Regional landslide susceptibility: spatiotemporal variations under dynamic soil moisture conditions

Quantification of landslide susceptibility variability in space and time in response to static and dynamic conditions is a fundamental research challenge. Here, we identify and apply new modeling and remote sensing observation techniques to statistically characterize susceptibility distributions under dynamic moisture conditions. The methods are applied at two study regions: Cleveland Corral, California, US and Dhading, Nepal. The results show that the temporal variability of safety factors is lower during the wet season than the dry season, but this variability, when scaled by mean seasonal stability, is constant annually. Relative variability differs by region with lower variability in Nepal, the highly susceptible region. L-Moment evaluations indicate that Nepal has a consistent, regional probability distribution, but that California has two distinct distributions. The variability in time is not normally distributed for either region. For both regions, transitional characteristic of safety factors show a strong power law relationship between the average duration and number of periods during which sites are highly susceptible. Because the mapped landslide locations typically had frequent crossings with brief unstable conditions, a consistent physical mechanism is pointed to as a possible cause of slope failure.