Abstract
ALOS/AVNIR-2 optical images were employed to extract slope failures which occurred in the 14 June 2008 Iwate-Miyagi Inland, Japan, earthquake. Since the earthquake occurred in the mountainous area, many landslides were caused. In this paper, landslide areas were extracted comparing the pre- and post-event ALOS/AVNIR-2 images by two methods. One method uses the difference of the normalized difference vegetation index (NDVI) obtained from the pre- and post-event images. Another method performs a supervised land-cover classification by maximum likelihood approach, and the difference of extracted bare-ground pixels between the two images was recognized as landslide. The digital elevation model (DEM) was also employed to reveal the relationship between the slope angle and the occurrence of landside, and to remove the extraction errors on non-sloping ground. Comparing with the visual inspection result, the accuracy of these image processing methods was demonstrated.
