Prediction of spatial distribution of debris-flow hit probability considering the source-location uncertainty

Abstract: Prediction of the extent and probability of debris flow under rainfall conditions can contribute to precautionary activities through risk quantification. To this end, quantifying the debris-flow risk against rainfall involves three components: predicting the debris-flow initiation locations under rainfall conditions, setting appropriate physical parameters related to debris-flow transportation, and evaluating the affected area using numerical simulation. In this study, we developed a logistic regression method that includes rainfall and topographic parameters as explanatory variables to quantify the probability of debris-flow initiation in an actual area with disaster record. Moreover, an objective parameter-set selection was introduced by evaluating the agreement between the simulation results with multiple parameters and the erosion/deposition area determined using the aerial light detection and ranging difference data after debris flow. Finally, by combining these results, we conducted a predictive debris-flow transport simulation using the initiation datasets generated by the logistic model. Therefore, the spatial distribution using the probability of the effects of debris-flow, which can be applied for risk quantification and evacuation optimization, was successfully obtained at 1-m resolution. Furthermore, a real-time hazard probability prediction system could be developed based on the presented simulation cost.