United States
Environmental Protection
Agency
Environmental Sciences Research   T, .
Laboratory                       /'
Research Triangle Park NC 27711
Research and Development
EPA-600/S3-82-011  Sept. 1982
Project Summary
Estimating  Cloud
Paramters  for  Neros I
F. M. Vukovich and D. P. Erlich
  Geosynchronous Orbiting Earth
Satellite infrared and visible imagery
were  combined with  surface and
upper-air meteorological observations
to determine cloud amounts and
cloud-top heights over the Northeast
Regional Oxidant Study grid for 1200,
1500, and 1800 EOT, on August 3,4,
and 13, 1979. Cloud amounts were
determined for cumulus clouds alone
and for all clouds. Cloud-top heights
were  determined specifically  for
cumulus clouds.
  A study was begun  to develop  a
model that could be used to estimate
the parameters of the cloud ozone
flux. Several models were developed
to estimate the average maximum
cloud vertical velocity; the best model
developed was a  multiple  linear
regression model.  The model input
parameters were the cloud-top height
and the cloud amount, which were
derived from satellite imagery. This
model yielded  an average correlation
coefficient of -0.78 and a root mean
square difference of ±0.8 m/s~1. On
the average, with the use of the multi-
ple linear regression model, there was
a 24% error in the estimated average
cloud vertical velocity. However, the
modeling results were not statistically
significant because of the limited data
available for developing the model.
The total number of data points was
nine, but only seven were useful.
  This Project Summary was devel-
oped by EPA's Environmental Sci-
ences Research Laboratory. Research
Triangle Park,  NC, to announce  key
findings of the research project that is
fully documented in a separate report
of the same title (see Project Report
ordering information at back).

Introduction
  In July and August of 1979, the U.S.
Environmental Protection Agency (EPA)
conducted the first phase of the North-
east Regional Oxidant Study (NEROS).
The primary purpose of the study wasto
measure concentrations of oxidant and
oxidant precursor on a regional scale in
the boundary  layer. From these data,
physical processes could be parameter-
ized in  numerical models and numeri-
cal model simulations evaluated.
  Solar radiation is a significant factor
in the formation of oxidants from oxi-
dant  precursors; the concentration of
pollutants within a layer of the atmos-
phere can be influenced significantly by
the vertical flux via cumulus cloud vent-
ing. To estimate the amount of solar
radiation penetrating the  boundary
layer, imagery from the Geosynchro-
nous Orbiting Earth Satellite (GOES)
was used to estimate cloud parameters
over grid squares approximately 20 km
by 20 km in  the Northeastern United
States.  The GOES  imagery  also was
used  to derive a physical relationship
between cumulus cloud-top growth and
vertical velocity within cumulus clouds
so that the vertical flux of oxidants and
oxidant precursors could be estimated.
  Another principal objective of this
research project was to determine cloud
parameters (spatial and vertical extent)
that could be used in modeling the pro-
duction of ozone in the  boundary layer
during specified periods in the NEROS
program. The  study area was bounded

-------
by 38° and 45°N latitude and 69° and
84°W longitude. Major objectives real-
ized during the  research project were:
  1. To determine the fractional cover-
    age of cumulus clouds (when
    these clouds alone existed) in each
    of the  1/4° longitude by 1/6° lati-
    tude NEROS grid squares, using
    GOES imagery and synoptic mete-
    orological observations for 1 200,
    1500,  and  1800 EOT on August 3,
    4, and 13(1979).
  2. To determine  the average height
    of cumulus cloud tops across each
    NEROS  grid  square using  the
    GOES infrared imagery and avail-
    able upper-air temperature profiles.
  3. To determine the fractional cover-
    age of all clouds (other than cumu-
    lus or multilayer clouds that might
    include cumulus) for each NEROS
    grid square.
  A secondary objective  of the project
was to study the period  during which
the so-called "cloud buster" experi-
ments  were  performed  to  determine
cloud  parameters such  as those dis-
cussed above; study of those parame-
ters relative to in situ measurements of
vertical velocity made in  cumulus
clouds  would help to determine  if  a
functional relationship could  be devel-
oped between the satellite cloud param-
eters and the cloud vertical velocities.
This relationship could then be used to
model the vertical ozone flow in cumulus
clouds. The specific  approach is out-
lined below.
  1. Available GOES infrared and vis-
     ible imagery in  or  very near the
    periods of the "cloud  buster"
    experiments  (i.e.,  1430  to 1530
     EOT on 22  August 1979, and 1400
    to 1700 EOT on 28 August 1979)
    were used to determine the frac-
    tional  coverage of cumulus clouds
    and the average  height of the
    cumulus clouds within a fixed 20-
    km  grid squared along aircraft
     transects  in southeastern  Pen-
     nsylvania and New Jersey.
  2. In situ measurements made from
    the aircraft of the peak vertical
     velocity in the cumulus clouds, the
     average vertical velocity in the
    clouds, the ozone concentration in
    and around the  clouds, and the
     upward- and downward-looking
     radiometer temperature in the vic-
     inity of the  clouds were deter-
     mined  along  the transects  from
     the data provided by EPA.
  3. The vertical velocity data provided
     from  the  in situ measurements
     were  compared  with the cloud
     parameters obtained from the
     GOES data to determine  if func-
     tional  relationships exist.


Determination of Cloud
Parameters for the
NEROS Grid
  The principal  data  used were the
GOES infrared and visible images ob-
tained for August 3, 4,  and 13 (1979).
GOES visible  images were available
over the region of interest on these days
at 1130, 1430,  and 1730 EOT.  The
GOES data  were collected in hard copy
image form  and on magnetic tape by the
Research Triangle Institute's  Satellite
Receiving Station in North Carolina.
  Synoptic  weather  data  for 1200,
1500, and  1800 EOT from the  first-
order, surface-synoptic weather sta-
tions in the  region were used, as well as
upper-air data for 0200, 0800, and
1400 EOT from the National Weather
Service  Upper-Air Stations  The 1400
EOT upper-air data were used exten-
sively because  they fell  within  the
period  1200  to 1800 EOT. These
weather data were obtained either from
the National Climatic Center in Ashe-
ville, NC or  from  EPA.
  The steps employed to determine
cloud amounts and cloud-top  heights
over the NEROS grid are as follows:
  1. To facilitate interpretation of the
    . satellite data, the gray scale of the
     GOES infrared and visible images
     was enhanced (i.e , the gray scale
     was confined to a range of temper-
     ature  and reflected radiation that
     gave the most useful information),
     using  the data on magnetic tape
     and the  facilities available at the
     RTI satellite receiving station.
  2. The enhanced GOES visible and
     infrared  (IR) images were photo-
     graphically  enlarged uniformly to
     further facilitate interpretation of
     the satellite data.
  3. A NEROS grid overlay was devel-
     oped on  transparent Mylar for the
     GOES imagery.
  4. The cloud amounts, cloud types,
     and cloud-base heights from sur-
     face synoptic data were plotted on
     another  transparent Mylar over-
     lay.
  5. An  analysis delineating areas of
     clear  skies, cumulus alone,  and
     multiple  cloud layers or clouds
     other  than  cumulus were devel-
     oped using  the GOES visible and
     IR  imagery  and the plotted cloud
     data  from the surface synoptic
     stations
  6  Analyses of cloud cover in areas of
     cumulus only,  and of multiple
     cloud layers or clouds other than
     cumulus, were  performed using
     the GOES visible and IR imagery
     and the surface synoptic cloud
     data
  7. An analysis of cumulus cloud-top
     temperature was performed using
     the  GOES  infrared  imagery and
     calibration data available from the
     GOES User's Guide.
  8  Cumulus cloud-top heights were
     derived  using the cloud-top tem-
     peratures combined with the radi-
     osonde data.
  9. The cumulus only cloud amounts
     (ac), cloud amounts for conditions
     other than cumulus alone (cra), and
     the cumulus-top heights (Hc) were
     selected at each of the NEROS grid
     points, formatted, and punched on
     computer cards.
  Cloud parameters from the  surface
synoptic data were plotted on transpar-
ent Mylar overlays using the GOES vis-
ible imagery  The plotted cloud data and
the GOES visible and infrared  images
were  then  used to define  regions of
cumulus alone, clear skies, and multiple
cloud layers.  The visible images were
used to interpolate  in  areas between
synoptic weather stations. The infrared
images were  examined along with the
visible  imagery to  determine if there
was a change in cloud  structure
between  synoptic stations that might
indicate a change in cloud type. Sim-
ilarly, an analysis of cloud amounts was
performed using the GOES visible imag-
ery, the plotted cloud data, and the anal-
ysis delineating areas of cumulus, clear
skies, and multiple cloud layers. Once
again, the visible imagery data were
used to interpolate in areas between the
synoptic weather stations.
  To obtain the cloud-top heights for the
cumulus clouds, the gray scale for each
of the GOES  infrared images was cali-
brated  with  respect  to temperatures,
using data available from the  GOES
User's Guide. For each GOES infrared
image, patterns of shades of  gray were
analyzed on transparent Mylar overlays,
and  the  patterns of gray scale were
assigned temperatures using the cali-
brated gray scale. The satellite tempera-
ture in the area of cumulus clouds (Ts)
and the temperature in the area of clear
skies (Ta) nearest the cumulus clouds
were determined. The cumulus cloud-

-------
top temperature can be estimated using
the following formula
         Tc = Ta - -1 (Ta - Ts)
                  Cc
where ac is the amount of  cumulus
clouds in tenths, and Tc is the cumulus
cloud-top temperature.
  The  cumulus  cloud-top height (Hc)
was determined using the cumulus
cloud-top temperature and the temper-
ature profile obtained from the radio-
sonde stations  nearest  the  cloud of
interest. In all cases, the 1400 EOT radio-
sonde  data  were  used. The  cumulus
cloud-top temperature and cloud-top
height were  also estimated  using the
1400 EOT radiosonde data and standard
techniques (i.e., a parcel of air was lifted
dry adiabatically to the lifting condensa-
tion level, then moist adiabatically to the
level of free convection and the equili-
brium  level  —  the level of  neutral
buoyancy; the cloud-top height was
assumed to be the height of the equili-
brium  level). These values were com-
pared  with  those values determined
using the satellite data.  When major
discrepancies (differences >  1000 m)
were found,  attempts were made to jus-
tify the differences. About 25 percent of
the data given satellite analysis could
not be  verified using  the calculations
from the soundings. These discrepan-
cies were generally due to the develop-
ment of isolated  large cumulus con-
gestus and cumulonimbus clouds.
  Afterwards, a transparent Mylar over-
lay of the NEROS grid was placed on the
analysis of   cloud amounts  and the
cloud-top height to determine those
parameters  at the grid points. These
data were formatted and  punched on
computer cards, which were delivered
to  EPA with an  explanation  of the
format.

Modeling  Cloud
Vertical Velocity
  A second objective of this project was
to develop a  model which would esti-
mate  the average maximum-cloud-
vertical velocity  using  parameters
obtained from satellite data:  the cloud
amount and cloud-top  height. The
model was to be used to  parameterize
the cloud vertical flux of  oxidants and
oxidants precursors. Since this was the
first attempt to develop such a relation-
ship, and since the data resources were
limited, the result of this study may be
considered asa guideline for more com-
prehensive  studies in  the future.  The
vertical  velocity data were obtained by
aircraft as a part of the so-called "cloud
buster" experiment. The purpose of this
experiment was to collect data for study-
ing the vertical flux of ozone in clouds.
  The principal data set utilized to deter-
mine the cloud amounts and cloud-top
heights for the  cloud buster experi-
ments  was the GOES infrared and vis-
ible imagery for 22 and 28 August 1979.
Since  the  aircraft data  for the  cloud
buster experiment were sampled from
1430 to 1530 EOT on 22 August and
from 1400 to 1700 EOT  on 28 August,
the GOES  visible and infrared images
for 1430 and  1500  EST,  respectively,
were used for this study. The GOES data
were collected in hard copy image form
and on magnetic tape by RTI
  Synoptic weather data from the first-
order  surface  weather  stations were
also used in the region and for the time
period  defined  by  the flight tracks (over
southeastern Pennsylvania and New
Jersey). The 1500 EOT weather data
were used in all cases,  and the 1400
EOT upper-air data were also used to de-
termine  cloud-top  heights.  Cloud
amounts and cloud-top heights were de-
termined in 20-km x 20-km squares
centered along the flight track of the air-
craft, using the methodology discussed
earlier.
  Various in situ  measurements were
made by an aircraft along the transects.
The aircraft data  were used to deter-
mine the  average maximum vertical
velocity in cumulus clouds over a 20-km
portion of the flight track coinciding with
the 20-km x 20-km region  where cloud
parameters were determined  using
satellite  data  (i.e., the  peak vertical
velocity was determined  for each cloud
in the  20-km portion of the flight track
and an average was computed over all
clouds  in  the  cell). The upward- and
downward-looking radiometer  temp-
eratures were  used to verify the exist-
ence of clouds.
  The data  reduction yielded nine inde-
pendent data points  from  the various
transects over  the two days. Two data
points  were obtained along a transect
that bordered  two distinct regions of
cloud amounts and cloud-top heights.
For that reason, it was difficult to specify
cloud amount  or  cloud-top height in
these cases. The values given are asso-
ciated  with a cloud  system with low
cloud-top  height  and an approximate
cloud amount of 0.3. The  other cloud
system had cloud-top heights on the
order of 5,000 m and cloud amounts of
approximately 0.6.  Because of the
problem  of selecting proper  cloud
amounts and cloud-top heights in this,
case, it was decided that these two data
points would be ignored in the analysis
that follows. Discarding these two data
points left only seven data points for the
analysis.

Results
  The following models were used with
the seven data points to develop a rela-
tionship  between the average  maxi-
mum vertical velocity in the cloud and
the cloud amount and top height derived
from satellite data: a multiple  linear
regression model; polynomial models in
which the average maximum cloud ver-
tical velocity was related to the  cloud
amount;  and the polynomial models in
which the average maximum cloud ver-
tical velocity was related to the cloud-top
height.   The  analysis indicated that
increasing the degree of  the polynom-
ials to a value greater than three did not
significantly improve the models. Table
1 lists statistics on the various models
including the correlation coefficient and
the root mean square difference (RMSD)
between  the estimated and the observed
average maximum cloud vertical veloc-
ity. The coefficients of the models were
determined via a standard regression
algorithm developed for the Tektronix
Model 4051 computer.
  The data in Table 1 indicate that both
the  cloud-top height  and the  cloud
amount are negatively correlated with
the  average  maximum  cloud vertical
velocity: as the cloud amount increases
or the cloud-top height increases,  the
average maximum cloud vertical veloc-
ity decreases. As the degree of the poly-
nomial models increased for both the
cloud amount and the cloud-top height,
the magnitude of the correlation coeffi-
cient increased and the magnitude-of
the  RMSD decreased. The  statistical
data in Table 1 suggest that the multiple
linear regression model yielded the best
relationship between the average maxi-
mum cloud  vertical  velocity and the
cloud amount and the cloud-top height.
  The specific form of the multiple lin-
ear regression model  for the average
maximum cloud  vertical velocity  (with
seven data points) is
    We = 7.8 - 4.4 ac - 0.0011 HT

where we is the estimated average max-
imum cloyd vertical velocity, ac is the
cloud amount, and HT is the cloud  top
height. Table 2 gives a comparison of
the estimated and  observed average
maximum cloud vertical velocities. Also
given is the residual and the RMSD. On
the average, the error in the estimated
                                                                                     ft US GOVERNMENT PRINTING OFFICE. 1982-559-017/083Z

-------
    Table 1.
Statistical Analysis Of The Various Model Types Yielding Estimates Of
The Average Maximum Cloud Vertical Velocity (We) As A Function Of The
Cumulus Cloud-Top Height (H-r) And The Cumulus Cloud Amount
Which Were Derived From Satellite Data.
                        Model
                                               R*
RMSD**
We
We
We
We
We

We
= 7.8 -4.4 ac -0.0011 H-r
= 7.7 -0.0014 «T
= 4.5-6.1 ac
= 0.05 WT - 0.000009 t-fi - 74.2
= 6.1 - 18.1 ac+ 17.4al
= 131 -0.15 Hi + 0 00006 H$ - 0.000000007 H?
= / - 19 ac + 20.8 al - 3.3 al
-0.78
-0.56
-0.67
-0.72
-0.67
-0.75
-0.75
±0.8
±1.1
±1.0
±1.0
±1.0
±0.9
+0.8
    *R is the correlation coefficient
    **RMSD is the root mean square difference between the observed and estimated
      average maximum cloud vertical velocity fm/s'1).
    average  cloud vertical velocity is 24%
    and the  error decreases as the vertical
    velocity  increases. However, it  should
    be noted that these results are not sta-
    tistically significant  because  of the
    limited data available for the develop-
    ment of  the model.
                             Conclusion

                               The fact that the cloud amount and
                             cloud-top height were negatively corre-
                             lated with the average maximum cloud
                             vertical velocity was surprising and may
                             be a  result of the small data set. How-
      F. M. Vukovich and D. P. Erlich are with Research Triangle Institute, Research
        Triangle Park, NC 27711.
      Terry L. Clark is the EPA Project Officer (see below).
      The complete report, entitled "Estimating Cloud Parameters for NEROS I,"
        (Order No. PB 82-186 552; Cost: $7.50,  subject to change) will be available
        only from:
             National Technical Information Service
             5285 Port Royal Road
             Springfield, VA22161
             Telephone: 703-487-4650
      The EPA Project Officer can be contacted at:
             Environmental Sciences Research Laboratory
             U.S. Environmental Protection Agency
             Research Triangle Park, NC 27711
              Table 2.    A Comparison Of The Ob-
                         served (We) And Estimated
                         (we)  A verage  Maximum
                         Vertical Velocities (m/s~*).
Wc
fm/sj
1.5
4.4
2.9
1.8
4.5
4.5
4.4
We
fm/sj
1.2
3.8
3.8
3.1
4.1
3.6
4.5
Residual
0.3
0.6
-0.9
-1.3
0.4
0.9
-0.1
              RMSD = ±0.8 m/s.

              ever, there  is  a possibility that the
              results  may be real. As clouds develop,
              both the horizontal and vertical dimen-
              sions increase (the cloud amount and
              the cloud-top height). Eventually, the
              cloud reaches a mature state where the
              vertical velocity begins to decrease and
              approaches zero or  begins  to become
              negative if hydrometers fall. Therefore,
              one can hypothesize that the maximum
              vertical velocity  in  isolated cumulus
              would be reached when the cloud is rel-
              atively  small  (i.e., in  its development
              stage and not in its mature stage). How-
              ever, such speculation can  be verified
              only if the data set were to be increased
              substantially.
United States
Environmental Protection
Agency
                     Center for Environmental Research
                     Information
                     Cincinnati OH 45268
                              Postage and
                              Fees Paid
                              Environmental
                              Protection
                              Agency
                              EPA 335
Official Business
Penalty for Private Use $300
       PS

-------