Martin Sauerbier
Project Manager
Photogrammetry MFB-GeoConsulting GmbH
Switzerland

Biography
Martin Sauerbier holds a doctoral degree in geomatics and worked as a research and teaching assistant at ETH Zurich, chair of photogrammetry and remote sensing, from 2001 to 2009. Since 2010, he is responsable for technical project management, software development and photogrammetry at MFB-GeoConsulting GmbH, the swiss distributor of Intergraph ERDAS remote sensing products. At MFB-Geo, he is involved in various projects related to earth observation and image processing.

Abstract
WorldView-2 Stereo Imagery for Glacier and Lake Monitoring

Nowadays, the behaviour of glaciers, glacial lakes and their impact are important topics in environment-related sciences with respect to climate change. Additionally, glacier lakes in several cases cause flooding, which depending on the landuse in the affected areas, may result in serious hazards for human life and infrastructure. Glacier and lake monitoring therefore are required in order to provide decision makers with current and accurate information on the situation. High-resolution optical spaceborne imagery so far was not largely deployed for such purposes, maybe due to cost and the available spectral coverage most of them provide. With WorldView-2, multispectral bands combined with high geometric resolution including stereo capability are available to users. In the following, an application of WV-2 stereo data within the frame of an Early Warning System for the Shaksgam Valley located in the Xinjiang province (China) is proposed. Although mapped as a lake in topographic maps, Lake Kyagar is not permanently filled with water, in several years the lake does not even emerge, e.g. in 2011. In other years, e.g. in 1971, 1997, 1999 and 2002, the glacial lake fills rapidly and causes major flood events by its outbursts, furthermore there are also years when the lake emerges but the water flows down the valley in a controlled manner without causing damage. The flood events have a strong impact to settlements and cities located downstream near the city of Yarkant. In order to provide a reference for future lake level measurements, a DTM was generated from 2 WV-2 stereo pairs using ERDAS LPS eATE. This new module allows for the automated generation of 3D point clouds from stereo imagery featuring high density and reliability. The DTM was generated for two pairs of stereo images, covering an area of 339 km2. In general, the result produced with eATE was quite accurate, though especially in areas with low contrast (snow cover) point density was decreased. We will present the derived DTM and discuss the shortcomings and future developments for automated image-based DTM generation in critical regions.