A Probabilistic Prediction Model for Heavy Rain-Caused Landslides in the Chongqing Region

Landslide (or rockslide) is a geological disaster that is mainly induced by strong precipitation, among a number of other natural inducing factors. Based on 1615 landslide cases, a statistical analysis is performed to find the relationship among the landslide occurrence time, rainfall 0-10 days ahead, and probability of landslides over the Chongqing region. The results show that 1) strong rainfall-caused landslides occur mainly on the day it rains or 1-2 days after the heavy rain, and as time goes on, the likelihood of the disaster reduces rapidly; 2) the heavier the rainfall, the closer the landslide time is to the precipitation time.A concept of "effective precipitation" is thus developed, and a categorical prediction model for heavy rain-caused landslides is established. Tests show that for categories III, IV, and V landslides, the model forecast accuracy arrives at 29.9%, 75%, and 100%, respectively. This indicates that the categorized probabilistic prediction can serve as a warning for the landslide prevention and mitigation.