was stored in two compressed gas cylinders at ground level. Piping
carried the gas through a linear mass flow meter (LKM) system to the
point of discharge into the stack. The LFM measured and displayed the
rate of gaseous tracer discharge via real- time digital
159
-------
T60
-------
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. Similarly, the oil consumption rate was logged to
document the quantity of oil-fog injected through the 91.4-m stack.
A nominal SF$ release rate of 1.26 g/s (10 Ib/hour) was used
for the Tracy experiment. The SFg and oil-fog releases commenced
approximately 30 minutes prior to the start of the sampling to ensure
that the tracer gas had actually reached the sampling grid when the
samplers were turned on. During the course of the experiment, the
91.4-m stack was vented through the use of a fan located at the bottom
of the stack. Only occasionally during the experiment was the unit
used to generate electricity. During these times the generation rate
was typically 20 MW.
5.2.2 Tracer Sampling and Analysis
Fifty-three syringe samplers were deployed by Rockwell
International (Cher 1984) to sample hourly concentrations. Sampling
sites were selected from 63 sampler locations (Figure 66).
Forty-three locations were specified to sample concentrations during
westerly winds and were used during the first few experiments. During
the course of the experiment, easterly winds were experienced
frequently so it was decided to select additional sites west of the
plant.
The samplers used in the program were sequential syringe samplers
manufactured by D&S Instruments. Nine 30 cm-^ syringes were housed
in each sampler. The syringe samplers functioned over a 9-hour period
with syringes sampling for consecutive 1-hour periods. Because of the
rough terrain, deployment of samplers required the use of either a
helicopter or a 4-wheel drive vehicle. t
In order to identify each sample by date, location, and time, a
numbering system was devised whereby each sampler, stake, and syringe
was given a unique number code. The numbers were printed on special,
double sided labels using a computer generated bar code system.
Labels were attached to the samplers, stakes, and syringes. Upon
deployment of each sampler, stake labels and sampler labels were
affixed to s. data sheet, which also contained information on the
syringe codes associated with the sampler. During analysis, a section
of the double-sided label from the stake, sampler, and each syringe
was transferred to the strip chart record containing the corresponding
trace. In this way, the strip chart record and the deployment data
sheet contained complete, redundant records of the deployment history
of each sample. When the strip chart records were read, all labels
were scanned with a bar code reader, and therefore all identifying
information was automatically transferred to the computer with
essentially no transcription errors.
A typical experiment ran from midnight to 9 A.M. Samplers were
loaded with syringes two days before the scheduled test. Loading
consisted of labelling the new plastic syringes, attaching the
161
-------
01
C
o
01
o
o
-------
needles, installing the units in the sampler, replacing any missing
septa, and setting the timers for the desired timing sequence and
starting time.
Deployment of the samplers occurred the day before the test and
exposed samplers were retrieved at the end of the test. For this
program a sufficient number of samplers were available so that
deployment and retrieval of samplers from the previous test could be
done simultaneously. Deployment and retrieval times were entered in
the data sheets for the two tests, and the stake identifying label was
also transferred to the deployment sheet at this time. To avoid
mixing samplers from different tests, the exposed samplers were marked
upon retrieval with an identifying tag. Analysis of the air samples
was usually performed on the same day as the test.
In addition to the syringe samplers, ARLFRD provided five
sequential bag samplers. These were collocated with five syringe
samplers for quality assurance purposes.
The analyses of the air samples were performed by gas
chromatography using an electron capture detector. Three Varian 3700
gas chromatographs were used, of which two were fitted with dual
columns to speed up the analyses. The columns were 3.175 mm x 1.829 m
stainless steel packed with molecular sieve 5A. The oven temperature
was 50°C, the detector temperature was 150°C, and the carrier gas was
nitrogen flowing at the rate of 50 cm3/min. Under these conditions,
the elution time of SFg and 02 were approximately 25 and 40
seconds, respectively. The entire analysis time (with backflushing
starting as soon as the oxygen peak began to elute) was approximately
2.5 min.
Each column-detector combination was calibrated using standards
obtained from compressed gas cylinders (Scott Specialty Gases)
containing manufacturer-certified SFg concentrations of 17, 83, 505,
and 1145 ppt SF6. Calibrations were repeated every two hours. The
response of the gas chromatographs remained constant within 10%.
Because the response of one of the gas chromatographs was slightly
non-linear, the response curves for all calibrations were determined
by fitting calibration data to a quadratic equation of the form:
concentration = ah(l+bh)
where a and b are calibration constants, and h is the peak height.
For the highest concentration used (1145 ppt) and the most non- linear
response gas chromatograph, the ratio of the quadratic term to the
linear term bh was approximately 0.15-0.20 or 15-20%. All calibration
constants were stored on disk using a DEC Professional 350 Computer.
The SFg peak heights were digitized electronically using a Science
Accessories Corporation sonic digitizer Model GP6-40 connected
directly to the computer. Peak heights were converted to
concentrations by applying the appropriate calibration constants.
163
-------
5.2.3 Plume Photographs
Two dedicated photographers took 5-minute exposures of the
oil-fog plume approximately every 15-minutes during the course of the
experiment. Their locations are depicted in Figure 67. The plume was
illuminated by two carbon arc lamps- one located south and one north
of the river. In addition, the ERT scientific observers took plume
photographs of interest from a number of locations.
5.2.4 Airborne Lidar Sampling
SRI International provided the ALPHA-1 airborne lidar to document
the three dimensional interaction of the oil-fog plume with the
terrain. The aircraft flew a "creeping-ladder" pattern from the stack
to approximately five to ten kilometers downwind. The flight pattern
legs were approximately at right angles to the mean plume direction.
The aircraft typically flew two three-hour missions during each
experiment. Facsimile lidar cross sections of the plume were
available at the end of each experiment for analysis by the project
scientists.
A separate report on the Alpha 1 observations was prepared by
Uthe and Morley (1984). Some of their material is presented in
Appendix C.
5.2.5 Meteorological Measurements
Table 15 lists the meteorological instruments used during the
November experiment. Their approximate locations are depicted in
Figure 67. The minisondes were released and tracked hourly on the
hour. The tethersondes were used to obtain vertical profiles of winds
and temperature up to 600 m. The tethersonde ascents occurred during
the first half-hour of each hour and the descents during the second
half. The tethersonde data were processed continuously so that the
data were available for near real-time analysis and operational
planning.
All data (including SFg) were achived and identified by the end
time (PST) of the averaging period, e.g., an hourly average wind
direction for the period 0200-0300" is identified as a 1-hour average
ending at 0300. Shorter-term measurements were also identified by the
end time of the measurement, e.g., Doppler wind data at 0200, 0220,
0240.
5.3 Preliminary Field Study Results
5.3.1 Summary of Data Base
Ten experiments were conducted for 73 hours during the period
7-20 November 1983. The use of 43 sampling sites was planned for the
first four days and 53 for the last six. Because of high winds and
snow, only 37 and 30 samplers were deployed for two experiments.
During the first four days, unfavorable weather conditions resulted in
164
-------
.
! :vg> \ W 7p--W: m^r
'!
h-7
'WV/'*f}\/SS&jg$^ \
\J'.v^oH;;jo:\ \f li ] .,
3
O
C
0)
a,
x
la
o
(-.
a,
•£>
0)
•x)
165
-------
TABLE 15. METEOROLOGICAL INSTRUMENTS AT THE
PRELIMINARY TRACY EXPERIMENT
Instrument
T-sonde (two)
Tethersonde (two)
Doppler acoustic sounder
To provide wind and temperature data
throughout the boundary layer (up to
3000 m) near the west and east
boundaries of the experiment-region.
To provide wind and temperature data up
to 600 m in (1) the flow upwind of the
stack and at (2) the flow upwind of the
bend to the north of the Truckee River.
To provide real-time information on the
vertical profile of winds near the
source location; data are archived
every 10 minutes.
Fixed meteorological sensors To provide
two 10-m towers
(winds and temperature)
two electronic weather
stations (winds and
temperature)
one 150-m tower east of
the stack (winds, tur-
bulence and temperature
at four levels -5m,
10 m, 100 m, 150 m)
Monostatic acoustic
sounders (two)
Optical anemometers (two)
(1) real-time information on
meteorological conditions in the
high terrain,
(2) data on the drainage flow, and
(3) data (historical only) rep-
resentative of plume conditions.
To provide information on the structure
of the boundary layer (1) upwind of the
source, and (2) upwind of the bend in
the river.
To provide information on the drainage
flow on the south side of target
mountain.
166
-------
few ground- level SFg concentrations. Table 16 Summarizes the
available concentration data from the entire Tracy experiment. A
total of 3167 SF$ concentrations are in the data base. All the
remaining measurements have also been delivered to ERT and are now
part of the Tracy Preliminary data base, which is summarized in
Table 17.
The complete data base was delivered to EPA accompanied by a
descriptive data report. The data base will be available from the EPA
Project Officer.
Maps of the SFg concentrations and the various meteorological
data were examined to assess which hours are sufficient for modeling.
From the 68-hour tracer data base, it was judged that approximately 34
"good" hours and 13 "marginal" hours* are available for modeling
purposes. Table 18 summarizes the modeling data base.
5.3.2 Overview of Results
The November experiment did achieve the program objectives:
(1) the Tracy site is feasible for the FSPS; and (2) the tracer gas,
meteorological and photographic data base are sufficient to design the
FSPS (see Section 5.6). Although the data base from the November
experiment is small compared to the CCB and HBR data bases, the
experiment did capture a wide variety of dispersion conditions.
Tracer gas concentrations were observed in the primary target areas
during stable conditions and also during windy, neutral conditions.
The data can be used to evaluate the CTMD modeling approaches and to
extend the hill and ridge data bases.
5.4 Example Results from Specific Experiments
To illustrate some of the experimental results, a few hours from
three experiment days are described qualitatively in terms of the
relationship among emissions, observed meteorological conditions and
subsequent ground- level tracer gas concentrations. Predominantly
stable atmospheric conditions occurred during the first two days and
windy, neutral conditions on the third day.
Experiment 5 (November 12, 1983) 0000-0500 PST
concentrations were observed in the high terrain of the
primary target areas during the course of Experiment 5. Figure 68
shows the geographical distribution of hourly concentrations observed
during the first five hours. In this figure, the dotted lines
indicate the position of Interstate- 80 and the Eagle Picher haul
road. The river flows from the lower left to the upper right and it
parallels Interstate-80. The X in the middle of the diagram
represents the location of the stack. The concentration range of
is indicated by the numerical symbols, defined as follows:
*The criteria for good hours are qualitative and subjective.
167
-------
TABLE 16.
SUMMARY OF ACQUISITION OF SF, CONCENTRATION DATA
o
Experiment
1 (A)
2 (B)
3 (C)
4 (D)
5 (E)
6 (G)
7 (H)
8 (1)
9 (J)
10 (K)
Date
Nov 7
Nov 8
Nov 9
Nov 10
Nov 12
Nov 14
Nov 15
Nov 16
Nov 18
Nov 19
Number of
Sampling
Sites
43
37
43
43
53
30
53
53
53
53
Sampling
Hours
9
9
9
9
9
9
9
9
9
9
Samplers
Analyzed
43
37
43
10
53
30
53
53
53
53
Hours
Analyzed
6
4
2
2
9
9
9
9
9
9
Total Number
of Samples
Analyzed
258
148
86
20
477
270
477
477
477
477
Total
68
3167
168
-------
TABLE 17. DATA BASE, PRELIMINARY TRACY EXPERIMENT
Information
Concentration
emission rate
150-m tower met data:
WS, WD, o0, w, ow,
T, AT from 5, 10, 100,
150-m, levels
10-m tower met data:
WS, WD, T
Doppler sounder data:
WS, WD
electronic weather station
data:
WS, WD
monostatic sounder
optical anemometers
tethersonde data:
WS, WD, T
T-sonde data:
WS, WD, T
Lidar data
Photographs
Base Log
Observer notes
Averaging Time
1-hour
15-min
5-min & 1-hour
5-min
10-min
1-hour
continuous
5-min
instantaneous
Comments
Available for 68 hours
instantaneous
continuous during each
mission
5-min
Two towers
T available from one
of the two stations
Facsimile record from
two locations
Path average wind
speed for two paths
Two profiles per hour
to 600 m at two
sites. RH
available from
Clark site,
pressure from site
west of plant.
One profile per hour
to 3,000 m at two
sites
Facsimile records and
aircraft position
(See Uthe & Morley
1984)
Five minute exposures
every 15 minutes
Available from
Project Officer
Definitions
WS wind speed
WD wind direction
-------
CO
H
E
t—'
K
a,
X
o
X
OS
a*
s
ly winds, neutral conditions
^J
c
CD
4J
CO
o
60
c
0
Li
j->
M
d variable winds; considerable
c
cd
jj
f*
bO
•H
•— i
^
Li
O
0)
cn
•rl
Li
CD
£
(X
sterly winds; no impact on samplers
ca
CD
^J
.c
60
.i-l
i— I
^
Li
a)
>
y and southerly winds throughout
!—(
Li
0)
^j
cn
cd
o>
4J
.c
60
uj
4J
C
0)
s
.1-i
Li
0)
&
CD
on primary target area as well as new
-f-
u
fcj
o
*O
CD
Li
3
Cfl
cd
CD
£
-U
cn
i
0)
-tj
0
CQ
0)
P,
I
CQ
60
.r<
C
r— t
t— 4
cd
vD
w
•d
o
o
erywhere after 0200
^
CD
j_)
cn
o
g
cd
fcf
w
•o
.1-1
S
•o
c
cd
^
•o
3
O
O
o
0
CM
O
.1-1
jJ
C
cn
•o
.r-l
3
gj
3
^J
C
jj
CO
.H
cn
u
0)
ex.
\o
GO
3
00
w
CD
4->
CD
60
C
•H
X
Li
cd
>
w
•o
c
.1-1
3
o
o
0
1
o
o
CO
o
0)
Li
CD
5j
•>^
Li
CD
CD
O
—4
H
c-1
O
cn 60
L, c
3 -H
o -*
X ffl
•o
•c o
cd u
> o
£ £
co o
O O
O O
£ £
O O
o o
£
CM
O
£ £ £ £ £
H r-l CO CM H
O O O O O
00 vO *«T CO vO
00
T-J
5
H
cn
Li
I
\O CO
o o
o o
o o
CO CO
oo oo
oo
o o
i i
CM O
O O
CO
CO 00
oo -v
-v O
ON r-l
o
I
o
O
CO
00
CM
o o o o o
i I I I I
O O r-l CM O
O O CM CM O
CO CO
CO 00
CO CO \D 00 CO
00 00 r-l H 00
>» I
ID IT)
I
i-l r-l H r-t
o
..-4
4J
CO
Li
-J
CO G)
e -i
PI -a
a)
ri
Ul
9?!
CQ OQ H OXr-ir>Sx{
V—' *—' N_^ V_^ \^*• \^' V^*- S^/ x^x
0^ CO *^ ^^ ^D r*"» 00 ^1^ ^?
o
•- u
•O 60
o
O CO
60 -i-l
III 0
CO J
04 O O
* * +
170
-------
Data for Hour Ending 0100 on 12-November-1983
2 1 *
J .3--"-^4
/' 9
.-&*'
xVQ.,---«'
^_/«,_
o o
Data for Hour Ending 0200 on 12-November-1983
0
2
30
,' 0
o o
Figure 68. Hourly SF6 concentrations November 12, 1983 0000-0500.
171
-------
Data for Hour Ending 0300 on 12-November-1983
1 1
...-5V p 0 o
~
/ °
o o
o S o
y
f
y
o
0
Data for Hour Ending 0400 on 12-November-1983
0 0 5 2 -
2 V'V
3
4 i
" ~: ~ 5 4 0 0
f"1 3 . <• *•
0 1
1 0
0
0
Figure 68 (Continued). Hourly SF6 concentrations November 12, 1983
0000-0500.
172
-------
Data for Hour Ending 0500 on 12-November-1983
z
>""-s ..............
5
00
.'*-.
— ^ ™
5 0
0
0
0
Figure 68 (Continued). Hourly SFg concentrations November 12, 1983
0000-0500.
173
-------
Concentration Range (ppt)
Symbol
0
1
2
3
4
5
6
7
8
9
During the first three hours, the highest concentrations were measured
at elevations typically 60-80 m above the top of the Tracy stack.
During the period 0300-0500, the highest concentrations were observed
along the valley floor in the gorge where the Truckee River bends to
the north. The highest concentrations, their location and elevation
above the stack base are (Cher 1984):
Time
(Ending Hour)
0100
Min
0
20
40
60
80
100
125
150
175
>200
Max
19
39
59
79
99
124
149
174
199
Sampler
Site
8
34
31
Cone.
ppt
211
178
159
R
km
6.2
3.6
3.2
e
deg
68
82
62
AZ
m*
169
108
170
0200
31
28
34
6
176
154
149
146
3.2
4.5
3.6
5.3
62
71
82
77
170
113
108
159
0300
1
2
6
310
169
119
4.6
5.0
5.3
95
90
77
158
170
159
0400
25
26
37
27
241
216
207
206
5.1
5.0
6.2
4.8
73
73
63
71
20
-15
-13
-8
174
-------
0500 26 190 5.0 73 -15
37 188 6.2 63 -13
25 171 5.1 73 20
*AZ measured from the base of stack. Stack height is about 91 m.
Figure 69 gives a 5-minute exposure of the oil-fog plume as taken
from Prospect hill (photography position #1, see Figure 67) at 0000
PST. Notice the dispersion of a stable plume from the Tracy stack
toward the east at an elevation near the top of the 150-m tower.
Figures 70 and 71 show 5-minute exposures, again taken front prospect
hill, at 0015 and 0030, respectively. The plume is apparently
producing ground-level concentrations on beacon hill. A 5-minute
exposure taken from Old Lonesome at the west end of the valley
(photography position #3) is shown in Figure 72. It illustrates the
extensive crosswind growth of the plume and shows some plume material
reaching the northwest sections of target mountain.
Hourly average wind directions and speeds taken at the 150-m
tower during the five hours are given in Figures 73 and 74. Winds
near plume height during the first three hours have a westerly
component, consistent with the observed concentrations, photographs
and observer comments. During the last two hours the winds have an
easterly component. Figure 75 gives 5-minute values of wind direction
and speed and ov from the 150-m le-'el of the tower. Again, the
winds are primarily westerly until about 0300, and then primarily
easterly. Notice the increase in turbulence after 0300. Figure 76
gives doppler profiles of wind direction and speed taken at 0020, 0130
and 0400.
A five-minute exposure of the oil-fog plume taken at 0130 from
Prospect Hill is shown in Figure 77. The plume is still being
transported toward the east, but notice that the top of the smoke
plume is below the top of the 150-m tower. Figure 78 shows the
geographic distribution of winds at the approximate elevation of the
plume above the base elevation of the 150-m tower for each of the five
hours. This figure and the doppler wind data illustrate the
complexity of the wind fields in the valley.
In any event, it is clear that the SFg plume was transported
directly to the samplers during the first three hours. Values of Hc
were calculated based on the elevation of beacon hill and using the
150-m tower data. A time series of 5-minute values of Hc are listed
along with other meteorological data in Table 19. The three hourly
Hc averages are 218,222 and 219 m. The elevations of the higher
concentrations are always less than Hc during these three hours.
Plume material was not transported to the samplers at the highest
elevations.
175
-------
§
o
o
00
CM
lib
176
-------
o
o
+J
fl
CO
00
u
o
I
o
00
•H
fc,
177
-------
o
o
+3
ffl
co
00
CM
0)
.Q
O
SB
cd
u
a)
2
CO
o
0>
4J
•r-l
•T-l
fc
0)
u
00
•H
Et.
1/8
-------
o
o
CO
oo
CM
a)
.o
o
s
I
O
u
u
o
d)
(D
tM
O
Li
Ed
179
-------
CM
II
II
K
.•....
'
I !
• 2
i !
a
1
I J
i '
i
8
R
I
It
u
o
f
O
2
0)
a)
ca
C
o
•o
co
0)
u
J!
> u>
180
-------
OL
I
£C
X
7 C
«- CM co 3 in
O O ^5 ^5 ^?
D
o
« o
" I
•
i
10
rg o
u>
s a IB
ea IA
3)
O
O
in
o
-o
-------
«.* i
LO
O
CO
d
q
LO
Q "ro I
> ^ i
> -o •
L_
O
(O
CO
O
CO
CO
8
CO
g
Is:
o
r-
CM
q
cvi
o
in
O
cn
o
to
o
CO
o
o
o
lJ
o
O
I
o
is
o
in
en
•o
Q
SJ
aO
•H
182
-------
400
375
350
325
300
275
250
225
200
175
150
125
100
75
50
-9.9
l.B
1. I
I ..3
-9.9
1.8
1.4
1.7
1.9
2.1
2.0
2.1
1.4
0.6
1.0
j. a. / a. a
999
265
247
249
999
203
222
208
213
229
235
236
214
157
82
O t S>0
400
180
270
360
0.0
2.0
HEIGHT
CM)
400
375
350
325
300
275
250
225
200
175
150
125
100
75
50
SPEED
(M/S)
-9.9
-9.9
-9.9
1.6
0.0
1.6
1.4
1.4
1.7
1.3
0.9
0.6
0.5
1.0
1.2
DIRECTION
(DEC FM N)
999
999
999
169
189
168
163
150
147
153
171
178
220
240
221
400 -
400
375
350
325
300
275
250
225
200
175
150
125
100
75
50
. x :i. :;>
-9.9
-9.9
4.3
4.2
4.5
4.0
2.9
3.0
2.5
2.2
2.1
2.4
2.4
1.6
1.5
t ISO
4.0 6.0 8.0 10.0 12.0
MIND SPEED (M/S)
90 180 270 360
0.0 2.0
999
999
183
179
182
172
151
126
109
82
88
89
88
72
67
AS O
400 _"
~" iTo' 6To ' 5to
WIND SPEED (M/S)
90 180 270
10.0 ' !?.(
360
350 ••
300 ..
250 ..
200 . .
150 ..
100 ..
•4-
t
50 f
?To 4.0 ?To 8.0 10.(
WIND SPEED (M/S)
17 o
Figure 76. Doppler sounder wind profiles.
183
-------
-u
cd
oo
er>
o
a)
U
0)
U
U
W
O
Q)
4)
I
•>
•rl
h.
2
2,
184
-------
, ;: • . • / y,7^>, • " \ „ • ~ '( Mf&. 'p V ^>J S i") Vv^'/VV-''! f' I *r "77^' v~f "~ ' ; ^ 'J\
•/] ^sn-> ^A4i'>w%^ ^F' ',"T/
.' -< ' vi V" ,.VsJ- V* i '(^'-^vV--^K-Mr r^" " • ^'>H^ /S2 - i; )s
1.' ' •: s' v; 1 A -1 '^ '' ^ - ^":7J//ii"^"^Tf V'7y "-vv - i RI;$V'
^^v^>l^>:;^ft|K^5W^
^:.. ,v. v. fit ^.^!/»rid J
•O^.-'j'i C'M /i:(((0 <^M:f^f U^sll^
w«y»^\f ^' - Ki" ^M^^'
^M^;Jiwm&^{ i' M IA1 ;< A ^ ^l% • r A'
W^wte^H^TT^rrrr^
J>A 0! \.'JZ\ Vil\^^v$v\ ok/vU^^HX, (^- / •' ( ] \
l^.p^^i^^
li|l«S!rIv* -i-VMi
-)? JJ-///('u^j ^(/^^S^^cv^ v ( p/^"^"""' _' l"J ! - ^"^-,< <
^>^:^4^41^|A:--f^^^^~ "^v" ^'' "~~ 1 * ^ ''^'" "^
[?^>i^4>^?»^
fctoibmw;
V'^'fi^'r-^rx''),)}^.;!'/ - , r"~i/^7 F '
^^T^ffi-f/,1'--;"
wov'^v/ (iv •••:
•Sr-^Si '-^^^ '"; '
{(l^y^^i^.S .
J'ii V/^flV'-V^" ^
^00
hi
,-vc
'{ '''
=^
Vf-^Lr. v-' ^-:- / ^'!' "
-^' ] --.j,^' •< f/c;:^^^1, «
)^.'K.- ;/'V^\ -' -•
J /V •— :/, i ^ /> 1 .,.
.-- .. «,;-.-. .— =---i , -,- -. r , -„,.
iSS'
^ «*
-'**\
".•''k-
.'••&'<
^ JJ ^ ^/ i
LT
0)
4J
cd
g
• iH
X
o
L<
o.
a
U)
• H
T3
y
•*-<
x;
a •
cd i -*
tj ,^
60 60
o -^
0) 0)
o jc
-------
T-=.--vfm-M--- "i*'-m,-\-i 4^A"iY^--?--"'-v- - -
\:O3tf'/-'-»;, ?'j: fU^MK ^i !•
- /r^-^/-/ ! , \U (|VA>jk-V(r- v' J V,
*-^T- •'•' ' m^^-i ""i r^
f|i;W-|telft^M^>
n.^y^v./g Lfefen\sr) T- r-^ H
;^: Jf^'DRli)PK^ ?' "= •-h,L': ;j
^»»fe>K^^^ai
' /,
X3i
i ,J*f - -
\' i
*y\
••\r
•a
c
U-l
o
•r* D.
TJ
(1)
O -1-1
•f-4 Cd
•s e
Q> •<->
a) X
U O
60 t-l
O O,
0) (X
O f?
186
-------
tn
•O
c
c x:
o oo
in F-I
.- O.
•O
0)
O -t->
.-<
-------
4J
of
en
•o
c
C A
O 60
•H *tH
4J 0)
a .c
.o
OS i-l
-rH Q.
•o
<]>
y *>
•rt a)
O. -r<
O) X
u o
60 U
O Q.
0) Oi
O 0)
•O
0)
-u
c
o
00
u
60
188
-------
4J
in
•O
c
o •
•4J
C J=
O 60
•*-4 -f-l
•U 4>
.»< 0)
^ E
•»J 3
01 r-l
•1 O.
•o
0)
O -P
•^ «
je 6
Q, .^i
-------
Hr
Sec
TABLK 19. FIVE-MINUTE METEOROLOGICAL DATA*
NOVKMBER 12, 1983, 0000 0300 PST
ws
WD
Ft-
Bulk H
0
0
0
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
3
300
600
900
1200
1500
1800
2100
2400
2700
3000
3300
0
300
600
900
1200
1500
1800
2100
2400
2700
3000
3300
0
300
600
900
1200
1500
1800
2100
2400
2700
3000
3300
0
1.1
.9
.5
.3
.3
.9
1.1
1.3
1.1
1.1
.5
.7
1.6
1.7
1.9
.7
.7
.1
.9
1.1
.7
.7
2.1
1.5
1.3
1.5
3.2
2.5
1.7
2.3
1.7
1.6
1.1
.7
1.7
1. 7
237.0
249.0
253.0
253.0
10.0
70.0
112.0
70.0
87.0
54.0
7.0
310.0
291.0
296.0
288.0
347.0
237.0
305.0
270.0
249.0
277.0
281.0
319.0
317.0
314.0
282.0
249.0
253.0
218.0
256.0
323.0
298.0
29.0
128.0
77.00
38.0
-.1
.1
.1
.1
.1
.1
-.1
-.1
.1
— 1
-.1
-.1
-.1
.1
.1
.1
.1
„ f 2
.1
.1
-.1
.1
-.1
.1
.1
.1
.1
-.1
.1
-.1
.1
— 1
.1
-.1
.3
- .1
.253
.291
.093
.046
.202
.330
.215
1.021
.167
.129
.286
.247
.360
.258
.345
.445
.097
.081
.245
.394
.349
.127
.451
.217
.145
.749
.743
.489
.383
.711
1.397
.360
1.092
.651
.730
1.311
.200
.120
.130
.140
.130
.130
.200
.200
.130
.200
.130
.200
.130
.130
.120
.120
.120
.130
.120
.120
.130
.280
.200
.130
.200
.200
.200
.280
.120
.280
.280
.360
.510
.200
.200
.200
178.0
198.0
143.0
166.0
187.0
246.0
254.0
240.0
242.0
259.0
253.0
267.0
247.0
238.0
229.0
211.0
230.0
253.0
220.0
188.0
230.0
159.0
226.0
233.0
227.0
238.0
206.0
227.0
216.0
219.0
224.0
182.0
222.0
179.0
260.0
213.0
1.1
1.2
.5
.6
.8
1.2
1.3
1.3
1.4
1.8
1.4
1.2
.9
1.0
1.3
.3
.9
1.0
.4
.1
. 7
.1
.8
1.2
1.1
1.2
1.7
1.5
1.1
1.3
1.0
.5
1.0
.3
1.2
.6
.2
.2
.1
.1
.1
.1
.1
.2
.2
.2
.1
.1
.1
.1
.1
. 1
. 1
.0
. I
. L
. L
.0
.1
.1
.]
.]
. 2'
.1
.1
.1
.2
.2
.2
.1
.1
.2
.0328
.0297
.0318
.0363
.0260
.0256
.0253
.0345
.0333
.0325
.0303
.0317
.0325
.0327
.0288
.0320
.0339
.0315
.0350
.0349
.0358
.0402
.0311
.0312
.0308
.0297
.0220
.0250
.0319
.0303
.0353
.0333
.0297
.0374
.0323
.0339
.0376
.0381
.0381
.0333
.0360
.0371
,0349
.0381
.0386
.0409
.0423
.0385
.0365
.0370
.0360
.0376
.0360
.0400
,0365
.0309
.0355
.0309
.0349
.0355
.0349
.0338
0327
.0390
.0395
.0400
.0360
.0321
.0338
.0326
.0284
.0327
*Collected at the upper level of the 150-m tower.
190
-------
During the last two hours, 0300-0500, the highest SFg
concentrations were measured in the gorge at the valley floor level,
yet the meteorological data suggest the plume was transported toward
the west. Figure 79, a 5-minute exposure taken from the west end of
the valley at 0315, shows the plume near the stack being transported
toward the west with considerable plume material remaining in the
target area. The Alpha-1 lidar observations* (Figure 80) also show
plume material at valley level in the gorge.
How did elevated plume material enter the gorge and produce high
concentrations on the valley floor? ERT observer comments suggest the
turbulent transport by drainage winds off beacon hill and target
mountain. The monostatic sounder at Clark also showed an increase in
turbulence after 0300. See also the 150-m tower ov data in
Figure 75. Figure 81 depicts a photograph of the monostatic sounder
facsimile record for 12 November. During 0000-0300 the sounder data
show decoupled layers with waves in the lower atmospheric boundary
layer. At 0300 the record shows the interaction of layers—implying
transport of elevated plume material from aloft—and complete mixing
by about 0440.
In summary, the first three hours of Experiment 5 illustrate
stable plume impingement conditions. The SF$ plume evidently
produced ground-level concentrations at samplers whose elevations were
below the calculated Hc. The hours 0300-0500 illustrate the
occurrence of high concentrations on the valley floor. The
concentrations evidently resulted from the turbulent transport by
drainage winds of elevated plume material.
Experiment 7 (November 15. 1983) 0300-0800
SF$ was observed by samplers located east and west of the TPP
stack during the morning of the 15th. Figure 82 shows maps of hourly
concentrations over the five hours from 0300-0800. Plan view maps of
the winds representative of 100 m and 150 m are given in Figures 83
and 84, Evidently, the winds in the valley switched back and forth
from westerly to easterly and then back to westerly during the five
hours. Figure 85 gives time series of wind direction and speed and
ow measured at the 150-m level of the tower. Notice the very low
values of ov. The highest concentrations, their locations and
elevations are (Cher 1984):
*See Appendix C for an explanation of the Alpha-1 observations.
191
-------
<*>
00
CM
H
U
I
co
CD
u
I
(ft
O
g
2
-------
193
-------
November 12
Time(PST)
600
• 400
200-
0200
0100
0000
•Time(PST)
600
0600
0500
0400
0300
600
01
0900
- Time (PST)
0800
0700
0600
Figure 81. Eagle Picher acoustic sounder records.
194
-------
Data for Hour Ending 0400 on 15-November-1983
0
0
1 ,' 0
o o / r, o
2 o cAi ° o
; 6
-..-'V 8| 5
0''
0 0
Data for Hour Ending 0500 on 15-November-1983
o
o o . . ' o
°
o
o o
. .....
Figure 82. Hourly SF6 concentrations November 15, 1983 0300 0800.
195
-------
Data for Hour Ending 0600 on 15-November-1983
° ° ° 0 -
0 o V o"
0 Ov' < 0
..
0 --" °
X
p
/
/ 1
/
o o
o
t
0
0
Data for Hour Ending 0700 on 15-November-1983
o o >--*
0 0 / „ 0
2 14£ °
3
1 6
0
0
Figure 82 (Continued). Hourly SF6 concentrations November 15, 1983
0300-0800.
196
-------
Data for Hour Ending 0800 on 15-November-1983
o o o ° '
0 ft 1 ,-1'" """*
0 o: i 0
° 2 I2?£l !
-'-- .............
2 3
0
Figure 82 (Continued). Hourly SFg concentrations November 15, 1983
0300-0800.
197
-------
M
•o
c
e
o
o •
r-t O
O
0) O
U 8
oo
-------
0)
T>
6
o
o
c
o
XI •
.1-1 <*)
U 00
*> 0>
01 r-l
•*4
•o -
in
O rH
a. a>
a) u
u e
tO 4)
o >
0) O
u z
•o
01
.1-1
,J
c
o
o
00
Ll
bO
199
-------
m
•G
C
o
o
.o •
•r< CO
u oo
at r-t
•«-l
T> -
J= U
O. (U
o >
-------
u
•O
• «-4
*
e
o
o
c
o
u oo
4J O>
01 r-1
•i-(
T) -
in
4) O
0 2
•a
0)
g
••-I
4->
c
o
ro
CO
a)
»-l
3
201
-------
tn
C
-»-*
»
e
o
o
I-l
c
o
••-I
4J
3
J3 •
•rH CO
[J 00
jj O>
Ul r-l
.«-*
•O -
in
O r-l
.1-*
J= fcl
a
-------
n
•O
E
i
o
u->
rH O
O
-
in
Q. 0>
td JD
t. e
60 0)
o >
0) O
O Z
00
u
10
.1-1
203
-------
: t ,%w >^f«Wf^ W-^
;-^ :-;',- ^>i/t 4:v^4^^^Kl/a^ ^ft
'-.'•^ '- ?"!«/ ^>%^»!^^^
f/a f ^^H-'^Wjnii/^
& ^^^Sft^
SIS4/-- --l-W^yT*-^-- -K+ • : / ,f—iv\-J \\,"_~, I
J)
W
TJ
g
o
m
r-i O
O
u co
JJ 0»
o> H
•*-i
•O -
m
O r-l
•VH
JS it
o. a)
u g
40
0) O
•O
0)
3
O
o
co
cu
60
204
-------
r—*
10
*r4
•o
0) O
O 2
•o
a
3
O
O
00
s
oo
205
-------
01
•o
e
i
o
in
r-l O
O
(w 00
o o
I
c o
o o
«* n
*> o
OO
•o -
in
a) A
U E
60 0)
o >
0) O
O 2
•o
a)
3
O
o
03
-------
in
TJ
C
f
o
1/1 •
r-l O
O
<4-i CO
O O
I
G O
O O
.1-1 CO
•^ o
.0 -
••-< in
(j 00
•o -
in
o tH
»iH
A U
CU V
0) A
t4 6
60 0)
o >
0) O
O 2
•o
a)
3
.•-I
*)
c
o
o
00
il
I,
-------
CN
d
o
c>
00
q
ci
q
d
SCM
O
O ci
I I
J I
s
o
o
e
I
O
ffl
-P
(0
•a
a)
o
oo
o
o
'*
o
.(J
o
oo
0)
u
•H
fa
Q "3 I O
> ® i n
5 T3 ! CO
O
o
CO
-------
Time
(Ending Hour)
Sampler
Site
Cone.
ppt
R
km
e
deg
AZ
ni*
0400
10
3
2
648
216
180
5.3
5.4
5.0
89
84
90
241
154
170
0500
2
3
10
417
313
269
5.0
5.4
5.3
90
84
89
170
165
241
0600
28
1
31
25
92
83
67
65
4.5
4.6
3.2
5.1
71
95
62
73
113
158
170
20
0700
56
28
1
22
53
126
96
77
76
76
5.3
4.5
4.6
4.1
5.8
231
71
95
103
246
254
113
158
56
224
0800
58
56
60
462
211
139
4.8
5.3
8.7
265
231
225
208
254
370
*AZ naasured from the base of stack. Stack height is about 91 m.
During the hour 0300-0400 the plume was transported directly (but
slowly) from the stack to the southwest corner of target mountain.
Figures 86 and 87 show photographs of the oil-fog plume taken from
Prospect hill at 0315 and 0330. The first five-minute exposure shows
a very stable plume being transported from the stack toward the east
at an altitude near the top of the 150-m tower. The second exposure
shows plume material in the valley south of beacon hill being advected
into target mountain. Figure 88 gives a photograph of the plume
209
-------
PI
O
CO
00
o
55
I
o
Q)
U
Dl
O
•H
0)
•H
fe
oo
u
ao
210
-------
o
PI
f)
O
-p
R)
oo
en
u
0)
1
00
o
u
•H
fa
211
-------
o
CO
CO
o
CO
oo
I
CO
0)
0)
Jj
w
O
0)
•rl
CO
00
212
-------
taken from position #3 at the vest end of the valley at 0300. The
exposure shows a plume with extensive crosswind diffusion being
transported toward the east.
Sampler 10, at an elevation of 241 m above the stack base,
measured a concentration of 648 ppt, the highest concentration
observed during the entire 10-day experiment. This peak
concentration, normalized by the SFg emission rate, is
3.1 ysec/m3. The calculated value of Hc is 245 m.
SFg tracer gas remained on target mountain during the hour
ending at 0500 despite the fact that winds gradually shifted to
generally easterly. A photograph (Figure 89) taken at 0430 from
Prospect Hill shows plume material overhead. A photograph (Figure 90)
taken at 0615 from the west end of the valley shows plume material
approaching from the east, eventually producing concentrations above
100 ppt at samplers west of the plant. By 0900 concentrations above
200 ppt were again observed in the target area.
Experiment 9 (November 19. 1983) 0000-0100
Figure 91 shows hourly tracer gas concentrations observed during
a windy neutral period. The photograph in Figure 92 illustrates a
persistent, coherent plume from the Tracy stack to the haul road south
of target mountain. The winds in the valley (Figure 93) were all
west-northwesterly at about 8 m/sec.
5,5 Summary of the Preliminary Experiment
The analysis of the three case study experiments and a
preliminary analysis of the other data show the occurrence of stable
plume impingement conditions. SF$ concentrations were observed
during stable conditions on target mountain and beacon hill. Although
the sampler coverage was relatively sparse, concentrations were also
observed during stable conditions in the hill 5610 complex northwest
of the plant.
The highest SFg concentrations were observed on the southwest
corner (samplers 10, 3, 2) of target mountain during Experiment 7.
The elevations of the samplers that captured plume material were a few
meters below the calculated hourly values of Hc. During other hours
of stable plume impingement conditions, plume material was observed to
stay below Hc. In short, it appears that the concept of a dividing-
streamline height will be useful to distinguish flow regimes and to
help simulate observed tracer gas concentration patterns in the Tracy
area.
Drainage winds and katabatic effects were seen to produce
ground-level concentrations on the valley floor in the gorge where the
Truckee River bends to the north. Observer comments, photographs and
acoustic sounder records all suggest the turbulent transport of "old"
plume material from aloft to the valley floor. The fumigation of
oil-fog by drainage winds was also observed on the south side of
target mountain. These katabatic effects were not observed at CCB and
HER and must be accounted for in the final design of the FSPS.
213
-------
o
rfi
•*
O
CO
CO
1
O
S3
cd
0)
u
s
o
•r-l
e
0)
00
0)
u
aO
214
-------
in
\o
o
a)
CO
00
O
s
en
CO
u
3
in
o
&
-------
Data for Hour Ending 0100 on 18-November-1983
0 °0°°
o o .-«••'"•« ....... ........
— kff' 1 0 0
/ °
0 0
o
0
0
Figure 91. Hourly SF& concentrations November 18, 1983.
216
-------
o
o
-p
(0
PI
00
00
u
4)
o
S5
CO
o
u
3
01
o
0
.1-1
Cx
2
217
-------
u
•o
-t-1
f
O
i/l •
H O
O
W-J t-t
O O
C O
O O
.rt o
4J O
.0 -
.t-l fO
b oo
JJ CT>
W t-t
•iH
•o -
oo
O r*
•f-4
£ Ll
Oi 0)
0) O
O 2
00
218
-------
The meteorological measurements depicted very complicated wind
flows during the November experiment. Horizontal and vertical wind
shears were common. These were probably caused by the combined
effects of the complex terrain and migratory anticyclones and cyclones
moving over the area in November. In summer and early fall we expect
the flows in the valley to be more dominated by drainage flows,
principally drainage down the Truckee River. In any event, during the
FSPS there must be a sufficient number of meteorological measurement
systems to provide information on the three-dimensional structure of
winds in the valley.
Finally, the November experiment produced a data base with a wide
variety of dispersion conditions--from windy, neutral cases to stable
plume impingement. About 34 hours were judged good for modeling. The
CTDM concepts will be tested using this data base.
5.6 Plans for the Full Scale Plume Study
The FSPS* will be conducted at the Tracy plant during the period
August 6, 1984 through approximately August 27, 1984. The
participants will include ERT, NOAA WPL, NOAA ARLFRD, and SRI
International. Table 20 provides a schedule of the daily experiments.
The FSPS will commence with two 4-hour shakedown experiments on
the 6th and 7th. These will start around 0300 PST and end around 0700
PST and will be conducted to test the equipment and the experiment
protocol. The shakedown experiments will be followed by 12 10-hour
tracer and simultaneous flow visualization experiments. These will
run primarily during the nighttime hours to capture stable conditions
and are scheduled to take photographic advantage of the full moon on
the llth.
The experimental methods of the FSPS are based on the preliminary
experiment conducted in November and to a large extent are similar to
the methods used and tested at CCB and HBR. The FSPS at the Tracy
station will include:
• Releases of SFg and oil-fog from the 300-ft stack and
CF$Br from various heights on the 150-m tower;
• Operation of 110 tracer gas samplers at 107 locations (four
will be on the 150-m tower);
*The detailed plans are described in the Work Plan for the Full Scale
Plume Study. ERT document P-B876-625, May, 1984.
219
-------
TABLE 20. FSPS SCHEDULE*
M 6 Aug 0300 - 0700**
T 7 Aug 0300 - 0700
T 9 Aug 2000 - 0600
F 10 Aug 2000 - 0600
S 11 Aug 2000 - 0600
W 15 Aug 2200 - 0800
T 16 Aug 2200 - 0800
F 17 Aug 2200 - 0800
M 20 Aug 2200 - 0800
T 21 Aug 2200 - 0800
W 22 Aug 2200 - 0800
S 25 Aug 0000 - 1000***
S 26 Aug 0000 - 1000
M 27 Aug ,0000 - 1000
*Schedule subject to change depending on weather conditions.
**0nly the fan for the 300-ft stack will be operating during the
first 11 experiments.
***The 120-MW unit will be operated during the last three experiments.
220
-------
• Fixed meteorological measurements:
a 150-m tower instrumented at six levels (instruments
will include sonic, propeller and cup-and-vane
anemometers and temperature and radiation sensors),
three 10-m towers instrumented at one level,
one 10-m tower instrumented at two levels,
three monostatic acoustic sounders,
two doppler acoustic sounders, and
two solar-powered electronic weather stations;
• Two tethersondes:
one operated at plume elevation to document
meteorological conditions representative of the
effective source height, and
one operated to measure vertical profiles of
meteorological parameters upwind of the source;
• Two radar balloon tracking systems:
one located near the west end of the valley to measure
the approach flow, and
one located north of the plant to document the winds
near potential impact areas;
• Ground-level tracer gas concentrations;
• Airborne and ground-based lidar measurements;
• Photographs (from five locations) and movies; and
• A command post near the 150-m tower that includes:
real-time display of data from the 150-m and 10-m
towers,
radio base station, and
- facsimile output of weather maps.
The meteorological data will be archived and displayed in real-time by
a system of onsite minicomputers. Real-time information on ambient
meteorological conditions will aid in understanding the dispersion
phenomena and will help the field managers maintain real-time
experimental control. The real-time operations management will be
supplemented by near real-time lidar data, a 48-hour turnaround on the
tracer gas concentrations, and a 48-hour turnaround on the
photographs. Provisions will be made for data analysis in the field
to help guide the experiment. Figure 94 illustrates the layout of the
FSPS.
Oil-fog and SFg will be released from the 91.4-m stack.
CF3Br will be released from one of the three levels (100, 120 or
140 m) on the 150-m tower. The tracer gases will be sampled at about
107 locations in the valley and on the mountains.
ARLFRD will operate 110 samplers during each approximately
10-hour experiment. All samplers will be used to get one-hour
averages. Four samplers will be operated at the 150-m tower. Each
10-hour experiment could produce 1,100 bags and 2,200 concentrations.
The samples will be analyzed and concentration maps produced within
about 48 hours after collection of the bags.
221
-------
illiillll
o
01
On
H
w
222
-------
The sampler network (see Figure 94) was designed principally from
the results of the preliminary experiment that was held in November
1983. It will measure ground-level tracer concentrations during
several plume dispersion conditions:
1) transport along known (observed) plume paths,
2) stable plume impaction,
3) channeling by major terrain features,
4) lee side phenomena,
5) katabatic fumigation,
6) recirculation in the valley, and
7) flat terrain versus hill effects.
Samplers on target mountain and beacon hill and the string of
samplers west of the plant are located to measure concentrations
during events similar to those observed in November and during
previous field experiments. The samplers on target mountain, beacon
hill and on the hill 5085 - hill 5610 complex are located to measure
concentrations during stable plume impaction conditions.
The meteorological towers, the acoustic sounder systems, the
radar wind systems and the tethersondes will provide information to
characterize the three dimensional structure of the wind and
temperature fields in the valley. The 150-m tower, which will be
instrumented with sonic, propeller and cup-and-vane anemometers, will
provide data on the winds and turbulence near the elevation of the
Tracy plume and data to define the meteorological conditions near the
height of the Freon emissions.
The lidar systems and photographs will document the path and
growth of the oil-fog (and coincidental SFg) plume upwind of the
major terrain elements. They will provide information to calculate
-------
SECTION 6
SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS FOR FURTHER STUDY
This Fourth Milestone Report documents the further evolution of
CTDM and presents a detailed mathematical description of the model
components, including those that explicitly account for the effects
that terrain has on plume dispersion. The latest version of CTDM has
been evaluated using an 80-hour subset of the SHIS #1 data base. The
report also describes an empirical modeling approach to the
Hogback Ridge data base. It provides an overview of the preliminary
dispersion experiment that was conducted at the Tracy Power Plant in
November, 1983 and presents plans for the Full Scale Plume Study.
6.1 Principal Accomplishments and Conclusions
The Complex Terrain Dispersion Model (CTDM)
Substantial progress has been made in the development of CTDM as
a method for simulating tracer gas concentrations observed at CCB and
as a practical regulatory model. The central feature of the model
still is its use of the concept of a dividing streamline to separate
the flow into two discrete layers. The upper weakly stratified layer
is handled by the LIFT component and the lower stable layer by the
WRAP component. Both model components include explicit mathematical
expressions that account for the important phenomena that control
dispersion in mountainous terrain:
• streamline contraction in the vertical,
• streamline distortion in the horizontal,
• acceleration of the flow, and
• changes in lateral and vertical diffusivities.
CTDM also includes a method to simulate the transition between the
upper and lower flows and subsequent ground-level concentrations and
to simulate the temporal variability of the Hc interface. The model
also includes new formulations for c*z and
-------
EPA Project Officer. The MDA was evaluated by comparing interpolated
winds and turbulent intensities to those estimated from lidar data and
plume photos. The analysis indicates that the MDA values are
generally appropriate for modeling, although photo and lidar estimates
were substituted in some cases when warranted. A subset of 80 hours
of SE'g and coincidental meteorological data was selected to evaluate
CTDM.
Investigations of Plume Growth
An analysis of SHIS #1 Tower A turbulence intensity measurements
with values of oz estimated from plume photographs and lidar data
suggests the following model of vertical dispersion of elevated
releases in the stable boundary layer:
a t
w _
(1 -H t/2T_ )
LJ
The dispersion time scale TL is given by
TL - a/aw
where the mixing length 9, is
_ _ , _
l I I
n s
and the neutral length scale 8,n and the stable length scale S.s
are given by
J,n = Tzr ; !LS = y2 ow/N
The variance of the vertical velocity fluctuations <*w and the
Brunt-Vaisala frequency N are evaluated at the release height zr.
The constants y and f were derived from surface layer
flux-gradient relationships and were found to have values of 0.27 and
0.32.
Evaluation of CTDM
The current version of CTDM was evaluated by comparing model
calculations to (1) observed SFg concentrations, (2) concentration
estimates based on a flat terrain model, and (3) concentration
estimates based on the COMPLEX I/II plume path assumptions. The
80-hour subset of the MDA was divided into four classes: neutral,
weakly stratified, impingement, and very stable. Performance
statistics were generated for each class as well as the entire
subset. The results show that CTDM simulates the observations better
than the other two approaches. CTDM has a weak overall bias toward
225
-------
overestimating the larger concentrations (m~ = 0.84), the COMPLEX
modeling approach generally overestimates by more than a factor of two
(mg = 0.42), and the flat terrain model generally underestimates by
more than a factor of two (m~ = 2.2). The noise in the CTDM
calculations is lower for the time- and space-paired statistics, which
indicates that CTDM is better able to simulate the observed
distribution of concentrations. These comparisons illustrate the
importance of including Hc and the stagnation streamline in a
modeling framework such as CTDM.
The performance of CTDM in each of the four classes indicates
that the model tends to overestimate the larger observed
concentrations for Hc greater than 1.25 times the release height
(the two "more stable" classes). These classes include the hours in
which the largest concentrations (scaled by the emission rate) were
observed at CCB. CTDM tends to underestimate in the other two "less
stable" classes, producing estimates that are 70% to 80% of the larger
observed concentrations. These results indicate that more work is
needed to further improve CTDM performance within each of the classes.
Model Performance for Various Quantities of Onsite MeteoroloRical
Data
The ability of CTDM to simulate the CCB observations with various
quantities of onsite meteorological data was evaluated by constructing
three alternate model input data sets: (1) 5-minute meteorological
data contained in the MDA; (2) an hourly data set constructed from
one-hour wind and temperature data measured at 10-m and 150-m; and (3)
an hourly data set constructed from wind and temperature data measured
at 10-m only. The results suggest that the 5-minute simulations
improve some individual hours but do not substantially change the
overall performance statistics. The results from the simulations
using the simplified meteorological input are inferior to those based
on the full MDA even though MDA wind directions were used for each.
The conclusion of this analysis is that onsite measurements of
turbulence intensity near release height and detailed vertical
profiles of wind and temperature are essential for accurately
simulating concentrations at a complex terrain site such as CCB.
Modeling the Hogback Ridge Data Base
To help understand the phenomena that control disperssion at the
ridge site a simple empirical modeling approach was taken. A model
was constructed by modifying the effective plume height as a function
of Hc. Model simulations were performed using a subset of the
SHIS #2 CF3Br data base. These empirical model calculations were
compared to observed CF3Br concentrations and to calculations made
with a flat terrain model and a model based on the half-height plume
path assumption. The empirical model performed better than the other
two models. Furthermore, for both CCB and HBR the flat terrain model
underestimated concentrations while the COMPLEX (or half-height) model
overestimated concentrations by roughly a factor of two.
226
-------
Fluid Modeling Facility Simulations
A series of tows was conducted in the EPA FMF salt-water-
stratified towing tank using, as the basic hill shape, the
fourth-order polynomial hill as used by Hunt and Snyder (1980).
Linear density gradients were established in the tank and the hill was
towed at a speed such that the Froude number Fr was 0,5. Since the
density gradient was linear and Fr = 0.5, the dividing-streamline
height was 0.5h. After that series was completed, a second series of
tows was conducted wherein the entire model was raised out of the
water to the point where the water surface was precisely at the
dividing-streamline height, i.e., the water surface was at half the
hill height. The model was towed at the same speed as in the
full-immersion tows, so that the Froude number with this now
half-height hill was unity, and all streamlines passed over the hill
top. The flat water surface thus forced a flat dividing-streamline
surface. The resulting surface concentration patterns were then
compared with the corresponding full-immersion tows. These
simulations, which are described in Appendix A, were performed to
answer the question: how good is the assumption of a flat
dividing-streamline surface?
The results suggest that this assumption is a reasonable
approximation to make, at least with regard to predicting the
locations and values of maximum concentrations and areas of coverage
on the windward side of the hill. When the stack heights are
relatively close to the dividing-streamline height, the lee-side
concentrations are also predicted reasonably well. These results were
used directly in the formulation of CTDM.
The FMF staff also conducted wind-tunnel studies (Appendix B) to
investigate the influence of an idealized three-dimensional hill on
ground-level concentrations from upwind sources in a neutral
atmospheric boundary layer and to locate the source locations where
this influence is the greatest.
The presence of the hill was found to influence the dispersion of
the plume to increase the maximum concentration in three ways. For
low sources, at moderate distances from the hill, the reduction in
mean wind speed allows the plume to reach the ground surface closer to
the source, producing higher concentrations than in the absence of the
hill. Plumes from higher sources can be thought of as being
intercepted by the hill. That is, the hill penetrates the plume where
the concentrations are greater than those that would occur at ground
level farther downstream over flat terrain. For yet higher sources,
the streamline convergence over the hilltop and the corresponding
downward flow in the lee of the hill brings the plume to the ground
more rapidly than over flat terrain. The maximum concentration for
these three regimes occurs upwind of the hilltop, near or on the
hilltop, or downwind of the hilltop, respectively. Terrain
amplification factors ranged from near 1.0 to 3.63. The region of
source locations that produced an amplification factor of 1.4 or more
extended to an upwind distance of 14 hill heights.
227
-------
6.2 Recommendations for Further Study
6.2.1 The SHIS #2 Data Base
Further refinements are recommended in screening and adjusting
the meteorological data obtained at the three primary towers at HER.
Once this is completed, the Quality Assurance Report for SHIS #2 can
be completed, and the modelers will have better guidance in
interpreting the data.
In addition to the sonic anemometer data used in the preliminary
modeling reported in this Milestone Report and the acoustic profile
data, the meteorological measurements from the 150-m and 30-m towers
(towers A and B) are essential to investigating the flow properties
below Hc. The temperature patterns and changes in the turbulence
properties will be investigated. A comparison of profiles of these
data within 30 m of the surface at these two towers should essentially
provide a description of the flow field within the "blocked" region
below Hc for nearly all releases below Hc.
For releases above Hc, comparison and integration of
tethersonde data, acoustic profile data, and measurements from
towers A and C are particularly important. Also, because of the
oil-fog plume rise so characteristic of many of the experiment-hours,
analysis of the photographs and the lidar scans of the visible plume
will be pursued to document and model the plume rise, and to model the
initial growth of the visible plume. Analysis of lidar scans and
photographs will also document patterns of streamline deflections in
the flow over the crest of HBR, and possibly the distribution of plume
material near the surface.
6.2.2 CTDM
Application of CTDM to the SHIS #1 data base has reached a stage
where the modeling framework is largely complete. Some effort will be
devoted to investigating the LIFT/WRAP transition zone, and fluid
modeling studies at the EPA FMF will help refine the formulation of
the flow in this region. In addition, it appears that the performance
of the model will be improved by including directional wind shear,
which so far has been ignored. Also ignored has been the variation of
the turbulence with height (away from the plume centerline).
Aside from these enhancements to CTDM, most of the work remaining
in applying CTDM to the SHIS #1 data base will focus on the
terrain-effects factors T^, T^, Tu, and T^. The present formulation
will be compared with computations performed using thin aerofoil
theory as well as results from fluid modeling simulations,. Also,
particular model parameters will be selected for optimization in order
to infer the terrain effect description which best matches the
observed concentrations and meteorological data. Presumably,
alternate choices of the terrain-effect parameters will be needed to
improve the model performance. The specification of T^ (either
T^z or T^y) is a good example. Our present calculations are made
with Tj_ = 1. Rapid distortion theory will be used to infer non-zero
values for T^. Optimized values of T^ will be compared with those
computed from the rapid distortion theory.
228
-------
Application of CTDM to other sites will require some
generalization of the code. A good test-case for such a
generalization is HER. The model will be altered so that the present
CTDM "assumptions" can be tested at HER. The experiments at the Tracy
Power Plant (FSPS preliminary and FSPS) also need to be considered in
structuring the general CTDM. New algorithms will surely be needed in
some circumstances, but we also need to know when the detail in the
present CTDM is most needed. A sensitivity analysis will help define
the conditions in which the terrain effects contained in CTDM have the
greatest and least impact on the magnitude of the concentration
estimates compared to estimates made by means of a simple
"flat-terrain" model. Key input parameters could then be designed to
signal a "CTDM" calculation or a "FLAT" calculation when the model is
applied in a regulatory permitting mode.
The development of a theoretical rather than an empirical model
for HBR will receive much attention in the next year. We first need
to identify the circumstances in which CTDM does as well as the
empirical model, and circumstances in which CTDM does much worse. The
cases in which CTDM fails must be analyzed in detail to develop a
rationale (theory) for modifying CTDM. In this way CTDM will increase
its range of applicability, and the need for empiricism will be
reduced. This process will begin by applying CTDM to the subset of 35
hours modeled in this Milestone Report. Although the meteorological
data are incomplete, we expect to learn something from this
preliminary modeling. We expect to develop a framework to simulate
the effects of an unsteady, "blocked" flow upwind of a ridge.
6.2.3 The FSPS Data Base
The FSPS at Tracy Power Plant has been designed to provide a data
base that will augment our understanding of how a plume in large-scale
stable flow interacts with topography and local slope flows to reach
the surface. The phenomena observed at this site will be described
and documented, and the relationship between these phenomena and
observed ground-level concentrations will be investigated. This
investigation will include comparisons with both the SHIS #1 and #2
results. We expect that the similarities among the FSPS and the SHIS
observations will identify those aspects of the FSPS data base that
can be modeled with CTDM directly. The dissimilarities will focus
attention on aspects that will require modification to CTDM.
229
-------
REFERENCES
Briggs, G.A. 1973. Diffusion Estimation for Small Emissions, ATDL
Contribution File No. 79, Atmospheric Turbulence and Diffusion
Laboratory.
Briggs, G.A. 1975. Plume Rise Predictions. Lectures on Air Pollution
and Environmental Impact Analyses. AMS, Boston.
Brighton, P.W.M. 1978. Strongly Stratified Flow Past Three
Dimensional Obstacles. Quarterly Journal of the Royal
Meteorological Society. 104; 289-307.
Businger, J.A. 1973, Turbulent Transfer in the Atmospheric Surface
Layer. In Workshop on Micrometeorology, AMS, Boston, 67-100.
Csanady, G.T. 1973. Turbulent Diffusion in the Environment.
D. Reidel Publishing Co., Dordrecht, Holland.
Deardorff, J.W. and G.E. Willis 1975. A Parameterization of
Diffusion into the Mixed Layer. J. Appl. Met.. 14: 1451-1458.
Drazin, P.G. 1961. On The Steady Flow of a Fluid of Variable Density
Past an Obstacle. Tellus. 13.: 239-251.
Gifford, F.A. 1980. Smoke as a Quantitative Atmospheric Diffusion
Tracer. Atmospheric Environment. 14: 1119-1121.
Holzworth, G.C. 1980. The EPA Program for Dispersion Model
Development for Sources in Complex Terrain. Second Joint
Conference on Applications of Air Pollution Meteorology, New
Orleans, LA. AMS, Boston.
Hunt, J.C.R., and R.J. Mulhearn 1973. Turbulent Dispersion from
Sources Near Two-Dimensional Obstacles. J. Fluid Mech., 61:
245-274.
Hunt, J.C.R. and W.H. Snyder 1980. Experiments on Stably and
Neutrally Stratified Flow Over a Model Three Dimensional Hill.
J. Fluid Mech.. 96: 671-704.
Hunt, J.C.R. 1981. Diffusion in the Stable Boundary Layer.
Atmospheric Turbulence and the Pollution Modelling. F.T.M.
Nieustadt and H. Van Dop (Ed.). D. Reidel, Holland.
230
-------
REFERENCES (Continued)
Hunt, J.C.R. 1982. Diffusion in the Stable Boundary Layer. In
Atmospheric Turbulence and Air Pollution Modeling. D. Reidel
Publishing Company, Dordrecht, Holland, 231-274.
Hunt, J.C.R., J.C. Kaimal, J.E. Gaynor and A. KorrelL 1983.
Observations of Turbulence Structure in Stable Layers at the
Boulder Atmospheric Laboratory. In Studies of Nocturnal Stable
Layers at BAO. Report Number Four, Jan. 1983. Available from
NOAA/ERL, Boulder, CO 80303, U.S.A., 1-52.
Kapsha, T.P., et al. 1976. Behavior of S02 Plumes from Tracy and
Fort Churchill Generating Stations Under Stable Atmospheric
Conditions in the Vicinity of Complex Terrain. Westinghouse
Electric Corporation, Pittsburgh.
Lavery, T.F., A. Bass, D.G. Strimaitis, A. Venkatram, B.R. Greene,
P.J. Drivas, and B.A. Egan 1982. EPA Complex Terrain Model
Development: First Milestone Report - 1981. EPA-600/3-82-036,
Research Triangle Park, NC. 304 p.
Lauwerier, H.A. 1954. Diffusion from a Source in a Skew Velocity
Field. Appl. Sci. Res.. 4, p!53.
Overcamp, T.J. 1983. A Surface-Corrected Gaussian Model for Elevated
Sources. J _._ of Climate and Applied Met. . 22: 1111-1115.
Pasquill, F., and F.B. Smith 1983. Atmospheric Diffusion, 3rd
Edition. Ellis Horwood Ltd, England.
Pearson, H.J., J.S. Puttock and J.C.R. Hunt 1983. A Statistical
Model of Fluid Element Motions and Vertical Diffusion in a
Homogeneous Stratified Turbulent Flow. JA Fluid Mech., 129:
219-249.
Riley, J.J., Liu, H.T. and Geller, E.W. 1976. A Numerical and
Experimental Study of Stably Stratified Flow Around Complex
Terrain. EPA Report No. EPA-600/4-76-021, Res. Tri. Pk., NC, 41p.
Sheppard, P.A. 1956. Airflow Over Mountains. Quart. J. R. _Met«jor.
Soc., 82: 528-529.
Stiyder, W.H. , R.E. Britter and J.C.R. Hunt 1980. A Fluid Mo_deling
Study of the Flow Structure and Plume Impingement on a
Three-Dimensional Hill in Stably Stratified Flow. Proc. Fifth
Int. Conf. on Wind Engr. (J.E. Cermak, ed.), 1: 319-329, Pergamon
Press, NY, NY.
231
-------
REFERENCES (Continued)
Snyder, W.H. and J.C.R. Hunt 1984. Turbulent Diffusion from a Point
Source in Stratified and Neutral Flows Around a Three Dimensional
Hill; Part II - Laboratory Measurement of Surface Concentrations.
Submitted to Atmospheric Environment.
Uthe, E.E. and B.M. Morley 1984. Alpha-1 Observations of Plume
Behavior for PMV&D Tracy Site. Electric Power Research Institute,
Palo Alto, California.
Venkatram, A. and R. Paine 1984. Development of a Model to Estimate
Dispersion of Elevated Releases in the Shear-Dominated Boundary
Layer. ERT Report No. P-B846-500. Prepared for the Maryland
Power Plant Siting Program.
Venkatram, A., D. Strimaitis, D. DiCristofaro 1984. A Semiempirieal
Model to Estimate Vertical Dispersion of Elevated Releases in the
Stable Boundary Layer. Atmospheric Environment. 18: 923-928.
232
-------
APPENDIX A
STABLE PLUME DISPERSION OVER AN ISOLATED HILL
RELEASES ABOVE THE DIVIDING-STREAMLINE HEIGHT
233
-------
STABLE PLUME DISPERSION OVER AN ISOLATED HILL
Releases above the Dividing-Streamline Height
by
William H. Snyder *
and
Robert E. Lawson, Jr. *
Meteorology and Assessment Division
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
December 1983
*On Assignment from the National Oceanic and Atmospheric Administration,
U.S. Department of Commerce.
23A
-------
ABSTRACT
A series of tows was conducted in a stably stratified salt-water
towing tank wherein the density gradient was linear and the
dividing-streamline height was half the hill height. Effluent was
released at three elevations above the dividing-streamline height.
Pairs of tows were made such that, in one tow, the hill (upside-down)
was fully-immersed in the water and the towing speed was adjusted to
provide a "natural" dividing-streamline surface. In the second tow of
the pair, the hill was raised out of the water to the point where only
the top half of the hill was immersed, thus, forcing a flat dividing-
streamline surface, while all other conditions were maintained
identical. Concentration distributions were measured on the hill
surface and in the absence of the hill. Concentration distributions
from each pair of tows were compared to ascertain effects of an
assumed flat dividing-streamline surface as is used in some
mathematical models. The results suggest that the assumption of a
flat dividing-streamline surface is a reasonable approximation to
make, at least with regard to predicting the locations and values of
maximum concentrations and areas of coverage on the windward side of
the hill. When the stack heights are relatively close to the
dividing-streamline height, the lee-side concentrations are also
predicted reasonably well. The apparent cause of the relatively poor
agreement between lee-side concentration patterns in the higher stack
cases is the presence of a hydraulic jump at the downwind base of the
hill in the full-immersion case which was absent in the half-immersion
case.
235
-------
1. INTRODUCTION
The structure of strongly stratified flows over a three
dimensional hill has been envisioned as composed of two layers: a
lower layer of essentially horizontal flow wherein plumes from upwind
sources impinge directly on the hill surface, and an upper layer
wherein plumes from upwind sources may pass over the hill top. This
basic concept was suggested by theoretical arguments of Drazin (1961)
and Sheppard (1956) and was demonstrated through laboratory
experiments by Riley et al. (1976), Brighton (1978), Hunt and Snyder
(1980), Snyder et al. (1980) and Snyder and Hunt (1983). Complex
terrain diffusion models utilizing this concept have been developed by
Hunt, Puttock and Snyder (1979) and, more extensively, by Lavery
et al. (1982), Strimaitis et al. (1982), and Lavery et al. (1983).
The basic parameter characterizing the flow structure is the Froude
number, F (see Snyder and Hunt, 1984).
We are concerned in this report with the upper-layer flow. The
basic assumption in this approach is that, in strongly stratified
flows (0
-------
hypothesis (balance of kinetic and potential energy), terminates on
the surface at the top of the hill, again suggesting that the
dividing-streamline surface cannot be flat. On the other hand,
concentration measurements on hill surfaces (Snyder and Hunt, 1984)
suggested that a flat.surface approximation may yield reasonable
estimates. From a practical viewpoint, the mathematical models are
vastly simplified if such an assumption yields reasonable estimates of
surface concentration. Hence, we attempt to answer the question not
from a detailed analysis of the shape of such a dividing streamline
surface, but from the more practical comparison of surface
concentration patterns.
A series of tows was conducted in the salt-water-stratified
towing tank using, as the basic hill shape, the fourth order
polynomial hill as used by Hunt and Snyder (1980). Linear density
gradients were established in the tank and the hill was towed at a
speed U such that the Froude number F ( = U/Nh, where N is the
Brunt Vaisala frequency) was 0.5. Effluent was released at heights of
Hs = 0.6h, 0.7h and 0.8h, and the resulting hill-surface
concentration patterns were measured. Since the density gradient was
linear and F-0.5, the dividing-streamline height was also 0.5h (Hc/h
= 1 - B'); since the effluent was released above Hc, the plumes did
not impinge directly on the hill, but instead pollutants reached the
hiil surface by the combination of streamline displacement and
diffusion. After that series was completed, a second series of tows
was conducted wherein the entire model (hill, baseplate and stack, as
a unit) was raised out of the water* to the point where the water
surface was precisely at the dividing-streamline height, i.e., the
water surface was at half the hill height. The model was towed at the
same speed as in the full-immersion tows, so that the Froude number
with this now half-height hill was unity, and all streamlines passed
over the hill top. The flat water surface thus forced a flat
dividing-streamline surface. The resulting surface concentration
patterns were then compared with the corresponding full-immersion tows
to ascertain the effects of a flat dividing-streamline surface.
*The model is routinely mounted upside down such that the baseplate is
submerged a few millimeters below the water surface. In discussion
of flow structure and plume behavior, however, we discuss the results
as if the model were right-side up.
237
-------
2. EXPERIMENTAL APPARATUS AND TECHNIQUES
Most of the details of the experimental apparatus and techniques
were given by Hunt and Snyder (1980) and in a laboratory report by
Hunt, Snyder and Lawson (1978). The basic method of making
concentration measurements was described by Snyder and Hunt (1984).
Only a brief overview is given here, but changes in the techniques and
apparatus as well as special features of these experiments are
described in detail.
A fourth-order polynomial hill (z(r) = h/(l+(r/L)4)) of height
24.3 cm was used in a stratified towing tank. The tank, 1.2 m in
depth, 2.4 m in width and 25 m in length, was stably stratified with
layered mixtures of salt water. This dye mixture was emitted at four
times the isokinetic rate from a bent-over "stack" of 0.635 cm o.d.
The stack exit was located 84.8 cm (3.5 h) upstream of the hill
center. The non-isokinetic effluent release rate was used to obtain a
realistic plume size and shape, i.e., in the isokinetic releases used
previously by Snyder and Hunt (1984), because of the nonexistent
approach-flow turbulence, the plumes tended to be exceptionally thin
and narrow. With the four-times-isokinetic rate, a weak but turbulent
jet was formed at the stack exit. This jet grew in size with
downstream distance to provide a plume with dimensions not
insignificant in comparison with, say, the hill height and hence, a
more realistic simulation of a typical atmospheric situation. The
maximum jet velocity at the hill center was estimated to be about
2 cm/s relative to the hill (Townsend, 1956), i.e., small compared
with the towing speed. Characteristics of the plume in the absence of
the hill were measured with horizontal and vertical rakes of sampling
tubes, and are also presented herein.
One hundred sampling ports were fixed on the hill surface,
distributed as shown in Figure A-l. In some preliminary tows, the
sampling ports (2.4 mm o.d.) protruded 2.5 mm above the smooth hill
surface, but as the plumes were spread broadly to cover most of the
hill surface (but very thinly in the direction normal to the surface),
narrow, clear wakes were observed down-stream of the protuberant
sampling tubes. This disturbance caused strong reductions in
concentrations measured at ports directly downstream from others,
e.g., ports along the 0° line (Figure A-l). The sampling ports were
then cut to be flushed with the hill surface, but some amount of
interference was still observed in the concentration distributions,
apparently due to the withdrawal of sample fluid through the ports.
After considerable experimentation, a final configuration was found
that displayed no interference: the ports were raised to the original
2.5 mm above the smooth hill surface and the hill was covered with
238
-------
312. 90°
-34-0,
180° |
.113.U211HimQ91.Q81lIl7iQei05.10ii.lQ3.
FIJOW '
DIRECTION \
2C.?U'ii01.
12. 101. 0 501. 502. 5Q3.5Qu.5055.065Q75a£S.0951Q5U512.513.5'.i-l.
801. 601. I
701> / I
802. 602.
612.
-45°
E
Pfl
POLYNOMIflL
HILL
FLUID
MODELING
SECTION
Figure A-l. Port locations and numbering system for lift study.
(Precise location is at decimal point. Dashed circle
marks half the hill height.)
239
-------
sharp-edged gravel of grain size (longest dimension) 2 to 4 mm. Also,
the sample withdrawal rate was reduced by a factor of about 4, to
13 cm-Vmin. This corresponds to a sampling "stream tube" diameter
of 1.3 mm'at the typical tow speed. The rough surface provided the
additional benefit of eliminating a viscous sublayer on the hill
surface, hence, avoided problems of molecular diffusion through this
viscous sublayer.
The concentration of dye in the collected samples were analyzed
on a Brinkman Model PC-600 probe colorimeter. The fiber-optics probe
was immersed sequentially into the sample test tubes. The wavelength
used was 570 nm. The output of the colorimeter, a voltage related to
the opacity of the solution being tested, was fed to a POP 11/44
minicomputer, where it was converted to a concentration in percent
dye. The conversion utilized a calibration curve formed by recording
the output voltage versus concentration for at least 12 "standards"
which consisted of accurately known dilutions of the same dye used for
the effluent source. A "Beer's Law" type of curve was best-fit to the
standards for use in converting the voltage from the unknown sample
into a dye concentration. Although the instrument required care in
use (by frequently checking the "zero transmittance" and "zero
absorbance" controls), frequent checks of the calibration showed
excellent repeatability. A typical calibration curve is shown in
Figure A-2. The concentration of dye in the effluent was occasionally
adjusted such that the majority of samples would be within the most
reliable range of the calibration curve, i.e., the vast majority of
samples analyzed were in the range of 0.005 to 0.25%.
The sample lines leading from the ports on the hill surface (or
rake) to the sample test tubes had to be filled prior to the beginning
of a tow in order for the vacuum-sampling system to work properly.
Because of the reduced sample flow rate, the volume of dye-free water
stored in the lines (and which was thereafter drawn into the test
tubes along with the dye samples) was approximately 25% of the total
sample collected. A first-order correction to the measured
concentrations was made to account for this dilution of the samples.
However, the precise volume of sample collected in each test tube
varied slightly from one tube to the next, so that the accuracy of the
concentration-measurement system is estimated to be in the range of
to 15%.
240
-------
A fOLCRL. it C 2. 1) ^.I^r 3:UD1 . POLrHFLL.31.
" 3CERFT. '4 t ?., 5) QTCON7 MGV-25-77
27-DEC-83
IP fOCD DT£.(^-J S3. 3231 . BRINKMHM . IPCGOOi
»&r!/SEC,OH..H2CCEP.=:0.5aCM.?'r-V:, «••
-------
3. PRESENTATION AND DISCUSSION OP RESULTS
3.1 Plumes in the Absence of the Hill
Vertical and horizontal concentration profiles measured at the
downstream position of the center of the hill (but in its absence) are
presented in Figures A-3 and A-4, respectively. Profiles were
measured for each of three stack heights, 0.1 h, 0.2 h and 0..6 h. For
the two lowest stack heights, the baseplate was raised out of the
water; for the highest stack height, it was submerged as it was for
the full-immersion tows with the hill (approximately 6 mm).
Concentrations have been nondimensionalized as x = CUooh^/Q,
where C is the dimensional dye concentration, Um is the towing
speed, h is the hill height, and Q is the volume flow rate of dye in
the effluent. The calculated plume parameters are listed in
Table A-l. The two higher-level plumes are essentially identical and
are very nearly Gaussian in shape. (Best-fit Gaussian curves; are
shown for comparison in the figures. The parameter Xmxg shown in
Table A-l is the maximum concentration in the best-fit Gaussian
distribution with the same standard deviation as the measured data.)
The lower-level plume, however, has clearly diffused to the water
surface (Figure A-3); a reflected-Gaussian profile assuming ain
effective stack height of 0.09 h appears to fit the data quite well.
The standard deviation of the vertical distribution az (calculated
assuming perfect reflection at the water surface), however, is 40%
larger than those of the higher level plumes. The standard deviation
of the lateral distribution cry of the lower level plume is also
somewhat larger (10%) than that of the upper level plumes. The
probable cause of this increased diffusion is the reduced stability
near the water surface, as shown in Figure A-5. Ideally, the density
gradient would have been perfectly constant all the way to the water
surface, so that the plume released at the lowest level (0.1 h) in the
half-immersion case would have been essentially identical in shape to
that released at the upper level (0.6 h) in the full-immersion case.
However, as will be shown later, these differences in approaching
plume shapes appeared to have little effect on the concentration
distributions measured on the hill surface.
3.2 Distributions on the Hill Surface
Figure A-6 shows side and top views of the plume surmounting the
hill, in this case with the hill fully immersed and the effluent
emitted at 0.6 h. These photographs show the plume to be deformed to
cover essentially the entire surface of the hill above half the hill
242
-------
LIFT .25 t 3. 1) LIFT STUDT, NO HILL2. VERT. PROF.. STK. HT.-2...
09-DEC-83 DTE=U5.6ML/MIN.DIL=865.8ML/MIN,HCRIT=. 5M, ci i: X P. . L'LK, T-0, 2= J 4. 58CMPBLCP
SCflL 1,0000, 0.000, 2*. ViO, 0.000
Q LIFT .21 (3, 1) LIFT STLIDT. NO HILL2, VERT. PROF. . 5TK. HT. =u. 8K K. i /?= 1 S. 77CK/SEC
07-DCC-83 DrC^U7. 7ML/MIN, DIL-905. 8ML/MIN, IICRIT-G. C-!!. n-vT: X-OH. GCt-'. T-0, Z-lPCLCfiL.
-------
A LIFT .26 (1. 3) LIFT STUDT.NO HI LL2. VERT. PROF. . STK. HT . -2 A3CK , T /:,- 1 :,. i Cf./L^C
scfiL ' 211.3000. o.'ooo, ' i'.'ooo', '"o'.ooo"""
LIFT STUDT.NO HILL2. LflT. PROF.,STK.HT.=U.96CM.T/S=15.77CK/SEC
DTE-i47.7ML/MIN,OIl.-905.8ML/MIN,IICRIT-0.5i:.riKt:i X-OU.CCM.T-D.2-17.
SCfiL 2U.3000. 0.000, 1.000, 0.000
O LIFT . ?li [ 1. 3) LIFT STUDT.NO HILL.VERT.PROF.,STK.HT.=1U.58CK,T/S=15.83CM/SfC
08-DEC 30
D LIFT .22 I 1. 3)
07-DCC-83
X
u
10 -
5
-.03
Y/H
Figure A-4. Horizontal concentration profiles measured at the
downstream position of the center of the hill, but in
its absence.
244
-------
File
No.
21
22
23
24
25
26
Horizontal/
Vertical
V
H
V
V
V
H
Hs/ll
0.2
0.2
0.6
0.6
0.1
0.1
Xmx
25.8
23.6
23.4
21.6
21.1
22.4
a
(cm)
1.07
2.36
1.07
2.28
1.50'
2.55
inixg
28.7
27.2
25.3
25.1
22.2
24.0
(a) Calculated assuming perfect reflection at the water surface and an
effective stack height of 2.2 cm (0.09 h).
245
-------
PL5RS .30 [3.
09-DEC-83
POLTHILL LIFT STUDT , WHTCR IIT . = 1 00. 8CK, MCTTLCH (PLE'OO)
TIME: 0610
SCRL 1.0000, 0.000. 21.500, 0.000
.8
.7
.6
.5
Z
\
N
«-»
1.06 1.07
1.08 1.09
SPECIFIC GRAVITY
1.1
1.11
Figure A-5. Density profile measured prior to Tow #25.
(Concentration distributions shown as lower plume in
Figures A-3 and A-4.)
246
-------
Figure A-6. side and top-views of plume released at 0.6 H.
F = 0.5, fully submerged.
247
-------
height and yet to be quite thin in the direction normal to the hill
surface (the dotted line on the photographs marks half the hill
height). Notice that, on the lee side, the plume sweeps to much lower
elevations, so that we may expect to see substantial surface
concentrations on the lee side below half the hill height.
Figure A-7 presents a scatter diagram showing the repeatability
of surface concentration measurements from one tow to the next. Tn
this case, both tows (numbers 10 and 11) were made with the hill
fully-immersed and the stack height fixed at 0.6 h. From the; diagram,
it may be seen that, for any port, repeatability is well-within a
factor of two. At large values of concentration, repeatability is
generally within 10%. The repeatability is less good at the lower
concentrations. A few points with the largest percentage differences
between the tows are marked with the port number (see Figure A-l);
these show that the largest differences occurred below half the hill
height, i.e., on the lower edges of the plume.
Figures A-8 and A-9 show the concentration distributions measured
on the hill surface in the full-immersed and half-immersed cases,
respectively. The most obvious difference between the two ca,5es is
the absence of lee-side concentrations below half the hill height in
the half-immersed case. Of course, in the half-immersed case,
concentrations at positions below half the hill height are zero,
because that portion of the hill is outside the water. In the
fully-immersed case, the plume diffused to some extent below half the
hill height around the upwind side, but also, as mentioned eai-lier,
this plume "hugged" the hill surface as it was swept down the lee side
to much lower elevation than the release height. Just beyond the
downwind base of the hill (not shown on the photographs), the flow
appeared to separate from the surface and rise abruptly in a hydraulic
jump (see Hunt and Snyder, 1980).
Side and top views of this plume in the half-immersed case are
shown in Figure A-10. These are to be compared with the
full-immersion case of Figure A-6. The comparison shows similar
behavior as discussed in the above paragraph.
Figure A-ll presents a scatter diagram comparing, on a
port-to-port basis, the surface concentrations measured in the half-
and fully-immersed cases. Measurements at points below half the hill
height are not included here because, of course, in the half-immersion
case, these ports were out of the water. Within the region of large
concentrations, the two cases compare quite favorably, the
half- immersed case yielding concentrations approximately 10 to 20%
larger than the fully-immersed case. In the region of low
concentration, quite large differences occur (worst case, a factor of
10). However, a close examination shows that in all cases where
concentrations differed by more than a factor of 2, the port locations
were very close to half the hill height i.e., either at 0.505 h or
0.59 h (port numbers are marked on Figure A-ll for those cases where
concentrations differed by more than a factor of 2).
248
-------
L1FOIF. 7 ( 3. 2) POLTHIL LIFT STUDY. COL 1. PGfU NO.: COL.2 TO U. CGMI. PRO1; i ITT.010
22-NOV-83 TO .012
100
I
s
o
cc
u.
X
O
1 10
CHI FROM TOW 11
100
Figure A-7. Repeatability of surface concentration measurements.
F « 0.5. HS/H =0.6, fully submerged.
249
-------
Figure A-8. Concentration distributions measured on the hill
surface. F = 0.5. HS/H = 0.6, fully submerged.
250
-------
Figure A-9. Concentration distributions measured on the hill
surface. F = 0.5, HS/H =0.6, half submerged.
251
-------
Figure A-10. Side and top views of plume released at 6.6 H.
F = 0.5, half submerged.
252
-------
LIFDIF. 7 ( li. 3) PGLTHIL LIFT STUDT. COL 1, PCRT NO.: COL.2 TO 4. CGNC. FfiCiM LIFT.OiC
22-NCV-83 TO .012
100
1=1
tu
o
-------
Note that the maximum hill-surface concentration is in the range
of 25 to 30, whereas the maximum in the plume in the absence of the
hill was in the range of 20 to 25. This suggests that even though the
plume shape was highly contorted through the streamline deformation
process, the basic diffusion process was limited by the presence of
the surface; this "reflection" at the hill surface resulted in a
maximum surface concentration that was 20 to 25% larger than that
observed at the center of the plume in the absence of the hill.
However, as mentioned earlier, the accuracy of the concentration-
measurement system was estimated to be in the range of +10 to 15%, so
that the above estimates of increased concentration should be taken
with caution.
Figures A-12 and A-13 show the concentration distributions
resulting from sources elevated at 0.7 h. As was the case with a
source height of 0.6 h, the contours on the windward side of the hill
are roughly circular, but, in this case, somewhat more elongated in
the streamwise direction. Again, of course, no concentrations were
observed below half the hill height in the half-immersion case; in the
full-immersion case, the plume "hugged" the lee side of the hill, but
was spread less broadly in the crosswind direction and lifted off the
surface into the hydraulic jump somewhat sooner.
Also, as was the case at a source height of 0.6 h, the maximum
concentration is about 10% larger in the half- than in the
full-immersion case. The location of the maximum, however, has moved
from the windward side to the lee side of the hill.
A scatter diagram comparing concentrations on a port-by-port
basis for the half- and fully-sumberged hills with a source height of
0.7 h is shown in Figure A-14. The correspondence between the full
and half-depth immersions is not as good as was the case with a source
height of 0.6 h. For the large concentrations, the agreement is
excellent, but for small concentrations, the scatter is quite large.
Port numbers where the concentrations differed by more than a factor
of two are marked on the figure. These show that the comparisons were
poor only in the elevation range of 0.5 h to 0.67 h, i.e., n&ar the
water surface in the half-immersion case. In the half-depth case,
concentrations were larger on the windward line of ports, apparently
related to the wider vertical diffusion of the plume to the water
surface, whereas concentrations elsewhere round the sides of the hill
were considerably smaller than in the full-depth case.
Figures A-15 and A-16 show the concentration distributions
resulting from sources elevated at 0.8 h. In this case, the
correspondence between the two distributions is not nearly as good as
was the case at the lower stack heights. whereas the locations of the
maximum concentrations are approximately the same, the values in the
full-immersion case are 55% larger and the area of plume contact on
the hill surface is nearly quadrupled over the half-immersion case.
Note that the location of the maximum concentration is well to the lee
side of the hill and that the area of plume contact is much smaller
than for the lower stack heights.
254
-------
Figure A-12. Concentration distribution measured on the hill
surface. F = 0.5, HS/H =0.7, fully submerged.
255
-------
\
\
\
\
\
Figure A-13. Concentration distribution measured on the hill
surface. F = 0.5, HS/H =0.7, half submerged.
256
-------
LITDir. B ( 2, 3) POLTHIL LIFT STUDY. COL 1, PORT NO. ; COL 2 TO 7, CONIC. T'l'ul', LITT.OiJ.
Ol-DEC-83 .017. .014. .015. .016. RND .018. RESPECTIVELY.
100
o
UJ
O
OC
Ul
5
0)
D
09
3
O
X
o
1 10
CHI, TOW 13, HALF SUBMERGED
100
Figure A-14.
Comparison of concentrations. Half submerged versus
fully submerged. F = 0.5, HS/H =0.7.
257
-------
\
\
Figure A-15. Concentration distributions measurea on the hill
surface. F = 0.5, HS/H = 0.8, fully submerged.
258
-------
\
\
Figure A-16. Concentration distributions measured on the hill
surface. F = 0.5, HS/H = 0.8, half submerged.
259
-------
A scatter diagram comparing these two cases is presented in
Figure A-17. The correspondence between the full- and half-depth
immersion cases is clearly poor, with the half-depth immersion tow
exhibiting surface concentrations that are generally less than half
those from the full-depth tow. Both of these tows were repeated and
the repeatability was found to be excellent, i.e., similar to that
shown in Figure A-7 for the lower stack height of 0.6 h.
The poor correspondence in this case may be partially understood
by examining the side- and top-view photographs of the plume as shown
in Figures A-18 and A-19. The top-views show that, in the half-depth
tow, the plume was much narrower on the lee side of the hill. The
side views show that, in the full-depth tow, the plume "hugged" the
hill surface on the lee side to elevations considerably lower than
half the hill height, then lifted off the surface and rose somewhat in
elevation with the hydraulic jump at the downwind base of the hill
(not seen in the photograph). This hydraulic jump was not present in
the half-depth tow, and the plume remained essentially at the water
surface. The presence of the hydraulic jump evidently affects the
flow structure on the lee side of the hill and, therefore, the
concentration distributions on the lee side.
260
-------
LJFDIF. C I U. G) POLTIIIL LIFT STUDT. COu 1, PORT NO.;CO^ 2 1C 7, CC'.MC. rf,3!
01-DEC-S3 .017. .014. .015, .015, fiN'D .018, RESPECT! V£LT.
100
Q
UJ
O
oc
111
2
CD
3
M
UL
<0
T-
I
»
I
U
1 10
CHI, TOW 14. HALF SUBMERGED
100
Figure A-17. Comparison of concentrations. Half submerged versus
fully submerged. F = 0.5, HS/H =0.8.
261
-------
Figure A-18.
Side views of plume released at 0.8 H. F = 0.5; upper,
fully submerged; lower, half submerged.
262
-------
Figure A-19.
Top views of plume released at 0.8 H. F=0.5; upper,
fully submerged; lower, half submerged.
263
-------
4. CONCLUSIONS
A series of tows was conducted in a stably stratified, salt-water
towing tank wherein the density gradient was linear and the
dividing-streamline height was half the hill height. Effluent was
released at 0.1 h, 0.2 h and 0.3 h above the dividing-streamline
height, i.e., at elevations above the base of the hill of 0.6 h, 0.7 h
and 0.8 h. Pairs of tows were made such that, in one tow, the hill
war fully immersed ir the water and the towing speed was adjusted to
provide a "natural" dividing-streamline surface (i.e., not flat) of
half the hill height (far upstream). In the second tow of the pair,
the hill was raised out of the water to the point where only the top
half of the hill was immersed; this, in effect, forced a flat
dividing-streamline surface (the water surface itself). In both tows
of each pair, all other conditions were maintained identical, e.g.,
towing speed, effluent release height and flow rate, and density
gradient. Concentration distributions were measured on the hill
surface and in the absence of the hill. The distributions from each
pair of tows were compared to ascertain the effects of an assumed flat
dividing-streamline surface.
Conclusions that may be drawn from this study are as follows:
1. Repeatability of concentration patterns from one tow to the
next, under ostensibly identical conditions, is regarded as
excellent, generally within +10%. The few points outside
this range (but generally well within a factor of two) were
found to be on the edges of the plume where the
concentrations were quite small and the turbulent
fluctuations naturally gave rise to this kind of variability.
2. At a stack height of 0.6 h (0.1 h above Hc), the maximum
concentration occurred on the windward side of the hill; the
plume was spread broadly to cover essentially the entire top
half of the hill, but its thickness in the direction normal
to the hill surface was small. Isoconcentration contours on
the windward side were roughly circular. Comparison of the
concentration patterns from the half- and full-immersion
tows showed remarkable similarities over the top half of the
hill. The locations of the maxima were essentially the
same, although the value of the maxima was 10 to 20% larger
for the half-immersion case. Regions of poor agreement
included a thin band just above the half-hill height (where
concentrations differed, in the worst case, by a factor of
about four) and, of course, everywhere below the half-hill
height.
264
-------
3. At a stack height of 0.7 h (0.2 h above Hc), the maximum
surface concentration occurred just downwind of the hill
center, with a value nearly the. same as with the stack.
height of 0.6 h. The plume was again spread broadly to
cover most of the top half of the hill. Isoconcentration
contours were again roughly circular on the windward side,
but somewhat elongated in the flow direction. Comparison of
the concentration patterns from the full- and half-depth
tows again showed remarkable similarities and quite good
agreement in the areas of higher concentration. Regions of
poor agreement covered a broader band just above the
half-hill height and, in that region, concentration values
differed by as much as a factor of 8.
4 At a stack height of 0.8 h (0.3 h above Hc), the maximum
concentration occurred well to the lee side of the hill and
its value was substantially smaller than the maxima observed
in the lower stack cases. Also in comparison with the lower
stack cases, this plume was quite narrow, and the area of
coverage on the hill surface was quite small. Comparison of
the concentration patterns from the full- and half-depth
tows showed that, in areas of large concentration, values of
concentration differed by roughly factors of two, In areas
of low concentration, values of concentration differed by up
to a factor of 10.
5. Observations of the flow structure revealed that a hydraulic
jump occurred near the downwind base of the fully immersed
hill, but was absent in the case of the half-immersed hill.
This hydraulic jump appeared to have a rather minor effect
on the streamline pattern near the surface on the lee side
of the hill, but, as pointed out by Snyder and Hunt (1984),
surface concentrations are extremely sensitive to the ratio
of the plume width to the normal distance of the plume
centerline from the hill surface when this ratio is near
unity. Hence, even a seemingly minor change in the
streamline patterns can have a major effect on the
concentration patterns.
6. Maximum concentrations on the surface under the "worst"
conditions were equal to or somewhat greater than those
observed at the plume centerline in the absence of the hill.
7. These results suggest that the assumption of a flat
dividing-streamline surface in a mathematical model is a
reasonable approximation to make, at least with regard to
predicting the locations and values of maximum
concentrations (and areas of coverage) on the windward side
of the hill. When the stack heights are relatively close to
the dividing-streamline height, the lee-side concentrations
would also be predicted reasonably well.
265
-------
Future work should include an improved concentration-
measurement technique to resolve the question of whether or
not hill-surface concentrations can exceed those at the
plume centerline in the absence of the hill (and by how
much).
266
-------
REFERENCES
Bass, A., Strimaitis, D.6. and Egan, B.A., 1981: Potential Flow Model
for Gaussian Plume Interaction with Simple Terrain Features, Rpt.
to Envir. Prot. Agcy. under Contract No. 68-02-2759, Res. Tri.
Pk., NC, p. 201.
Brighton, P.W.M., 1978: Strongly Stratified Flow Past Three-Dimensional
Obstacles, Quart. J. Roy. Meteorol. Soc., v. 104, p. 289-307.
Drazin, P.G., 1961: On the Steady Flow of a Fluid of Variable Density
Past an Obstacle, Tellus, v. 13, no. 2, p. 239-51.
Hunt, J.C.R., Puttock, J.S. and Snyder, W.H., 1979: Turbulent
Diffusion from a Point Source in Stratified and Neutral Flows
around a Three-Dimensional Hill: Part I: Diffusion Equation
Analysis, Atmos. Envir., v. 13, p. 1227-39.
Hunt, J.C.R., Richards, K.J. and Brighton, P.W.M., 1984: Stratified
Shear Flow over Low Hills: II. Stratification Effects in the
Outer Flow Region, To be submitted to Quart. J. Roy. Meteorol.
Soc.
Hunt, J.C.R. and Snyder, W.H., 1980: Experiments on Stably and
Neutrally Stratified Flow over a Model Three-Dimensional Hill, J.
Fluid Mech., v. 96, pt. 4, p. 671-704.
Hunt, J.C.R., Snyder, W.H., and Lawson, R.E. Jr., 1978: Flow Structure
and Turbulent Diffusion around a Three-Dimensional Hill: Fluid
Modeling Study on Effects of Stratification; Part I: Flow
Structure, Rpt. No. EPA-600/4-78-041, Envir. Prot. Agcy., Res.
Tri. Pk., NC.
Lavery, T.F., Bass, A., Strimaitis, D.6., Venkatram, A., Greene, B.R.,
Drivas, P.J. and Egan, B.A., 1982: EPA Complex Terrain Modeling
Program: First Milestone Report - 1981, Rpt. No. EPA-600/3-82-036,
Envir. Prot. Agcy., Res. Tri. Pk., NC, p. 304.
Lavery, T.F., Strimaitis, D.G., Venkatram, A., Greene, B.R.,
DiCristofaro, D.C. and Egan, B.A., 1983: EPA Complex Terrain
Model Development: Third Milestone Report-1983, Rpt. No.
EPA-600/3-83-101, Envir. Prot. Agcy. Rpt., Res. Tri. Pk., NC, p.
271.
267
-------
Riley, J.J., Liu, H.T. and Geller, E.W., 1976: A Numerical and
Experimental Study of Stably Stratified Flow Around Complex
Terrain, Rpt. No. EPA-600/4-76-021, Envir. Prot. Agcy., Res. Tri.
Pk., NC, p. 41.
Sheppard, P.A., 1956: Airflow over Mountains, Quart. J. Roy. Meteorol.
Soc., v. 82, p. 528-9.
Snyder, W.H., Britter, R.E. and Hunt, J.C.R., 1980: A Fluid Modeling
Study of the Flow Structure and Plume Impingement on a
Three-Dimensional Hill in Stably Stratified Flow, Proc. Fifth
Int. Conf. on Wind Engr. (J.E. Cermak, ed.), v. 1, p, 319-29,
Pergamon Press, NY, NY.
Snyder, W.H. and Hunt, J.C.R., 1984: Turbulent Diffusion from a Point
Source in Stratified and Neutral Flows around a Three-Dimensional
Hill; Part II: Laboratory Measurements of Surface Concentration,
Atmos. Envir. (to appear).
Strimaitis, D.G., Venkatram, A., Greene, B.R., Hanna, S., Heisler, S.,
Lavery, T.F., Bass, A., and Egan, B.A., 1982: EPA Complex Terrain
Model Development: Second Milestone Report - 1982, Rpt. No.
EPA-600/3-83-015, U.S. Envir. Prot. Agcy. Rpt., Res. Tri. Pk., NC.
Townsend, A.A., 1956: The Structure of Turbulent Shear Flow, Cambridge
Univ. Press, Cambridge, England, p. 315.
268
-------
APPENDIX B
DISPERSION FROM A SOURCE UPWIND OF A
THREE-DIMENSIONAL HILL OF MODERATE SLOPE
269
-------
DISPERSION FROM A SOURCE UPWIND OF A
THREE-DIMENSIONAL HILL OF MODERATE SLOPE
Roger S. Thompson
and
William H. Snyder*
Meteorology and Assessment Division
Atmospheric Sciences Research Laboratory
Environmental Protection Agency
Research Triangle Park, NC 27711
May 1984
*0n assignment from the National Oceanic and Atmospheric
Administration, U.S. Department of Commerce.
270
-------
1. INTRODUCTION
The series of experiments performed in 1980 at Cinder Cone Butte
(CCB), Idaho, by Environmental Research and Technology, Inc.
(Strimaitis et al., 1983 and Lavery et al., 1983) provide a data base
for development and evaluation of dispersion models for complex
terrain situations. As a follow-up and companion effort, fluid
modeling studies have been performed at the EPA Fluid Modeling
Facility using both a stratified towing tank and a meteorological wind
tunnel. The work described in this report was inspired by an earlier
case study of a one-hour period of the field study for which the
atmosphere was neutral. That wind-tunnel study (see Appendix A of
Lavery et al., 1983) provided good agreement between laboratory and
field measurements of concentrations. However, the maximum observed
concentration on the hill surface was found to be only about 10%
greater than that observed from the same source over flat terrain;
that is, the influence of CCB in increasing the maximum observed
ground-level concentration (glc) was not dramatic. Thus, a set of
experiments to determine the influence of the hill on the maximum glc
and to locate the source positions where this influence is the
greatest was a natural continuation of the fluid modeling effort. The
data base for mathematical modeling evaluation is extended by these
measurements. All measurements were made under simulated atmospheric
conditions.
The nearly-axisymmetric CCB shape was replaced with a truly
axisymmetric hill having a simple mathematical formula. The average
silhouette of CCB was best fit by the formula to maintain a close
relationship with the previous efforts: the maximum slope of CCB was
matched in the model.
Since, for upwind sources, the maximum ground-level concentration
can be expected to occur from sources on the centerline, this study
was restricted to only those source positions. The maximum
concentrations from these sources are expected to occur along the
centerline through the source and hill center; thus, measurements were
limited to centerline sampling port locations.
The measure of the influence of the hill on the maximum glc was
computed as a "terrain amplification factor" A. This factor is
defined as the ratio of the maximum glc observed with the hill present
to the maximum observed in the absence of the hill. The location of
the maxima are not considered in this evaluation.
271
-------
2. EXPERIMENTAL DETAILS
The model selected for this study is an idealization of Cinder
Cone Butte (CCB), an isolated three-dimensional hill about 100 meters
high. As described in the First Milestone Report (Lavery et al.,
1982), CCB has a double peak but is roughly axisymmetric. Typical
maximum slopes are about 25 degrees.
Profiles of the shape of CCB along radial lines for azimuths of
0, 10, 20, .... and 350 degrees were graphed. Several functional
forms were considered as approximations to the average silhouette of
CCB. The function that was found to best fit the hill shape is
h(r) = {(H+«)/(l+(r/L)4) - & 0 < r < R
0 r > R
where: r is the radius,
h is the local height,
H is the height at the center,
6 is a vertical offset for a finite model,
L is the radius for which h = H/2 if 6 =0,
and R is the radius of the modeled area.
The vertical offset was included in the formula to avoid a
discontinuity at the edge of the model. For CCB, H = 100 m,
L = 250 m, R = 500 m, and 4 = 6.25 m were chosen. This gives a
maximum slope angle of 24.4 degrees. Comparison of this function with
CCB silhouettes at intervals of 90 degrees in the azimuth are shown in
Figure B-l.
Another appealing reason to use this hill shape is that Hunt et
al. (1978) previously used the same functional form (however, with
H/L = 1 which gives a maximum slope angle of 45 degrees) in
experiments that established a large data base, which is available for
comparison. Also, the Fourier transform of this function should be
obtainable for use in mathematical models.
A scale model of the idealized CCB was vacuum molded of acrylic
plastic over a wood form. The height of the model was 15.5 cm which
corresponds to a scale of 1:640 based on the 100 m height assumed for
CCB. The radius of the model was 0.78 m. The vertical offset in the
shape function prevented a discontinuity in the surface at the edge of
the model. If the model had followed the formula exactly there would
have been a discontinuity in the slope at the model edge of 2.7
degrees. But the edge was faired during construction.
272
-------
IOC
?no 300
R, METERS
U'lO
Figure B-l. Silhouettes of Cinder Cone Butte and the profile selected
for this axisymmetric hill study.
273
-------
The model was covered with gravel of the same size used in
roughening the test section floor for boundary layer generation. The
same boundary layer generating devices used in the study described in
the Third Milestone Report (Lavery et al., 1983) were again used for
this study. The model was located with its center at a distance of
13.8 m downstream of the entrance to the test section.
Mean velocity and turbulence intensity measurements were made
with hot-film anemometers (TSI model 1241 probes and model 1053B
anemometers). A bent-over brass tube (o.d. =0.32 cm) was used to
emit a non-buoyant tracer gas (methane, 99% pure). The emission rate
was 2950 cm^/min which gave an exit velocity of about two times the
local speed. Samples were drawn through ports located on the hill
surface or through sampling rakes, positioned with the instrument
carriage, for locations upwind or downwind of the hill or in the
absence of the hill. Flame ionization detectors (Beckman model 400
Hydrocarbon Analyzers) were used to measure the sampled
concentration. Two-minute averages were obtained of values sampled at
rates of one sample per second for the flame ionization detectors and
500 samples per second for the anemometers with the laboratory's
minicomputer.
When necessary to consider actual concentrations rather than
terrain amplification factors or other ratios of concentrations, a
non-dimensional concentration x was calculated according to
X = CU(H)H2/Q
Where C = the measured concentration,
U(H) = the approach wind speed at z = H,
H = the hill height, and
Q = the emission rate of tracer.
For visualization, plumes were generated with cotton tipped glass
tubes of liquid titanium tetrachloride (marketed as "smoke sticks" by
E. Vernon Hill, Inc.). Photographs of these plumes were taken with a
Graphlex 4X5 camera fitted with a Polaroid back and type 55 P/N film.
Two 500 w photolamps were directed at the hill from elevated positions
downwind of the hill on either side of the test section. A
"time-averaged" photograph was obtained by making eight one-second
exposures with an interval of a few seconds between the exposures.
274
-------
3. MEASUREMENTS AND RESULTS
Vertical and lateral profiles of mean velocity and turbulence
intensity were obtained in the wind tunnel test section before
installing the model. The vertical profiles were measured at
distances of 10 and 14 m from the beginning of the floor roughness
(11.6 hill heights upstream and 14.2 hill heights downstream of the
position selected for the hill center). These profiles define the
approach wind field and demonstrate the uniformity of the wind tunnel
flow over the portion of the test section used for this study (Figures
B-2 and B-3). Lateral traverses were made at the 10 m distance at
heights of one-half and one times the hill height. Along these
traverses, the mean velocity and turbulence intensities varied by less
than 4%.
Fitting a power-law profile to the approach flow mean velocity
gives an exponent of 0.166, that is, u(z)« z°-166. A logarithmic
fit gives z0 = 0.2 mm or 0.0013 times the hill height and
u*/k =0.51 m/s.
As required for computation of A, ground-level concentrations
were measured for sources in the absence of the hill. Stack heights
of 0, 40, 80, 120, 160, and 200 mm were selected to be integer
multiples of 1/4 of the anticipated hill height of 160 mm; however,
the true height of the model came out to be 155 mm. These same stack
heights were used with the model in place and are referred to as stack
heights of 1/4, 1/2, 3/4, 1, and 1 1/4 hill heights. The distances
from the hill center to the source were also scaled by the hill
height.
The reference (no hill) data base consists of longitudinal
profiles of glc for the various stack heights (Figure B-4) normalized
(Berlyand, 1975) using the maximum glc (Xmax^ an<* tlie distance to
that maximum (xmax') as similarity parameters (Figure B-5). The
dependence of these parameters on the source height Hs was
determined to be
X aH1'15.
max s
The maximum concentrations from these profiles were used to calculate
the terrain amplification factors.
275
-------
7 --
6 --
A 10 M FROM STflRT OF ROUGHNESS
D l"i M
5 --
T"
S
6
3 -•
2 --
.25
.5 75
U IZ) /U (H)
A D
AD
A
1.2E
Figure B-2. Vertical profiles of the mean air speed in the test
section of the wind tunnel. H = Hill height.
276
-------
7 --
5 --
BE
so
O
A UVU, 10 M
A i.,1 Vu, i 4 y
D V'/V, iO K
• V'/v, 114 f
O uVU, 10 M
* W'/U, 114 M
A
Figure B-3. Vertical profiles of local turbulence intensities in the
test section of the wind tunnel.
277
-------
c
t\i
LJ
A
—T£i i r
2i
A
O
i
j.
A
o
Figure B-4. Longitudinal profiles of ground-level concentration from
point sources in the absence of the model.
278
-------
o
A
O
O
O
L1
O
a
A
A
A
A ,
A
A
c
A
Figure B-5. Longitudinal profiles of ground-level concentration from
point sources in the absence of the model plotted in
similarity coordinates.
279
-------
With the hill in place, glc's were measured on the hill via
surface ports on the hill and both upwind and downwind of the hill
with sampling rakes. Sources were located at distances of 4 to 16
hill heights upwind of the center of the model and at heights of 1/4
to 1 1/4 times the hill height. Complete longitudinal profiles of the
glc were obtained for selected source positions; but for most
positions, only sufficient measurements were made to find the maximum
glc.
The presence of the hill may be considered to influence the
transport and dispersion of the plume to increase the maximum glc in
three ways. For low sources, at moderate distances from the hill, the
reduction in mean wind speed allows the plume to reach the ground
surface closer to the source, producing higher concentrations than in
the absence of the hill. Plumes from higher sources can be thought of
as being intercepted by the hill. That is, the hill penetrates the
plume where the concentrations are greater than those that would occur
at ground level farther downstream over flat terrain. For yet higher
sources, the streamline convergence over the hilltop and the
corresponding downward flow in the lee of the hill brings the plume to
the ground more rapidly than over flat terrain. The maximum
concentration for these three regimes occurs upwind of the hilltop,
near or on the hilltop, or downwind of the hilltop, respectively. The
distance from the source to the maximum glc was always observed to be
less in the presence of the hill. No amplification factors less than
1.0 were observed in this study; that is, the hill's presence never
influenced the dispersion from these upwind sources to produce a
maximum glc that was less than would occur over flat terrain.
Hunt et al. (1979) presented a theoretical argument for the
presence of a "window" of upwind source locations that produce a
terrain amplification factor of 2 or greater for three-dimensional
terrain objects. They show that for sources within this window, the
maximum concentrations occur on or near the hill.
A series of longitudinal profiles of glc for all source heights
at a. distance of six hill heights upwind of the hill center exhibit
these types of influence (Figure B-6). The Hs = 0.25 H source has
its maximum glc occur near the upwind base of the hill; a low plume
being slowed by the presence of the hill. The maximum glc for the
Hs = 0.75 H stack occurs on the hill top; the hill intersecting the
plume. The Hs = 0.5 H stack produces a large region of nearly
constant concentration over the upwind face and crest of the hill; a
combination of the first two types of influence. Two nearly equal
maxima occur for the Hs = 1.0 H stack, one on the hill top and one
downwind of the hill. The maximum glc for the Hs = 1.25 H stack
clearly occurs downwind of the hill. Figure B-7 gives photographs of
the Hs = 0.5 H and Hs = 1.25 H plumes to illustrate their behavior.
All of the above concentration profiles exhibit a rapid reduction
in concentration at approximately 1 1/2 to 2 hill heights downwind of
the hill center. The flow separated from the hill surface at this
point and concentrations just downstream were determined by the
concentrations in the recirculating flow.
280
-------
A-
A
A.
A
A
X/L'
r
^ _ C"
Figure B-6. Longitudinal profiles of glc for sources located six hill
heights upwind of the hill center. Hill profile shown
below; note exaggerated scale.
281
-------
a) H /H = 0.50
b) HS/H = 1.25
Figure B-7. Visualization of plumes from sources at x/H = -5.
282
-------
Terrain amplification factors for all source positions are
presented in Figure B-8. Lines of constant A (= 1.4 and 2.0) are
shown as dashed lines. The solid lines divide the area into three
regions, where the source produces its maximum glc upwind of the
hilltop, on the hilltop (between the peak and the separation point),
and downwind of the hill.
Terrain amplification factors were computed for a hill of similar
functional form but much greater slope (45°) using data from Snyder
and Britter (1979) who report on a continuation of the work by Hunt
et al. (1978). That study used a boundary layer of approximately the
same depth but it was generated with a rougher surface (larger size of
gravel coating) and a smooth hill that was one and a half times the
height of the hill used in the present effort. The rougher surface
produced turbulence intensities a factor of two larger than in the
present study. Surface concentration profiles were obtained for two
source positions: release heights of one-half and one times the hill
height at a distance of 3.62 hill heights upwind of the hill center.
The terrain amplification factors for these source locations were
computed to be A = 4.00 and 1.27 for the stacks one-half and one times
the hill height, respectively. Values for the present study were
calculated from releases at a slightly greater upwind distance (4 hill
heights). They were 3.63 and 1.27 for the same stack heights,
respectively.
The earlier work with the scale model of Cinder Cone Butte (see
Appendix A, Lavery et al., 1983) also provides a point for
comparison. The source was located 10 hill heights upwind, if we
consider the hill height to be 100 m. However, it was not directly
upwind of the hill center but on a line 3° from the mean wind
direction. The source height was 0.42 H. The maximum concentration
occurred at a position toward the side of the hill about halfway up.
The terrain amplification factor was computed to be 1.08. This falls,
between the values for the present study of 1.0 and 1.55 for stacks at
the same upwind distance but heights of 0.25 H and 0.50 H,
respectively.
283
-------
\ •
l
284
-------
4. SUMMARY AND CONCLUSIONS
A wind-tunnel study was performed to investigate the influence of
an idealized three-dimensional hill on ground-level concentrations
from upwind sources in a neutral atmospheric boundary layer and to
locate the source locations where this influence is the greatest.
The presence of the hill was found to influence the transport and
dispersion of the plume to increase the maximum glc in three ways.
For low sources, at moderate distances from the hill, the reduction in
mean wind speed allows the plume to reach the ground surface closer to
the source, producing higher concentrations than in the absence of the
hill. Plumes from higher sources can be thought of as being
intercepted by the hill. That is, the hill penetrates the plume where
the concentrations are greater than those that would occur at ground
level farther downstream over flat terrain. For yet higher sources,
the streamline convergence over the hilltop and the corresponding
downward flow in the lee of the hill brings the plume to the ground
more rapidly than over flat terrain. The maximum concentration for
these three regimes occurs upwind of the hilltop, near or on the
hilltop, or downwind of the hilltop, respectively. Terrain
amplification factors ranged from near 1.0 to 3.63. The region of
source locations that produced an amplification factor of 1.4 or more
extended to an upwind distance of 14 hill heights.
285
-------
REFERENCES
Berlyand, M.G., 1975: Contemporary Problems of Atmospheric Diffusion
and Air Pollution, Hydromet. Press, Leningrad, USSR, 448p.
Hunt, J.C.R., Puttock, J.S. and Snyder, W.H., 1979: Turbulent Diffusion
from a Point Source in Stratified and Neutral Flows Around a
Three Dimensional Hill: Part I: Diffusion Equation Analysis,
Atmos. Envir., v. 13, p. 1227-39.
Hunt, J.C.R., Snyder, W.H, and Lawson, R.E., 1978: Flow Structure and
Turbulent Diffusion around a Three-Dimensional Hill: Fluid
Modeling Study on Effects of Stratification: Part I: Flow
Structure, Rpt. No. EPA 600/4-78-041, Envir. Prot. Agcy., Res.
Tri. Pk., NC, 96p.
Lavery, T.F., Bass, A., Strimaitis, D.G., Venkatram, A., Greene, B.R.,
Drivas, P.J. and Egan, B.A., 1982: EPA Complex Terrain Model
Development: First Milestone Report - 1981, Rpt. No. EPA
600/3-82-036, Envir. Prot. Agcy., Res. Tri. Pk., NC, 304p.
Lavery, T.F., Strimaitis, D.G., Venkatram, A., Greene, B.R.,
DiCristofaro, D.C. and Egan, B.A., 1983: EPA Complex Terrain
Model Development: Third Milestone Report - 1983, Rpt. No. EPA
600/3-83-101, Envir. Prot. Agcy., Res. Tri. Pk., NC, 271p.
Snyder, W.H. and Britter, R.E., 1979: Aspect Ratio Study: Unpublished
In-house Data Report, Envir. Prot. Agcy., Res, Tri. Pk., NC.
Strimaitis, D.G., Venkatram, A., Greene, B.R., Hanna, S.R.,
Heisler, S., Lavery, T.F., Bass, A. and Egan, B.A., 1983: EPA
Complex Terrain Model Development: Second Milestone Report -
1982, EPA 600/3-85-015. Envir. Prot. Agcy., Res. Tri. Pk., NC
375p.
286
-------
APPENDIX C
ALPHA-1 OBSERVATIONS OF THE
TRACY OIL-FOG PLUME*
*This appendix is a condensation of a report by Uthe and Morley (1984)
287
-------
C.I Introduction
During the November experiment the ALPHA-1 airborne lidar mapped
aerosol plume behavior in the vicinity of the Tracy Generating
Station. The ALPHA-1 was based at the Reno airport and was operated
by SRI International under the sponsorship of the Electric Power
Research Institute.
An oil-fog tracer aerosol was generated and emitted froit. the
Tracy Station with a plume sufficiently dense for photographic
purposes. The ALPHA-1 mapped the plume behavior by making cross-plume
traverses above the plume at various downwind distances from the Tracy
Station. Typically, the plume traverses were made at downwind
increments of 0.5 nmi from the Tracy Station to 6 nmi downwind. The
aircraft typically operated at an altitude of 10,000 ft MSL, although
lower altitudes were flown during cloudy conditions. The mountainous
terrain and nighttime conditions prevented data collection during
times when cloud bases were below 7500 ft. The experimental data
collection typically began at about midnight and extended to
0900 PST. The ALPHA-1 normally made two flights, each between three
and four hours in duration.
The ALPHA-1 operation consisted of transmitting 1.06 ym
wavelength laser pulses at a repetition rate of 5/second (horizontal
resolution of about 12 m depending on aircraft ground speed). See
Uthe et al. (1980a and 1980b) for a description of the ALPHA-1
system. Received energy was detected, logarithmically amplified, and
digitized at 20 ns (3 meter) intervals. Lidar signatures were stored
on nine-track 1600 bpi magnetic tape and were also used to generate
real-time facsimile pictorial displays of plume, atmosphere and
terrain structure. Aircraft location was determined both by a LORAN-C
and RNAV system. The RNAV was programmed to provide distance and
bearing to the Tracy Station. Aircraft location and time information
were recorded on magnetic tape for each laser firing. The RNAV data
also were plotted on the facsimile recorder to provide the information
needed to direct the lidar data collection. The ALPHA-1 logarithmic
amplifier saturates for signals greater than about four orders of
magnitude (40 dB) . The ground returns are normally saturated,,
Because of the dense aerosol plume, plume returns near the source were
near saturation of the receiver amplifiers. Therefore, near-surface
plume returns were difficult to separate from ground returns. To
solve this problem, the transmitted energy was decreased by a factor
of 2 (3 dB), providing more dynamic range between plume and ground
returns. However, the reduced transmitted energy also decreased
sensitivity to low-density atmospheric aerosol features. A log of
data collection times is presented in Table C-l.
The ALPHA-1 data records are stored on nine-track 1600 bpi
magnetic tape in binary forms. The data consists of 1000 8-bit lidar
backscatter values for each laser firing. Also recorded in each block
of data are the time and location. The magnetic tape records are
maintained at SRI for future analysis.
288
-------
TABLE C-l
LOG OF ALPHA-1 DATA COLLECTION
TRACY GENERATING STATION, NOVEMBER 1983
Date
(Nov 83)
5
6
7
8
9
12
12
14
14
15
15
15
16
17/18
18
19
19
Time
(PST)
— —
1125-1140
0155-0420
0015-0325
0240-0530
0035-0340
0500-0720
0040-0335
0435-0800
0025-0340
0435-0735
2135-2330
0235-0240
2225-0130
0320-0345
0025-0130
Flight Time
(hr)
1.2
1.0
3.5
3.7
3.4
3.6
2.8
3.4
3.8
3.7
3.4
2.5
0.5
3.5
0.8
1.5
1.8
Remarks
ferry
test flight
2nd flight
2nd flight
2nd flight
clouds and turbulence
2nd flight/clouds
rain and snow
ferry
Total
44.1
289
-------
The ALPHA-1 magnetic tapes were used to generate facsimile
displays that depict plume, atmosphere, and terrain structure. The
displays also provide time and aircraft location. Tables presenting
information needed to interpret the lidar pictures are given in
Section C.2 and the lidar pictures are reproduced in Section C.3. The
reproduced pictorial records are reduced in size and the originals
provide a better indication of plume behavior. These original
facsimile records are available on a loan basis from SRI.
C.2 Date, Time, Altitude, and Location Data
This section provides a listing (Table C-2) of date and time
information needed to interpret the ALPHA-1 facsimile data plots
presented in Section C.3.
The listing provides the page number of Appendix B of the
original report by Uthe and Morley (1984) that presents facsimile
plots for each of the 87 data tapes collected during the Tracy
program. Tape number is identified on the lower left hand side of the
facsimile plots. Picture number refers to the data presented within
vertical solid lines and each picture presents an individual ALPHA-1
cross section. The ALPHA-1 was turned off during times between the
individual cross sections. Time (PST) is given for the first time
mark of each picture. Time marks are identified in the key to reading
the lidar cross sections presented in Figure C-l. In Section C.3 only
the facsimile plots representative of the three case studies described
in Section 6 of this Milestone Report are given.
Aircraft location is determined by an area navigation (RNAV)
system that provides information on bearing and distance to a selected
waypoint position. The waypoint was established at the Tracy Station,
and distance and bearing to the station are plotted on the facsimile
data displays. Figure C-2 presents a map of the Tracy area with
ALPHA-1 coordinates of distance (nmi) and bearing to the Tracy Station
superimposed. LORAN-C data on longitude and latitude were also
recorded but not plotted on the displays. Figure C-l is an example of
ALPHA-1 backscatter signatures processed as a facsimile record for
Tape 45 (page 8-23). Terrain, plume, and atmospheric features are
identified. Location of time marks, altitude scale, and aircraft
location plots are given as a key for interpreting the cross sections
presented in Section C.3. The altitude scale is 600 m/division for
all data.
290
-------
TABLE C-2
DATE AND TIME DATA FOR ALPHA-1 DATA COLLECTION
AT THE TRACY GENERATING STATION, NOVEMBER 1983
Gray Scale Depiction
Gray Scale Depiction
Page * Tape
B-2
1
B-2 2
B-3 3
B-3 4
B-4 5
Picture
1
1
2
3
4
5
6
-7
8
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
1
2
3
4
5
6
7
L/clUC/
Time (PST)
6 Nov 1983(1)
1125
7 Nov 1983(2)
0156
0159
0202
0204
0207
0212
0216
0219
0223
0225
0227
0230
0233
0236
0242
0245
0253
0256
0259
0303
0307
0310
0314
0317
0321
0325
0329
0332
0338
0342
Page
B-4
B-5
B-5
B-6
B-6
Tape Picture
6 1
2
3
4
5
6
7
8
7 1
2
3
4
5
8 1
2
3
4
5
6
7
8
9
9 1
2
3
4
5
6
7
10 1
2
3
Time (PST)
0346
0349
0354
0356
0359
0401
0404
0406
0409
0411
0414
0416
0419
8 Nov 1983
0014
0017
0020
0023
0026
0029
0031
0034
0037
0042
00^5
0048
0054
0059
0102
0105
0109
0113
0116
(1) Location from Reno Vortac (test flight).
(2) All following locations from Tracy Power Plant.
* Page number of the original report by Uthe and Morley (1984)
291
-------
TABLE C-2 (Continued)
Gray Scale Depiction
Picture
B-6 10
Time (PST)
B-7 11
B-7 12
B-8 13
B-8 14
B-9 15
B-9 16
4
5
6
7
1
2
3
4
5
1
2
3
4
5
6
1
2
3
4
5
6
7
1
2
3
4
5
6
7
8 Nov 1983
0119
0123
0128
0131
0138
0141
0144
0149
0154
0200
0203
0206
0216
0219
0222
0227
0331
0235
0243
0246
0251
0253
0257
0300
0303
0306
0310
0312
0316
1
2
3
4
5
6
0326
9 Nov 1983
0244
0246
0250
0252
0255
0258
Page Tape
B-9 16
B-9 17
B-10 18
B-10 19
B-ll 20
B-ll 21
Picture
7
8
9
10
1
2
3
4
5
6
7
8
1
2
3
4
5
6
1
8
1
2
3
4
5
6
7
8
1
2
3
4
5
6
1
2
3
4
uaue/
Time (PST)
9 Nov 1983
0302
0304
0307
0309
0312
0314
0317
0319
0322
0325
0328
0330
0336
0342
0345
0348
0352
0354
0358
0401
0406
0409
0412
0414
0417
0420
0422
0425
0434
0438
0442
0446
0449
0453
0459
0506
0509
0513
292
-------
TABLE C-2 (Continued)
Gray Scale Depiction
Page Tape
B-ll 21
B-12 22
B-12 23
B-13 24
8-13 25
B-14 26
B-14 27
B-15 28
Picture
5
1
2
1
2
3
4
C
_>
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
1
2
3
4
uate/
Time (PST)
9 Nov 1983
0516
0522
0528
12 Nov 1983
0035
0038
0042
0046
0049
0057
0102
0106
0110
0113
0116
0120
0124
0128
0131
0135
0138
0143
0146
0150
0153
0157
0203
0207
0213
0216
0220
0223
0228
0231
0235
0242
: :
Page Tape Picture
Brl5 28 5
B-15 29 1
2
3
4
5
B-16 30 1
2
3
4
5
B-16 31 1
2
3
4
B-17 32 1
2
3
4
5
B-17 33 1
2
3
4
5
6
B-18 34 1
2
3
4
5
B-18 35 1
2
3
4
5
Date/
Time (PST)
12 Nov 1983
0247
0253
0258
0302
0307
0311
0317
0322
0329
0334
0338
0503
0506
0512
0519
0525
0532
0536
0539
0543
0549
0554
0558
0602
0605
0609
0613
0617
0621
0625
0630
0636
0641
0644
0647
0650
293
-------
TABLE C-2 (Continued)
Gray Scale Depiction
Page Tape
B-18 35
B-19 36
B-19 37
B-20 38
B-20 39
B-21 40
B-21 41
B-22 42
Picture
6
1
2
3
4
1
2
3
4
5
6
7
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
1
2
3
Date/
Time (PST)
12 Nov 1983
0655
Gray Scale Depiction
Picture
1
2
3
4
1
2
3
4
5
6
7
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
1
2
3
0702
0709
0712
0718
14 Nov 1983
0043
0046
0049
0052
0055
0100
0105
0110
0114
0119
0125
0130
0134
0139
0142
0147
0151
0200
0204
0208
0213
0216
0222
0226
0231
0234
0240
0246
0250
0252
B-22 42
B-22 43
B-23 44
B-23 45
B-24 46
B-24 47
4
5
6
Date/
Time (PST)
Nov 1983
0256
0259
0303
B-25 48
I
2
3
4
5
6
7
8
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
1
2
3
4
5
6
1
2
3
0313
0317
0320
0323
0326
0330
0334
0336
0439
0441
0444
0467
0451
0454
0459
0502
0506
0509
0513
0516
0521
0524
0528
0539
0547
0551
0555
0558
0602
0605
0608
0624
0626
0629
294
-------
TABLE C-2 (Continued)
Gray Scale Depiction
Gray Scale Depiction
Page Tape
B-25 48
B-25 49
B-26 50
B-26 51
B-27 52
B-27 53
Picture
4
5
6
7
8
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
8
9
1
2
3
4
5
6
7
1
2
1
2
3
4
5
Date/
Time (PST)
14 Nov 1983
0632
0634
0637
0640
0642
0647
0650
0652
0655
0658
0701
0704
0707
0712
0715
0717
0720
0722
0724
0726
0728
0730
0734
0738
0741
0743
0746
0750
0753
0757
0800
15 Nov 1983
0026
0028
0031
0033
0036
Page Tape Picture
B-27 53 6
7
8
9
B-28 54 1
2
3
4
5
6
7
B-28 55 1
2
3
4
5
6
7
B-29 56 1
2
3
4
5
6
7
B-29 57 1
2
3
4
5
6
B-30 58 1
2
3
4
5
6
Date/
Time (PST)
15 Nov 1983
0039
0042
0044
0047
0051
0055
0057
0100
0105
0108
0111
0115
0119
0122
0124
0127
0130
0136
0139
0143
0146
0149
0152
0155
0158
0202
0205
0209
0212
0215
0219
0226
0229
0233
0237
0241
0245
295
-------
TABLE C-2 (Continued)
Gray
Page
B-30
B-31
B-31
B-32
B-32
B-33
B-33
Scale Depiction
Tape Picture
59 1
2
3
4
5
6
60 1
2
3
4
5
6
7
8
61 1
62 1
2
3
4
5
6
63 1
2
3
4
64 1
2
3
4
5
65 1
2
3
4
5
6
Date/
Time (PST)
15 Nov 1983
0249
0253
0256
0301
0305
0309
0313
0317
0320
0323
0326
0328
0332
0334
0339
0437
0440
0444
0448
0452
0455
0459
0505
0510
0515
0521
0525
0530
0534
0538
0542
0546
0552
0557
0600
0603
Gray Scale Depiction
Page Tape Picture
B-34 66 1
2
3
4
5
6
7
B-34 67 1
2
3
4
5
6
7
B-35 68 1
2
3
! 4
5
6
7
B-35 69 1
:f 2
3
4
5
B-36 70 1
i 2
3
4
5
6
7
8
B-36 71 1
2
3
Date/
Time (PST)
15 Nov 1983
0606
0609
0612
0615
0618
0621
0624
0633
0636
0638
0641
0644
0649
0651
0656
0701
0704
0707
0710
0713
0716
0719
0722
0725
0729
0732
2137
2139
2142
2146
2151
2154
2157
2201
2206
2209
2213
296
-------
TABLE C-2 (Continued)
Gray
Page
B-36
B-37
B-37
B-38
B-38
B-38
B-39
Scale Depiction
Tape Picture
71 4
5
6
72 1
2
3
4
5
6
7
73 1
2
3
4
5
6
7
74 1
2
75 1
2
76 1
2
3
4
5
6
7
77 1
2
3
4
5
6
7
Date/
Time (PST)
15 Nov 1983
2219
2222
2229
2233
2238
2240
2243
2245
2249
2252
2259
2302
2306
2309
2313
2317
2320
2325
2328
16 Nov 1983
0235
0237
17 Nov 1983
2228
2232
2236
2239
2243
2247
2250
2254
2257
2300
2304
2307
2311
2315
Gray Scale Depiction
Page Tape Picture
Date/
Time (PST)
B-39
78
B-40
79
B-41 81
B-41 82
1
2
3
4
5
1
2
3
4
17
Nov 1983
2319
2324
2331
2335
2340
2345
2349
2353
2357
B-40 .79
B-40 80
5
6
1
2
3
4
5
6
7
8
9
18 Nov 1983
0001
0004
0008
0011
0013
0016
0018
0024
0026
0029
0031
1
2
3
4
5
6
7
8
9
1
2
3
4
5
6
7
8
0034
0038
0040
0043
0046
0049
0054
0057
0059
0103
0105
0108
0110
0112
0115
0117
0123
297
-------
TABLE C-2 (Continued)
Gray Scale Depiction n t / Gray Scale Depiction n t- /
Page Tape Picture Time (PST) Page Tape Picture Time (PST)
18 Nov 1983
B-41 82 9 0126*
B-42 83 1 0131*
B-42 84 1 0322*
2 0327*
3 0330*
4 0336*
5 0339*
B-42 85
B-43 86
B-43 87
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
8
1
2
3
4
5
19 Nov 1983
0023
0025
0028
0030
0033
0036
0038
0042
0054
0059
0101
0105
0107
0110
0112
0114
0118
0122
0125*
—
0130
*
Cross section along axis of the plume.
298
-------
O
o
O
(H
O
(X
o
•p
X
I
u
00
299
-------
§
4->
I/I
X
O
•P
nj
•H
O
O
O
D.
rt
rt
-------
C.3 ALPHA-1 Facsimile Records for the Case Study Experiments
Discussed in Section 5
301
-------
302
-------
303
-------
304
-------
305
-------
306
-------
307
-------
308
-------
309
-------
310
-------
311
-------
Ill
312
-------
313
-------
314
-------
315
-------
316
-------
317
-------
318
-------
REFERENCES FOR APPENDIX C
Uthe, E.E., N.B. Nielsen, and W.L. Jimison 1980a: "Airborne Lidar
Plume and Haze Analyzer (ALPHA 1)," Bu 11. Amer. Meteoro1. Soc.,
61, 1035-1043.
Uthe, E.E., W.L. Jimison, and N.B. Nielsen 1980b: "Development of an
Airborne Lidar for Characterizing Particle Distribution in the
Atmosphere," Final Report, EPRI No. EA-1538, SRI International,
Menlo Park, CA.
Uthe, E.E. and B.M. Morley 1984. Alpha-1 Observations of Plume
Behavior for PMV&D Tracy Site, prepared for Electric Power
Research Institute, Palo Alto, California.
U S GOVERNMENT PRINTING OFFICE-559-013/20008
319
-------
|