Application of Fuzzy Logic GIS to Modelling Coseismic Landslide Susceptibility in the Southern Alps, New Zealand, from a Potential Alpine Fault Earthquake

Thursday, 18 December 2014
Tom Robinson1, Timothy R Davies1, Thomas M Wilson1, Caroline Orchiston2 and Theodosios Kritikos1, (1)University of Canterbury, Christchurch, New Zealand, (2)University of Otago, Department of Tourism, Dunedin, New Zealand
Recent earthquakes such as the 1999 Chi-Chi and 2008 Wenchuan events have demonstrated that the hazard from large earthquakes in mountains is not simply that of strong ground shaking. Coseismic landsliding can be as devastating as, or more devastating than, the initial earthquake. In mountainous areas with high seismic hazard, understanding the potential scale and spatial distribution of coseismic landsliding is therefore vital to fully describing the earthquake hazard. Currently however, estimating coseismic landslide susceptibility requires either: a substantially complete coseismic landslide inventory from an historic event in the region; densely spaced, detailed geotechnical data; expert knowledge; or some combination of these. In regions where these are not available, estimating the extent of coseismic landsliding is not currently possible.

This study uses statistical analysis of substantially complete coseismic landslide inventories from Northridge, CA and Wenchuan, China to identify common regional factors which appear to control the spatial distribution of landsliding in both locations. These factors were: shaking intensity (MMI), slope angle, distance to faults and streams, and slope position. Factors such as slope curvature, slope aspect, and elevation played no consistent role in the formation of landslides. Combining these observations with fuzzy logic in GIS we are able to successfully model landslide susceptibility for the 1999 Chi-Chi, Taiwan earthquake. This suggests that modelling susceptibility for a given earthquake scenario is possible, using observations of historic events in similar geotectonic environments. Applying the result to a potential M8 Alpine fault earthquake in New Zealand yields a susceptibility map for the entire South Island. High susceptibility is modelled across an area >50,000 km2, predominantly focussed on the western rangefront of the Southern Alps. Landsliding therefore has the potential to be widespread, presenting a range of major hazards.