Abstract:
In a first coarse-scale global feasibility study, we assess the potential of various grid-scale soil water estimates for the probabilistic modeling of hydrologically triggered landslides, using Soil Moisture Ocean Salinity (SMOS), Soil Moisture Active Passive (SMAP) and Gravity Recovery and Climate Experiment (GRACE) remote sensing data, Catchment Land Surface Model (CLSM) simulations and various data assimilation products. Surface soil moisture retrievals from SMOS or SMAP satellite observations are only available for less than 20% of the globally reported landslide events, because they are intermittent and the observed microwave data cannot easily be converted to soil moisture in regions with complex terrain. Monthly total water storage estimates from GRACE cover 75% of the reported landslides, but their coarse spatio-temporal resolution is not ideal for landslide prediction. Spatio-temporally complete CLSM simulations reveal that soil water estimates can distinguish between stable slope and landslide conditions in a probabilistic way. By assimilating SMOS and/or GRACE data into CLSM at 36-km resolution for the period 2011-2016, the landslide probability estimates based on moisture percentiles increase relative to model-only estimates when and where model-only moisture conditions are not wet already at observed landslides. The 9-km SMAP Level 4 data assimilation product more generally updates the soil water conditions such that the landslide probabilities increase at most observed landslides for the period 2015-2019.