Modeling Flood Risk for South Korea

Abstract:

Catastrophic flood events have caused significant losses for South Korea each year. It is very important to generate high resolution flood return period map for the government and insurance company to evaluate the flood risk. This research was initiated to achieve this goal. A 2000 year spatial distributed stochastical rainfall was generated by analyzing the historical rainfall of South Korea using principle component analysis. A rainfall-runoff model and a routing model were calibrated by driving the model with historical forcing and calibrated against gauge observations. The calibrated model was used to couple with the stochastical forcing to generate 2000 year discharge and runoff. The flood maps with different return periods were generated by numerically solving the shallow water equations using finite volume method on GPUs. The results of this research showed a reasonable flood map in South Korea, compared with the observed data. Further, this research could be used as an important reference for the government and insurance companies for risk management purpose.