Spatially explicit shallow landslide susceptibility mapping over large areas

Recent advances in downscaling climate model precipitation predictions now yield spatially explicit patterns of rainfall that could be used to estimate shal- low landslide susceptibility over large areas. In Cali- fornia, the United States Geological Survey is explor- ing community emergency response to the possible effects of a very large simulated storm event and to do so it has generated downscaled precipitation maps for the storm. To predict the corresponding pattern of shallow landslide susceptibility across the state, we have used the model Shalstab (a coupled steady state runoff and infinite slope stability model) which susceptibility spatially explicit estimates of relative potential instability. Such slope stability models that include the effects of subsurface runoff on potentially destabilizing pore pressure evolution require water routing and hence the definition of upslope drainage area to each potential cell. To calculate drainage area efficiently over a large area we developed a parallel framework to scale-up Shalstab and specifically intro- duce a new efficient parallel drainage area algorithm which produces seamless results. The single seamless shallow landslide susceptibility map for all of Califor- nia was accomplished in a short run time, and indicates that much larger areas can be efficiently modelled. As landslide maps generally over predict the extent of in- stability for any given storm. Local empirical data on the fraction of predicted unstable cells that failed for observed rainfall intensity can be used to specify the