Control of style-of-faulting on spatial pattern of earthquake-triggered landslides

Abstract

Predictive mapping of susceptibility to earthquake-triggered landslides (ETLs) commonly uses distance to fault as spatial predictor, regardless of style-of-faulting. Here, we examined the hypothesis that the spatial pattern of ETLs is influenced by style-of-faulting based on distance distribution analysis and Fry analysis. The Yingxiu-Beichuan fault (YBF) in China and a huge number of landslides that ruptured and occurred, respectively, during the 2008 Wenchuan earthquake permitted this study because the style-of-faulting along the YBF varied from its southern to northern parts (i.e. mainly thrust-slip in the southern part, oblique-slip in the central part and mainly strike-slip in the northern part). On the YBF hanging-wall, ETLs at 4.4-4.7 and 10.3-11.5 km from the YBF are likely associated with strike- and thrust-slips, respectively. On the southern and central parts of the hanging-wall, ETLs at 7.5-8 km from the YBF are likely associated with oblique-slips. These findings indicate that the spatial pattern of ETLs is influenced by style-of-faulting. Based on knowledge about the style-of-faulting and by using evidential belief functions to create a predictor map based on proximity to faults, we obtained higher landslide prediction accuracy than by using unclassified faults. When distance from unclassified parts of the YBF is used as predictor, the prediction accuracy is 80 %; when distance from parts of the YBF, classified according to style-of-faulting, is used as predictor, the prediction accuracy is 93 %. Therefore, mapping and classification of faults and proper spatial representation of fault control on occurrence of ETLs are important in predictive mapping of susceptibility to ETLs.