Identifying key predictors for the susceptibility of Himalayan glacial lakes to sudden outburst floods

Despite being a rather rare phenomenon when compared to the occurrence rates of other alpine hazards (e.g. landslides, avalanches), glacial lake outburst floods (GLOFs) pose a significant threat to downvalley communities in glaciated mountain areas. Characteristically high peak discharge rates and flood volumes, documented to have reached 30,000 m³/s and > 50 million m³ in the past century, not only provide GLOFs with a landscape-forming potential but also killed a reported global total of > 12,000 people and caused severe damage to infrastructures. Extensive glacial covers and steep topographic gradients, coupled with rapidly changing socio-economical implications, make the Hindu-Kush-Himalaya (HKH) a high priority region for GLOF research, even though recent studies suggest an annual occurrence rate of 1.3 GLOFs per year across this range during the past three decades. So far, GLOF research in the greater HKH region has been predominantly focused on the classification of potentially dangerous glacial lakes derived from analysing a limited number of glacial lakes and even fewer reportedly GLOF-generating glacial lakes. Moreover, subjectively set thresholds are commonly used to produce GLOF hazard classification matrices. Contrastingly, our study is aimed at an unbiased, statistical robust and reproducible assessment of GLOF susceptibility. It is based on the currently most complete inventory of GLOFs in the HKH since the 1980’s, which comprises 38 events. In order to identify key predictors for GLOF susceptibility, a total of 104 potential predictors are tested in logistic regression models. These parameters cover four predictor categories, which describe each glacial lake’s a) topography, b) catchment glaciers, c) geology and seismicity in its surroundings, and c) local climatic variables. Both classical binary logistic regression as well as hierarchical logistic regression approaches are implemented in order to assess which factors drive susceptibility of HKH glacial lakes to sudden outbursts and whether these are regionally distinct.