A Statistical Model for the Impact of the 1999 Chi-Chi Earthquake on the Subsequent Rainfall-induced Landslides

博士 === 國立成功大學 === 地球科學系碩博士班 === 101 === A statistical model is proposed for the temporal influence of the 1999 Chi-Chi earthquake on the subsequent rainfall-induced new landslides in the Jhuoshuei River Watershed of Central Taiwan. Based on the landslide inventory via satellite image classifications...

Full description

Bibliographic Details
Main Authors: Shou-HengLiu, 劉守恆
Other Authors: Ching-Weei Lin
Format: Others
Language:en_US
Published: 2013
Online Access:http://ndltd.ncl.edu.tw/handle/87947964808203115173
Description
Summary:博士 === 國立成功大學 === 地球科學系碩博士班 === 101 === A statistical model is proposed for the temporal influence of the 1999 Chi-Chi earthquake on the subsequent rainfall-induced new landslides in the Jhuoshuei River Watershed of Central Taiwan. Based on the landslide inventory via satellite image classifications and daily precipitation records from 1996 to 2008, the parameters of the model functions were obtained by genetic algorithm. The proposed model gives estimation from daily precipitation with an average error of 13.91% of new landslide density to the variation analysis result from landslide inventory over 15 image dates. The results of the model estimation reveal a continuously decaying pattern of earthquake influence on the subsequent rainfall-induced new landslides, an average daily 0.004‰ background new-landslide density, and a 4.97‰ new-landslide density caused by the earthquake directly within the study area. Also, the estimations of the models suggest that there are influences and latency of antecedent rainfall among the occurrence of new landslides. Estimations from time-decay earthquake-impact-factor models show consistent results on the decay constant of 0.0012, which is equivalent to a 578-day (1.58-year) half-life for the influence of an earthquake. Model validations show more representative estimations on short-term (i.e., days) precipitation records and decreasing function for the impact factor of antecedent rainfall. However, constant-impact model of antecedent rainfall is recommended if there is no sufficient observations.