Detection and volume estimation of large-scale landslides based on elevation-change analysis using DEMs extracted from high-resolution satellite stereo imagery

A new technique for quickly assessing extensive areas of large-scale landslides that uses digital elevation models (DEMs) extracted from high-resolution satellite images is presented in this paper. The proposed technique observes the elevation changes by using multitemporal DEMs. Five-meter-resolution DEMs from SPOT-5 images are applied to two large-scale landslide disasters: the landslides triggered by the 2004 Mid Niigata prefecture earthquake (magnitude 6.8; a moderate-topography area in Central Japan), and the landslides caused by the 2004 typhoon Mindulle (a steep-topography area in Central Taiwan). Both events yielded elevation changes in excess of 10 in. We assess the DEMs produced by the proposed method and their landslide application. We find three main results. 1) The elevation difference error increases with the slope angle. The root-mean-square error was 4-5 m on slopes lower than 30 degrees in both areas, whereas it was 5-6 in for slopes that exceeded 30 degrees in moderate topographies and 5-9 m in steep topographies. 2) The proposed technique well delineated the large-scale landslides. The total rate of successful area detection was over 70% for slopes under 40 degrees but under 40% for slopes that exceeded 40 degrees. 3) The landslide volume could be roughly estimated in units of 100 x 10(3) m(3). The developed technique well supports damage assessments of large-scale landslides because the location, depth, and volume can be quantitatively determined by remote sensing.