9 Alaska SAR Facility: The US Science Center for Sea Ice SAR Data
197
and old or multiyear ice (ice that has survived a summer melt) (Kwok et al.1992). Some
programs additionally address the classification of pancake ice and brash ice, which
often appear at the ice margin. Finally, ice type issues are confusing in summer conditions when the thinner ice species disappear and the thicker change in form and surface character (Gogineni et al. 1992; Holt and Digby 1985).
Ice concentration, that is, one minus the fraction of the ice-covered seas which is actually open water, is a variable often mentioned and rarely if ever effectively defined or
conclusively determined. The expected behavior of the open water fraction has not been
seriously examined. Using SAR data it is possible to determine some kinds of open water
areas, e.g., a freezing lead in which Langmuir cells are organizing the newly formed ice
bits, but there are areas of ambiguity, e.g., smooth open water looks the same as smooth
new ice within the noise limits of most SAR systems. Various ambiguities of open water
determination are characteristic of all satellite systems including visible-light, infrared,
and passive microwave. An advantage of sequential SAR observations is that open water
production due to deformation and divergence can be resolved with good accuracy if
repeat data of sufficient timeliness can be obtained.
Ice motion is the key sea ice variable obtainable from SAR data (Holt et al.1992), second
in overall importance only to ice extent as a polar ocean surface variable. The method is
simple in principl: features in the backscatter field in the SAR image are tracked in images
acquired with an acceptably short time interval so that the motion of interest is resolved.
At ASF the ice tracking concept has had a trial using a dedicated processor called the Geophysical Processing System (GPS); results from it have been discussed at length in Stern et
al. (1995). The GPS addressed ice motion in the Beaufort Sea, primarily in winter; there
were the conditions for which good quality algorithms were available (Holt et al.1992).
RADARSAT SAR data are expected to increase the capability to monitor sea ice
through an increase in coverage arising from the wider swath in the ScanSAR mode
(see Table 2) and the use of the onboard recorder. Additionally, improvements in the
analysis algorithms have extended the ice tracking capability seasonally and spatially.
Consequently, the RADARSAT Geophysical Processing System (RGPS) aims at determining and analyzing ice motion for all the Arctic Ocean ice for all seasons (Kwok et
al. 1995), although summer ice edge kinetics are still problematic. The RGPS will convert Lagrangian ice motion data into derived fields of ice age and thickness (for firstyear ice) for the entire Arctic Basin; this aggressive program is the first use of satellite
radar data to obtain a large-scale geophysical product on a routine basis.
The attempts so far to resolve ice motion in the Southern Ocean have been restricted to special periods and locations, due both to tracking algorithm challenges and to
data availability. These constraints are also being removed for the Southern Ocean, however, and in coming years it is likely that an Antarctic RGPS (ARGPS) will be implemented. The ice velocity and deformation environment in the Southern Ocean are
appreciably different from those in the Arctic, however, and the required data products
for ARGPS will be different from those of RGPS.
The ice-covered seas are locations of operations including shipping, fishing, and
petroleum exploration and production (see, e.g., Mulherin et al. 1996). In all there areas
ice is a key impediment and often a hazard. To operate safely and efficiently in these
regions it is important to have information on ice conditions in the immediate area of
operations, several-day ice forecasts for a larger region, and general forecasts for basinwide areas. This information is best acquired through data gathering and modeling of
197
and old or multiyear ice (ice that has survived a summer melt) (Kwok et al.1992). Some
programs additionally address the classification of pancake ice and brash ice, which
often appear at the ice margin. Finally, ice type issues are confusing in summer conditions when the thinner ice species disappear and the thicker change in form and surface character (Gogineni et al. 1992; Holt and Digby 1985).
Ice concentration, that is, one minus the fraction of the ice-covered seas which is actually open water, is a variable often mentioned and rarely if ever effectively defined or
conclusively determined. The expected behavior of the open water fraction has not been
seriously examined. Using SAR data it is possible to determine some kinds of open water
areas, e.g., a freezing lead in which Langmuir cells are organizing the newly formed ice
bits, but there are areas of ambiguity, e.g., smooth open water looks the same as smooth
new ice within the noise limits of most SAR systems. Various ambiguities of open water
determination are characteristic of all satellite systems including visible-light, infrared,
and passive microwave. An advantage of sequential SAR observations is that open water
production due to deformation and divergence can be resolved with good accuracy if
repeat data of sufficient timeliness can be obtained.
Ice motion is the key sea ice variable obtainable from SAR data (Holt et al.1992), second
in overall importance only to ice extent as a polar ocean surface variable. The method is
simple in principl: features in the backscatter field in the SAR image are tracked in images
acquired with an acceptably short time interval so that the motion of interest is resolved.
At ASF the ice tracking concept has had a trial using a dedicated processor called the Geophysical Processing System (GPS); results from it have been discussed at length in Stern et
al. (1995). The GPS addressed ice motion in the Beaufort Sea, primarily in winter; there
were the conditions for which good quality algorithms were available (Holt et al.1992).
RADARSAT SAR data are expected to increase the capability to monitor sea ice
through an increase in coverage arising from the wider swath in the ScanSAR mode
(see Table 2) and the use of the onboard recorder. Additionally, improvements in the
analysis algorithms have extended the ice tracking capability seasonally and spatially.
Consequently, the RADARSAT Geophysical Processing System (RGPS) aims at determining and analyzing ice motion for all the Arctic Ocean ice for all seasons (Kwok et
al. 1995), although summer ice edge kinetics are still problematic. The RGPS will convert Lagrangian ice motion data into derived fields of ice age and thickness (for firstyear ice) for the entire Arctic Basin; this aggressive program is the first use of satellite
radar data to obtain a large-scale geophysical product on a routine basis.
The attempts so far to resolve ice motion in the Southern Ocean have been restricted to special periods and locations, due both to tracking algorithm challenges and to
data availability. These constraints are also being removed for the Southern Ocean, however, and in coming years it is likely that an Antarctic RGPS (ARGPS) will be implemented. The ice velocity and deformation environment in the Southern Ocean are
appreciably different from those in the Arctic, however, and the required data products
for ARGPS will be different from those of RGPS.
The ice-covered seas are locations of operations including shipping, fishing, and
petroleum exploration and production (see, e.g., Mulherin et al. 1996). In all there areas
ice is a key impediment and often a hazard. To operate safely and efficiently in these
regions it is important to have information on ice conditions in the immediate area of
operations, several-day ice forecasts for a larger region, and general forecasts for basinwide areas. This information is best acquired through data gathering and modeling of