United States
Environmental Protection
Agency
Environmental Sciences
Research Laboratory
Research Triangle Park NC 27711
Research and Development
EPA-600/S3-84-044 Apr. 1984
Project Summary
Determination of Cloud
Parameters for NEROS II from
Digital Satellite Data
Jan L Behunek, Thomas H. Vender Haar, and Pat Laybe
The U.S. Environmental Protection
Agency (EPA) requires statistical de-
scriptions of total cloud amount,
cumulus cloud amount and cumulus
cloud top height for certain regions and
dates as input for their regional-scale
photochemical oxidant model of air
pollution. These statistics are used to
parameterize the presence of sunlight
for photochemical reactions, and to
diagnose vertical transport of pollutants.
The EPA Northeast Regional Oxidant
Study (NEROS II) supplied the case
studies for the digital satellite data
needed to derive these statistics.
The report provides users of the
results with dates, times, and regions
analyzed, output formats used, and
discusses special conditions within the
output data. The work reported here
demonstrates the viability of deriving
total cloud amount, cumulus cloud
amount, and cumulus cloud top height
characteristics from digital satellite
data.
This Project Summary was developed
by EPA's Environmental Sciences
Research Laboratory. Research Triangle
Park. NC. to announce key findings of
the research project that is fully docu-
mented in a separate report of the same
title (see Project Report ordering
information at back).
Introduction
The U.S. Environmental Protection
Agency (EPA) requires statistical descrip-
tions of total cloud amount, cumulus
cloud amount and cumulus cloud top
height for certain regions and dates as
input for their regional-scale photochem-
ical oxidant model of air pollution. These
statistics are used to parameterize the
presence of sunlight for photochemical
reactions, and to diagnose vertical
transport of pollutants.
The EPA Northeast Regional Oxidant
Study (NEROS II) supplied the case
studies for the digital satellite data
needed to derive these statistics. Digital
satellite data transmitted from the
Geostationary Operational Environmental
Satellite (GOES) were analyzed with the
help of Colorado State University's (CSU)
Interactive Research Imaging System
(IRIS) to generate the required diagnoses
of total cloud amount, cumulus cloud
amount, and cumulus cloud top height.
Synoptic rawinsonde data also were used
to translate cloud top temperatures to
cloud top heights. The IRIS also produced
the quantitative cloud field statistics
necessary to manipulate the digital data.
Digital satellite data were essential to
the success of the study because they
provide uniform spatial coverage (unlike
surface station data). Furthermore, the'y
are readily processed by human-computer
interactive techniques, which are faster
than manual analyses of hard copy
satellite images. The digital satellite data
also produce more precise cloud top
height determinations, due to the avail-
ability of a wide range of infrared digital
counts, rather than the narrow range of
gray-shades seen on hard copy images.
Procedure
The cloud field statistics produced for
EPA were derived by a two step process.
The first step involved displaying the
digital satellite data with the IRIS and
-------
generating values for total cloud amount,
cumulus cloud amount, and cumulus
cloud top temperature. The second step
was to convert the cumulus cloud top
temperature values to cloud top height
values by referring to rawinsonde-
derived profiles of atmospheric tempera-
ture versus height.
GOES digital satellite data in the visible
channel and infrared window (transparent
to the earth's surface where cloud free)
channel were obtained by CSU from the
National Oceanic and Atmospheric Ad-
ministration's Environmental Data Infor-
mation Service (NOAA/EDIS). The GOES
data archive is maintained for NOAA/EDIS
by the Space Science and Engineering
Center at the University of Wisconsin.
The visible and infrared data were
preprocessed within the IRIS to decode
the format and then displayed in image
form on a video monitor. The horizontal
resolution of the visible data was 1 km at
the subsatellite point (75°W, 0°N),
whereas the resolution of the infrared
data was 4 km by 8 km. These resolution
values were slightly larger at the latitudes
and longitudes of interest for this study.
The satellite-derived cloud statistics
were output in gridded form so that they
could be easily read into EPA's pollution
model. The values of total cloud amount and
cumulus cloud amount apply to individual
grid cells, whereas the cloud top heights
are mean values for each cell. In general,
each output grid cell had dimensions of
1/6 degree latitude by 1/4 degree
longitude. Each grid was made rectangu-
lar with respect to lines of latitude and
longitude to facilitate participation of the
computer in generating statistics.
Navigation of the satellite data was
a routine, but important, part of the
analysis. Navigation is the creation of a
mapping between points in the coordinate
system of the satellite image and geogra-
phic points on the earth Precise naviga-
tion of the data ensured that the geogra-
phic location of each grid cell was known.
The accuracy of the navigation of each
image was checked by overlaying graphics
plots of coastlines, lakes, and reservoirs
based upon the navigation calculations.
Any discrepancies between these plots
and visibly identified landmarks indicated
a need to revise the navigation. Finally,
care was taken to make sure that cloud
features within the visible and infrared
images were aligned.
Results and Discussion
The satallite-derived cloud statistics
included (1) percentage total cloud cover,
(2) percentage cumulus cloud cover, and
(3) a frequency distribution of cumulus
cloud top heights. Several limitations were
encountered in screening the data.
The identification of regions with cloud
cover and regions with no cloud depended
on the ability of the scientist involved to
select a brightness threshold for the
visible channel that delineated those
areas. The first constraint on this process
was related to the size of the geographic
area seen within a single field of view of
the satellite's visible sensor. In general,
the smallest cloud feature resolved by the
visible sensor was approximately 2 km2 in
area at the latitudes and longitudes
studied.
A second constraint on the cloud
amount analysis was the difficulty
associated with identifying optically thin
cirrus clouds. Cirrus clouds sometimes
are so thin that they do not reflect an
amount of visible radiation distinguishable
from that reflected by the surface
The ability of CSU to identify cumulus
clouds and distinguish them from other
types of clouds was most affected by the
presence of multilayer clouds. Cirriform
cloudiness generated from cumulonimbus
anvils often conceals new cumulus
growth beneath it. The only readily
available source of information on the
presence of such concealed cumuli is
surface station data. However, no attempt
was made to incorporate surface station
cloud reports into this study.
A second limitation that sporadically
affected cloud type identification was the
requirement to calculate cloud statistics a
few times after the sun had set at the
location of interest This fact rendered the
visible data useless for all times after
2000 EOT. After 2000 EOT all of the
statistics were derived from infrared data
alone.
The determination of cloud top heights
was impacted by several constraints. The
first constraint was related to thermal
emission by atmospheric water vapor.
Water vapor does have a small emittance
in the 11 /ym infrared channel used inthis
study, and the absorption and subsequent
radiation by water vapor between the
target (cloud or ground) and the radiometer
can cause the target to appear colder than
it really is. The impact of this process on a
satellite-derived surface temperature has
a magnitude of approximately 4°C
A second influence on the calculated
cloud top temperatures was the response
time of the infrared sensor Normally, that
sensor does not respond immediately to a
rapid change in temperature as it scans
across a scene Therefore, the temperature
near the edge of a tall, cold cumulus cloud
was likely to be overestimated The
magnitude of this error ranges from 1 tc
5°C, depending on the actual cloud top
and background temperatures and the
configuration of clouds.
Another limitation that created difficul-
ties in calculating cumulus cloud top
temperatures was related to the occasional
presence of multilayer clouds. Due to
their large size, infrared pixels can
contain cumulus clouds and clouds of
another type. Such pixels yield cloud top
temperatures that are not entirely repre-
sentative of cumulus clouds. In order to
minimize this problem, a cloud top
temperature was not calculated from any
infrared pixel for which the total cloud
cover was composed of less than 70
percent cumulus cloud.
The final sources of error in deriving
the cloud top height were related to the
rawmsonde data used to translate cloud
top temperatures. The few times rawin-
sonde data were not available from the
Bureau of Reclamation, the conversion of
temperatures to heights was accomplished
using rawinsonde data from the single
nearest time rather than from the
observation times surrounding the analysis
time. Secondly, rawinsonde data has a
much lower horizontal resolution than
the resolution of the required temperature-
height analyses. The horizontal resolution
of the synoptic rawinsonde network
ranged from 225 to 450 km, whereas the
resolution of the temperature-height
analyses was approximately 20 km.
A constraint that affected all of the
cloud statistics produced for EPA was the
misrepresentation of cloud locations due
to the viewing angle between the satellite
and the clouds. A cloud top that is signifi-
cantly above the earth's surface appears
displaced away from the satellite along
the line of sight. The effect is greater for
deep clouds than for shallow ones. The
impact on the cloud statistics was limited
to erroneous placement of clouds near
the edge of one grid cell into a neighbor-
ing grid cell. When the effects of all the
limitations are quantified and summed,
the error involving cloud top heights can
be assessed This error was assessed by
CSU to be 462 meters.
Conclusions and
Recommendations
The work reported here has demon-
strated the viability of deriving total cloud
amount, cumulus cloud amount, and
cumulus cloud top height statistics from
digital GOES data. The interactive
technique developed at CSU allowed the
rapid processing of large quantities of
satellite and rawinsonde data in order to
accomplish that goal The successful
-------
execution of the contracted tasks is sig-
nificant in that it should allow the EPA
model, into which the current results are
to be entered, to perform a more accurate
and comprehensive simulation of air
pollution than has been accomplished
previously. The true worth of the work
reported here must be determined by the
improvement that the cloud field analyses
causes in the EPA model.
Several additional tasks should be
executed to enhance the usefulness of
satellite data for modelling air pollution.
The first of these tasks is to compare the
satellite-derived statistics with statistics
derived from other data platforms. In the
near future, CSU and EPA scientists will
compare cumulus cloud top heights
derived from aircraft-borne LIDAR data
and from GOES data, and the results will
be published. That study should partially
satisfy the need for mtercompanson
A second desirable accomplishment
would be to further investigate the impact
of cirrus clouds on the calculated cloud
statistics. In particular, the effect of the
transmittance and emittance of cirrus on
satellite-derived cumulus cloud top
temperatures requires further research.
Although Reynolds and Vender Haar and
others have addressed this problem, it
still is far from solved.
Other applications of satellite data are
possible. For instance, the vertical
velocities of convective clouds may be
inferred from satellite data to provide
details of pollutant transport. Also, the
motions of small cumuli may be observed
by satellite to quantitatively diagnose
horizontal transport processes. Similarly,
features within the atmospheric moisture
field may be followed using multispectral
satellite data. The pursuit of these
applications could prove beneficial to EPA
or to others concerned about air pollution
Jan L. Behunek, Thomas H. Vender Haar, and Pat Laybe are with Colorado State
University, Fort Collins, CO 80523.
Terry L. Clark is the EPA Project Officer (see below).
The complete report, entitled "Determination of Cloud Parameters for NEROS II
from Digital Sate/lite Data," (Order No. PB 84-162 601; Cost: $8.50. subject to
change} will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
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
•ft US GOVERNMENT PRINTING OFFICE, 1984 — 759-015/7662
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
------- |