xEPA
Unrted States
Environmental Protection
Agancy
Environmental Sciences Research
Laboratory
Research Triangle Par* NC 27711
EPA-600/9-81-020
April 1981
Research and Development
On-Site
Meteorological
Instrumentation
Requirements to
Characterize &
Diffusion from
Point Sources *
Workshop Report
o?
^
> . >
-------
; RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the MISCELLANEOUS REPORTS series. This
series is reserved for reports whose content does not fit into one of the other specific
series. Conference proceedings, annual reports, and bibliographies are examples
of miscellaneous reports.
-------
EPA-600/9-81-020
April 1981
ON-SITE METEOROLOGICAL INSTRUMENTATION REQUIREMENTS
TO CHARACTERIZE DIFFUSION FROM POINT SOURCES
Workshop Report
By
David Strimaitis
Gale Hoffnagle
Arthur Bass
ENVIRONMENTAL RESEARCH & TECHNOLOGY, INC.
696 Virginia Road, Concord, Massachusetts 01742
Contract No. 68-02-3282
Project Officers
D. Bruce Turner
John S. Irwin
William B. Petersen
Meteorology and Assessment Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the U.S. Environmental
Protection Agency, and approved for publication. Approval does not
signify that the contents necessarily reflect the views and policies
of the U.S. Environmental Protection Agency, nor does mention of trade
names or commerical products constitute endorsement or recommendations
for use.
11
-------
ABSTRACT
Results of a workshop entitled "On-Site Meteorological
Instrumentation Requirements to Characterize Diffusion from Point
Sources" are sunmarized and reported. The workshop was sponsored by
the U.S. Environmental Protection Agency in Raleigh, North Carolina,
on January 15-17, 1980. Its purpose was to provide EPA with a
thorough examination of the meteorological instrumentation and data
collection requirements needed to characterize airborne dispersion of
air contaminants from point sources and to recommend, based on an
expert consensus, specific measurement techniques and accuracies.
Secondary purposes of the workshop were to (1) make recommendations to
the National Weather Service (NWS) about collecting and archiving
meteorological data that would best support air quality dispersion
modeling objectives and (2) make recommendations on standardization of
meteorological data reporting and quality assurance programs.
This report was submitted in fulfillment of Contract
No. 68-02-3282 by Environmental Research & Technology, Inc. (ERT)
under the sponsorship of the U.S. Environmental Protection Agency.
This report covers the period 26 September to 30 August 1980 and work
was completed as of 1 September 1980.
iii
-------
CONTENTS
ABSTRACT
TABLES vi
SYMBOLS AND ABBREVIATIONS vii
ACKNOWLEDGEMENTS x
1 . INTRODUCTION 1
2. ESSENTIAL (PRIORITY 1) MEASUREMENTS 3
3. DESIRABLE (PRIORITY 2) MEASUREMENTS 13
4. HELPFUL (PRIORITY 3) MEASUREMENTS 16
5. RECOMMENDATIONS FOR THE OFFICE OF THE FEDERAL
COORDINATOR FOR METEOROLOGY 18
6. ARCHIVING AND RETRIEVAL OF DATA 22
7. WORKSHOP PARTICIPANTS 23
REFERENCES 28
APPENDICES
A DATA REQUIREMENTS
B MEASUREMENT TECHNIQUES - DERIVED VARIABLES
C MEASUREMENT TECHNIQUES - METEOROLOGICAL VARIABLES
-------
TABLES
Number Page
2-1 Measurements Required for Priority 1 Variables 10
2-2 Interdependence of Priority 1 Variables 12
3-1 Dependence of Priority 2 Variables on Measurements 15
4-1 Dependence of Priority 3 Variables on Measurements 17
5-1 Recommended Low-Level Measurements and Derived Quantities 19
5-2 Desirable Low-Level Measurements 20
5-3 Recommended High-Level Measurements from Remote Sensors 21
VI
-------
SYMBOLS AND ABBREVIATIONS
c Pollutant concentration
c Specific heat of air
de/dz Wind direction shear with height
e Water vapor pressure
g Acceleration due to gravity
h Surface layer depth
h Convective mixing depth
h Mechanical mixing depth
h0 Average height of roughness elements
k von-Karman constant
n Frequency in cycles per unit time
u, v, w Orthogonal wind speed components
u; uv, us Mean wind speed; vector, scalar
u^ Friction velocity
w^ Convective velocity scale
ZQ Surface roughness length scale
z Effective surface roughness length scale
CQ Drag coefficient
2
C Temperature structure parameter
2
C Velocity structure parameter
D Thermal diffusivity
Fr Froude number
H Surface heat flux
r
Km>KH Turbulent momentum and heat exchange coefficients
L Monin-Obukhov length scale f
vn
-------
SYMBOLS AND ABBREVIATIONS
Q_ Maximum surface temperature flux
in
Q Surface temperature flux
Q Radiant flux density
R Climatological representativeness
R Primitive representativeness
R Spatial representativeness
Rig Bulk Richardson number
Rif Flux Richardson number
Ri. Gradient Richardson number
T Temperature
T Dew point temperature
T^ Surface layer temperature scale
Xg Temperature dissipation length
AT/AZ Temperature gradient over a layer
e Eddy dissipation rate
Q Wind direction
0 Potential Temperature
X Area density of roughness elements
P Density of air
O Standard deviation of temperature
y, z Horizontal and vertical plume spread parameters
ae, v, u Standard deviation of azimuthal wind direction/wind
,;>, w Standard deviation of elevation angle/vertical speed
speed components
Standard de
components
viii
-------
SYMBOLS AND ABBREVIATIONS
o(l/u) Standard deviation of the reciprocal of wind speed
fluctuations
T Reynolds shear stress
SBL Surface boundary layer
PEL Planetary boundary layer
ix
-------
ACKNOWLEDGEMENTS
This workshop was sponsored by the U.S. Environmental Protection
Agency under Contract 68-02-3282, administered by D. Bruce Turner,
John Irwin, and William Petersen of the EPA Division of Meteorology.
Many individuals contributed to the content of this report during
the course of the workshop. The authors are especially indebted to
Dr. Todd Crawford, Maynard Smith, and Thomas Lockhart, the three panel
chairmen, for their efforts in collecting, organizing, and integrating
the conclusions and recommendations of their panels. Their panel
reports form the basis of this document.
We give special thanks to Dr. Frank Pasquill, a consultant to
ERT, for his active participation in the workshop and his assistance
in reviewing and distilling the workshop findings.
-------
1. INTRODUCTION
A workshop entitled "On-Site Meteorological Instrumentation
Requirements to Characterize Diffusion from Point Sources" was held in
Raleigh, North Carolina, on January 15-17, 1980. The workshop was
sponsored by the U.S. Environmental Protection Agency (EPA) and
arranged by Environmental Research & Technology, Inc. The purpose of
the workshop was to provide EPA with a thorough examination of the
meteorological instrumentation and data collection requirements needed
to characterize airborne dispersion of air contaminants from point
sources, and to recommend, based on an expert consensus, specific
measurement techniques and accuracies. Secondary purposes of the
Workshop were to (1) make recommendations to the National Weather
Service (NWS) about collecting and archiving meteorological data that
would best support air quality dispersion modeling objectives and
(2) make recommendations on standardization of meteorological data
reporting and quality assurance programs.
A representative group of 31 scientists drawn from regulatory
agencies, industries, universities, research laboratories, and
consulting firms participated in the workshop. The workshop was
conducted in three panels for the majority of its work, with plenary
sessions for coordination.
• Panel A, chaired by Maynard Smith, evaluated Data
Requirements, including model input variables, data
representativeness, accuracy requirements, averaging and
sampling periods, data archiving, and recommendations to the
NWS.
• Panel B, chaired by Todd Crawford, considered the
computations and data required for Derived Variables used to
describe turbulence and diffusion in the planetary boundary
layer.
• Panel C, chaired by Tom Lockhart, reviewed measurement
techniques for the requisite Meteorological Variables.
-------
Summary reports from each of these panels have been integrated as the
basis of the workshop recommendations. The individual panel reports
by the panel chairmen are included as working papers in the
appendices. As such, they have undergone minimum editing.
Meteorological measurements discussed by the workshop are used in
point source diffusion modeling and in analysis of air quality
monitoring data (e.g., as discussed in the Modeling Guideline [EPA
1978a] and the Monitoring Guideline [EPA 1978b]). The workshop
recommendations are intended to help the modeler choose the
instrumentation and analysis techniques required to measure accurately
the variables needed for his model rather than to help him choose the
"best" model. Most variables previously found useful for dispersion
modeling were considered, and reasonably attainable accuracies were
estimated.
The meteorological variables were divided into tnree priority
classes: (1) essential, (2) desirable, and (3) helpful. These
classes reflect a judgment about desirable progress in
state-of-the-art modeling and should not limit the modeler in
selecting the variables for specific applications.
The workshop results de-emphasize relying on the
Pasquill-Gifford-Turner (PGT) stability classification as a primary
means of characterizing atmospheric diffusion; rather they emphasize
relying upon on-site measurements of turbulence
-------
2. ESSENTIAL (PRIORITY 1) MEASUREMENTS
The quantities that have been identified as essential to further
progress in atmospheric dispersion modeling are:
• mean horizontal wind speed (u) and mean wind direction (6)
at 10 meters (m),
• mean horizontal wind speed and mean wind direction at source
height,
• harmonic mean wind speed (i/u) between source height and
effective plume height,
• standard deviations of horizontal and vertical wind
fluctuations (e.g.,0^,0, ) at 10 m,
• temperature difference (AT) between 2 and 10 m in
unobstructed surface layers,
• temperature (T) at 2 m,
• heights of the convectively mixed layer (h ) and the
mechanically mixed layer (h ),
• average surface roughness length (ZQ),
• friction velocity (u^),
a bulk Richardson number (Rir.)
D' '
• Monin-Obukhov length scale (L),
• turbulent heat flux (HU),
o convective velocity scale (w^), and
• the water vapor content (ew) between source height and
effective plume height (if the water content of source
emissions is important).
Some of these quantities are primary—direct averages or
statistical evaluations of direct measurements—while others are
derived from combinations of the primary quantities.
Wind Measurements
For horizontal wind speed, the scalar mean (u_) and the
s
resultant vector mean (uv) should be reported. (,The harmonic mean
-------
is, however, preferred for calculating initial dilution and
plume rise. These variables are defined by
Mean wind direction should be reported as the hourly resultant vector
direction (6 « arc tan v/u). Satisfactory measurements of wind speed
and wind direction can be made with cup anemometers and wind vanes,
with wind vanes and front-mounted propeller anemometers, or with two
or three propellers mounted on orthogonal axes (three propellers to
include vertical wind velocity).
The standard deviations of wind fluctuations (a ,a , or a
cv, a ) should be calculated directly, not by a range estimator.
For sufficient accuracy (^5%), the computation should be based on
digital or mathematically equivalent analog techniques.
At monitor sites likely to experience intermittent pollutant
concentrations, wind statistics based on a 3-minute sampling duration
(or shorter) for a, should also be gathered. These aQ values can
o o
then be combined to match pollutant release intervals in order to
model time-integrated concentration fields.
Cup or propeller anemometers must be accurate to within 0.2 m/s
+5% of the wind speed, with a start speed of <0.5 m/s, and a distance
constant (63% recovery) of 515 m. Wind vanes must have a resolution of
1° and an accuracy of 5°. Delay distance (50% recovery) must be
<5 m with a damping ratio of >0.4 .
Sixty or more samples will estimate hourly means to within
5-10%. Sample averaging time should be 1-5 seconds, with a response
time of 1 second. Three hundred and sixty or more samples will
estimate the hourly standard deviation within 5/-10X. The need for 360
-------
samples for calculating hourly standard deviations was questioned
during post-workshop discussions. The response from Panel C members
stated that 360 samples is a conservative number based on substantial
experience. In some instances fewer samples would be needed, but
there is no general guide for identifying those situations prior to
sampling.
Applied techniques for obtaining winds aloft include tracking
superpressure balloons, and using Doppler sodar systems.
Ascending free balloons provide data on flow trajectories over a
finite time interval. Since the balloon does not exactly follow an
air parcel when vertical motion is involved, the results must be
interpreted cautiously.
The Doppler sodar technique may be used in either the monostatic
or bistatic mode. Measurements of wind components aloft are accurate
to about 0.5 m/s. These measurements routinely provide information to
heights of 500 m, and sometimes to heights above 1,000 m.
Temperature Measurements
Ambient air temperature should be measured at 2 m , consistent
with the World Meteorological Organization (WHO) standards for ambient
temperature measurements. It is recognized that some derived
parameters are based on temperatures at 10 m, but this value may be
calculated from the 2 m temperature and the 2-10 m temperature
difference. The ambient temperature measurement should be accurate to
0.5°C, with a resolution of 0.1°C.
Temperature differences should be measured between 2 and 10 m in
an unobstructed surface layer. Measurements should be made with an
independent probe system that makes no reference to other measurements
of ambient temperature or to the standard deviation of temperature
fluctuations (OT). properly shielded and aspirated, a temperature
probe system can attain 0.1°C accuracy, and 0.02°C resolution
within the present state-of-the-art. Standards of probe selection,
calibration, and maintenance for research efforts can yield 0.05°C
accuracy, and 0.01°C resolution. The time constant (63%) of the
-------
temperature probe should be 1 minute, and the sample averaging time
should be greater than or equal to half the time constant.
In siting the probes, care must be taken to maintain the
integrity of the ambient environment, particularly the surface, in the
itanediate vicinity of the tower. This is especially important for
temperature probes 2 m above the ground. Site preparation usually
requires substantial modification of the surface at a monitoring
site. Panel C did not make specific recommendations about site
preparation techniques to insure representative measurements.
It is recognized that 0.1°C accuracy is insufficient for
distinguishing atmospheric stability in some applications. Greater
accuracy is needed to enhance the usefulness of temperature difference
measurements.
Mixed Layer Heights
Mixed layers may be maintained convectively by surface heating or
maintained mechanically by wind-generated turbulent mixing. A mixed
layer is classified as either convective or mechanical as one or the
other mechanism is dominant. When both mechanisms are active, the
buoyancy driven mixed layer usually dominates.
The convective layer height (hfi) should be measured routinely
by a monostatic sodar with vertical beam. Objective rules have been
developed to assist the operators in evaluating hc, and automatic
and pattern recognition methods are under development. Sodar
resolution is about 10 m, and the useful range varies from 50 to
1,000 m. The sampling time constant is typically 10 seconds.
Convective mixed layer heights derived from sodar and from rawinsonde
profiles generally agree to within 50 m.
When the mixed layer height exceeds the range of the sodar
system, other methods are available. Aircraft equipped with a fast
response temperature sensor readily provide temperature profiles to
greater heights. Tethersondes may be used in moderate to light wind
conditions where they are not an aviation hazard. A slow-rise
radiosonde balloon may be used when tracked by double theodolite
observations. These tracked balloons also provide data on winds aloft,
-------
Bistatic sodar can be used to derive the mechanically mixed layer
height, h , in the stable boundary layer. But the method is of
limited value because the characteristic value of h lies between
0 and 200 m, while the minimum range of the technique is 50 m. The
mechanically mixed layer height is best measured with temperature or
velocity turbulence data from an instrumented tower, supported by
sodar.
Roughness Length
The roughness length (ZQ) at a site may be derived from wind
speed profile data under neutral atmospheric conditions or estimated
from a description of surrounding surface features. Once estimated,
the roughness length may be assumed to be fixed. Consequently, only a
limited record of wind profile data is needed.
If ZQ is to be derived from wind profile data over relatively
flat, homogeneous terrain, wind measurements should be made at three
heights between 2-20 m, most desirably with 5% wind speed accuracy,
10-second response time, 30-second sample averaging time, and 1-hour
sampling duration. If, however, the surface is patchy with large and
varied roughness elements so that it is difficult to determine the
proper instrument heights, ZQ can be estimated indirectly and ZQ
values should depend upon wind direction. The estimate should take
into account the size and distribution of roughness elements. Several
methods for estimating ZQ were referenced by Panel B, but none has
been shown to be completely satisfactory. Fortunately, dispersion
modeling is not very sensitive to ZQ. jt is adequate to estimate
20 to within a factor of 2.
Bulk Richardson Number, Friction Velocity, Monin-Obukhov Length,
Heat Flux, and Convective Velocity
The derived quantities Rig, u-f| L, HF, w* can be
estimated from temperature and wind measurements.
-------
The bulk Richardson number, RiB> is derived from measurements
at 2 m and 10 m:
„. . g • (Z10) (AT + F )
TIO ir-r*
\uio)
where g is the acceleration due to gravity, T,Q (°K) is the
ambient temperature at 10 m, and T is the dry adiabatic lapse rate.
This value of Rig ia representative of the surface boundary layer
only.
The Monin-Obukhov length scale, L, may be evaluated indirectly
from the value obtained for Rig. Assuming the validity of surface
layer similarity theory relationships, L can be obtained by iterative
approximation to:
z
L
where F and G represent integral forms of the empirical flux-profile
relations (Benoit 1977, Irwin and Binkowski 1980). Both F and G are
functions of Z/ZQ, z/L, and zQ/L. Since the bulk Richardson
number is derived from measurements at 2 m and 10 m, z in this
application is 10 m. This procedure for obtaining L has been compared
with direct measurement results, and shown to be within a factor of 2.
The surface friction velocity u^ is readily calculated once L
is derived. By definition, the function F, used in the iterative
technique for obtaining L from profile measurements, is the ratio of
the mean wind at the upper measurement height (10 m) to the friction
velocity. Therefore
F(zio/zo' zio/L' VL)
-------
Similarly, the surfce heat flux EL, £g calculated from L and
u*. Using the definition of the Monin-Obukhov length,
Hj, . -ul P Cp T1Q/(k g L)
Finally, the convective velocity scale, w+y is derived from the
heat flux and the convective layer height, h by:
10
10
Measurements of incomming solar radiation or net radiation can also be
used to estimate w^ (venkatram 1978, Briggs 1975).
Water Vapor Content Measurements
Water vapor content, ew> is derived in part from measurements
of temperature dew point, Td. Either of two dew point sensor
techniques may be used: the dew cell technique or the cooled-mirror
technique. A measurement accuracy of ^1.5°C over the temperature
range -30° to +20°C is adequate. As a rule, the time constant of
the system is not critical—1 to 30 minutes. For some applications
such as cooling tower fog formation, a short time constant may be
required.
The dew cell technique with a resistance thermometer element has
a characteristic time constant of 12-15 minutes; miniature versions
with thermistors are much faster. For measurements at source height
or above, it is important to measure the corresponding ambient
temperature so that the water vapor content of the air may be
calculated.
Summary
The basic measurements required for Priority 1 quantities are
presented in Table 2-1 with information on instrument heights,
accuracy, resolution, sampling times, and response time.
-------
cn
u
S3
CD
r-4
QU
S
<3
en
1
H
CD
CO
c
O
a
CO
CD
OS
vl
CO
C
o
cj
CD
O
C
CO
4J
CO
• ia
a
• H
H
60
e
• H
60
CO
V-i
CD
>
CO
•o
c
o
CJ
CD
CO
in
i
Ji'S
*>£
CO 60
e c
O -H
o a
a
cu a
o n
1 C
u
CO
•f*4
a
CO
•a
c
o
u
CD
1 CO
Ut
1
f— 1
CD
4J
3
c
•l-l
e
*•*
CO
^3
c
o
CJ
CD
CO
o
en
CO
73
e
0
CJ
CO
o
••a
1
CO
•o
C
O
u
cu
CO
o
-H
1
CO
CD
U
3
C
•t-l
a
o
i
CD
(0
C
O
Q.
CO
cu
iW
0
CN 0)
**» S
A |'w
cu
3
C
s
•-*
CO
TJ
c
o
CJ
CD
9)
o
f)
CO
1?
O
VI
CO
•O
CD
cx
CO
o
CO
cu
u
OS
o
7 §
N OS
o
fe
u
hJ
g
u
PS
co
z
u
Q£
CJ
CM
o
ago
o o
«-*!—( O
o
o
u"> O O
tO -H
CD CD
co co iJ >
O «0 CJ
CM U-l CO f-4 t-4
• A CD
O O < OS O
a
o
m cj
I *A
O O
in -H -*
o
o
"5
0)
CD
u
3
O
CO
CD
cj
3
O
CO
« CJ
!-l 0, r-4
a
o
1— 1
1
CN
a
o
o
o
^
CN
1
O
8
o
o
CM
1
O
g
t-4
o.
ft
9
i—i
~ a.
3
CO
e
CO
CD
31
o
60
C
o
o
u
c
CD
c
CD
e
Q)
la
3
en
(0
0)
-e-
M
+CD
C
CD
3
cr
CD
la
O
G
O
-a
a>
u
3
cr
cu
06
10
-------
All the derived quantities depend, ultimately, on the primary
measured quantities; some derived quantities depend on others derived
first. Table 2-2 lists all the Priority 1 quantities, the time
intervals for their characterization, and the quantities from which
they are determined.
11
-------
IU
O
CO
C.
CO
w
,J
CO
en
u
•O .U
e u
2.2
~
• -I AJ
•sa
03
"O 3
>*4 03
•a -H
(U M
O
a S
o o
f-l 03
• • ••
CD CD
§
o o
co •-«
a s
CD
a
o
7
a
o
f>J
•t
a
o
i_H
8
CM
a ••
j: 3
S
eN
^x
H
^
8
S
a
tN ej *~>
v-x J3
05
H o a
<3 J= J3
* ta *
3 EC 3 0)
12
-------
3. DESIRABLE (PRIORITY 2) MEASUREMENTS
In addition to the Priority 1 variables, six other variables were
recommended as desirable for more accurate modeling in certain
situations:
• standard deviation of the reciprocal of the wind speed at
source height (o(l/u)),
• mean wind direction at plume height (9 ),
• standard deviation of horizontal and vertical wind
fluctuations at source height (VQ,O$),
• temperature gradient from source height to effective plume
height (AT), and
• solar radiation (SR).
Wind Measurements
The Priority 1 wind measurements at source height will provide
the data necessary to estimate cr(l/u), 0Q, and o^ at source height.
Remote sensing or tall tower observations of u and 6 are required for
9 at plume height.
Computational procedures and sampling duration for a(l/u) are
similar to those for OQ and o^ already discussed in Section 2. It is
recommended that the standard deviation for a(l/u) be computed from
1-minute averages of u to retain only the lower frequency variations
in wind speed which affect plume rise.
Temperature Measurements
For sources of modest effective plume height, the temperature
gradient from source height to plume top may be obtained from an
instrumented tower, with measurement accuracies of 0.1 C. For tall
sources, discrete temperature sounding profiles are accurate enough to
resolve major temperature gradients through the layer of plume rise.
Such temperature profiles, obtained once per hour, will also help
resolve h and h . f
m c
13
-------
Solar Radiation
Global sun and sky radiation should be measured with pyranometers
with 50 W/m^ accuracy, a resolution of 10 W/m , and time constant
(63%) of about 5 seconds. A complementary net radiometer may have a
response time between 0.5 and 1.0 minute. Radiation measurements may
be used to estimate w^.
Summary
Six additional modeling variables are desirable. Their
derivations require measurements of winds aloft (at plume height), of
temperature profiles to plume heights, and of solar radiation (SR)
measurements at the surface. Table 3-1 lists the Priority 2
variables, the time interval for characterization, and the quantities
from which they are immediately determined.
14
-------
2
W
z
o
C
CM
at
P-
o
z
O
•I-I
a
e
o
(U
u
^4
3
O
CO
• •
3
0)
B
3
t-l
a
••
CD
•»
3
CD
O | CD O
«!
en
15
-------
4. HELPFUL (PRIORITY 3) MEASUREMENTS
Several further variables are helpful when modeling in complex
terrain or when considering long-range transport. These include the
mean vertical velocity (w~), the Froude number (Fr), the geostrophic
wind (U ), and the state of the ground over which the plume
O
travels, w requires measurements of vertical velocity and Fr requires
vertical profiles of temperature and velocity to heights exceeding
local terrain elevation. Geostrophic winds should be calculated from
three-hourly NWS observations near the site. Table 4-1 summarizes the
Priority 3 variables.
16
-------
CO
u
C
OS
terrain
X
cu
a
§
u
o
l-l
3
iu
CD
CO
9
00
C
•H
i— <
V
•o
o
a
terrain
X
(-1
0)
a
o
u
S
o
a
co
e
o
-------
5. RECOMMENDATIONS FOR THE OFFICE OF THE FEDERAL
COORDINATOR FOR METEOROLOGY
Translating the recommendations of this workshop into specific
requirements for new source evaluations would be very costly. It is
suggested that the National Weather Service assist as much as possible
by altering, where feasible, its current data acquisition and
processing procedures and instrumentation.
In the list of recommendations, Panel A has suggested only those
items for which major needs exist, and which would have considerable
transfer value from NWS sites to pollutant source locations.
Recommendations which would be incompatible with aircraft
operations have been avoided because many of the NWS stations are at
airports. The list is therefore restricted to low-level measurements
(xlQ m), or measurements from remote sensors„
Special efforts should be made to obtain more frequent radiosonde
data. A series of at least four per day, for periods of approximately
3 to 4 months in various sections of the U.S., would be very helpful.
They would be of maximum benefit if the program were conducted on a
firm schedule, published well in advance.
The list of recommendations is presented in Tables 5-1, 5-2, and
5-3. The tables detail recommended low-level quantities, low-level
measurements desirable at locations relatively free of wakes and
runway surfaces, and recommended high-level measurements from remote
sensors.
18
-------
CM
O) OI
4-1 -H
3 03
JJ 4J tO
•H -i-l 01
4J
03 U-4 01
.0 •* .£3
3
to d^iD
D •-1
CQ 3
V4 O
to O JS
0)
>
o
w a, •
M O 3
H u 0
M
£_*
122
^
f—v
ex
Q
td
^>
i_4
ft!
W
a
a
z
en
»"H E-* 01
1 Z 00
lO Cx2 C
)!tf^ (Q
w w .c
»_3 OCt O
PQ J3
< C/3 U-4
H •< 0
•g m
a.
»—3 ?S
W H
§
|
_]
S
Q
Z
§
S
o
o
u
60
G
•H
03
03
rt)
y
O
V4
P>4
Q
4J
M
Q
r*
00
C OI
•r^ j:
03 4«)
03
01 l>4
U OI
0 >
I-i O
a
03
tQ 01
4J 3
tO r-t
•o to
^
Jj
01 G
u to
r-l OI
"O
O T3
jj 4J
G
O
4J
10
e
M
4J
03
C
t— 1
a 01
a
O 03
4J
tJ
M i***
0 0
09 ^
C 03
01 01
03 l-l
£.
01 4J
J>
O I-i
Oi |j 0)
a. 3
e o
l-l >-l
/-«s
lA
vO
•
O
V
4J
01 3
T3 ^
O •
^ CD
Q*
o
O 4J
4J ~~^
M (30
O G
03 -^
e o.
o> 0
03 tO
T3
0*
^> V><
O 01
U 4J
a. u
a 01
<-> A
TJ
1-1
(0
•a
c
to
4J
09
>,
r— 1
M
3 •
O 03
*C H
O
OI *H
•U U
3 to
O.-H
a >
O 01
O T3
•o
)•*
to
•a
c
to
4^
03
>%
p-4
l-l
3
O OB
j= e
0
4-1 U
3 tO
Q.-H
e >
O 0)
O "O
\ 3
^4
0
U-4
03
^
O
03
a
01
03
0)
S
S
0)
01
3
r-H
fl3
^
^
»-H
l-l
3
O
(-;
C
tB
01
B
•a
0)
• H
a"
01
r— 1
aj u
C C
O OI
.^ a
t-i O>
•rJ -r<
T3 3
•o a-
-------
o
•H
>
60 -O
c
•H T3
CO Id
CO CO
01 t3
U d
O <0
04 CO
I-i
O
IM
t>0
c
en
co
0)
u
o
w
O. to
01
CU 3
4-1 l-l
• H CO
en >
• H
3 I-i
« I
OS -H
l-l
o
TJ
C
O
•1-1
4J
03
4J
e
cu
M
4J
CO
d
M
cu
V.
•l-l
3
%
2
cu 3
J3 C
C h
C8
O «
0)
>, "
CU CJ
60
-i C
3 -H
0) -i-l
D 3
CO
•a cu
i-i ^
3 W
O 3
C
(0
3
Ol
s
o
<— 4
L
a
03
13
"O y
0) -r-4
en co
CU
c* 4J
H O
20
-------
TABLE 5-3
RECOMMENDED HIGH-LEVEL MEASUREMENTS
FROM REMOTE SENSORS
Quantity Comment Priority
u above 100 m An elevated wind speed measurement
at some standard (agreed-upon)
height
hc The depth of the convective mixing
layer, as available from acoustic or
other sounders
d6/dz Measurements of wind shear in the first
2000 meters would be helpful
21
-------
6. ARCHIVING AND RETRIEVAL OF DATA
Most primary data will be collected by an automated system, and
will be initially formatted autographically and/or digitally,
depending on site instrumentation. Autographic data should be reduced
to digital fora to eliminate difficulties inherent in their archiving,
retrieval, and use. In order that data be available to users, a
responsive data archiving and retrieval system is needed.
The following actions are suggested.
• Fix the retrieval and user requirements for data quantities,
including both derived and measured quantities. (There is
no point, for example, in archiving data which lose their
utility beyond the real-time application.)
• Establish the data base access strategy to optimize file
design.
• After quantities to be archived are determined, choose a
format which accommodates all data, is site-independent, and
is compatible with the data base access strategy.
• Preprocess data on-site as completely as possible, but keep
enough detail to permit subsequent processing. Retain
complete details for periods of special interest.
• Establish quality assurance guidelines for preprocessing and
on-site data collection processes. Pre-archival processing
could include flagging of suspect data to warn the user.
• Maintain site and instrument descriptions, and documentation
of processing techniques for each site as an integral part
of the system.
22
-------
7. WORKSHOP PARTICIPANTS
Panel A
Mr. Maynard Smith
Meteorological Evaluation Services
134 Broadway
Amityville, NY 11701
Mr. Richard Bailey (Observer)
NWS, Meteorology Services Division W11X1
8060 13th Street
Silver Spring, MD 20910
Dr. Gary Briggs
ATDL-NOAA
P.O. Box E
Oak Ridge, TN 37830
Mr. Edward W. Burt
OAQPS (MD-14)
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Bruce A. Egan
Environmental Research & Technology, Inc.
3 Militia Drive
Lexington, MA 02173
Mr. Danny Fulbright
National Climatic Center
Federal Building
Asheville, NC 28801
23
-------
Mr. John S. Irwin
Meteorology and Assessment Division, ESRL
U.S. Environmental Protection Agency (MD-80)
Research Triangle Park, NC 27711
Dr. Frank Pasquill
37 Arbor Lane
Winnersh, Wokingham, Berkshire
England RGll 5JE
Dr. S. Sethuraman
Brookhaven National Laboratory
Upton, NY 11973
Dr. Isaac Van der Hoven
Air Resources Environmental Laboratory
NOAA, ERL
8060 13th Street
Silver Spring, MD 20910
Panel B
Dr. Todd V. Crawford
Savannah River Laboratory
E.I. duPont de Nemours
Aiken, SC 29801
Dr. S.P.S. Arya
Associate Professor of Meteorology
Department of Geosciences
North Carolina State University
Raleigh, NC 27607
Dr. Francis Binkowski
Meteorology and Assessment Division, ESRL
Environmental Protection Agency
Research Triangle Park, NC 27711
24
-------
Professor Robert Bornstein
Department of Meteorology
San Jose State University
San Jose, CA 95192
Dr. Tom Horst
Atmospheric Sciences Department
Battelle Pacific Northwest Labs.
Battelle Blvd.
Richland,WA 99352
Mr. Einar Hovind
North American Weather Consultants
600 Norman Firestone Road
Goleta, CA 93017
Dr. Douglas G. Smith
Environmental Research & Technology, Inc.
3 Militia Drive
Lexington, MA 02173
Dr. A. Venkatram
Ontario Ministry of the Environment
Air Resources Branch
880 Bay Street
Toronto, Ontario M5S 1Z8
Canada
Dr. Allen Weber
Savannah River Laboratory
E.I. duPont de Nemours
Aiken, SC 29801
25
-------
Panel C
Mr. Thomas J. Lockhart
Meteorology Research Inc.
Box 637
464 West Woodbury Road
Altadena, CA 91001
Mr. Robert C. Beebe, COM
Supervisor, Data Management Section
Tennessee Valley Authority
346 Evans Building
Div. Water M
Knoxville, TN 37902
Mr. James L. Dicke
SRAB - MD 14
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Peter L. Finkelstein
EMSL - MD 75
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Mr. Benjamin R. Greene
Environmental Research & Technology, Inc.
696 Virginia Road
Concord, MA 01742
Dr. J. Chandran Kaimal
WPL - R45X7
NOAA, ERL
Boulder, CO 80303
Dr. Paul B. MacCready, Jr.
Aerovironment Inc.
145 Vista Avenue
Pasadena, CA 91107
26
-------
Mr. Daniel A. Mazzarella
Science Associates, Inc.
P.O. Box 230
Princeton, NJ 08540
Dr. James D. McTaggart-Cowan
Atmospheric Environment Service
Fontaine Bldg., 13th Floor
Ottawa, Ontario, K1A OH3
Canada
Mr. William Petersen
Meteorology and Assessment Division, ESRL
Environmental Protection Agency (MD-80)
Research Triangle Park, NC 27711
Mr. Robert N. Swanson
Pacific Gas & Electric Company
215 Market Street, Room 453
San Francisco, CA 94106
EPA Project Officer
Mr. D. Bruce Turner
Research Meteorologist
Meteorology and Assessment Division, ESRL
Environmental Protection Agency (MD-80)
Research Triangle Park, NC 27711
ERT Project Manager
Mr. Gale F. Hoffnagle
Environmental Research & Technology, Inc.
3 Militia Drive
Lexington, MA 02173
27
-------
REFERENCES
Benoit, R. 1977. On the Integral of the Surface Layer Profile-Gradient
Functions. J. Applied Meteorology, 16(8);859-860.
Briggs, G.A. 1975. Plume Rise Predictions. Lectures on Air Pollution
and Environmental Impact Analysis, September 29-October 3; Amer.
Meteor. Soc., pp. 59-111.
EPA 1978a. Guideline on Air Quality Models. EPA-450/2-78-027.
Environmental Protection Agency, Office of Air Quality Planning
and Standards, Research Triangle Park, NC.
EPA 1978b. Ambient Monitoring Guidelines for Prevention of
Significant Deterioration (PSD).EPA-450/2-78-019.
Environmental Protection Agency, Office of Air Quality Planning
and Standards, Research Triangle Park, NC.
Irwin, J. and F. Binkowski 1980. Estimation of the Monin-Obukhov
Scaling Length Using On-Site Instrumentation. Atmospheric
Environment (in press).
Venkatram, A. 1978. Estimating the Convective Velocity Scale for
Diffusion Applications. Boundary Layer Meteor, 15;447-452.
28
-------
APPENDIX A
DATA REQUIREMENTS
CONTENTS
1. INTRODUCTION A-2
2. SPECIFIC VARIABLES REQUIRED FOR PLUME RISE
AND DISPERSION MODELING A-3
2.1 Quantities Measured at Different Heights or
Height Intervals A-3
2.2 Quantities Measured at the Ground at
Designated Elevations A-9
3. GENERAL METEOROLOGICAL MEASUREMENTS A-14
4. DATA REPRESENTATIVENESS A-16
4.1 Types of Representativeness A-16
4.2 Flat Terrain A-17
4.3 Complex Terrain A-18
4.4 Urban/Rural Interaction A-18
4.5 Land/Water Interaction A-19
4.6 Length of Record A-20
5. RECOMMENDATIONS TO THE OFFICE OF THE FEDERAL
COORDINATOR FOR METEOROLOGY A-21
6. ARCHIVING AND RETRIEVAL OF DATA A-25
REFERENCES A-26
PANEL A MEMBERS
Chairman: Maynard Smith
Gary Briggs
Edward Burt
Bruce Egan
David Fulbright
John Irwin
Frank Pasquill
S. S ethuRaman
Isaac Van der Hoven
Observer Richard Bailey
A-l f
-------
1. INTRODUCTION
Panel A was asked to review modeling data requirements, a task
which we interpreted as including both current and future modeling
needs. We were also asked to consider the representativeness of such
data, as well as accuracy, time averaging, period of record, and
archiving. Finally, Panel A was expected to recommend changes and
additions to National Weather Service (NWS) measurements that could
help in providing the data needed by air pollution specialists.
Panel A developed a series of tables listing the key variables in
air quality modeling. For this initial review, the problem was
treated as though the site in question were flat and uncomplicated,
ignoring the possible horizontal variations which could make a single
data set unrepresentative, but taking full account of the need for
representativeness in the vertical. For each variable or derived
quantity, we chose appropriate heights for the measurements, specified
appropriate averaging times, and considered the priority of the
measurement with respect to one or more applications.
We decided that it would be unproductive to define the required
accuracy of the measurements, since it would involve the desired
accuracy of the predicted ground-level concentrations, by no means a
stationary target. Instead we listed the sensitivity of the
dispersion property to the measured variable in two ways, first as the
index in a power law (e.g., a a CQ*), and also on a simple
relative scale where a value of 1 meant high sensitivity (as
concentration to u) and 3, low sensitivity.
The difficult problem of data representativeness was addressed by
considering three types of sites in addition to the idealized flat
terrain location: (1) rough, complex terrain, (2) a land-water
interface, and (3) an urban-rural boundary area.
A-2
-------
2. SPECIFIC VARIABLES REQUIRED FOR PLUME RISE
AND DISPERSION MODELING
Tables 1 and 2 list the variables and derived parameters
important in modeling plume rise and dispersion. From the basic
quantities, other hybrid parameters can be developed or estimated.
The tables are supplemented by descriptions and comments, where
necessary.
In the columns designated as sensitivity, Exp denotes the
approximate quantitative effect of the variable on the application in
question. An Exp entry of -1, for example, means that concentrations
would be inversely proportional to this variable. An entry of 1/3
indicates that the concentrations or plume rise would vary as the 1/3
power of the variable. The second entry, Rel, is a qualitative
estimate of the effect of the variable on the modeling results,
measured as (1) important, (2) moderate, or (3) minor. The entry
designated as Priority means the quantity is (1) essential,
(2) desirable, or (3) helpful in modeling atmospheric dispersion.
2.1 Quantities Measured at Different Heights or Height Intervals
Table 1 lists variables that change significantly with height and
which must at least be estimated at varying elevations above ground.
Each of these variables is discussed in the following sections.
u:
The mean wind speed, u, affects diffusion in two important ways:
plume rise and longitudinal dilution. The sensitivity of plume rise
to u varies with stability. In stable conditions it depends
approximately on u , while in neutral conditions it may vary as
__2
u ; for unstable conditions the sensitivity is intermediate. For
all conditions, the dependence of concentration on longitudinal
dilution is simply u . In addition, a mean wind speed is needed
for the bulk Richardson number, a highly desirable parameter for
A-3 f
-------
9i
M
o.
u
c
•prf
X
« 3
a; e
8'
.5 f
oo « e c:
c u •- o
••- u 1 O
IS C H
» U 91 -O
5 § S^
Ob
3
'00
U
a -o
V
b 11
a. -a
« c
u o
«
V O
V
$5
00 O
•58-
• r-4 U
09 U
(U
X u
p- X
(Q V
« o
(3 o
CO
H
O
H
Z
^ r-l —<
W
nJ
cn
b*
M
O
H
S
CO
w
2
w
n
o
H
CO
§ 2
'S S
O.
a]
•*-<
Q
• H 41 0) 01
oS I I
5^-t F^ l-l
a. a a.
H
Z
O- - "
o o
-4 00
•&!
. |
C U
ti a
a ja
0) O
o >«-i £
>-. D 00
O O 01
A-4
-------
CM
Cd
02
H
cn
a;
o
M
£«M
5>
W
O
U
a
H-l
vs
W
Q
C-l
0
Q
S3
O
tti
U
Cd
a
C~«
OS
Id
Q
H
3 "
•< •>
w «
03
S I
O
H
CO
Cd
h-4 u*
H o
*""^ U
H -c
2> oo
•^ 01
3 x
O*
M
c
V
1
4J
o
u
0.
c
0
..4
a
0)
3
a
£
at
•^
-1
-fl
V
a
• «4
£
OB
e
o
4J
a)
—
«
^
«
1
a
tl
w
c
GO
3
_
C
O
aj c -H a)
« « .rf B S •* «
»0 B3 •+ «1rt
p *-4 vu U *J M
« w M » - a. j
« C5 4) to CO
*M O OD u W
4) *•-> b * CW •< <
+ Ed O + * -f
------
0)
+ C
-J **-i O 1
O *»J > 41
U O G O O ••-( * —• U 4t
-o ,.H e u o e
»(J UN U4JU -M ++ UUV
MC 5 •-• tQ"4tB 4j 01 c a a c u e a
coxa BUS a — a x> « o o o 4* « «. 3
.,4 ».,4 u w jj — i u jo *j u u CD to 4J qcafio*
ca^»*q aurM co«co tj u ^ u ea .^ 4) --4
3 C --- N 41 «— 41 U 0) CO) CU V CU B « U <0 T3
^tQCBb M«<0 U ^ U C 9 <0 (0 U C 3 ^ W
ako*JV S*13" £.jOa ON° al a Q c3 Socw
+
-* -. % >\ >s >s >s
3 3 3 3 3J 03 0] 03 03
CA VI CO « -J _1 _J J J
oa
•H *4 4t UUOCJS
-J at a ^ a jz £ js js
00
01
•••4
3
-4
«
u
(S
3
O
.C
—
at
•*^
e
o
u
ca
u
3
00
e
a
CO
a)
O
z
A-5
-------
characterizing the surface layer and vertical dispersion near the
ground. In this application, the vertical dispersion at a given
distance can depend on u~^ in very unstable conditions.
Adequate wind speed measurements for all these purposes can be
made at a single location free of interference from structures. For
bulk Richardson number determination, a 10 m measurement height is
preferred. For plume rise and dilution, the preferred height is the
final plume height. However, for normal accuracy this is not
necessary.
There is little wind shear above 50 m in unstable conditions, so
u at source height or the top of a tower is adequate. In neutral
conditions, extrapolation of the logarithmic profile law from these
heights is adequate, provided that the roughness length of the surface
near the source is reasonably estimated (this only requires a
description of the ground cover in the area). In stable conditions,
wind shear can be large above this measurement height, but plume rise
depends only on u~^'^. Except for elevated structures or terrain,
high rising plumes in these conditions rarely affect the ground until
breakup fumigation occurs; the resultant ground concentration then
depends on the inverse of the total height of the plume, but is just
as sensitive to other factors less well known. Therefore, u at source
or tower height is adequate for most site and accuracy requirements.
For low level sources, large wind shear near the ground needs to
be accounted for in modeling. But u as a function of z can be
calculated from u at a single height, such as 10 m, in combination
with either the bulk Richardson number or some other determinant of L,
the Monin-Obukhov length.
For hourly wind speed averages, about 60 samples of the velocity
per hour are needed. It is preferable to average (1/u) rather than
inverting (u) for plume rise and dilution.
In low wind speed convective conditions, fluctuations in plume
rise due to variations in u can significantly increase vertical
dispersion. One-minute averages adequately monitor these
fluctuations, and the variance
-------
fluctuations in u affect plume rise and transport significantly and
thus are critical.
6:
The 1-hour average horizontal wind direction, 6, has two
important uses: the placement of the plume in the x-y plane, and in
the determination of the direction shear in the atmospheric layer in
which the plume is dispersing.
For the placement of the plume, 6 measured near the source at a
height of 10 m and at the top of the meteorological tower should be
sufficient for modest source heights and close distances (<10 km) in
cases other than complex terrain. In complex terrain, 6 measurements
at additional sites may be required. A special case is the direction
of travel of elevated plumes as related to underlying or intercepting
high terrain. For plume travel of distances greater than 100 km, a
wind field derived from standard NWS measurements of wind profiles may
be required.
The importance of wind direction shear with regard to crosswind
dispersion from tall stacks is primarily limited to downwind distances
of 10 km or more. Also, complex terrain often exhibits very drastic
shears in shallow layers close to the ground and measurements at
additional sites may be required.
<
V V (°v aw):
Standard deviations of wind fluctuations are needed (1) to
directly characterize the dispersion parameters (a and tf_) and
(2) in place of u* and Hp to characterize the boundary layer
structure as affected by roughness and thermal stratification. The
grade (1) sensitivity reflects the expected linear proportionality
between ay or az and aQ or C^ .
Modeling in the context of chemical transformation should refer
strictly to instantaneous concentrations and hence to the
instantaneous dimensions of the plume. For a power plant plume, these
dimensions are first determined (i.e., at short range) by the
i'
A-7
-------
buoyancy-induced entrainment. Only with increasing distance are the
appropriate o values affected by ambient turbulence. Then, however,
they obviously must be less than the total values derived by adding
2 2
the buoyancy-induced entrainment a to the a evaluated from
2 .
1-hour averages of a and a^ (e.g., instantaneous a is
9 2 2 2 Z
always somewhere between a , and a , + a , where a ,
} zb zb zw' zb
refers to the buoyancy-induced entrainment and a refers to the
Z *v
1-hour average of o^).
It was agreed that the priority should be lower at the stack or
source height than at 10 m for both parameters. Given cr at the low
level, a useful extrapolation may be made to higher levels in some
cases, using the known dependence of this variable on z/L and z/h .
Then for estimates of short-range dispersion of a power plant plume,
the buoyancy-induced entrainment is more important than ambient
turbulence.
It was also recognized that, with sampling time specified,
requirements for specifying suitable reading intervals and smoothing
time and time-constant remained to be specified to ensure adequate
derivation of the total variance within some reasonable limit. But
these aspects were deferred for consideration by Panel C.
AT/Az:
The temperature gradient, AT/AZ, is needed for two purposes:
plume rise in stable conditions and characterization of the surface
layer. Plume rise is not very sensitive to this quantity (being
^1 / o
proportional to (AT/AZ) ), but an estimate is needed over the
layer of plume rise. The best AT/AZ for this purpose is from the top
of the stack to the top of the plume. Elevated plumes in stable
conditions do not usually reach the ground until inversion breakup
fumigation occurs, and there is probably much more uncertainty about
o than about plume height. In this case, a determination of AT/AZ
from routine NWS radiosonde soundings will be adequate for most
sites. The most obvious exception is the source in a deep valley;
since a is constrained by the valley walls, inversion breakup
y c
fumigation may be of critical interest, but temperature soundings
A-8
-------
outside the valley are irrelevant. For this purpose, a study of the
vertical temperature structure at the specific site may be necessary.
The temperature gradient near the ground is needed for
determining the bulk Richardson number, in turn a convenient
determinant of L, which very effectively characterizes vertical
diffusion near the ground. In addition, AT/AZ below 30 m correlates
well with H_ in unstable conditions. In these conditions,
'10
concentrations can vary with |AT/Azl * . The recommended height
interval for these purposes is AT between 2 m and 10 m. When there is
a tower at the site, AT between the tower top and 10 m should also be
provided as a supplementary stability measure. One-hour averaging
times are convenient.
Fr:
The Froude number, Fr, is a measure of the importance of density
stratification on flow in complex terrain. It has been shown to be
valuable in fluid modeling studies during stable conditions where it
helps specify the flow pattern over, around, and downwind of a hill as
a function of hill size, flow velocity, and density (temperature)
gradients. In the atmosphere, velocity and density gradients are not
often linear in complex terrain. This requires that care be taken in
evaluating (and therefore detecting) multiple layering of the
atmosphere. When multi-layering occurs, the appropriate Froude number
will depend on layer properties at release height, effective plume
height, and at heights in excess of the terrain elevation.
2.2 Quantities Measured at the Ground at Designated Intervals
Table 2 lists a set of important measurements or derived
variables which have little relation to height. It is sufficient to
evaluate these quantities at the surface or at a designated
elevation. Each is discussed in the following sections.
A-9
-------
u*-
" •
The friction velocity, u^ £S needed to predict plume rise in
neutral conditions, and it also provides one way to estimate the
Monin-Obukhov scaling length. For plume rise, it is the most
convenient means for estimating t at plume height during neutral
conditions, and also characterizes the magnitude of vertical
velocities that cause an elevated plume to reach the ground. Plume
rise depends approximately on u^4/3 (Briggs 1975). For this
purpose, it is usually adequate to apply the logarithmic wind profile
law for neutral conditions, u* , 0.4ut/ln(zt/zo), provided
that the surface roughness length, ZQ can be reasonably estimated,
where z^ ^3 Che reference height of the upper wind speed
measurement, say at stack height. A factor of 2 accuracy is
acceptable for the ZQ estimate; this can be achieved by an
experienced observer or obtained from a careful description of the
ground cover surrounding the site. The cover at a radial distance of
roughly 10 to 40 z is most relevant. The wind speed used should be
that at the top of a meteorological tower, u, since it is less
influenced by the immediate surrounding surface than a lower
measurement.
If the Monin-Obukhov length, L, is not determined using the bulk
Richardson number, it is usually determined from u* and a heat flux
measurement, that is, from L ™ -ug copT/(k g Hp). From its
definition, L is proportional to u^J, so L is quite sensitive to
the surface friction velocity. The vertical dispersion from a source
near the ground is highly correlated with L, but at most it depends on
| L! to the plus or minus 1/2 power. From this, we would say that QZ
is only fairly sensitive to u^ por this purpose, L may be
adequately estimated from u at 10 m using known surface layer profile
relationships. This latter analysis may require iterative techniques
since u^/u £3 a function of ZQ/L and z/L.
Hp, L and w*;
Surface heat flux, HFf is a basic parameter characterizing the
boundary layer structure in the surface layer in conjunction with u^
A-10
-------
through the Monin-Obukhov length scale, L. It is important over the
rest of the convective boundary layer in conjunction with h through
the convective velocity, w .
Turbulence intensities and dispersion rates are well correlated
with L and w , though to an extent that is not true of H itself;
* F
in convective conditions velocity variances and a behave as
1/31/2 Z
HI and Hi , respectively, at most. In very stable
conditions, field studies indicate that with passive surface releases,
1/2
the sensitivity of a^ becomes greater than | Hp . One might
have expected H to be an important determining factor for h at
night but apparently this factor is overshadowed by u ,
Considerable spatial variation of H and the dependent
r
parameter L are to be expected as a result of transitions in surface
characteristics (especially albedo and wetness). For this reason, a
direct measurement, which is difficult to make, at a single point is
of dubious value; some of the indirect measures, such as solar
insolation, might actually be more broadly representative. There are
several variables that can be used as indicators of or contributors to
heat flux. These include net radiation, sky radiation, sky cover,
solar elevation, state of ground, and evapotranspiration. None of
these in itself can be substituted for the heat flux. Only sky cover
and solar elevation have been used systematically (with other
variables) to replace the heat flux factor for estimating dispersion.
The roles of the other variables for estimating the heat flux warrant
further study. H can also be determined from variances of
turbulence quantities that are directly influenced by H such as
or, in convective conditions, o .
The eddy dissipation rate, £ , characterizes the high frequency
(small eddy size) part of velocity spectra, and in dispersion models
is used chiefly as a factor influencing plume rise in neutral and
unstable conditions. Plume rise depends on e."^'^, approximately, so
estimation of e within +50% is usually adequate. It is believed that
this accuracy can be obtained from knowledge of surface parameters,
. A-ll
-------
especially u^ and Hp, for the conditions of interest. There is a
need for further research on the dependence of e on height and on its
variation within large convective structures (it is known to be much
larger in thermals than in downdrafts, and some models require
separate values).
The importance of the convective mixing depth, h , is that it
limits the vertical dimension of the pollutant cloud, and in so doing
reduces the dilution of concentration with downwind distance. It also
may inhibit or stop plume rise so that higher maximum downwind
concentrations result from an elevated plume. Used with a measure of
surface heat flux, it characterizes the whole convective layer,
including turbulent velocities, h has a definite diurnal
variation, increasing in depth after daybreak, reaching a maximum in
the early afternoon. The mixing layer typically decays at dusk to be
replaced by the characteristic nighttime, ground based radiation
inversion layer except during cloudy and windy conditions.
V
The mechanical mixing layer, h , is generally a nocturnal
feature, driven by strong wind shear near the ground. Near-surface or
small stack emissions may be trapped in the layer with no vertical
growth beyond h but can at the same time exhibit considerable
lateral meandering. The net downwind concentration effect on
low-level sources generally results in higher downwind concentrations
than would be experienced in the daytime convective mixed layer. Near
an urban area, h may sometimes merge with h over the city
because of the urban heat island effect. This can possibly result in
a nighttime rural elevated plume being prevented from reaching the
rural surface, and, upon being transported towards an urban area,
essentially being fumigated to the heat island ground surface. After
leaving the urban area it may be lifted again above the rural h .
r m
In complex terrain h can deepen as a result of the downslope air
A-12
-------
drainage within valleys. Furthermore, side valleys, canyons, and
compensating flows can generate overriding or undercutting outflows
that may cause plumes at different elevations to move in different
directions. No model relating h to other more easily measured
variables has gained acceptance, but it seems likely that h is most
strongly correlated with u^.
The standard deviation of the near surface (10 m) temperature is
of interest for the validation of second order closure models. It can
be used for an estimate of surface heat flux, primarily under unstable
conditions. The low frequency contribution to the variance
represented by the gradual trend in temperature as the hour progresses
should be eliminated.
A-13
-------
3. GENERAL METEOROLOGICAL MEASUREMENTS
Other sections of the report deal with variables specifically
required for the modeling of plume rise and dispersion. For certain
applications, such as long-range transport or the establishment of
relations with more general meteorological patterns, additional data
are required. These quantities are listed in Table 3 and are
generally self-explanatory. Specification of the horizontal wind and
temperature fields clearly is related to representativeness of the
data (discussed in more detail in the next section), and the
geostrophic wind may be a convenient substitute for upper wind
measurements, especially if long-range transport is involved.
Solar radiation and net radiation are included because they may
already be included in the measurements made at some standard
observation stations, and because they can serve as an indicator of
heat flux. The state of the ground also exerts important control on
the heat flux.
A-14
-------
•2
4)
§
a
m
o
c
0
•H e
a a
0 U
a u
w 8
* 'H
CO
a
D
c
3
I 3
>•
Il
JJ
e
V
?
,
w
01
o.
8
•-
41
hi
«J
a
a>
"
a
a
a
c
3
IH
C
V
0
+
M
a
8
u
u
01
§
1
•o
3
O
U
U
0
•H
U
a-
w
u
a
to
c
01
h
jj
91
to
C
eg
i
c
o
0
T3
a
9
M
3
«
V
M
0
3
>*-<
%
S
U
lu
41
•J
Cl
1
j:
o
o
o
1
i*
o
u*
en
H
3
cn
I
c
a
• r4
a
W
03
H
O
O
O
(£.
O
wl o u 9
tl
4J
w
c
T)
Tl
• H
O
b
CL
J!
c
a
u
S.
oo
Q
&
n
c
o
eo a
c o
85
tn ai
o a
01
u b
£ 3
a*
o v
-C
H
•^ 13 r-*
u e
-------
4. DATA REPRESENTATIVENESS
Transport and diffusion models can simulate atmospheric processes
accurately only if the input data faithfully represents the area and
volume in which the processes operate. While the dangers of
inadequate input are clear, nonrepresentative data are too easily
tolerated.
We recommend that the terms "representativeness" and
"representative siting" be given practical, mathematical definitions.
These definitions will permit comparisons between data collected at
different sites. Research is required to develop acceptable threshold
values for these differences. The threshold should be greater than
the "functional precision" of the instrument (Hoehne 1977). The
functional precision is the root mean square of differences between
two identical sensors mounted without influencing one another, at the
same height and as close to each other as practical (within 10 m).
The difference between the functional precision of identical types and
different types describes the difference in instrument performance.
4.1 Types of Representativeness
Primitive representativeness (R_) is equivalent to correlation
of time synchronous data. The flow field might be called frozen where
every point changes in the manner point A changes at the time point A
changes. Hence, another point B would have a data set identical to
point A. For wind data:
=
b , a
for any sampling time, method of averaging time, or location.
Climatological representativeness (Rc) £3 based on the shape of
the distribution over a long period of time. This type of
representativeness may be dependent on time over short periods, but
A-16 f
-------
the distribution of average values at point B is represented by the
distribution measured at point A.
For this type of representativeness
(u, ) (u )
ZD « _ a
N *- N
SK
K(ub) = K(ua)
where
N = number of sampling intervals within the period of record
SK = skewness
K = kurtosis
Spatial representativeness (R ) can be considered a special
s
case of primitive representativeness. Measurements at point A should
be representative of the spatial average over some horizontal area on
the order of 1 to 10 km. This is usually the representativeness
referred to when characterizing upward heat flux with a minimum of
instrumentation over a surface area of great variability.
4.2 Flat Terrain
Panel A assumed that primitive representativeness is valid to
distances of approximately 20 km within flat terrain. In practice,
certain variables may not meet this assumption. Upward heat flux and
net radiation are typically so sensitive to surface characteristics
that spatial representativeness may be very difficult to attain with a
single measurement site. It is also possible that measurements of
bulk Richardson number may require more than one site.
A-17
f
-------
4.3 Complex Terrain
Primitive representativeness is extremely limited in complex
terrain. Hence, the meteorological monitoring should be determined on
a case-by-case basis. For instance, the placement of an elevated
plume, horizontally and vertically, requires detailed knowledge of the
flow field. Even then, time lapse photographs of continuous smoke
releases may still be required to define the trajectory*. At present,
mathematical modeling of flow fields is inadequate for the definition
of plume trajectories. The specification of the profiles of
turbulence intensities would also require measurements along the plume
trajectory since the extrapolation of such quantities from
near-surface measurements is unreliable. In deep valley situations,
the temperature lapse rate over the depth of the buoyant plume rise is
needed for proper modeling of inversion breakup fumigation.
In summary, because of current inability to reliably characterize
wind flow and turbulence structure in complex terrain, these variables
must be measured. The degree of specification is partly
model-sensitive. However, even if a model is incapable of handling
complex flow (such as curved trajectories), the knowledge of such
trajectories may be important if the terrain height is greater or less
than the effective source height. The wind flow structure is further
complicated by (1) upslope and downslope winds, (2) wind channeling
effects, (3) lee wave phenomena (density stratification), and
(4) divergence.
4.4 Urban/Rural Interaction
Plumes from sources near urban areas may have both urban and
rural characteristics. Although "near" has not been defined formally,
sources within 3 km downwind of an urban area seem to be influenced,
depending on the size of the city. For these cases, meteorological
*Available models cannot consistently and accurately duplicate real
flow fields, thus estimating maximum short-term (1-hour)
concentrations due to elevated plumes remains inadequate.
A-18
-------
measurements should be taken at two locations, urban and rural. The
measurements will help model dispersion when the source is upwind or
downwind of the city. The special circulations that can be induced by
the urban heat island during low wind conditions should also be
considered. The priority of the variables listed in Tables 1, 2,
and 3 are the same.
4.5 Land/Water Interface
The unique meteorological features of a shoreline require special
on-site measurement considerations. There are three main effects of
the shoreline environment on the dispersion characteristics of the
atmosphere: (1) mesoscale recirculation due to sea (or lake) and land
breezes (Lyons 1973), (2) changes in atmospheric stability caused by
the development of internal boundary layers due to changes in
roughness and heating (Raynor et al. 1979, SethuRaman 1980), and (3) a
combination of these two.
For example, a stable stratified air mass over water can become
unstable over land during daytime conditions and onshore flow. This
will cause releases from a coastal source to behave like a stable
plume, fumigating further downwind. Because internal boundary layers
with steep slopes can develop in the leading edge, such fumigation
could occur near the boundary depending on the effective source
height. For these reasons, a model that uses meteorological
measurements made at one location at a coastal site will not be able
to predict the downwind ground-level concentrations due to the changes
in stability.
It is recommended that the measurements for flat terrain be made
at a minimum of two locations, one at the shore representative of the
over water conditions for onshore flow, and the other about 1 km
inland where the conditions are representative of the air mass inside
the internal boundary layer. For the site inland, near surface
measurements up to a height of 10 m will suffice. For sites where
chemical transformations are significant, humidity measurements will
be important at least at one level. Additional monitoring is
necessary over water for diffusion during land-water-land cases, such
A-19
-------
as near lakes and rivers. Where the shoreline terrain is more
complex, including bluffs, measurements at additional locations will
be necessary and should be determined on a site-specific basis.
4.6 Length of Record
At least one year of synchronous data, collected at two sites, is
necessary to assess spatial representativeness, assuming the data
record is reasonably complete. However, to establish whether a year
of data is typical or climatologically representative, more than one
year of data is required. It is therefore recommended that 2 or
3 years of data be used to determine climatological
representativeness, with the possibility of re-analysis as new data
become available.
A-20
-------
5. RECOMMENDATIONS FOR THE OFFICE OF THE FEDERAL
COORDINATOR FOR METEOROLOGY
Translating the recommendations of this workshop into specific
requirements for new source evaluations would be very costly. It is
suggested that the National Weather Service assist as much as possible
by altering, where feasible, its current data acquisition and
processing procedures and instrumentation.
In the list of recommendations, Panel A has suggested only those
items for which major needs exist, and which would have considerable
transfer value from NWS sites to pollutant source locations.
Recommendations which would be incompatible with aircraft
operations have been avoided because many of the NWS stations are at
airports. The list is therefore restricted to low-level measurements
("-10 m), or measurements from remote sensors.
Special efforts should be made to obtain more frequent radiosonde
data. A series of at least four per day, for periods of approximately
3 to 4 months in various sections of the U.S., would be very helpful.
They would be of maximum benefit if the program were conducted on a
firm schedule, published well in advance.
The list of recommendations is presented in Tables 4, 5, and 6.
The tables detail recommended low-level quantities, low-level
measurements desirable at locations relatively free of wakes and
runway surfaces, and recommended high-level measurements from remote
sensors.
A-21
-------
^
4J
•H
O
• H
U
CU
ft)
•a
CM
4) 4)
4J >H
3 «
•U 4J «
•l-l -ft 0)
O) 14-1 4)
.0 -H J3
05 _
D
ID O JS
CO
Cu •
U
M O 3
H •" O
K-l
H
4?
C3
Cr
Q
w
i-i
fVj
w
Q
Q
5=
to
H 1>
5S W>
S3 S
60
C
«t-4
M
(0
a>
o
O
£
0}
44
«
0
r*
to
c
w o
ex
05
«J 0)
4J 3
TO •— i
*O CO
5>
Vj
01 C
4-1 CT3
f-i a
< a
H
2
vO
C
O 0)
O -O
•a
u
(Q
05
>%
3
o
a>
o
O ^
Crf ^0)
(X 5
e o
o
V
4J
w
O 4)
M iJ
^^ 4J
S C
o o
UJ -H
u
^ nj
oj -H
U T3
o w
H «
A-22
-------
*\
"«
o
•H
0)
o;
Vi
>>
.-I
tA
3
03
2
W
H
Z
f*3
S
§
£j
w
u
o
j
Cd
S?
v«
to cu
Z >
w
•H
4-
c
C!
6
^ »
D
M o
o *•
U-l <
3
^
i
j
60
C 2
•^ S
« .2
CO *"
« tJ
O oj •
0 <-> 0
M 0 *J
ft, CO •-* fl)
*" j
i
j M
0 U)
>
cd
> -u
T3
A-23
-------
TABLE 6
RECOMMENDED HIGH-LEVEL MEASUREMENTS
FROM REMOTE SENSORS
Quantity
Ci above 100 m
Comment
de/dz
An elevated wind speed measurement
at some standard (agreed-upon)
height
The depth of the convective mixing
layer, as available from acoustic or
other sounders
Measurements of wind shear in the first
2,000 m would be helpful
Priority
1
A-24
-------
6. ARCHIVING AND RETRIEVAL OF DATA
Most primary data will be collected by an automated system, and
will be initially formatted autographically and/or digitally,
depending on site instrumentation. Autographic data should be reduced
to digital form to eliminate difficulties inherent in their archiving,
retrieval, and use. In order that data be available to users, a
responsive data archiving and retrieval system is needed.
The following actions are suggested.
• Fix the retrieval and user requirements for data quantities,
including both derived and measured quantities. (There is
no point, for example, in archiving data which lose their
utility beyond the real-time application.)
• Establish the data base access strategy to optimize file
design.
• After quantities to be archived are determined, choose a
format which accommodates all data, is site-independent, and
is compatible with the data base access strategy.
• Preprocess data on-site as completely as possible, but keep
enough detail to permit subsequent processing. Retain
complete details for periods of special interest.
• Establish quality assurance guidelines for preprocessing and
on-site data collection processes. Pre-archival processing
could include flagging of suspect data to warn the user.
• Maintain site and instrument descriptions, and documentation
of processing techniques for each site as an integral part
of the system.
A-25
-------
REFERENCES
APPENDIX A
Briggs, G. A. 1975. Plume Rise Predictions. Lectures on Air
Pollution and Environmental Impact Analyses, Sept. 29-Oct. 3j
Amer. Meteor. Soc., pp. 59-111.
Hoene W. E. 1977. progress and Results of Functional Testing
(supplement to NOAA Technical Memorandum NWS T&EL-12). NOAA,
National Weather Service. NOAA TM NWS T&EL-15, April.
Lyons, W. A. And H. S. Cole 1973. Fumigation and Plume Trapping on the
Shores of Lake Michigan During Stable Onshore Flow. J. Appl.
Meteor. 12: 494-510.
Raynor, G. S., S. SethuRaman and R. M. Brown 1979. Formation and
Characteristics of Coastal Internal Boundary Layers During
Onshore Flows. Boundary-Layer Meteor. 16; 99-103.
SethuRaman, S. and G. S. Raynor 1980. Comparison of Mean Wind Speeds
and Turbulence at a Coastal Site and Offshore Location. J. Appl.
Meteor. 19 (in press).
A-26
-------
APPENDIX B
MEASUREMENT TECHNIQUES - DERIVED VARIABLES
CONTENTS
1. INTRODUCTION B-2
2. PRESENTATION OF PANEL CONSENSUS B-3
Exhibit 2.1 Wind Sigroas (a^r a^, a^, aQl a ) B-4
Exhibit 2.2 Roughness Length (z ) B-6
Exhibit 2.3 Richardson Number (Ri) B-8
Exhibit 2.4 Friction Velocity (u^) B-10
Exhibit 2.5 Convective Velocity Scale (vr.) B-ll
Exhibit 2.6 Monin-Obukhov Length (L) B-12
Exhibit 2.7 Heat Flux (H_) B-13
£
Exhibit 2.8 Pasquill-Gifford Dispersion Classes B-16
3. DISCUSSION OF DERIVED VARIABLES B-18
REFERENCES B-31
PANEL B MEMBERS
Chairman: Todd Crawford
< Pal Arya
Frank Binkowski
Robert Bornstein
Thomas Horst
Einar Hovind
Doug Smith
Bruce Turner
A. Venkatram
Allen Weber
B-l
-------
1. INTRODUCTION
The objective of Panel B was to summarize the derivations and
characteristics of dispersion-related variables derived from basic
meteorological measurements. Characterization of these variables
included assessments of accuracy, use, and influence on the results of
diffusion calculations.
Section 2 is a condensed summary of the panel's conclusions in
the form of individual exhibits or tables which focus on the
individual parameters. A more detailed presentation of Panel B's
discussions and recommendations follows in Section 3.
B-2
-------
2. PRESENTATION OF PANEL CONSENSUS
Evaluation of dispersion-related variables derived from basic
measurements followed a flexible format. This format included
consideration of:
• possible modeling use,
• basic measurements required,
• accuracy of the derived variable,
• influence of the derived variable on diffusion calculations,
• method of deriving variable, and
• recommendations.
This format has been used as a device for presenting a summary of the
panel's conclusions. Eight exhibits cover wind field standard
deviations, roughness length scale, Richardson number, friction
velocity, convective velocity scale, Monin-Obukhov length scale, heat
flux, and Pasquill-Gifford dispersion classes.
B-3
-------
EXHIBIT 2.1: WIND SIGMAS (a . a
Possible
Modeling Use;
Basic Measurement!
Height:
Location:
Averaging Time
for basic
measurement:
Ground Level
Release
Elevated
Release
K-Theory for a? Plume Rise1 and
Lagrangian Similarity Theory Statistical Theory
for crz, Statistical Theory
for °y (see Pasquill 1975)
Wind speed and direction
10 m Plume Centerline2
Representative Surface (not near terrain)
1-5 sec
1-5 sec
Response Duration
(Time Constant): 1 sec
Sampling
Duration3.
Accuracy:
Accuracy of Derived
Variable
1 sec
3 minutes or greater depending on travel time to
receptor
u, v, w 0.2 m/sec + 5% (max)
8, * 3o
5-10%
7-30%
Influence of Derived Variable
on Diffusion Calculation;
Through 0 and
(see Table 2 in Pasquill 1975).
Method of Deriving Variable;
Statistical Analyses of Measurement Records for each of the respective
components .
Recommendations ;
#1. NWS should use a 1-hour sampling duration for wind direction and wind
speed. They should report OQ based on the same 1-hour sampling duration.
B-4
-------
#2. The variables OQ and a^ should be calculated directly, rather
than by using a range estimator.
#3. It is recommended that a sampling duration shorter than one hour for
both means and standard deviations of wind components be considered when
modifications of NWS recording and archiving procedures are planned. The
principal use of short-term values (e.g., 3-15 min) is for models designed
to assess trajectories and impacts of short-term (often accidental)
pollutant releases. Thus, the archive recordings could be recycled every
one to four weeks to avoid expensive modifications in long-term archival
data sets.
//4. For industrial or other sites likely to experience noncontinuous
releases of pollutants, statistics based on a 3-minute sampling duration
for OQ (at least) should also be permanently archived. This will allow
a1 a for this short sampling duration to be combined and appropriately
matched with release duration in modeling time-integrated concentration
fields.
Footnotes
*tfu is usable for modification of plume rise estimates for puff
releases.
Measurements could be made at release height if they were the only
feasible alternative.
o
JThree minutes based on historical estimates of a~ from original
Forton experiments and other experiments designer to utilize lateral
P-G dispersion rates. Idaho experiments on diffusion under low wind
speed conditions have also illustrated the utility of this approach
(Sagendorf and Dickson 1974).
^Specific cases will vary depending on degree of long wave convective
contribution to the turbulence measurement that can be identified as
a separate component by trend analysis.
B-5
-------
EXHIBIT 2.2: ROUGHNESS LENGTH (z )
0'
Ground Level
Release
Possible Modeling
Use; _ Lagrangian similarity theory
Basic Measurement Wind speed
Height: Minimum of three heights
between 2-20 m
Location: Representative of the surface
Averaging time: 30 sec
Ins trumerit
response duration: 10 sec
Sampling duration: 1 hour
Accuracy:. 5%
Accuracy of the derived variable; within a factor of 2
Use of the derived variable
in diffusion calculations;
(1) In Lagrangian similarity relations for plume height and ground-level
concentration.
(2) In the similarity relations for u*} HF, L, etc. and in conversion
between P-G stability classes and Rig or L.
Influence of derived variable on the
accuracy of diffusion calculations;
Since only log ZQ £s involved in various similarity relations, diffusion
calculations are not very sensitive to ZQ and its determination within a
factor of 2 may be adequate.
Method of deriving the parameter;
(1) For neutral stability, by fitting the logarithmic wind profile law
i
u = iT ln 7"
K ZQ
to the measurements of u versus z.
c
B-6
-------
(2) An alternative method is to measure the average height (hQ) and
area density of predominant roughness elements and then use an appropriate
empirical relation for determining ZQ. xhis may be preferred for tall
roughness as in an urban setting.
Recommendations;
(1) For determining ZQ from the mean wind profile measurements,
anemometers should be placed above the tops of the roughness elements.
ZQ should be determined separately for different wind directions.
B-7
-------
EXHIBIT 2.3: RICHARDSON NUMBER (Ri)
Ground Level
Release
Elevated
Release
Possible Modeling
Use:
Basic Measurements:
Lagrangian similarity
K-theory
statistical
-tower obs .
of tFw1 & w'T"
(2) Ri^ - tower obs.
of u&T at two levels
(3) Rig - tower obs.
of u&T at h, and T
near surface.
plume rise plus
statistical
K-theory
(1) for Ri'B (local):
T at two levels and u
in between from tall
tower, aircraft, or PBL
model
(2) for Rig (SBL)
tower obs or u&T at h
and T near surface.
Time constant etc: that appropriate for fluctuating or mean quantities
- see Exhibit 2.6 for L, Monin-Obukhov length.
Accuracy of derived variable: derivable from errors in fluctuating or
mean quantities.
Influence on
diffusion:
to obtain oy and oz; K;
or P-G class
Method of deriving
variable
in order of decreasing
theoretical strength, but
increasing practicality;
(2) Ri.
(3) Ri
B
u'w 3u/3z
& Az A0
T 1
T (Au)2
Z10 9(Z10)- 0(22)
r / / w2
(1) in stable PBL case
with low level jet;
Ri'B (local)
(2) Convective SBL:
Ri (SBL)
Recommendations;
(1) The SBL value of the bulk Ri should be used with surface releases.
(2) The local PBL value of the bulk Ri should be used with elevated
releases during stable conditions with low-level jets (LLJ's).
B-8
-------
(3) The SBL value of the bulk Ri should be used to estimate PEL mixing
during convective SBL conditions.
(4) To obtain SBL bulk Ri values, the NWS should measure AT between 2 m
and 10 m.
B-9
-------
Possible Modeling
Use:
Basic Measurement;
EXHIBIT 2.4: FRICTION VELOCITY
Modeling releases within the surface layer defined
by z
L.
is input for Lagrangian similarity
theory or K-theory.
u* obtained from u (10 m) with a drag coefficient
from a detailed field experiment; u^ = Cn u (10 m)
Influence of derived c a
variable on
dispersion d£
calculation; du.
(surface releases)
2: sensitive when u^ is small
u* *
cau*2 (fugitive emissions from saltation)
dc . , ,
-3— a u^; sensitive when u is large.
*
Method of Deriving
Variable:
u*2 a T/p = - u'w' (coiistant flux layer)
(1) Eddy correlation to determine - u'w1
Location: 10 m, Reading interval: 1/5 Sec,
Sampling time: 60 min; Accuracy: 20%
(2) Drag plate measurement of T (shear stress)
Location: surface, Reading interval: depends on
instrument sensitivity.
Sampling time: 60 min; Accuracy: 40%
(3) Profile measurements of temperature and velocity
Location: 2 or 3 levels below 10 m
Reading interval: I min, Sampling time: 60 min
Accuracy: less than direct methods
Recommendations; Profile measurements of temperature and velocity
constitute the most practical method to derive u^ for dispersion
calculations.
B-10
-------
EXHIBIT 2.5: CONVECTIVE VELOCITY SCALE w
Possible Modeling
Use:
Basic Measurement:
Used primarily in modeling pollutant releases into
the mixed layer from sources above 0.1 h (see
Deardorff and Willis 1975, Venkatram 1980)
(1) Surrogate methods are well suited for this
application; e.g., tower 10 m high equipped with
platinum wire thermometer to measure QT> which can
be used to compute QQ.
(2) Wa = p(t) Qm; w* can be estimated from
solar radiation or mixed layer height (see Venkatram
1978).
Influence of derived
variable on diaper-
aion calculation;
c a 1/w* (approximately)
Concentration is sensitive
small .
to
when w* is
Method of deriving
variable:
w* -^- Q0 nc; , v0 - w a. 0
(1) E&iy correlation measurement of w'T1
Location: 10 m, Reading interval: 1/5 sec
Sampling time: 60 min; Accuracy: 15%(?)
(2) Profile measurements of temperature and
velocity
Location: 2 to 3 levels below 10 m, Reading
interval: 1 min.
Sampling time: 60 min; Accuracy.: 20% (?)
(3) Surrogate methods of measuring w'T1
(See Businger 1973 and Wyngaard et al. 1978).
Recommendations; Surrogate methods to measure QQ are well suited for
most applications. Acceptable estimates of w* can be derived from this
Q0. Optical and acoustic remote sensing techniques to measure C^2
and Cn2 are suitable for deriving routine estimates of w*. If
nothing else is available, incoming solar radiation or net radiation can
be used to derive acceptable estimates of w*.
B-ll
-------
EXHIBIT 2.6: MONIN-OBUKHOV LENGTH (L)
Ground Level
Release
Elevated
Release
Possible Modeling
Use;
Basic Measurement;
Height:
Location:
Response time:
Sampling
duration:
Accuracy
Influence:
Method of Deriving
variable:
Similarity Theory Statistical Theory
Wind speed, temperature, and temperature difference
10m
Representative surface
(Primary variables)
(Primary variables)
(Primary variables)
Through profile equations
1 min
3 min - 1 hour
u: 5%; 0.2 m/sec threshold
T, AT: .01°C
(1) direct; L =
kg
(2) indirect; - - Ri
k F
•R
'h h '0'
,ZQ, L , :
7 z '
/h , h, 0
I T~~ -r »
by successive approximation\
zo L L
Rec ommenda t i on s; (1) NWS add AT to routine observation.
(2) Industrial sites: Measure Rig
B-12
-------
EXHIBIT 2.7: HEAT FLUX (H )
y
Possible Modeling
Use:
Basic Measurement;
Height:
Location:
Sampling
duration:
Measurement
Technique;
1
T
PC,
1/3
, Mixed-layer similarity
L » -u*3 pC-T/kgHp Monin-Obukhov similarity
Lower boundary conditions for numerical PBL models
10 m or less
Representative surface
20-60 «in
(1) Eddy correlation Hp = Pcp w'T1 =-pC_ u* T*
Instrument Response: nz/u<3, unstable
nz/u<5-, neutral
nL/u<10, stable
Accuracy: 10%, homogeneous terrain
Comments: requires research grade instrumentation
(2) Dissipation Technique:
_ 2
X
0
kz
where,
1.48 (l-9Z/L)-l/2-> z/L<0
2 (0.74+4.7Z/L>, Z/L>0
a. Direct Dissipation:
^ ™ V -_. / » _ \ »
L0
at
Instrument Response: 103 - 10^ Hz
Accuracy: 20%, limited by knowledge of $x
Comments: Requires highly specialized instru-
mentation and analysis.
B-13
-------
b. Inertial subrange dissipation:
SQQ(kl) = 0.4X0 el/3k! -5/3
Instrument response: 10 Hz
Accuracy: as for 2a
Comments: Requires research grade instrumentation
c. Structure Function
[T(x) - T (x+r).]2 » cT2 r 2/3
XQ- 0.44 CT2 (cv2) 1/2
Instrument response: ?
Accuracy: As for 2a
Comments: May use path averaging of optical
scintillations
3. Profile Method:
U*kz £9
u T a —-—— —
* * *. 3 z
h
_ 01)/(ln(z2/zi) +
Instrument response: 1 min
Accuracy: 20%
Comments: Requires a minimum of 2 temperature
measurements (10 m and 2 m)
(4) Surrogate Method: HF =» f<^T)
Instrument response: as for eddy correlation
Accuracy: depends on knowledge of f(o~)
Comments: Easier than eddy correlation since do
not need to measure w.
(5) Surface Energy Budget:
Soil Heat Flux » Net Radiation + Sensible Heat
Flux + Latent Heat Flux
Instrument Response: 1 min,-
B-14
-------
Accuracy: 20% ?
Comments: Requires measurements or assumptions
equally as restrictive as other methods
Influence of derived
variable on accuracy
of model variables; E(w*) ^ E (Hj-)/3
E(L) ^ E (HF)
where E is "error"
Recommended Method; Profile or surrogate method
B-15
-------
EXHIBIT 2.8: PASQUILL-GIFFORD STABILITY CLASS
Ground Level
Release
Elevated
Release
Possible Modeling
Use;
Basic Measurement;
Height of
Measurement
Location:
Interrogation
Interval:
Time Constant:
Representative
Interval:
Accuracy:
Gaussian type statistical
N/A (Surface variable)
Representative surface
N/A
N/A
1 Hour
Wind Speed: 0.2 m/s +5%
Cloud Cover: 20%
Day: No error
Hour: No error
Ceiling Height: 20%
Alternative -
Upward Heat Flux: 20%
Incoming Solar Radiation: 20%
Accuracy of Derived
Variable; P-G: One class
Smith P: 0.2-0.4 Variations in P (range 0-7)
Influence of Derived
Variable on Diffusion
Calculation*: at 500 m:
Factor of 1.8 to 2.7
at 10 km:
Factor of 1.7 to 5.4
B-16
Factor of more than 100
Factor of 1.7 to more
than 4
-------
Method of Deriving
Variable; Processing of Input Variable Values in Computer
Program
Recommendation;
In general, other more direct measures for estimating dispersion (one for
horizontal, one for vertical) are preferred. However, for the time period
where models that require the P-G class as an input variable are in use,
the F.B. Smith (1972) technique should be used to generate the continuous
variable P (which can be converted to P-G class., if necessary).**
*Using P-G curves
**Beeause of the range of dispersion that may be associated with any given
condition, air quality estimates made using this variable should have
error bounds reported with the estimate.
B-17
-------
3. DISCUSSION OF DERIVED VARIABLES
Wind Sigmas (au> a^, ow, aQ, a^)
The opening discussions of derived variables were focused upon the
standard parameters derived from measurements of wind velocity
components. These include a , a , a , o_, and o, , which are
often treated as if they were direct measures of turbulence, rather
than as statistically derived parameters. The relationships between
the rectilinear components and the magnitude of the mean wind vector
u and the polar component forms are:
a = a ~ u
u v ~ r
°w = ur °*
Studies of 0.2-second observations for 40-minute periods by Weber
et al . (1975) show that for wind speeds between 2 and 10 m/sec, the
following correlations were very good (regardless of stability):
°/a vs u and a/a vs u*
On the other hand, 0 Afy or a /a appears to be a weak
function of stability:
o^/o =1.5 for stable, a /a ^1 for nonstable.
aA is most important near elevated sources and for evaluation of
extreme cases which are of importance in regulatory and operational
decisions. For predicted averages or modeling ensembles, the u*
estimate is a more effective parameter to establish than a .
B-18
-------
The adequacy of all of these relationships for wind speeds less
than 2 m/sec was questioned. The use of velocity component range to
estimate values for any of the c's was recommended only as a last
resort (when specific statistical calculations were impracticable).
The ratio of the range to the standard deviation has been found to be
a function of both stability and measurement height. This suggests
that large errors may result unless corrections are undertaken. A
range estimator is sometimes, however, the only practical method when
large quantities of historical strip chart data are the only records
of wind variation. Use of Pendergast and Crawford (1974) results can
help reduce the error in this process.
A question about the importance of sensor response to the
precision of the concentration estimate to be derived from use of
these o's was raised. As long as the sampling time exceeded the plume
travel time to the receptor of interest, the uncertainty introduced by
smoothing the high frequency contributions is not often significant
for Gaussian model estimates. Where the high frequency component is
of particular interest, a synthetic restoration technique (e.g.,
Pasquill 1976) may be used.
The question of recording both mean wind component values and
turbulence parameters for a time interval shorter than an hour was
considered. Comparison of exposures due to, and with historical
modeling of, puff releases as well as for releases under low wind
speed conditions with wide ranging wind direction changes (Sagendorf
and Dickson 1974) have shown the utility of these intermediate
averages. A 3-minute period was tentatively suggested as a useful
duration. This corresponds to the period for which the a
predictions were made in the studies leading to the original P-G
dispersion curves (Pasquill 1976). Specific applications and
intrumentation will determine the selection of this intermediate
period for particular sites. The 3-minute period represents a
compromise that allows an adequate number of high frequency digital
samples (of order 1 to 0.1 Hz) to be used in the computation of this
intermediate mean and variance so that sampling errors are often
tolerable. In addition, hourly averages can be simultaneously
recorded or can be estimated from 20 of these ^-minute values.
B-19
-------
The relative value of measuring a or a4, rather than
estimating it from heat flux or other surface boundary layer (SBL)
parameters was discussed. The reliability of sonic anemometry was
discussed in this regard; it appeared that for a (single vertical
w
axis) measurements, the reliability of sonic measurements is not yet
equivalent to that of mechanical wind sensors.
The Lagrangian time scale is recognized to be important
(especially with increasing time of travel), however, at present there
is a large difference in the value as determined from diffusion or
turbulence measurements. Until this difference is resolved, it may
not be appropriate to perform routine estimates. Further research is
needed to determine the ultimate reliability and usefulness of
turbulence-derived estimates of this parameter.
Roughness Parameter (z.)
Q_
Together with u^> ZQ incorporates the effect of the
small-scale surface irregularities on turbulent transfer processes in
the atmospheric surface layer. It is usually involved in the integral
forms of the flux-profile relations in the surface layer. In
diffusion models, only the Lagrangian similarity theory relations
explicitly involve ZQ or rather log ZQ.
Since z is basically defined as a parameter in the logarithmic
wind profile law
) (B-l)
for a neutral surface layer over a hompgeneous surface, it is best
determined from wind profile measurements in the part of the surface
layer above the tops of the various roughness elements.
An indirect way of determining ZQ) which does not require wind
profile measurements, is through the use of previously developed
empirical relations between z and 3Ome of the measurable physical
characteristics of the surface, such as the mean height (h ) and
area density of the predominating roughness elements. For example, on
B-20
-------
the basis of profile derived z values for various types of natural
surfaces, Lettau has proposed an empirical relation
log ZQ - 1.19 log hQ-1.24 (B-2)
in which z and h are both expressed in cm. According to
Equation B-2, the ratio ^Q/zQ varies from 4 for tall forests to 27
for sands. Actually, z also varies with the area density and shape
of roughness elements and, for very flexible roughness, on the mean
wind speed near the top of the elements. But these factors are
implicitly involved if one is using an empirical relation of the type
in Equation B-2 based on the actual observations involving the same
type of surface elsewhere.
There are not enough direct (based on wind profile) measurements
°f Z0 over urban areas, from which an empirical relation of z- as
a function of h and area density (A) can be obtained. Some
guidance is given by wind tunnel measurements by Counihan (1971) using
Leggo blocks as roughness elements (model buildings). His results
show that ZQ/h0 when plotted as a function of A, at first
increases, attains a maximum value for certain density, and decreases
monotonically with the further increase in A. More measurements of
this type are needed for other types of roughness elements. These can
be used to test some of the relations that have been proposed (Lettau
1969; Businger 1974) on the basis of purely geometrical considerations.
The foregoing discussion mainly applies to a more or less flat
and homogeneous surface. Many natural surfaces of interest in
diffusion modeling are not homogeneous. In a strict sense, z may
not be defined if the wind profile in the neutral surface layer
deviates too much from its logarithmic fora. ' When the roughness
characteristics of an inhompgeneous surface are to be expressed in
terms of single length parameter, for the sake of simplicity, an
effective roughness z^ n,ay be defined. For example, Fiedler and
Panofsky (1972) defined z as "that roughness length which
homogeneous terrain would have in order to produce the correct
space-average downward flux of momentum near the ground, with a given
(space-averaged) wind." They suggested that an/Instrumented aircraft
B-21
-------
could be used for measuring the area-averaged u and u for
determining z . The above concept of effective roughness can be
used for a rolling terrain in which the topographical features are not
more than 100 m or so. The roughness parameter z or z would
0 Oe'
not be relevant in a more complex (hilly) terrain. For that matter,
the similarity theory relations in which this parameter is involved
would not be valid either.
In some diffusion calculations, a water (large lake or sea)
surface may be involved. Unlike that over a land surface, z £s a
variable parameter which depends on the wave field, which in turn
depends on the stress (T) on the surface or u fetch etc. For a
fully developed wind wave field (waves in equilibrium with the
prevailing winds), Charnock (1955) proposed the relation
2
. - bi-
o g
where b is a constant (b = 0.02). From a large number of observations
over the oceans, the above relation has been shown (Garratt 1977) to
work well within a factor of two. It has the simplicity of not
requiring any measurements of the sea state (wave height, phase
velocity, etc.).
Richardson Number (Rif> Ri, , Rip)
The vertical gradient of potential temperature is sometimes used
to determine atmospheric stability for diffusion computations. This
parameter, however, is not a complete description of atmospheric
stability as it ignores the effects of wind shear. The nondimensional
flux Richardson number Rif> which arises from the turbulent energy
equation, however, does incorporate both effects and is given by
B-22
-------
Rif- c T T: „ (B-4)
p
This parameter represents the ratio of the rate of dissipation of
turbulent kinetic energy by buoyancy to its rate of production by the
shear stress. Its sign is dependent on that of H_ and it is
positive/zero/negative in stable/neutral/unstable atmospheres,
respectively.
Evaluation of Ri, in the surface boundary layer (SBL) requires
measurements of the turbulent heat flux Hp, the Reynolds shearing
stress T , as well as the vertical wind shear 3u/3z. Of course, the
vertical heat flux (a'w'T1 ) can be parameterized by U^T* and the
shearing stress (au'w1) can be parameterized by u^, if these
parameters can be obtained by some other means.
If gradient transport theory is used, S.i, can be given by
Ri . (B-5)
but this form requires knowledge of KTI/KM which is somewhat
difficult to determine, as is discussed below. Thus the gradient form
of the Richardson number defined as
Ri - Ri. i -f 3Q/aZ , (B-6)
f *H T Ou/3z)2
is in a form which can be evaluated by observations of T and u at two
levels of a. tower from the finite difference form of the Richardson
number Ri given by
o
Ri. - | , AZ (B-7)
A T (Au)2
B-23
-------
However, small errors in the velocity observations, especially in low
speed conditions, can lead to erroneously large values of Ri, and thus
the bulk form of the Richardson number Ri £3 evaluated from
B
. A free.) - e(z))] (M)
8 T
where h is the depth of the SBL or convective boundary layer and Z is
some convenient level near the surface, such as 2 m. For surface
releases it is recommended that Ri be used as a stability parameter
which can be used to evaluate the Monin-Obukhov length in a SBL (Irwin
and Binkowski 1980).
In modeling elevated releases in the region of the planetary
boundary layer (PBL) above the SBL, Ri. has been used to estimate the
local stability of the PBL even though there is no theoretical basis
for using Ri in the PBL. In addition, problems arise in the
evaluation of Equation B-7 in the area of low level jets associated
with surface and elevated stable layers, as in this region the shear
goes to zero and Ri becomes an extremely large positive value.
However, observations of the mixing in the region of the low
level jets has shown that mixing is great in this area. Similar
results have been found using turbulent closure models. Thus perhaps,
the following form of Ri could be_ used to evaluate the stability of
the PBL under these conditions:
RiB
where u £s the wind halfway between the levels of the two
temperature measurements. This situation could arise during lofting
plume 'situations.
A problem also arises during daytime convective conditions when a
constant potential temperature PBL caps a superadiabatic SBL. Use of
Equation B-7 to represent the local stability of the PBL will give a
B-24
-------
Ri=sO, but convective elements passing through the PEL from the SBL
will be causing a great deal of mixing. Thus, the amount of mixing in
the PBL under these conditions might better be estimated from the SBL
value of Ri ag determined from Equation B-8. This situation could
B
arise during fumigatron conditions.
Friction Velocity (ui)
u* is the relevant velocity scale for modeling vertical
dispersion in the surface layer, i.e.,z»0(L). However, it is
not clear that u^ determines the crosswind spread in the surface
layer. For emissions generated by saltation of loose particles on
roads or tailings, u determines the emission rate. There is some'
3
evidence to show that the emission rate is proportional to u^.
Eddy correlation techniques to measure -u'w' can be used in field
studies. For operational use, profile measurements of temperature and
wind at two or three levels below 10 m can be used to derive
acceptable estimates of u
Convective Velocity Scale (w.)
w* determines vertical and horizontal spread of plumes released
in the mixed layer. There is some evidence to show that a in the
v
surface layer is also proportional to w
As w^ a QQ' , surrogate methods to estimate QQ, the
surface temperature flux, are well suited for most applications. For
example, the relationship between a and Qn in the unstable
boundary layer can be exploited for deriving acceptable values of
w*. Even an aircraft equipped with a simple temperature sensor can
be used to measure a which is related to Q by
where z £3 the altitude of the aircraft traverse, z should be
a j, a
less than 0.1 h for the relationship to hold. This method of
B-25
-------
measurement (field studies only) has the advantage of being able to
yield an area averaged QQ or w^ which is what an elevated plume
"sees."
Recent studies indicate that acceptable estimates of Qn and
hence w^ can be obtained from rugged optical and acoustic sensors to
measure c£ and C2.
T n
Simple modeling (see Venkatram 1978) shows that w^ can be
expressed as
AQ 1/2F
m
(f)
where 0 is the maximum surface temperature flux whose half period
is IT. The variable A varies within very narrow limits and can be
essentially treated as a constant. This is very useful as we can
estimate w from 0 with little knowledge of other boundary layer
ul
variables. 0 can be estimated from surrogates such as the incoming
solar radiation if direct measurements are not available.
The measurement of h has been discussed by another group.
Monin-Obukhov Length (L)
Within the surface layer, the dynamic effects of heat cannot be
neglected as in the dynamic sublayer beneath it. Consequently,
additional parameters are available to form velocity, length, and
temperature scales. The parameters can be combined to form a length
scale L = -(u^T)/(k g w'T1), the Monin-Obukhov length.
Similarity theory based on this scale and u^ and T^ indicates
that nondimensional profiles of velocity, temperature, and local
statistical quantities should be universal functions of the
nondimensional height, z/L.
iy
Since the length scale may also be defined as L = (u£ T)/
(k g T*), its value can be computed from basic turbulence
measurements used to evaluate shear stress and heat flux. This is the
direct measurement method.
B-26
-------
The indirect computation relies on measurements of wind speed and
temperature (as well as temperature difference) in the surface layer.
Specifically, the bulk Richardson number is used in conjunction with
known integrals from the profile laws of similarity theory (Paulson
1970; Nickerson and Smiley 1975; Bemoit 1977) through the relation:
* - Ri —
L R1B G
where
L' L u*
* h fO, . A0
ST~» 7> ^' r^
Z0 L L T*
and where h is layer over which Ri £3 measured. The solution for L
B
is obtained through successive approximations, and current techniques
show (Irwin et al. 1980) that the indirect procedure produces an
estimate within 25% of that obtained using the direct methods for L>h
with research grade instrumentation (i.e., Kansas, Minnesota). When
L
-------
Gifford (1961) modified Pasquill's plume spreading data into a
family of curves of standard deviation o and cr t also expressed
y z
as a function of the same stability categories and distance from the
source. The P-G curves were later published by Turner (1970) in a
workbook on dispersion estimates.
With the rapid increase in application of this popular scheme for
obtaining a and o particularly after the introduction of the
"STAR" computer program, there also appears to have been a tendency
among the users to overlook the limitations of this system imposed by
the original data base. The original Pasquill curves had a rather
limited data base, dealing with vertical plume spread data at
distances less than 1 km and with speculative extrapolations for
stable conditions at distances greater than 1 km. The intended use of
the Pasquill curves were to provide tentative estimates of dispersion
parameters in the absence of turbulence-related observational data
(wind fluctuations).
Because of the increased use (and misuse) of the P-G
classification system in critical environmental impact evaluations,
the system has been subjected to increased scrutiny by the technical
and legal community. This workshop represents the fourth scientific
group which has conducted a critical review of the applicability of
the P-G curves and classifications during the past 3 to 4 years. A
summary of some recommendations which have emerged from these meetings
are given below.
At the 1977 Chicago Specialists Conference on EPA Modeling
Guideline (Roberts 1977), the working group II-4 stressed that a
number of questions regarding the use of P-G curves and stability
classification deserve serious attention and systematic study.
Significant discrepancies were noted between P-G stability categories
and those derived from tower AT. The latter was recognized as not
being representative of the key factors responsible for turbulence.
The P-G curves of o were also questioned strongly with suggestions
Z
that for tall stack emissions, stability A and, to some extent,
stability B curves may result in large overestimates of short-term
ground-level concentrations. It was also recommended that "STAR"
B-28 f
-------
program stability classifications must be supplemented by detailed
specification of vertical wind and temperature profiles, including
mixing height data.
At the 1977 AMS Workshop on Stability Classification Schemes and
Sigma Curves (Hanna et al. 1977), recommendations were made that a
y
should preferably be estimated from wind direction fluctuation o., as
6
recommended earlier by Pasquill (1961). Adjustments should also be
made to account for surface roughness and urban heat island effect by
changing the stability class by one unit in the direction of unstable
to account for these effects. Regarding cr it would be preferable
z
to use cr, as a measure of vertical dispersion; AT is only useful in
stable conditions and only'when measured sufficiently near the surface.
The EPA 1979 Workshop on Atmospheric "Dispersion Models in Complex
Terrain (Hovind et al. 1979) expressed the concern of determining
terrain effects upon a and a . Regarding future attempts to
improve Gaussian models in complex terrain applications, it was
suggested that on the basis of field studies a matrix be developed
which will consider effects of changes in model variables, such as P-G
stability classification, upon proposed improvements in the models.
Rather than rely on the single P-G stability class for both
horizontal and vertical dispersion estimates, separate dispersion
related variables for horizontal and vertical dispersion are disoussed
elsewhere in this document. However, before models which use these
<
variables directly become available as "guideline" models for
regulatory purposes, it is still desirable to have an hourly P-G
stability class so that current models may be used. Also, after new
models become available, it may still be desirable to have input
information for old models so that they may be executed for comparison
purposes.
The seven class stability value can be inferred from hour of day
and day of year (so that solar elevation angle can be obtained), and
from hourly observations which include cloud cover, ceiling height,
and 10 meters wind speed. Pyranometer measurements can replace the
cloud observations and solar angle during the day.
B-29
-------
An alternative has been suggested by F. B. 'Smith (1972) which
determines a continuous stability variable, P, which can have real
values between 0 and 7. Any value of this variable- can be converted
to the 7 class stability value above. Variables required are 10 meter
wind speed and upward heat flux. Alternate variables for the upward
heat flux are incoming solar radiation by day and cloud amount at
night.
B-30
-------
APPENDIX B
REFERENCES
Benoit, R. 1977. On the Integral of the Surface Layer
Profile-Gradient Functions. J. Applied Meteorology,
^6(8):859-860.
Businger, J. A. 1973. Turbulent Transfer in the Atmospheric Boundary
Layer. Workshop on Micrometeorology. D. A. Haugen, ed., American
Meteorological Society.
Businger, J.A. 1974. Aerodynamics of Vegetated Surfaces. Heat and
Mass Transfer in the Bioshpere, Vol. I: Transfer Processes in
the Plant Environment. Scripta Book Co. Washington, D.C.:
Scripta Book Co.
Charnock, H. 1955. Wind Stress on a Water Surfce. Quart. J. Roy
Meteor. Soc., 8^; 639-642.
Counihan, J. 1971. Wind Tunnel Determination of the Roughness Length
as a Function of the Fetch and the Roughness "Density of
Three-Dimensional Roughness Elements. Atmos. Environ., ^:637-642.
Deardorff, J. W. and G. W. Willis 1975. A Parameterization of
Diffusion into the Mixed Layer. Journal of Applied Meteorology,
14:1451-1458.
Fiedler, F. and H.A. Panofsky 1972. The Geostrophic Drag Coefficient
and the Effective Roughness Length. Quart. J. Roy. Meteor. Soc.,
98:213-220.
Garratt, J.R. 1977. Review of Drag Coefficients Over Oceans and
Continents. Mon. Weath. Rev., 105;915-929.
Gifford, F. A. 1961. Use of Routine Meteorological Observations for
Estimating Atmospheric "Diffusion. Nuclear Safety, 2^:47.
Hanna, S. R., G. A. Briggs, J. Deardorff, B. A. Egan, F. A. Gifford,
and F. Pasquill 1977. AMS Workshop on Stability Classification
Schemes and Sigma Curves - Summary of Recommendations. Bulletin
of the American Meteorological Society, 58:1305.
Hovind,'E. L., M. W. Edelstein, and V. C. Sutherland 1979. Workshop
on Atmospheric "Dispersion Models in Complex Terrain. U.S.
Environmental Protection Agency, Research Triangle Park, NC. EPA
600/9-79-041.
Irwin, J. and F. Binkowski 1980. Estimation of the Monin-Obukhov
Scaling Length Using On-Site Instrumentation. Atmos. Environ.,
in press.
Lettau, H. 1969. Note on Aerodynamic Roughness - Parameter Estimation
on the Basis of Roughness Element Description. Journal of
Applied Meteorology, JJ:828-832.
B-31
-------
Nickerson, E.G. and V.E. Smiley 1975. Surface Layer and Energy Budget
Parameterizations for Mesoscale Models. J. Applied Meteorology,
14(3):297-300.
Pasquill, F. 1961. The Estimation of the Dispersion of Windborne
Material. Meteorological Magazine, ^2 (1063):33-49.
Pasquill, F. 1975. The Dispersion of Material in the Atmospheric
Boundary Layer - the Basis for Generalization. Lectures on Air
Pollution and Environmental Impact Analyses, Sept. 29 - Oct. 3.
American Meteorological Society.
Pasquill, F. 1976. Atmospheric Dispersion Parameters in Gaussian
Plume Modeling. Part II. Possible Requirements for Change in
the Turner Workbook Values. EPA-600/476-030b. U.S.
Environmental Protection Agency, Research Triangle Park, NC.
Paulson, C. A. 1970. The Mathematical Representation of Wind Speed
and Temperature Profiles in the Unstable Atmospheric Surface
Layer. Journal of Applied Meteorology, 9_:857-861.
Pendergast, M. M. and T. V. Crawford 1"74. Actual Standard Deviations
of Vertical and Horizontal Wind Direction Compared to Estimates
from Other Measurements. Proceedings of Symposium on Atmospheric
Diffusion and Air Pollution. American Meteorology Society.
Roberts, J. J. 1977. Report to the U.S. Environmental Protection
Agency of the Specialists' Conference on the EPA Modeling
Guidelines. U.S. Environmental Protection Agency, Research
Triangle Park, NC.
Sagendrof, J.F. and C.R. Dickson 1974. Diffusion Under Low Wind
Speed, Inversion Conditions. NOAA Technical Memo. ERL-ARL-52.
Air Resources Laboratory, Idaho Falls, Iowa.
Smith, F. B. 1972. A Scheme for Estimating the Vertical Dispersion
of a Plume from a Source near Ground Level. Proceeding of the
Third Meeting of the Expert Panel on Air Pollution Modeling
NATO/CCMS Report No. 14. Brussels.
Turner, D. B. 1970. Workbook of Atmospheric Dispersion Estimates.
Publication No. AP-26. U.S. Environmental Protection Agency,
Research Triangle Park, NC.
Venkatram, A. 1978. Estimating the Convective Velocity Scale for
Diffusion Applications. Boundary-layer Meteorology, 15:447-452.
Venkatram, A, 1980. Dispersion from an Elevated Source in a Convective
Boundary Layer. Atmospheric Environment, 14:1-10.
B-32
-------
Weber, A., J.S. Irwin, J.P. Kahler, and W.B. Peterson 1975.
Atmospheric Turbulence Properties in the Lowest 300 Meters. U.S.
Environmental Protection Agency Monitoring Series. EPA
600/4-75-004, July, pp. 153.
Wyngaard, J.C. and F.F. Clifford 1978. Estimates of Momentum, Heat,
and Moisture Fluxes for Structure Parameters. J. Atmosperic
Sciences, 35:1204-1211.
B-33
-------
APPENDIX C
MEASUREMENT TECHNIQUES - METEOROLOGICAL VARIABLES
CONTENTS Page
1. INTRODUCTION C-2
2. SURFACE AND TOWER MEASUREMENTS C-9
2.1 Wind Speed and Direction C-9
2.2 Temperature C-ll
2.3 Humidity C-12
2.4 Pressure C-14
2.5 Precipitation C-14
2.6 Solar Radiation C-15
3. REMOTE MEASUREMENTS C-18
3.1 Convective Mixed Layer Height C-18
3.2 Mechanical Mixing Layer Height C-21
3.3 Flow Field Aloft C-22
REFERENCES C-25
PANEL C MEMBERS
Chairman: Thomas Lockhart
Robert Beebe
James Dicke
Peter Finkelstein
Benjamin Greene
Chandran Kaimal
Paul MacCready
Daniel Mazzarella
James McTaggart-Cowan
William Petersen
Robert Swanson
C-l
-------
1 . INTRODUCTION
Two general instrument classifications are made: conventional
tower or mast-mounted instrument and the remote instrument or
non-ground-attached system (balloons, aircraft, etc.). The following
variables were considered.
Scalar mean, harmonic mean, and vector mean wind speed
(us, uh, uv) (m/s)
Wind direction (6,) (degrees arc)
Wind components
along mean wind component (u) (m/s)
crosswind component (v) (m/s)
vertical component (w) (m/s)
Temperature (T) (°C)
Temperature difference (AT) (°C)
Dew point temperature (T,) (°C) or relative humidity
Radiant flux density (Q*) (W/m2)
Atmospheric pressure (p) (kPa)
Precipitation (P) (mm)
Convective mixing layer height (h ) (m)
Mechanical mixing layer height (h ) (m)
The following conclusions apply to measuring each of these variables:
1) Data should represent a 60-minute interval labeled by the
ending time (Julian date, hours, and minutes in local
standard time in the range 0001 to 2400).
2) The number of samples taken and the process by which they
are combined are functions of the variables and the data
application (described with each variable). No less than
half the prescribed number of samples may be used to
represent the hour.
C-2
-------
3) Siting of the instrument is often the most important
consideration and should be made by an experienced
meteorologist.
4) A vigorous quality control and assurance program is required
to achieve valid data to the specifications described below
and in a sufficient quantity to represent long periods of
time (months to a year) without bias.
In addition to these general recommendations, a summary of
measurement-specific information is presented here by variable
category. The summaries include acceptable levels of accuracy,
resolution, and response time which are considered both achievable and
sufficient for hourly dispersion applications. Detailed discussions
of the measurement techniques are addressed in Sections 3 and 4 for
surface and tower measurements, and remote measurements,
respectively.
Flow Measurements
Tower and mast-mounted (10 m is standard height for wind
measurements):
Anemometers (cup or propeller) (for ug, uy, u^, u, v, w)
Starting Speed (within accuracy) <0.5 m/s
Distance Constant (632 recovery) < 5 m
Accuracy <0.2 m/s +57. of speed
Wind Vanes (for 0, , cr , a, )
t) T
Starting Speed (10° displacement) < 0.5 m/s
Damping Ratio >0.4
Delay Distance (50% recovery) < 5 m
Accuracy < 5°
Resolution < 1°
03
-------
Sampling
Sixty or more samples will estimate the mean within 5
to 10%.
Three hundred and sixty or more samples will estimate
the standard deviation within 5 to 10%.
Note: the effective resolution of 360 samples of a
1° measurement resolution is 0.05° =
1
Also the resolution is the best
estimate of accuracy of a relative measurement such as
the distribution about a mean.
Computations
Wind Speed (horizontal)
Both the scalar mean (u ) and resultant vector
s
average (u ) should be reported. The harmonic
mean (u, ) is preferred for initial dilution and
plume rise calculations.
V
«— N ' _1
- (1 y I }
v N ifei u ;
s >
1
Wind Direction (horizontal)
The directions should be reported as the hourly
resultant vector direction
9 = arc tan v/u when u is oriented north
C-4
-------
Standard Deviations
For any variable, X, representing one of at least
360 samples of horizontal wind direction (0), vertical
wind direction (), vertical wind speed (w), scalar
wind speed (u ), or the reciprocal of the scalar wind
s
speed (1/u ) the standard deviation may be calculated
from
1/2
Note that the 360° to 1° crossover in 9 must be
accounted for when d is computed from a single vane
system.
When digital methods are used, the accuracies mentioned
above will be achieved. The adequacy of analog
computers for this purpose was not estimated.
Indirect Measurements
The only commercially available monitoring system for indirect
flow measurements is the Dbppler acoustic radar. Other techniques are
used experimentally. Accuracy of the 20-minute mean orthogonal
components u, v, and w is estimated at 0.5 m/s. Resolution for a
w
is 0.1 m/s. Reliable range is 50 m to 600 m with 1,000 m exceeded
under ideal conditions. Height resolution is 30 m.
C-5
-------
Temperature Measurements
Ambient Temperature (T)
When a good-quality aspirated radiation shield is used
measurements with a variety of transducers can be calibrated and
operated to provide
Accuracy <0.5°c
Resolution <0.1°c
Time Constant (63%) ^1 minute
Temperature Difference (AT or T - T ,)
z2 zl
Two identical aspirated radiation shields equally exposed will
provide a AT (a relative measurement) value to accuracies proportional
to the effort exerted in selection and calibration. Resolution should
be one-fifth the accuracy.
Readily Research
Obtained Effort
Accuracy <0.1°c <0.05°C
Resolution <0.02°c <0.01°C
'Time Constant (63%) 1 minute
Dew Point Temperature (T )
d
Humidity may be measured in a variety of ways. Methods other
than dew point measurements should meet the equivalent dew point
accuracy stated here. The saturated salt technique is acceptable for
all dispersion applications. It provides:
Accuracy < 1.5°c
Range - 30°c to +20°C
Resolution < 0.1°c
Time Constant (63%) ^ 30 minutes
c
An accuracy of l°c will probably result from careful use.
C-6
-------
Sampling
The sample averaging time need be no shorter than half the time
constant. Shorter intervals may be used for conformity with other
variables.
Solar Radiation
A measure of global sun and sky radiation can be made using
good-quality pyronometers which provide the following performance:
Accuracy < 50 w/m2
Resolution < 10 w/m2
Time Constant (63%) ^ 5 sec
Atmospheric Pressure
The standard atmosphere pressure for the station elevation will
often be of sufficient accuracy to represent true pressure for
dispersion applications.
Accuracy £1 kPa (10mb)
Resolution <0.2 kPa (2 mb)
Precipitation
Measurements are highly sensitive to wind speed. Use of a shield
and placement in a sheltered location is highly desirable.
Accuracy 1-6% depending on precipitation rate
Resolution 0.2 mm water equivalent
C-7
-------
Height of the Mixed Layer
Convective Mixed Layer
From the position of a monitoring requirement the monostatic
acoustic radar (or Doppler version) was considered as the only
technique commercially available.
Accuracy ~50 m
Resolution ~10 m monostatic radar
~30 m Doppler
Range 50 m - 500 m usually
50 m - 800 m often
50 m - 1,000 m sometimes
Time Constant ~ 10 sec monostatic radar
~ 20 minutes Doppler
Mechanical Mixed Layer
Since the range of this phenomenon is approximately 0-200 m, it
may sometimes be estimated from tower data. Remote techniques may be
used (such as acoustic radar) if they are modified for limited range.
Such instruments are not currently available commercially.
C-8
-------
2. SURFACE AND TOWER FLOW MEASUREMENTS
2.1 Wind Speed and Direction
Wind speed and direction, or wind velocity (a vector) can be
measured with accuracy acceptable for dispersion modeling with
commercially available instrumentation.
Surface wind velocity measurements should be measured at 10 m.
The area surrounding the instrument should be clear of obstructions to
flow for a distance equal to 10 times the height of such
obstructions.
Wind speed and direction instrumentation may also be mounted on
open towers at any desired height. The instrument should be mounted
on a boom which holds it at a sufficient distance from the tower such
that it is outside the major influence of the tower. A useful
rule-of-thumb for this is a distance of 2 times the diameter of the
tower. Instruments should not be mounted on chimneys or other solid
structures.
Satisfactory wind speed and direction measurements can be made
using cup anenometers and wind vanes, wind vanes with propeller
anemometers mounted on the front, or with two or three (for vertical
wind speed) propellers mounted on orthogonal axes.
Wind vanes should have a starting speed of 0.5 meters per second
or less, a damping ratio of at least 0.4,
and a delay distance of 5 meters or less, which refers to responses
resulting from 10° deflections. For such vanes, an absolute
accuracy of 5° with a resolution of 1° can be expected with
reasonable care and an acceptable quality assurance program.
Anemometers should have a starting speed of 0.5 meters per second
or less, and distance constant of 5 meters or less. An accuracy of
0.2 m +5% of the measured value can be expected with an acceptable
quality assurance program.
C-9
-------
For better definition of the flow field, the use of several
appropriately sited sets of instruments is highly recommended.
With normal atmospheric variability, a set of 60 samples should
give a mean value within 5 to 10% of the true value. Approximately
360 samples are required for a similar accuracy for standard
deviations.
Wind speed should be reported as vector, scalar, and harmonic
means. The difference between vector and scalar means should be noted
to indicate major fluctuations in the wind directions over the
sampling period. Wind directions should be reported as the vector
mean direction.
Digital computation is the preferred method for determining
standard deviations of wind speed and direction. The adequacy of
analog computers for this purpose cannot be estimated by the panel.
Vertical Velocity
Vertical velocity is measured most conveniently with a vertical
propeller anemometer, mounted apart from any other instrumentation.
Because the instrument is often operating at the lower extreme of its
range, and because the elevation angle of the wind vector is
frequently slight, meaning that propellor is operating at yaw angles
where it is least accurate, the accuracy of the vertical velocity
measurement may be poor. (Careful alignment of the propeller axis
with respect to the vertical is therefore imperative.) Bivanes may be
used to measure the elevation angle, but they require frequent
maintenance and calibration, do not give reliable information during
precipitation, and are, in the opinion of the panel, more of a
research than an operational instrument.
Sonic, hot film, and other such anemometers should also be
considered research tools, excellent for micro-turbulence studies, but
not desirable or practical for air pollution diffusion studies at this
time.
The recommendation of the panel to the NWS is that they adopt the
F460 anemometer and vane system as soon as practical.
C-10
-------
2.2 Temperature
The state-of-the-science for temperature measurement independent
of the medium is well advanced. Good quality linear thermistors or
platinum resistance devices are quite adequate and cost effective for
measuring this variable in air. One could use quartz crystal
thermometers if greater transducer accuracy were desired, but they are
relatively expensive and not as reliable for long-term use, making
them more suitable for research purposes. Thermocouples may be
suitable for ambient temperature measurement, but are not recommended
due to the increased difficulty in operating them. In any case, it
should be recognized that the limitations placed on the accuracies for
T or AT stem primarily from the difficulties in adequately shielding
the sensor.
Ambient temperature can be measured to an accuracy of 0.5°C
with a resolution of 0.1°C, provided a good-quality aspirated
radiation shield is used. Experience has shown that under all
conditions, a downward-facing shield with the sensor unable to sense
the ground but being in a flow in excess of 2.5 meters per second"!
provides for the greatest accuracy (McTaggert-Cowan and McKay 1976).
The sensor should be lagged to provide a time-constant of
approximately 1 minute when in the shield-flow. If this is done, a
60-minute average value is obtainable with a minimum of 60 equally
spaced samples. In maintaining consistency with World Meteorological
Organization (WHO) standards for ambient temperature, we recommend
that the shield be located at the 2 m height. It is recognized that
some models may be based on temperatures at 10 m but this value may be
readily calculated by using the more critically needed variable of AT
measured between 2 and 10 m.
AT should be measured with a completely separate system from T.
Two and ten-meter heights for the shields are considered to be the
standard elevations for this measurement in addition to being the most
readily available, although it is recognized that in some applications
the lower-level shield may be located at 10 m or any other elevation.
If this latter situation occurs, it must be recognized that the
accuracy and resolution figures below must be related to a
C-ll *
-------
considerably greater shield-separation due to the less rapid change of
temperture with height at that level. The accuracy and resolution
that can be readily obtained for AT are 0.1°C and 0.02°C,
respectively. Considerable care will allow these measurements to be
reduced to 0.05°C and 0.01°C. The time constant and sampling
interval are the same for AT as for T. Because of this, aTj a
variable required in some models, cannot be meaningfully measured with
the above T or AT systems. A much shorter time-constant device will
be required which would not be suitable for standard network use but
could be used in research applications. Hot-wire probes, sonic
thermometers, platinum wire systems, and small bead thermistors would
be most likely choices in this case (Kaimal 1975).
Siting for temperature variable measurements must be done
carefully. Care must be taken to maintain the integrity of ambient
environment, particularly the surface very near the tower.
To maintain precision and accuracy in these systems, a rigorous
quality control program is required. This program should consist of
frequent operational checks and routine shield maintenance and sensor
calibrations dependent on the environment at the site.
2.3 Humidity
Humidity, defined as vapor pressure, represents a variable
constituent of air. By volume, water vapor can vary from 0 to 4%.
Relative humidity, one of the terms that describe humidity, is a
dimensionless number determined by dividing the vapor pressure by the
saturation vapor pressure and multiplying by 100 to obtain a
percentage. Dew point, another term describing humidity, is defined
as the temperature to which air must be lowered in order for it to be
saturated; it is the preferred measurement for tower installations.
For a 10 m or taller tower, the present choice for a dew point
sensor is between the dew-cell technique and the cooled-mirror
technique. A measurement accuracy of j^l.5°C over the temperature
range of -30° to 20°C is recommended. The time constant of the
system can be anywhere from 1 to 30 minutes, but for some applications
such as fog formation in cooling tower studies, it may be necessary to
f
C-12
-------
use an instrument with a time constant considerably shorter than the
30-minute maximum.
The dew cell, a generic term that describes an instrument which
operates on the principle of the saturation vapor pressure of lithium
chloride, is moderately priced and less complex than the cooled-mirror
device. It should be installed in a weather shield that offers
protection from solar radiation and wind. Some miniaturized sensors,
designed without sufficient protection from the cooling effect of wind
or the heating effects of solar radiation are susceptible to large
errors. Lithium chloride, the salt used in this sensor, undergoes
recognizable phase changes (Acheson 1963), but over the range of most
interest, these are not too troublesome. At some locations where
atmospheric pollution is high from "poisoning" chemicals, the dew cell
error increases. Frequent cleaning and recharging with fresh lithium
chloride solution is essential. Typically, the dew cell, when
equipped with a 100-ohm resistance thermometer element, has a time
constant of 12 to 15 minutes, but miniaturized sensors with
thermistors are much faster.
Cooled-mirror dew point instruments offer greater accuracy and
faster response characteristics. Typically, the accuracy is ^0.5°C
while the resolution is 0.1°C. Unlike dew cell sensors which the
user must house in a weather shield, cooled mirror devices are
supplied in a housing and may include sensors for both dew point and
temperature. The latter must not be used for T or AT measurements,
but may be used for calculating relative humidity. In some cooled
mirror instruments the "standardizing" process is automatic. This
procedure involves heating the mirror surface periodically to clear it
of frost and dew deposits. In others standardizing is a manual
operation. Occasional cleaning of the mirror surface is necessary,
especially in polluted atmospheres. The cooled mirror technique is
considered a fundamental measurement technique.
Psychrometers which incorporate a dry bulb and a wet bulb are a
basic measurement technique. For checking, the sling psychrometer, or
better, the motor-operated psychrometer is an easy-to-use instrument
but requires some skill and care. The motor-aspirated unit of the
Assman type can be obtained with specially calibrated thermometers
r
C-13
-------
that have a resolution of 0.05°C and a basic accuracy of 0.1°C.
Psychrometers built with platinum wire sensors, thermistors, etc.,
have been used on some tower installations but users experience
maintenance problems and operational problems, especially in
near-freezing conditions. The psychrometric technique is not
recommended for operational towers installations but may have limited
applications at research sites.
2.4 Pressure
For diffusion calculations, pressure can be measured with almost
any pressure instrument. It probably could even be estimated from a
standard atmosphere and the site elevation. If pressure is measured
in a machine processable data base, an accuracy of +_ 1 kPa (10 mb) is
adequate. The expression for the representative period may be a
single instantaneous sample (p) or the arithmetic mean of any number
of samples (p).
2.5 Precipitation
There are two basic designs of recording precipitation gauges,
identified by the principle on which they operate: the weighing gauge
and the tipping bucket gauge. With modifications, the weighing gauge
can be adapted for remote transmission; the tipping bucket requires no
modification.
In both of the gauges, precipitation can be resolved to 0.2 mm,
or 0.01 inches, water equivalent. Unlike indicating gauges which
require a measuring stick calibrated to resolve amounts to 0.2 mm
and/or 0.01 inches, recording gauges are advantageous because they
document the total amount of precipitation and the beginning, ending,
and rate of precipitation, when the amount exceeds the resolution of
the instrument.
The weighing gauge is usable for collecting frozen precipitation,
but in below-freezing conditions, the bucket must be charged with
antifreeze as well as an evaporation-retarding oil in keeping with
accepted practice. The use of a shield and/or the installation of the
C-14
-------
gauge in a sheltered location is essential for the collection of
representative snow samples and is desirable year-round. The NWS
reports (Weiss 1961) that with a 20 mph wind, the catch of snow in a
gauge can be reduced by 50%'
Tipping bucket rain gauge accuracies for liquid precipitation
decrease as a function of rate-of-fill. Accuracies are 1% for
rainfall rates of 1 inch/hour or less; 4% for rates of 3 inches/hour;
and 6% for rates up to 6 inches/hour. Greater inaccuracies occur when
the buckets are not properly sited. Heating systems for tipping
bucket gauges used in below-freezing conditions are of limited value;
they must be used with care and with thermostatic control to avoid
evaporation of the sample.
The automatic wet/dry precipitation collector represents a
specialized type of nonrecording gauge designed for acid-rain studies
and other programs involving the chemical or radioactive analysis of
precipitation. A sensor that detects the onset and cessation of
precipitation automatically moves a lid or cover over the collector to
minimize evaporation. The significant factor in this type of
collector is the ratio between the amount of precipitation and the
polluting ingredient. Dry fallout is also of interest for some
studies.
It should be emphasized that while the siting of a gauge is
important, the representativeness of a measurement at one location
must be weighed against multiple or network measurements.
2.6 Solar Radiation
Generally, solar radiation is meaured in units of energy flux,
either in W/m2 or cal/cm2-min. It is also described in terms of
percent of sunshine and effective cloud cover. When solar radiation
is measured within specific narrow-band wavelengths, it is useful for
determining the turbidity and amount of precipitable water—indices of
atmospheric pollution and determining the potential for photochemical
reactions. Broad-band measurements of energy flux are of significance
in the processes related to turbulence and diffusion.
C-15
-------
Short wave radiation is represented by wavelengths less than
4 micrometers (urn), long wave radiation by wavelengths from 4 ym to
100 urn.
Both total solar and net radiation measurements have been used in
determining atmospheric stability. Since both are relatively
inexpensive and easy to install and maintain, installing one set of
sensors should receive serious consideration. One set of measurements
is likely to be representative of a large area unless there are
significant variations in topography or vegetative cover.
Siting of the sensor(s) must allow for an unrestricted view of
the sky in all directions during all seasons, including the lowest
solar elevation angle reasonably possible. Sensor height should be
within 2-5 m. The surface below the net radiometer must be
representative of a broad area around the site.
Pyranometers measure the total sun and sky shortwave radiation.
They incorporate either a thermopile or a silicon photovoltaic cell as
a sensor. The silicon cell covers wavelengths from 0.35 to 1.15 ym,
the thermophile from 0.28 to 2.80 ym. This is the range covered by
most NWS measurements. Not all pyranometers are
temperature-compensated but those so designed are preferred. Standard
units are compensated within ^1.5% over the range -20° to +40°C.
The sensitivity of a typical 48-junction thermopile "black and white"
pyranometer is 11 yV per W/m^, or 7.5 mV per calorie/cm^-nun.
Since absolute values of solar radiation have limited -value, it is
important to translate the data to integrated figures, typically for
1-hour periods.
When gathering data on problems involving solar radiation as a
important factor in photochemical reactions, the spectral pyranometer,
designed with two hemispheres, is the better selection. The
substitution of one of the hemispheres with a Schott glass filter in a
defined wavelength permits measurement of yellow, orange, red, or dark
red bandwidths. For problems related to ozone, an ultraviolet
radiometer for the spectral range of 0.295 to 0.385 ym would be the
better choice. The choice of instrument depends on the application.
Net radiometers are designed to measure both the incoming (mostly
short wave) radiation and the outgoing (mostly long wave) radiation.
C-16
-------
Typically, they cover the wavelengths from 0.3 to 60 ym. Polyethylene
hemispheres are used to cover and protect the thermopile sensors.
Most thin hemispheres must be internally pressurized to maintain their
shape and to minimize condensation. In a recent design change, the
hemispheres were constructed of heavy-duty plastic which attenuates
the signal by 4% but has the advantage of eliminating the need for
external sources of nitrogen gas. An in-line dessicator tube is
sufficient to handle the condensation problem. The sensitivity of one
typical domestic-made net radiometer is 5 uV per W/m2; an imported
unit has a sensitivity of 0.5 mV/mW/cm^. The time constants of
these units are 30 and 60 seconds, respectively, with typical signal
ranges of -5 to 10 mV. The negative value represents outgoing
radiation; the positive value represents incoming radiation.
The installation of a net radiometer is quite different from that
of a pyranometer. The radiometer is installed with the aid of a
bubble level on an arm that is south-facing, usually about 1 meter
above the ground surface so that there is an unobstructed view of the
total incoming and total outgoing radiation. By contrast,
pyranometers should be located on a level roof mount so that no
obstruction casts a shadow on the sensor at any time. The hemispheres
on the net radiometers and pyranometer must be kept free of foreign
material and dust.
C-17
-------
3. REMOTE MEASUREMENTS
3.1 Convective Mixed Layer Height
The height of the convective mixed layer is the height to which
material released in the atmosphere's boundary layer will spread under
unstable and neutral conditions. This mixed layer is capped by a
region of stable temperature profile, usually by a profile so stable
that the temperature actually increases with height (an inversion
layer). Then the top of the mixed layer can be measured if the base
of the capping inversion can be identified.
In principle, it should be possible to detect the inversion base
by measuring the temperature variation with height using in-situ
sensors mounted on aircraft, free balloons, and tethered balloons.
This approach, through straightforward, requires much manpower, is
intermittent and expensive if monitoring is to be carried out for a
long period. An alternate approach, which lends itself to unattended
continuous operation, takes advantage of the scattering properties of
the mixed layer-capping inversion interface. This interface is a
region of sharp gradients in temperature and in humidity and aerosol
concentration. Remote sensing techniques (sodars, lidars, and FM-CW
radars) provide echo signatures from which hc can be extracted;
sodars are the most economical and commercially available.
The key to interpreting backscatter returns from remote sensors
is the recognition that the sharpest inhomogeneities in temperature
and humidity are found at a height slightly above the mixed
layer—capping inversion interface. The base of the strongest echo
layer is therefore used as the reference level for h These
returns often show gravity waves and gust fronts (such transients at
the capping layer must be recognized when interpreting the
representativeness of single soundings with in-situ sensors).
Vertical profiles of temperature are readily accomplished with a
light plane spiraling up and then down. A fast response temperature
sensor is installed for recording temperature versus pressure altitude
C-18
-------
by chart or manually. Helicopters do the job in regions where
airspace is more restricted. Small radio-controlled drone aircraft
with telemetering temperture and pressure sensors have been used for
the same purpose.
A type of radiosonde balloon telemetering temperature is commonly
used. Altitude cannot be accurately ascertained from time and ascent
rate, so either double theodolite observation or a telemetering
pressure system must also be used. Since standard radiosondes have
poor vertical sampling rates, slow rise balloons with appropriate
sensors are used for deriving high-resolution temperature profiles.
Tracked balloons also yield wind data.
Tethersondes, carrying aloft sensors whose readings are
telemetered to the ground, are convenient to use in moderate and light
wind conditions in regions where they are not an aviation hazard.
Temperature profiles are easiest to obtain, but turbulence can also be
measured. The profiling is most commonly performed by raising and
lowering the balloon, but it can be done by using multiple sensors
spaced below a stationary tethered balloon.
A monostatic sodar with vertically pointing beam, with
time-height echo amplitude displayed on a facsimile recorder, is
widely used to find hcm It aiso shows the overall situation in the
lower atmosphere, showing multiple layers, clarifying the evolution of
mixing processes, etc. A shearing echo is recognizable by being
predominantly horizontal. Its presence is attributable to small-scale
temperature variations associated with turbulence within a stable
region. It designates a blocking layer from the standpoint of
vertical diffusion—its turbulence is of a scale and intensity and
intermittency which is ineffective in causing rapid vertical material
transport. During unstable convective conditions, the lower portion
of thermals are identifiable on the record by "spikes" rising from the
surface, while the shearing echoes delineating mixing height ride
horizontally (or in gentle waves) overhead. Objective rules have been
developed which permit hc to be derived, approximately hourly, by a
human operator. Automatic pattern recognition might permit the task
C-19
-------
to be done automatically, but this is not yet a proven technique.
Numerous comparisons between h ascertained by sodar and by
temperature profiles have shown satisfactory agreement between the two
methods—generally within 50 m. The upper heights at which h£ can
be observed by sodar depend on the meteorological situation, the
characteristics of the sodar unit, and the ambient noise. A height of
500 m is usually reached, 800 m is often reached, and 1,000 m is
sometimes reached. The period of greatest uncertainty in h
estimation is the three hours prior to sunset when the capping
inversion either disappears or becomes too weak to be detected. If
the more advanced Doppler sodar system is available, yielding directly
turbulence intensity versus height, chart interpretation of hc £s
not necessary.
FM-CW radar presents a similar facsimile display from which h
can be found. The echos are primarily from humidity variations. The
radar reaches as high as needed for boundary layer monitoring, but the
device is specialized, expensive, and not commercially available.
A backscattering lidar system records echos from aerosol
particles, and displays them on a facsimile recorder analogous to
sodar and radar. Since high aerosol content is characteristic of the
mixed air mass (expecially its top) and low content characterizes the
region aloft, the interface region and h can easily be found on the
display. The lidar reaches as high as needed. The device is
specialized, expensive, and not in commercial production.
All these methods will operate in altitudes unreachable by normal
towers. Aircraft and balloons readily achieve any height desired.
Tethersondes are limited by practical considerations of lift, payload
and the wind drag of balloon, instruments, and tether, as well as by
Federal Aviation Administration regulations. Heights of about 300 m
are very common, 1,000 m heights are sometimes handled, and greater
heights are feasible if economically justifiable. Sodar is generally
satisfactory to above 500 m, and sometimes beyond 1,000 m—with the
height increasing as the technology evolves. The other indirect
probes can reach beyond 2 km.
C-20
-------
3.2 Mechanical Mixing Layer Height
In a stably stratified boundary layer there is often enough wind
shear to produce vertical mixing over its entire depth. The height to
which the mixing operates is not determined solely by the temperature
profile, but rather, in a complex way, by both the temperature and
velocity fields, and is influenced by surface roughness as well. The
nonlinear interactions can make the situation nonstationary.
The height of the mechanical mixing layer is best defined by the
vertical profile of the vertical component of turbulence. The
vertical profile of temperature represents a satisfactory alternative,
if interpreted correctly. The strongly stable portion of the lower
boundary layer designates a region into which cold surface air has
been mixed; the temperature profile indicates that there was mixing
into this region, but does not verify that mixing is going on at the
present moment. When considering continuous releases and hour-long
pollutant concentrations, the time variations of mixing properties are
not of primary importance. Consequently, the profiles of turbulence
or temperature can give the information needed on the height to which
diffusion reaches upward from the surface. Conceptually, all the
techniques available for ascertaining h can be applied to finding
nm, although the aircraft method and free balloon method are
scarcely "practical" when dealing with height only up to 50 m or so.
For low altitudes, the use of instrumented towers is much more
feasible than they are for the hc case. Tethersondes can
conveniently yield temperature profiles, if intermittent data are
satisfactory.
Sodar records can be interpreted to give h Here, the upper
boundary of the strong echo region (rather than the base) is assumed
to be the top of the mechanical mixing layer. Conventional sodars,
which have been developed to emphasize the higher altitudes, can only
show surface-induced mixing rising above the minimum range of about
35 m. It is anticipated that better information to heights down to
10-15 m can be derived from units tailored to the purpose of finding
021
-------
^m. The units would operate at high frequency, with shorter pulse
lengths, and the facsimile display would utilize an expanded vertical
scale. Interpretation of b^ frOm sodar is occasionally ambiguous at
this stage of development of the technique—primarily because of the
complex nature of the phenomenon being observed.
3.3 Flow Field Aloft
All the probing devices discussed for finding hc can
conceptually be adapted to the more demanding task of monitoring the
flow field aloft (mean and turbulence). Diffusion modeling is not
limited to heights below hc> since effective stack heights can
exceed hc^ and diffusion through higher layers can also be important
in long-range transport.
For case study projects, instrumented aircraft can be
particularly effective. Aircraft such as those developed and operated
by National Center for Atmospheric Research combine flow probes with
inertial navigation systems to measure mean wind as well as turbulence
in all components and all important scales. An aircraft's mobility
allows the acquisition of data throughout a large three-dimensional
volume, with the limitation that flights cannot ordinarily be below
150 m (with much higher threshold, because of safety, in complex
terrain). The same airplane that monitors the atmosphere can also
monitor a pollutant plume or tracer.
Ascending free balloons tracked by double theodolite are useful
in finding the wind field—to the extent that a spot observation
represents events over a finite time interval. Tracking of "zero
lift" balloons can yield information on turbulence and flow
trajectories, but the technique is extremely difficult. Tracking of
superpressure balloons is a more applied technique, and yields data on
flow trajectories. Since the balloon does not exactly follow an air
parcel when vertical motion is involved, the results must be
interpreted cautiously.
C-22
-------
For a complex terrain research program, a network of
double-theodolite tracked balloons could be used for finding the wind
field. A network of tethersondes with wind sensors can also be used.
Turbulence sensors can also be added to monitor mixing factors
directly. Operation of a network of highly instrumented tethersondes
is a formidable undertaking.
Wind and turbulence fields can also be monitored by Doppler
versions of the indirect probing techniques: sodar, FM-CW radar, and
lidar. Other radars have occasionally been used to observe the motion
of clear air into which chaff has been introduced to give
back-scattering. The relative economy of sodar makes this technique a
good candidate for observing the flow field for both case study
research programs and long term monitoring programs for the altitudes
pertinent to high plume dispersion.
Recent studies comparing various flow measuring devices with the
BAO 300 m instrumented tower at Boulder, Colorado, have shown that
accuracies of Doppler sodar systems are within 0.5 m/sec for
measurement of wind components aloft. Also, there is initial evidence
that certain velocity variations can be continuously monitored
directly giving some turbulence quantities of importance in diffusion
modeling. Investigators are trying to derive temperature gradient
information (only to the broad accuracy limits required when
temperature gradient data are used in plume rise equations) from
profiles of wind and turbulence. If successful, Doppler sodar may be
capable of providing all the meteorological inputs needed for high
stack dispersion model calculations.
There are two basic types of Doppler sodar systems: monostatic,
where all antennas are located adjacent to each other (sometimes
mounted on one trailer) and bistatic, where antennas are separated
horizontally by distances comparable to the maximum height of
observation. In complex terrain, not only must the logistics and
complexity of installation be considered, but also the problem that
echoes from nearby elevated terrain can swamp the echoes from
atmospheric scattering unless the antenna sidelobe suppression is well
C-23
-------
handled. Doppler sodar appears capable of giving mean and turbulent
flow information well suited to diffusion modeling, routinely to
heights of 500 m, and sometimes to heights above 1,000 m.
Another remote sensing technique for indirect probing is the
two-ended laser anemometer. It does not have an extensive history for
validation, but does provide a method for obtaining integrated wind
speed components normal to the path along path length. It is used to
measure wind speeds across valleys and will determine mean speeds well
under 1 m/sec.
C-24
-------
REFERENCES
APPENDIX C
Acheson, D. 1963. Some Limitations and Errors Inherent in the use of
the Dew Cell for Measurement of Atmospheric Dew Points. Monthly
Weather Review, May.
Kaimal, J. C. 1975. Sensors and Techniques for the Direct Measurement
of Turbulent Fluxes and Profiles in the Atmospheric Surface
Layer. Atmospheric Technology. Instruments and Techniques for
Probing the Atmospheric Boundary Layer. NCAR, Boulder, Colorado,
McTaggart-Cowan, J. D., and D. J. McKay 1976. Radiation Shields - An
Intel-comparison. Atmospheric Environment Service, Downsview,
Ontario, Canada.
Weiss, L. L. 1961. Relative Catches of Snow in Shielded Gauges at
Different Wind Speeds. Monthly Weather Review, 69.
C-25
-------
TECHNICAL REPORT DATA
(Please read Inmrucnons on the reverse before completing)
REPORT NO.
EPA-6QO/9-81-020
3. RECIPIENT'S ACCESSION-NO.
' TITLE AND SUBTITLE
ON-SITE METEOROLOGICAL INSTRUMENTATION REQUIREMENTS
TO CHARACTERIZE DIFFUSION FROM POINT SOURCES
Workshop Report
5. REPORT DATE
April 1981
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
David Strimaitis, Gale Hoffnagle, Dr. Arthur Bass
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Environmental Research & Technology, Inc.
696 Virginia Road
Concord, Massachusetts 01742
10. PROGRAM ELEMENT NO.
CDTA1D/04-0262 (FY-81)
11. CONTRACT/GRANT NO.
68-02-3282
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
RTP, NC
PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Results of a workshop entitled "On-Site Meteorological Instrumentation
Requirements to Characterize Diffusion from Point Sources" are summarized and
reported. The workshop was sponsored by the U.S. Environmental Protection Agency in
Raleigh, North Carolina, on January 15-17, 1980. Its purpose was to provide EPA with
a thorough examination of the meteorological instrumentation and data collection re-
quirements needed to characterize airborne dispersion of air contaminants from point
sources and to recommend, based on an expert consensus, specific measurement technique^
and accuracies. Secondary purposes of the workshop were to (1) make recommendations
to the National Weather Service (NWS) about collecting and archiving meteorological
data that would best support air quality dispersion modeling objectives and (2) make
recommendations on standardization of meteorological data reporting and quality
assurance programs.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATi Field/Group
IS. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)
UNCLASSIFIED
21. NO. OF PAGES
124
20. SECURITY CLASS (This page I
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
-------
3'
5-
<
0)
5- *
C
VI
a
> me
•821
3 ^'
S 5
0
oo
it? II?
1x8 ^5.5
PI
O
O
I
00
!•—
I
o
N>
O
n S-
3 ;
s •
ii
3 <
s S-
= •
3§
*f
3 •
5?
si
a
a
a
u
CD
0)
o
en i m ^ "^ m
• -o« 3 =
> » S <.
. 3 CB -.
Ul O O O
OJ< S 3
01 03
3 »
3
o 5
n
-n -g
o o
A (A
"» M
o> a
a! a>
3
a
------- |