Risk Assessment for Landslides Using Bayesian Networks and Remote Sensing Data

The use of land-based data processing capabilities and analysis has improved due to an increase of publicly available datasets with more spatial coverage, finer resolution, and better accuracy. In this study, LiDAR derived information such as a digital terrain model (DTM) and a canopy height model (CHM) from a selected area of the Oregon Coast Range, is used to develop a set of hazard and risk index maps. The manipulation of these models resulted in three maps named “Physical Model”, “Vegetation Density” and “Wetness Index” that were combined with an existing landslide susceptibility map known as “SLIDO”. These maps served as input to a Bayesian Network capable of assessing the state of risk of a slope in “Prognosis” and determining required conditions to achieve a prescribed risk condition in “Diagnosis”. Preliminary results showed that 96% of 1m deep catalogued landslides present a risk index of 0.5 or higher.