Rapid post-earthquake modelling of coseismic landslide intensity and distribution for emergency response decision support

Abstract

Abstract. Current methods to identify coseismic landslides immediately after an earthquake using optical imagery are too slow to effectively inform emergency response activities. Issues with cloud cover, data collection and processing, and manual landslide identification mean even the most rapid mapping exercises are often incomplete when the emergency response ends. In this study, we demonstrate how traditional empirical methods for modelling the total distribution and relative intensity (in terms of point density) of coseismic landsliding can be successfully undertaken in the hours and days immediately after an earthquake, allowing the results to effectively inform stakeholders during the response. The method uses fuzzy logic in a GIS (Geographic Information Systems) to quickly assess and identify the location-specific relationships between predisposing factors and landslide occurrence during the earthquake, based on small initial samples of identified landslides. We show that this approach can accurately model both the spatial pattern and the number density of landsliding from the event based on just several hundred mapped landslides, provided they have sufficiently wide spatial coverage, improving upon previous methods. This suggests that systematic high-fidelity mapping of landslides following an earthquake is not necessary for informing rapid modelling attempts. Instead, mapping should focus on rapid sampling from the entire affected area to generate results that can inform the modelling. This method is therefore suited to conditions in which imagery is affected by partial cloud cover or in which the total number of landslides is so large that mapping requires significant time to complete. The method therefore has the potential to provide a quick assessment of landslide hazard after an earthquake and may therefore inform emergency operations more effectively compared to current practice.