複数洪水イベントの組み合わせによる洪水予測に適したRRIモデルパラメータの最適化手法に関する検討

抄録

The water level rises rapidly during the flood due to the topographical characteristics of the small mountainous river basins. Therefore, a flood forecasting system that can accurately predict the water level is required. In constructing a model, estimating parameters that greatly affect the reproducibility of the model has traditionally required time and effort. The purpose of this study is to apply the SCE-UA method to the RRI model and propose an optimization parameter identification method with high prediction accuracy for large, medium and small flood events without the user’s hydrological background. The RRI model is applied to thirteen floods in the dam basin with the SCE-UA method and optimized simultaneously the multiple flood events focusing on the runoff ratio using four types of evaluation methods considering the reproducibility of the rising and peak parts of the water level from the perspective of flood forecasting. Furthermore, the reliability of the identified parameters was verified at large, medium, and small another flood events．