GIS-based landslide hazard zonation mapping using statistical approaches

Conventional approaches to solve slope failure problems involve using deterministic and probabilistic models to calculate the factor of safety for slopes. These models usually require a large amount of field and laboratory data on soil properties and slope geometry. This requirement makes the determination of conventional factors of safety highly unrealistic if they are required over large areas. An efficient screening methodology is to create a hazard zonation map in which regions are classified according to varying instability potential. This dissertation describes the process of using a database of slope failure details and the power of a GIS to analyze digital data for various causative factors of slope instability in order to create a landslide prediction model for Northwest Arkansas. First an extensive field investigation was performed to identify causative factors. Then data of all causative factors were obtained from remotely sensed imagery or publicly available database. The importance of each causative factor was evaluated by using statistical weighting techniques and a thematic map was created for each causative factor. A hazard zonation model was then created by overlaying all thematic maps together. All of this information was visualized and analyzed using a GIS. The model's general applicability was tested on another area with similar geoenvironmental conditions with good results.