Estimating the timing and location of shallow rainfall‐induced landslides using a model for transient, unsaturated infiltration

TLDR

Shallow rainfall‑induced landslides commonly occur under conditions of transient infiltration into initially unsaturated soils. The authors developed a two‑layer infiltration model—an unsaturated zone above a saturated zone—implemented within a GIS framework to predict landslide timing and location. The model couples analytical solutions for transient, unsaturated vertical infiltration above the water table with pressure‑diffusion solutions below, linking them through a rising transient water table, and uses the resulting pore pressures in one‑dimensional slope‑stability calculations. The simplified scheme delivers faster spatially distributed simulations, accurately computes cell‑wise factors of safety during rainstorms, reproduces observed attenuation of pore pressures, predicts realistic instability timing (≈7 h earlier than observed), and identifies primary landslide source areas near Seattle.

Abstract

Shallow rainfall‐induced landslides commonly occur under conditions of transient infiltration into initially unsaturated soils. In an effort to predict the timing and location of such landslides, we developed a model of the infiltration process using a two‐layer system that consists of an unsaturated zone above a saturated zone and implemented this model in a geographic information system (GIS) framework. The model links analytical solutions for transient, unsaturated, vertical infiltration above the water table to pressure‐diffusion solutions for pressure changes below the water table. The solutions are coupled through a transient water table that rises as water accumulates at the base of the unsaturated zone. This scheme, though limited to simplified soil‐water characteristics and moist initial conditions, greatly improves computational efficiency over numerical models in spatially distributed modeling applications. Pore pressures computed by these coupled models are subsequently used in one‐dimensional slope‐stability computations to estimate the timing and locations of slope failures. Applied over a digital landscape near Seattle, Washington, for an hourly rainfall history known to trigger shallow landslides, the model computes a factor of safety for each grid cell at any time during a rainstorm. The unsaturated layer attenuates and delays the rainfall‐induced pore‐pressure response of the model at depth, consistent with observations at an instrumented hillside near Edmonds, Washington. This attenuation results in realistic estimates of timing for the onset of slope instability (7 h earlier than observed landslides, on average). By considering the spatial distribution of physical properties, the model predicts the primary source areas of landslides.