Global Soil Water Estimates as Landslide Predictor: The Effectiveness of SMOS, SMAP, and GRACE Observations, Land Surface Simulations, and Data Assimilation

Abstract This global feasibility study assesses the potential of coarse-scale, gridded soil water estimates for probabilistic modeling hydrologically triggered landslides, using Soil Moisture Ocean Salinity (SMOS), Active Passive (SMAP), and Gravity Recovery Climate Experiment (GRACE) remote sensing data; Catchment Land Surface Model (CLSM) simulations; six data products based on assimilation SMOS, SMAP, and/or GRACE observations into CLSM. SMOS or SMAP (~40-km resolution) are only available less than 20% globally reported landslide events, because they intermittent uncertain in regions with complex terrain. terrestrial storage include 75% landslides but have coarse spatial temporal resolutions (monthly, ~300 km). CLSM simulations added advantage complete coverage, found to be able distinguish between “stable slope” (no landslide) conditions landslide-inducing a way. Assimilating increases probability percentiles relative model-only at 36-km resolution period 2011–16, unless content is already high (≥50th percentile). The Level 4 product (at 9-km resolution, 2015–19) more generally updates toward higher probabilities similar majority where cannot easily converted moisture owing