LAUSR

Glacial lake is an important water resource. But glacial lake outburst floods (GLOFs) are destructive not only to property and infrastructure but also to people living in the regions. GLOFs prediction and risk evaluation are critical for preventing and mitigating the negative impacts. This paper proposes a prediction model for the possibility of GLOFs, which emphasizes the selection of easily available predictors. Taking 29 glacial lakes in the Himalayas as samples, the Geodetector is used to detect 4 selected predictors: the width of dam crest, the ratio of freeboard to dam height, the area of glacial lake and the area of mother glacier. The result shows the ratio of freeboard to dam height has the largest q-value of 0.3229. In the interaction detector, the width of dam crest and the ratio of freeboard to dam height had the highest explanatory power of 0.7667 after the interaction. The GLOFs probability prediction model correctly classifies 78% of drained lakes and 90% of undrained lakes, for an overall accuracy of 86%. Taking Amazhibu Tsho as an example, calculate variation in the probability of GLOF with different predictors. The results can provide practical and efficient references for local government and people.