2014 GSA Annual Meeting in Vancouver, British Columbia (19–22 October 2014)

Paper No. 19-6
Presentation Time: 9:15 AM

COMBINING SPECTRAL, TOPOGRAPHIC AND SAR COHERENCE DATA WITHIN AN OBJECT BASED CLASSIFICATION ENVIRONEMENT FOR THE AUTOMATIC CLASSIFICATION OF DEBRIS COVERED ICE


ROBSON, Benjamin Aubrey1, NUTH, Christopher2, DAHL, Svein Olaf1, HÖLBLING, Daniel3, STROZZI, Tazio4 and NIELSEN, Pål Ringkjøb1, (1)Geography, University of Bergen, Fosswinckelsgate 6, Bergen, 5007, Norway, (2)Department of Geosciences, University of Oslo, Postboks 1047 Blindern, Oslo, 0316, Norway, (3)Department of Geoinfomatics -Z_GIS, University of Salzburg, Schillerstrasse 30, Salzburg, 5020, Austria, (4)GAMMA Remote Sensing, Worbstr. 225, Gümligen, 3073, Switzerland, benjamin.robson@uib.no

Up to date and accurate glacier outlines are a necessity for many applications within glaciology. Remote Sensing provides a good means for extracting glacier outlines in a semi-automatic or automatic way. Debris covered ice however remains a bottleneck in such automated techniques due to the spectral similarity of glacier debris to the surrounding bedrock, previous efforts have used topographic parameters, however accurate DEMs are often lacking for many glaciated regions. Here, for the first time, we combine SAR coherence data with Landsat 8 imagery, and topographic parameters derived from the SRTM DEM in an object based classification of the Manaslu area of Nepal. SAR coherence proves useful at differentiating glacier ice when under a thick debris cover, however the classification struggles in areas of stagnant or very slow moving ice which have a low loss of coherence, similarly steep valley sides and moraines often have large losses of coherence. Utiling spectral and topographic data in combination with SAR data therefore provides the best result. Overall the inclusion of SAR coherence data in combination with spectral and topographic information provides a marked improvement to automating the delineation of debris covered glaciers. Future classifications would benefit from having spectral, topographic and SAR data from a more similar time period.
Handouts
  • Combining Spectral, Topographic And SAR Coherence Data.pptx (19.3 MB)