The key to advancing the prediction and predictability of many landslides is the transient response of the subsurface moisture to spatio-temporal patterns of precipitation that satellite platforms such as the TRMM and the future GPM provide. These precipitation data, however, are only useful for landslide hazard forecasting if physically-based hydrology models that are capable of capturing these transient dynamics are used to interpret the spatially complex precipitation. Using models to forecast landslides over large regions in reasonable detail requires knowledge of topography, vegetation/land cover, and rainfall at resolutions of at least large hillslopes. NASA data have facilitated the necessary terrain assessment by deploying a network of sensors to observe topography (SRTM) and vegetation/land cover and its change (MODIS, LANDSAT) over large regions with sufficient spatial resolution and adequate revisit intervals. This proof-of-concept study will test the following hypotheses: 1. Because TRMM/GPM data capture complex spatial patterns in precipitation and provide frequent observations, they can be used with a distributed hydrology model to capture the spatio-temporal distribution of soil moisture at resolutions consistent with relevant slope stability indices. 2. Detailed process modeling of spatiotemporal patterns of infiltration and lateral moisture redistribution using TRMM/GPM precipitation input as forcing significantly improves the skill of predictions of landslide timing, location and spatial extent. The proponents will focus on small tropical montane watersheds of the Luquillo Experimental Forest (LEF) in northeast Puerto Rico. This site is rich in data and its landslides have been extensively studied. It is well sampled by TRMM observations and precipitation ground truth is available from raingages and meteorological radar. Satellite precipitation will be used to drive an advanced distributed model of hydrology and vegetation (tRIBS-VEGGIE) in order to predict regions of high landslide risk and the potential impact. Predictions will be verified against observations.