[A-HW20] Hydrology & Water Environment

[AHW20-P05] Hydrological Hindcasts and Parameterizations for Flash Floods Real-Time Forecasting in the Chugoku Region of Japan

*Josko Troselj¹, Han Soo Lee¹ (1.Graduate School of Advanced Science and Engineering, Hiroshima University)

Keywords:extreme river discharge, 2018 Japanese flash floods, CDRM hydrological model, SCE-UA optimization method, real-time flood forecasting

As a result of climate change, unprecedented heavy rainfall disasters are increasingly occurring worldwide. It is therefore important to develop Early Warning Systems for real-time forecasting of extreme river water levels and discharges. This study applies the Cell Distributed Runoff Model version 3.1.1 (CDRM) hydrological model calibrated by the Shuffled Complex Evolution optimization method developed at the University of Arizona (SCE-UA) for seven first-class river basins in the Chugoku region of Japan, to obtain hydrological parameterizations and hindcasts of river mouth discharge hydrographs during the Heavy Rainfall Event of in July 2018. This study evaluates hypothesis that calibrated river basin parameters from historical extreme rainfall-induced extreme river discharge events can be used to accurately forecast extreme river discharge hydrographs from future events of similar scales by introducing an innovative historical extreme rainfall event classification methodology based on similarities between the events on the way that the classification criterion is meaningful for future real-time forecasting application. Using this approach, this study produced cumulative ensemble mean validation results with very high reproducibility metrics for the two evaluated methods, using 5 and 7 calibrated parameters. This methodology can therefore be used for accurate real-time forecasting of future flash floods caused by extreme rainfall in the region and similarly also worldwide.