 |
||||
|
||||
|
|
|
 |
||
|
||||
      |
GLERL 2002 Milestone Reports
GLERL 2002 Milestone Home
GOAL: SUSTAIN HEALTHY COASTS
OBJECTIVE 2: Promote Clean Coastal Waters to Sustain Living Marine Resources and to Ensure Safe Recreation, Healthy Seafood and Economic Vitality
PM: Number of coastal and Great Lake states provided with improved predictive capabilities and understanding of coastal processes
Milestone: Complete and report on investigations of classifying and mapping Great Lakes ice cover using satellite synthetic aperture radar (SAR) data.
Scientist: G. Leshkevich
Purpose: The purpose of this study was to develop algorithms to classify and map Great Lakes ice cover using satellite SAR data. The goal was to collect calibrated polarimetric SAR signatures (C-band) in the field (Lake Superior) to form a library of backscatter signatures for different ice types. This knowledge is important for the development of objective, automated machine interpretation and mapping of Great Lakes ice cover.
Effort: In cooperation with a colleague at the Jet Propulsion Laboratory, calibrated polarimetric C-band backscatter signatures (0-60 degree incidence angles) from different ice types on Lake Superior were recorded using a C-band scatterometer mounted on a U.S. Coast Guard icebreaker during February and March, 1997. Preliminary analysis of ERS-2 and RADARSAT satellite SAR data recorded during the field experiment has been conducted using the library of signatures collected.
Customers: This study was funded through the NOAA Coastal Ocean Program. Users of the results of this study would include National Ice Center (NIC), U.S. Coast Guard, and National Weather Service.
Significance: This was the first attempt to measure polarimetric C-band backscatter signatures from different freshwater ice types in the field in order to develop a library of signatures (lookup table). This unique data set can be used to machine classify and map Great Lakes ice cover from calibrated satellite C-band SAR data including ERS-2, RADARSAT, and (recently launched) ENVISAT data.
Success: Calibrated ERS-2 satellite SAR data recorded at the time of the experiment was analyzed using signatures for a subset of ice types and open water from the library of C-band signatures collected in the field. Most of the ice cover in the two ERS-2 scenes were classified using the subset, however open water was not classified in the scenes. After a more thorough calibration of the ERS-2 scenes (correcting for power loss) and accounting for local incidence angle, a better classification of the scenes resulted compared to ground truth data collected during the field experiment. Open water was classified after the re-calibration.
Owing to the delayed development of calibration algorithms for RADARSAT ScanSAR Wide A data, RADARSAT data collected during the experiment was subsequently calibrated and preliminary analysis has been completed. Classification of a scene encompassing the same area as the ERS-2 scene was performed using the same open water and ice types. Although most of the scene was classified, comparison with the ERS-2 classified scene showed some differences that may be attributed to different times of data acquisition, different polarizations, or calibration accuracy.
Next Steps: Working cooperatively with the John Hopkins University Applied Physics Laboratory, calibration software for the RADARSAT ScanSAR Wide A block-averaged data received in near real-time from the Canadian Ice Center via the North American ice link was acquired and implemented at GLERL. This will allow calibration of this modified (block averaged) imagery and testing of an ice classification product using the C-band signature library. Further testing and validation of the ice classification algorithm using RADARSAT and the dual polarized ENVISAT data are planned.
Milestone Home CoastWatch Program Page
Last updated: October 4, 2002 mbl