Modeling Slope Stability in Honduras

Abstract

Hurricane Mitch (1998) was the deadliest storm to strike the western hemisphere in over 200 yr, and landslides were a significant component of the storm impact, inflicting direct damage on steep lands and triggering mudslides that devastated valleys below. The stability index mapping (SINMAP) model was applied to a 46.1‐km2 agricultural area in Central Honduras to predict the spatial distribution of shallow debris slides based on the infinite slope stability model and a steady‐state hydrology module. The region was surveyed for geology and depth of bedrock, and 63 soil samples were collected on which bulk density, saturated shear strength, angle of soil friction, and saturated hydraulic conductivity were measured. A digital elevation model (DEM) was developed from digitized elevation contours. The natural variability of soil properties was accounted for in a set of model simulations based on parameter distributions, and results were presented in a distributed probability map (DPM) for slope failure that, in turn, was used to identify the spatial scale of variability captured by the model. Ripley's K‐function for distribution of a spatial point process was applied to mapped landslides and to simulated data based on the DPM. Observed landslide locations were more tightly clustered than those predicted by the physically based model, indicating that fine‐scale physical phenomena not captured by the model likely play a role in slope failure. Aggregation of observed landslides on a scale of 150 m made the comparison with model predictions more consistent. Both parameter variability and inventory scale are important considerations in the evaluation of a slope stability model.