Identification and Characterization of Potential Ground Targets for Satellite Microwave Sensor Calibration Monitoring

1

H13C-0444
Locating Stable Calibration Targets
Identification and Characterization of Potential
Ground Targets for Satellite Microwave Sensor
Calibration Monitoring
M.H. Savoie1(savoie_at_nsidc.org), E. Njoku2, M.J.
Brodzik1, K. Knowles1, R.L. Armstrong1
http//nsidc.org
Antarctic Experimental Area
Central Africa Experimental Area
Introduction
These four-panel images provide basic information
describing the suitability of given footprint
locations (at 3x3 EASE-Grid cell spatial
resolution) as calibration targets. The upper
left panel shows the general Earth brightness
temperature level across the experimental area.
For Antarctica, the images are of 36 GHz
horizontally polarized brightness temperature
data from ascending passes. The brightness
temperatures are around 165 K. For Central
Africa, 6.9 GHz vertically polarized data are
shown from descending passes. The brightness
temperatures are around 280 K. The 3x3
footprint statistics are computed at all grid
cell locations within the experimental area. By
displaying these statistics in two dimensions we
can determine, over a large experimental area, a
good reference target location as one that is
dark (low value) in both the Standard Deviation
of Daily Footprint Mean and Average Daily
Footprint Standard Deviation. Brightness
temperatures at such location vary little in time
and in space The green markers represent
pre-selected locations that were examined to
determine their quality as reference targets. In
the Antarctic, the green marker represents the
location of the Dome C experimental site. We can
see by simple examination that the Dome C target
may not be in the best location from a purely
satellite sensor point of view, but still may
represent a reasonable target. It also has the
advantage of being a well-instrumented site. The
Salonga target location is shown as the green
marker in the Central Africa images. Both the red
and green markers represent locations with small
annual variation in the mean brightness
temperature of the footprint, but the green
target represents a much more spatially
homogenous target since the red target footprint
includes a water body.
 AMSR-E is one of a series of Earth-orbiting
microwave radiometers that include SMMR, SSM/I,
TMI, AMSR, and Windsat. Onboard calibration
systems with each instrument allow ground-based
processing to derive brightness temperatures of
the Earth. However, a variety of post-launch
artifacts such as uncorrected attitude errors,
instrument misalignment, thermal gradients, and
component degradation, create a need for
fine-tuning of the calibration in the data
production system.  The challenge of post-launch
tuning is finding suitable targets. Reference
targets should be homogenous over a large enough
surface to resist nonlinear averaging and antenna
pattern effects, while remaining stable and
well-known in their brightness temperature
characteristics. Calm ocean scenes have been
used to cross-calibrate the SSM/I, TMI and AMSR-E
sensors at the cold end (80-150 K) of the typical
Earth-view brightness temperature range.
   However, even with excellent cold end
calibration, such that ocean geophysical products
are accurate, the warmer temperature calibration
may be inaccurate, causing errors in the
geophysical retrievals over land. Some previous
studies have investigated suitable targets at the
mid-range (200 K) and warm end (250-300 K) of
the Earths emission range. However, the optimal
locations and dimensions of such targets have not
been studied adequately from a satellite sensor
viewpoint. Here, using AMSR-E EASE-Gridded
brightness temperatures, optimal ice sheet
locations are investigated for the mid-range, and
tropical forest locations for the warm end.
Definitions
Experimental Area A large region selected to
search for stable targetson this poster it is a
100x100 array of EASE-Grid 25km cells. The
footprint statistics are calculated at each
EASE-Grid point across the entire experimental
area, and displayed as images in the panels on
the right. Footprint A set of grid cells,
examined together, meant to represent a
homogenous region from where the satellite
sensors would receive a signal. For this poster,
we only examine an array of 3x3 EASE grid cells
as a footprint. For mathematical simplicity,
assume these footprints are described by an array
of N9 grid cells with indices j. Footprint Mean
The mean of the brightness temperatures of each
grid cell in a footprint.   Footprint Standard
Deviation Square root of the variance or sample
standard deviation, for the set of grid cells in
a footprint. This statistic shows the spatial
variance at an instant in time or how the grid
cells change brightness temperature over the
footprint. We need this to be small to have a
homogenous target.   Average Daily Footprint
Mean Time average of the Footprint Mean. These
presented data are for AMSR-E from 02-Jun-2002
through 04-Nov-2004. Average Daily Footprint
Standard Deviation The average of each
Footprint Standard Deviation over a time series.
Sites that are spatially homogenous, or have low
brightness temperature gradients across a
footprint, will have low values of Average Daily
Footprint Standard Deviation. If a site has a
large value, it means that often the footprint
covers a region with a large brightness
temperature gradient and would not make a
suitable reference target. This is evident in the
images at water boundaries. Standard Deviation
of Daily Footprint Mean Standard deviation of
the footprint mean over a length of time. Shows
how much chronological variation a footprint has.
Sites with a low standard deviation of the daily
footprint mean when the time period is long, are
sites that generally have a low seasonal
brightness temperature variation.
These plots detail data used in the above 4-panel
images. A time series of brightness temperatures
(above) and standard deviations of the footprint
(below) is shown for each marker above. The
average of the brightness temperature plot yields
the pixel value for the Average Daily Footprint
Mean Brightness Temp. at the marker location.
While the standard deviation of the same plot
gives the value for the Standard Deviation of
Daily Footprint Mean image. The average of the
lower plot shows the pixel value for the Average
Daily Footprint Standard Deviation. These plots
give another visual representation of what
constitutes a stable reference target. Its
clear that the Central Africa red location has a
much larger spatial variability than the green
location by its consistently large standard
deviation of the footprint and would be
unsuitable as a calibration target.
Note the mid-latitude plots are interrupted
because the satellite overpasses do not occur
every day as they do at high latitudes.
Conclusions
Contacts savoie_at_nsidc.org 1 National Snow and
Ice Data Center, University of Colorado, Boulder
2NASA Jet Propulsion Laboratory, Pasadena,
CA Funding for this work has been provided by
NASA NAG5-9412 "Standard Global Snow Products
Derived from Satellite Remote Sensing" and NASA
NAG5-11107 "Validation of AMSR-E Snow Products.
We have presented a method for systematic
investigation of brightness temperature spatial
and temporal variability over land targets, for
example ice-sheet and forest sites, in order to
characterize these targets and provide a method
for selecting optimum sites for long-term
spaceborne radiometer system calibration
monitoring over land. Such sites may be
instrumented and maintained as calibration sites
for future missions such as CMIS and Hydros.