Combining rainfall parameter and landslide susceptibility to forecast shallow landslide in Taiwan

Abstract

Catastrophic landslides and debris slides triggered by typhoons such as Typhoon Morakot (2009) have occurred more frequently in the recent years, and caused many casualties and much economic loss in Taiwan. For the purpose of reducing the damage and preventing loss of life resulting from geological hazards, this study collects multiple period landslide inventories which contain the information of occurrence time, location, magnitude, rainfall intensity, accumulated rainfall to establish the rainfall threshold for shallow landslides on a regional scale. This study applies the concept of a hazard matrix which combines the magnitude (landslide ratio of slope units) and the possibility of occurrence (historical disaster records) to set up the early warning thresholds. Accordingly, the critical rainfall thresholds are build up based on the R24 (24 hours cumulated rainfall) and I3 (3-hour mean rainfall intensity) of historical records. A validation result shows the model can predict the possible sediment hazard on the hillslope 2~9 hours before occurrence of landslides. The web-GIS based early-warning system is also developed to display the real-time rainfall data and assess the warning signal immediately for disaster prevention through increasing the response time.