Efficacy of Mathematical Models Using Physical Soil Factors in Determining Landslide Hazard

Abstract

This research evaluates and compares the use of a terrain stability model with a statistical method in developing landslide susceptibility maps. The conceptual approach was to model slope stability in a geographical information system for a wide area. The study area is a 75.4 km² area due east of Tomales Bay in Marin County (California) characterized by rugged terrain underlain by competent bedrock to areas of hilly terrain with chaotic bedrock. The primary methods were a mathematical model (SINMAP – Stability Index Mapping) based on the infinite slope equation and a steady state hydrology and statistical models based on binomial variance (using logistic regression). The terrain stability model worked well in rugged terrain with well-defined drainages where SINMAP could model threshold saturation and resulting instability but did not work well in areas with less-rugged terrain and immature drainages. In such areas the statistical models offered greater detail regarding areas of potential hazard given the variance of the sample.