AUTOMATED MAPPING OF SCREE-COVERED GLACIERS USING SPACE IMAGERY

M.F. Buchroithner, T. Bolch

Dresden University of Technology, Institute for Cartography, Dresden, Germany

manfred.buchroithner@tu-dresden.de

Many glacier tongues in the Himalaya are heavily covered by supra-glacial debris. This makes the automated mapping of the actual ice snout by space imagery difficult due to the similar spectral signal of the surrounding debris. The most significant features which distinguish the debris-covered glacier parts are their typical surface characteristics, such as a rough surface and frequent ablation ponds. These characteristics are caused by the movement and the melting of the glacier ice. Recent glacier recession results even in an increasing debris coverage. However, also within the debris-covered glacier tongues are significant differences. E.g. the surface especially of the distal part of the glaciers looks smoother and shows no significant indices for movement.

Therefore, multi-temporal ASTER stereo-imagery would represent an ideal tool to develop an automated way to delineate the extents of the active and inactive glacier. Combining ASTER´s thermal information with various shape parameters derived from stereo models, the active and the inactive parts as well as the marginal moraines could be obtained. Automated velocity measurement using ASTER data of 2001 – 2003 verified and improved the morphometry based glacier mapping. Clearly, mainly due to the resolution of the ASTER DEM (30m) this concept is only promising for large glaciers, such as the Khumbu glacier at Mt. Everest, and the accuracy is in the order of 5%. In order to further improve methods of pattern recognition are included in the model. The first results are promising. In addition, when high resolution DEMs (e.g. form the TanDEM-X-Mission) will be widely available, the accuracy will be sufficient for real automated glacier monitoring, including smaller glaciers.