PROPOSAL FOR METHOD FOR GENERATING PROBABLISTIC FLOOD EVENT SET FROM PRECIPITATION DATA

Abstract

 Although quantitative flood risk assessment owned by General insurance company is an important issue for them, the problem is that the knowledge of models for quantitatively assessing flood risk is shallow in some countries / regions.

 In this research, we propose that the method for quantitatively assessing flood risk with holding Java, Indonesia up as an example by using precipitation data instead of the hazard module as a part of conventional natural catastorophe model used for natural disaster risk analysis in general insurance companies. First of all, the river discharge is calculated by RRI model from GSMaP. Secondly, the dependencies between points are calculated using Heffernan-Tawn model which is known as a conditional probability model. These dependencies are used to calculate the simultaneous excess probability of river discharge. We visualize the simultaneous excess probability and confirm that this explains the spatial dependencies of a probabilistic flood event.