Probabilistic analysis of landslide hazard: considering the dependence between hazard components

Abstract

This study reported a landslide hazard pattern considering the dependence of hazard components, with the example of Heilongjiang Province. The spatial and magnitude probabilities of landslides were first constructed to develop a potential link with the annual occurrence rate. Thereafter, landslide hazard considering component dependence was derived based on a modified Poisson model, and presented via the exceeding probability under the scenarios of combining two magnitudes (greater than or equal to 30,000 and 100,000 m3) and four periods (1, 3, 5 and 10 years). Results show that the Poisson model refined is not only a temporal probability function conditional on the magnitude and space, but can capture the hazard definition accurately. Moreover, the exceeding probability is related to the location of the raster, magnitude and the time interval of the landslide. Geographically, the areas with high exceeding probability are concentrated in the central and southeastern parts of the study area. Further, landslide probability dynamically varies with the magnitude and time interval, increasing gradually with the time interval, but decreasing with the landslide magnitude. This study is significant to resolve the conflict between the independence assumptions of the hazard calculation and the conditional attributes in the hazard definition.