SPATIAL PATTERNS AND TIMING STATISTICS OF RAINFALL-INDUCED DISCRETE LANDSLIDE EVENTS AT THE CATCHMENT SCALE

Rainfall induced shallow landslides often exhibit a wide size distribution with landslide volumes spanning several orders of magnitude. In addition to key attributes for landslide triggering namely, slope, soil type, cover and hydrologic conditions, landslide statistics are influenced by the nature of triggering described as progression of local and often benign failure events that may culminate into abrupt mass release. We developed a catchment scale hydro-mechanical model that computes temporal and spatial distributions of hydrologic attributes (infiltration, surface runoff routing, subsurface flows) that affect force balance, soil strength and load redistribution among soil elements. A unique feature of the model is ability to track local failure initiation and propagation preceding rapid landslides where load redistribution may exceed mechanical threshold. The model was applied for two event-based landslide inventories located at the foothills of the Swiss Alps. We made use of precipitation radar data to simulate triggering events and compared resulting (localized) frequency/magnitude statistics with observations. We also investigated effects of rainfall variations in space and time on localization and landslide statistics. Results for homogeneous rainfall fields and for spatially and temporally variable rainfall affecting landslide localization patterns are compared also for occurrence dynamics (timing during a storm). Effects of soil heterogeneity and vegetation cover were included in the scenario generation. The model provides highly resolved susceptibility maps with spatial density of discrete events (of certain size/volume) and temporal susceptibility that may evolve with storm duration and intensity.