 |
||||
|
||||
|
|
|
 |
||
|
||||
      |
GLERL 1999 Milestone Reports
GOAL: ADVANCE SHORT-TERM WARNING AND FORECAST SERVICES
OBJECTIVE 3: ENHANCE OBSERVATIONS AND PREDICTION
PM: Severe Thunderstorm Warning Lead Time And Accuracy And Other PM’s
Milestone: Complete improvements to the algorithm developed for Great Lakes ice cover classification and mapping using new field data and/or satellite SAR (Synthetic Aperture Radar) data.
SCIENTIST: G. Leshkevich
Purpose: Due to the size and extent of the Great Lakes and the variety of ice types and features found there, the timely and objective qualities inherent in computer processing of satellite data make it well suited for monitoring and mapping ice cover. However, during winter months cloud cover over the Great Lakes impairs the use of satellite imagery from passive sensors operating in the visible, near infrared, and thermal infrared regions and passive microwave data currently lacks the spatial resolution required for Great Lakes ice cover monitoring and analysis. The all-weather, day/night viewing capability of satellite Synthetic Aperture Radar (SAR) makes it a unique and valuable tool for Great Lakes ice identification and mapping providing that data analysis techniques can be developed. The European Remote-Sensing Satellite (ERS-1) SAR with vertical polarization launched in 1991 and more recently RADARSAT, an operational satellite carrying a SAR operating at 5.3 GHz (C-Band) with horizontal polarization launched in 1995, provide an opportunity for this development.
Using airborne and shipborne data as "ground truth", preliminary computer analysis of ERS-1 and RADARSAT ScanSAR narrow images of the Great Lakes using a supervised (level slicing) classification technique indicates that different ice types in the ice cover can be identified and mapped. During the 1997 winter season, shipborne polarimetric backscatter data were acquired using the Jet Propulsion Laboratory (JPL) C-band scatterometer, together with aerial reconnaissance data, surface-based ice physical characterization measurements, and environmental parameters, concurrently with RADARSAT and ERS-2 satellite overpass. The scatterometer data set, composed of over 20 ice types or variations measured at incident angles from 0o to 60o for all polarization’s, was processed to radar cross-section and establishes a library of signatures (look-up table) for different ice types to be used in the machine classification of calibrated satellite SAR data. This method is used to obtain ice classification maps from ERS-2 SAR data.
This milestone was undertaken to improve, validate, and complete the initial development of the ice classification and mapping algorithm developed using ERS-2 satellite SAR (Synthetic Aperture Radar) imagery and bring it to an operational state. The goals were to validate the algorithm using shipborne observations and add missing ice types to the library of signatures previously developed. In addition, the algorithm was to be tested using operational RADARSAT ScanSAR Wide A imagery which was projected to be calibrated by the start of the project. Used on operational RADARSAT imagery, the algorithm would improve ice observation and prediction on the Great Lakes. The resulting data can be used for near real-time ice mapping (navigation and ice breaking) and in ice forecasting and prediction models, as well as for climatological and other uses.
Efforts: Arrangements for ship (ice breaker) support on Lake Superior were made with the U.S. Coast Guard. In addition, calibration efforts for RADARSAT ScanSAR Wide A imagery were being advocated and moved forward at the Jet Propulsion Laboratory. As a result, a set of calibration imagery was collected over the Canadian Boreal "Taiga" Forest.
Customers: Users of the satellite SAR ice classification/mapping algorithm would include the National Ice Center and the U.S. Coast Guard and National Weather Service via dissemination through the NOAA CoastWatch Great Lakes Program.
Significance: This effort would bring the ice algorithm to an initial usable or operational state using operational RADARSAT SAR imagery received via the North American Ice Link (with Canada).
Success: This project was not completed because of two major factors: 1) Last year was the mildest ice year on the Great Lakes in recent history, especially on Lake Superior, which developed little more than shore ice. This resulted in cancellation of the field work for validation. 2) A change in priorities at the Jet Propulsion Laboratory postponed the RADARSAT ScanSAR Wide A calibration effort. RADARSAT International (RSI) did not have a calibration algorithm for its ScanSAR Wide A imagery. Calibrated SAR imagery is needed for the ice classification/mapping algorithm, testing and further development with RADARSAT imagery could not proceed.
Next Steps: It is expected that RADARSAT ScanSAR Wide A imagery will be calibrated by the upcoming ice season (by the Alaska SAR Facility and/or Canadian agencies). Pending a more normal ice season on Lake Superior and with calibrated RADARSAT ScanSAR Wide A imagery, efforts to validate and improve the Great Lakes ice algorithm can be expected to continue. In addition, the fully polarimetric data set collected can be tested on ENVISAT dual-polarized SAR imagery when available.
Last updated: July 9, 2002 mbl