Development of an urban landslide cellular automata model: a case study of North Vancouver, Canada

Many GIS-based landslide models require detailed datasets that are ideally collected from field measurements, which can incur high costs for carrying out surveys. Even when the data is on hand, implementing physics-based slope stability techniques can be difficult. Common research practice uses differential equations to characterize the dynamic flow of a landslide, but it is often laborious without making substantial simplifications. A possible solution is to implement a cellular automata modeling approach, which represents both spatial and temporal components, to simulate the dynamics of the landslide propagation process. In this study, a simplified cellular automata model is developed for the effective prediction of landslide runouts, where the data requirement is a high resolution digital elevation model (DEM). Parameters, such as slope and slope curvature features, are derived from the DEM and coupled with logistic regression. The developed model is implemented on the Patrick and Dawson-Chu Slide in North Vancouver, Canada. The results from this study site were favorable, given almost 90% agreement between simulated landslides and data obtained for real landslides. In addition, sensitivity analysis was performed on the initiation sites to test the model logic and outputs of the landslide flow.