A REGIONAL MODEL OF LANDSLIDE SUSCEPTIBILITY USING LANDLAB AND MACRO-SCALE HYDROLOGIC SIMULATIONS

Landslides are key landscape processes shaping steep terrain, yet remain elusive to predict. Physically-based numerical models of hillslope stability provide a path forward, but are challenging to parameterize and apply at regional scales, where field investigation is often unfeasible. We employ the newly developed Landlab platform, an open-source, Python-based earth systems modeling environment, to deploy a numerical model of hillslope stability. An infinite-slope limited-equilibrium landslide model predicts spatial and temporal probability of failure over large regions driven by groundwater recharge derived from the Variable Infiltration Capacity (VIC) model, a distributed macro-scale hydrology model. Uncertainties in local soil conditions, such as soil properties, are addressed through Monte Carlo simulations of the factor-of-safety stability index. Predictions are further refined by an optional geomorphic soil depth evolution model that estimates distributed soil depth with greater spatial variability than conventional soil survey map units. The coupled hydro-geomorphic model advances the characterization of the landslide probability without detailed data collection, and captures more of the uncertainty intrinsic in regional landslide assessments. We demonstrate the model in North Cascade National Park Complex, a rugged terrain with nearly 2,700 m (9,000 ft) of vertical relief, covering 2,757 km² (1,064 mi²) in northern Washington State, U.S.A. The results will aid resource management decision-making in current and future climatic and landscape conditions.