Landslide hazard analysis – a case study in WuShe reservoir catchment

Abstract:

A complete landslide inventory covering a long time span is the foundation for landslide hazard analysis. Previous studies only estimate landslide susceptibility because enough landslide inventory was usually unavailable. This study collects the spot image of ten events from 1994 to 2009 of WuShe reservoir catchment in central Taiwan. All of landslide inventories were manual interpretation from spot image. The recent and expanded landslide inventory due to the event was identified from the landslide inventory of pre and post event. The recent landslide inventory was used to conduct landslide hazard evaluation.

This study follows the landslide hazard analysis framework of CNR-IRPR in Italy, which had demonstrated how the spatial probability, temporal probability and size probability of each slope-unit were calculated. Landslide hazard in WuShe reservoir catchment was obtained following the procedure. The preliminary validated result shows that the prediction of landslide hazard in high landslide hazard region was better, when compared with landslide susceptibility.

However, the reliability of temporal probability and size probability were related to the number of landslide inventories. When no landslide was recorded, or lack of suitable landslide inventory, would lead to the incorrect analysis of landslide hazard. The event based spatial probability was affected by rainfall distribution, but the rainfall distribution of different typhoons were usually highly variable. Next phase of this study would attempt to discuss these issues and develop a suitable framework for landslide hazard analysis in Taiwan.