NH42A-07:
Analyzing Shallow Landslides and Rainfall Conditions in Taiwan: An Application of the Soil Water Index
Thursday, 18 December 2014: 11:50 AM
Chi-Wen Chen1, Hitoshi Saito2,3 and Takashi Oguchi3,4, (1)University of Tokyo, Natural Environmental Studies, Bunkyo-ku, Japan, (2)Kanto Gakuin University, College of Economics, Yokohama, Japan, (3)University of Tokyo, Center for Spatial Information Science, Kashiwa, Japan, (4)University of Tokyo, Natural Environmental Studies, Kashiwa, Japan
Abstract:
Rainfall-induced landslides are significant natural hazards in Taiwan. This study analyzes 282 shallow landslides in Taiwan during 2006-2013 to identify the rainfall conditions responsible for landslides. We used the soil water index (SWI) which can represent the conceptual soil water contents as influenced by present and antecedent rainfall. SWI is used by the Japan Meteorological Agency to assess landslide hazards in Japan. Previous studies show that SWI can successfully predict the occurrence of landslides but only for Japan. Therefore, this study examines whether SWI can be also applied to Taiwan. We used the landslide data in 2006-2012 for analyses and those in 2013 for verification. The values of SWI before the rainfall events which triggered landslides were used as the indicator of the antecedent rainfall condition. We found that under different values of SWI before rainfall events, the rainfall conditions needed for triggering landslides, such as the rainfall intensity and duration, are different. Then we classified rainfall condition for triggering landslides into two types, short duration-high intensity (SH) and long duration-low intensity (LL), based on SWI and the principal component analysis. The SH type is associated with a rapid increase of SWI with short duration, and the LL type is with a gradual rise and subsequent constancy of SWI. Based on this result, we modeled the general trend of changes in SWI for the two types. We then verified the model by analyzing 19 landslides that occurred in 2013 with 14 SH types and five LL types. We also checked hourly changes in SWI for these 19 events and found that they all followed the general trend of the inferred SH and LL curves. Based on these results, it would be possible to predict a landslide of the SH or LL type, at an early stage of a rainfall event. Our results indicate that SWI is applicable to Taiwan in assessing regional landslide hazards.
