230
C. BERTOIA, J. FALKINGHAM, F. FETTERER
was viewed on two different occasions. The area coverage with ERS-1 was usually too
small to be operationally useful, although in a small case study Fetterer and Gineris
(1992) found vectors from SAR to be much more reliable than those from AVHRR
images. While AVHRR images provide a wider view of ice conditions (under clear sky
conditions), errors due to inaccuracy of geolocation and the coarse sensor resolution
were unacceptably high, especially under light wind conditions.
CIS uses an ice tracking algorithm called TRACKER that has similar area-matching
lineage to the one used by NIC. A form of this algorithm was used for ice tracking with
AVHRR imagery (Heacock et al. 1992). The algorithm has undergone development to
address operational issues such as time and cost limitations, varying needs for accuracy, and varying operator skill levels (Marko and Gower 1992). In an operational demonstration, ice tracking in the Gulf of st. Lawrence was attempted with ERS-1 data; however, the satellite three-day ground track repeat period was too long to successfully capture rapidly moving ice. A demonstration within the confines of Lake Superior was
much more successful. Research sponsored by CCRS and the Canadian Coast Guard to
derive ice pressure products from tracked ice displacement as an aid to navigation is
currently underway for the Gulf of st. Lawrence. These products will include a modeled forecast component.
Initial runs of TRACKER at CIS with RADARSAT ScanSAR indicate that the wide
swath eliminates many tracking problems; however, both NIC and CIS consider the value of SAR-derived ice motion vectors to be in regional or tactical scale support for maritime operations. For basin scale ice drift, a dynamic/thermodynamic model such as
the Polar Ice Prediction System (PIPS; Preller and Posey 1989) for the central Arctic and
a free drift model at the ice edge perform sufficiently well (when unencumbered by
coastal dynamics). Models also have the advantage of providing the predictive capabilities desired by the operational centers, who are interested in where the ice is going
to go, not in where it has been.
10.9
Problems and Potential
Operational determination of the extent and thickness distribution of sea ice has been
undertaken by several countries since the turn of the century. Since that time, operational sea ice mapping techniques have evolved as sensor capabilities for imaging ice
have improved, but these techniques have been based primarily on the reasoning developed for the interpretation of visible and infrared imagery. These techniques will need
refinement in order for the ice centers to meet their operational requirements using
active radar imagery.
There is enormous potential for the application of SAR interpretation techniques to
operational analysis of sea ice in the next few years. At the time of this writing, operational use of data from the Canadian RADARSAT satellite has just begun. The future of
space borne SAR for monitoring ice conditions looks relatively optimistic with several
missions well into the planning stages. Canada plans to launch RADARSAT -2, a copy of
its predecessor, around the year 2000. Long lead research for RADARSAT-3, a new-generation SAR satellite scheduled for launch in 2005, has commenced. ESA intends to
launch Envisat in 1998 with a C-band ''Advanced'' SAR (ASAR) that will have a swath
width of about 400 km and resolution similar to ERS-1 and RADARSAT. NASA has
C. BERTOIA, J. FALKINGHAM, F. FETTERER
was viewed on two different occasions. The area coverage with ERS-1 was usually too
small to be operationally useful, although in a small case study Fetterer and Gineris
(1992) found vectors from SAR to be much more reliable than those from AVHRR
images. While AVHRR images provide a wider view of ice conditions (under clear sky
conditions), errors due to inaccuracy of geolocation and the coarse sensor resolution
were unacceptably high, especially under light wind conditions.
CIS uses an ice tracking algorithm called TRACKER that has similar area-matching
lineage to the one used by NIC. A form of this algorithm was used for ice tracking with
AVHRR imagery (Heacock et al. 1992). The algorithm has undergone development to
address operational issues such as time and cost limitations, varying needs for accuracy, and varying operator skill levels (Marko and Gower 1992). In an operational demonstration, ice tracking in the Gulf of st. Lawrence was attempted with ERS-1 data; however, the satellite three-day ground track repeat period was too long to successfully capture rapidly moving ice. A demonstration within the confines of Lake Superior was
much more successful. Research sponsored by CCRS and the Canadian Coast Guard to
derive ice pressure products from tracked ice displacement as an aid to navigation is
currently underway for the Gulf of st. Lawrence. These products will include a modeled forecast component.
Initial runs of TRACKER at CIS with RADARSAT ScanSAR indicate that the wide
swath eliminates many tracking problems; however, both NIC and CIS consider the value of SAR-derived ice motion vectors to be in regional or tactical scale support for maritime operations. For basin scale ice drift, a dynamic/thermodynamic model such as
the Polar Ice Prediction System (PIPS; Preller and Posey 1989) for the central Arctic and
a free drift model at the ice edge perform sufficiently well (when unencumbered by
coastal dynamics). Models also have the advantage of providing the predictive capabilities desired by the operational centers, who are interested in where the ice is going
to go, not in where it has been.
10.9
Problems and Potential
Operational determination of the extent and thickness distribution of sea ice has been
undertaken by several countries since the turn of the century. Since that time, operational sea ice mapping techniques have evolved as sensor capabilities for imaging ice
have improved, but these techniques have been based primarily on the reasoning developed for the interpretation of visible and infrared imagery. These techniques will need
refinement in order for the ice centers to meet their operational requirements using
active radar imagery.
There is enormous potential for the application of SAR interpretation techniques to
operational analysis of sea ice in the next few years. At the time of this writing, operational use of data from the Canadian RADARSAT satellite has just begun. The future of
space borne SAR for monitoring ice conditions looks relatively optimistic with several
missions well into the planning stages. Canada plans to launch RADARSAT -2, a copy of
its predecessor, around the year 2000. Long lead research for RADARSAT-3, a new-generation SAR satellite scheduled for launch in 2005, has commenced. ESA intends to
launch Envisat in 1998 with a C-band ''Advanced'' SAR (ASAR) that will have a swath
width of about 400 km and resolution similar to ERS-1 and RADARSAT. NASA has