2014 年 70 巻 4 号 p. I_397-I_402
We present here a novel system that predicts stream flood condition on a real-time. The system drives LSM-coupled distributed hydrological model to simulate river discharges as well as basin's water balance components, such as soil moisture, on hourly basis. In the meantime the system evaluates past Quantitative Precipitation Forecast (QPF) performance with respect to C-band radar derived observed rainfall for generating ensemble QPF members. The system then simulates ensemble stream flow forecast by taking real-time hydrological model output as initial condition and ensemble QPF members as forcing. The system is applied to the Upper Tone river basin, showing promising performance for the recent three typhoon cases. Being integrated with data-archive, this approach can be operationally used and its application to other river basins is widely possible.