|
|
|
Nowcast Products
WINDSAT
|
Click on the image to view the large picture:
|
 |  |
WINDSAT data are being processed by the Remote Sensing Division of the Naval Research
Laboratory (NRL) to produce sea ice retrievals for evaluation and analysis
at NIC. WINDSAT sea ice concentrations are retrieved using a similar methodology as the
SSM/I NASA Team algorithm and employing spectral polarization ratio (PR) and
spectral gradient ratio (GR) from brightness temperatures (TBs) measured by 18.7V,
18.7H, and 37V channels. The tie-point TBs for the first-year and multi-year ice
in the Arctic region are determined by matching PR - GR values between SSM/I and
WINDSAT data. For the Antarctic region, a proxy algorithm to SSM/I NASA Team 2
is used. The tie-point TBs for "Type A' " are chosen at the local maximum in the joint
probability density distribution in the PR - GR space near "Type A". The tie-point
TBs for proxy "Type C" are determined by selecting a point along the second ridge
between "Type B" and freezing "Open Water" in the joint probability density distribution
in the PR - GR space. The tie-point TBs for freezing "Open Water" near ice edges are
determined by the low 25 percentile of the minimum TBs from all WINDSAT orbits from
February 2003 to November 2005.
A weather filter set is employed to reduce spurious ice retrievals over open water
due to weather contaminations from wind, rain, cloud liquid water, water vapor and
other atmospheric interference. The filter set is determined by relating threshold
GR(37v19v) and GR(23v19v) values to PR(19v19h), coupled with relating threshold
GR(37v19v) values to GR(23v19v). The filter set first excludes "open water" points,
then identifies "sea ice retrievable" points and leaves a small fraction of
"undetermined" points in between.
To further reduce spurious ice retrievals, WINDSAT sea ice concentrations are
retrieved in climatologically possible sea ice regions with identified contaminations,
up to 40%, from land and permanent snow and ice. Possible sea ice regions are defined
by monthly maximum sea ice extent using historical sea ice data retrieved from SMMR
and SSM/I from 1978 to 2003. Contaminations from land and permanent snow and ice to
seawater are quantified by convolving WINDSAT antenna gain function within 200 km
diameter with 1-km resolution land-cover databases. A two-dimensional Gaussian
distribution with half-power beam width of 35 x 53 km was used as the target antenna
gain function to obtain beam-averaged WINDSAT TBs, and subsequently the derived sea
ice concentrations. Climatology and contaminations are part of the Level 2 products.
The Level 3 daily products are obtained by bin averaging Level 2 products into 25x25
km Equal-Area Scalable Earth Grid. They include land and permanent snow and ice
contaminations. The contaminations are used to classify pixels as "land" for 61 - 100%
contamination, "coast" for 41 - 60%, "near-coast" for 21 - 40%, "far-coast" for 1 - 19%
and "open water" for 0% contamination. In the images shown here, when sea ice
concentration at a pixel exceeds certain value (as a conservative function of
contaminations from land and permanent snow and ice) in the "far-coast" and "near-coast"
zones, the pixel is displayed as ice. Otherwise, the pixel with non-zero ice concentration is masked as "land", where
low ice concentration may exist. The "land" also represents "coast" zone within and
outside the possible ice regions. The legend "oc" represents "open water" in the
climatologically ice free areas and "open water" determined by the weather filter set
within the possible sea ice areas. It also represents "near-coast" and "far-coast" zones
in the climatologically ice free areas. Finally, the legend "nd" represents pixels
without observations as well as pixels "undetermined" by the weather filter set within
the possible ice regions.
Reference:
Guoqing Lin, Li Li and Peter W. Gaiser, Sea Ice Retrievals from WindSat Data , Proceedings
of the IEEE International Geoscience And Remote Sensing Symposium, Denver,
Colorado, July 31-August 4, 2006.
If you have any comments, please E-mail the National Ice Center Liaison
|
|
