抄録
The long-and short-term runoff model (LST model) is applied to hydrological data of six mountainous river basins in Okayama Prefecture. The model is calibrated and verified in each basin and the overall efficiency is evaluated for each. The model performs well in all the basins with good accuracy.
Real-time flood forecasting with LST model is carried out using a simple method, in which an updated state variable is given by back calculation. The applicability of the simple method is tested extensively using data for 7-18 years in 6 basins. The efficiency of this simple method is also compared with that of extended Kalman filter in each basin. The simple method is shown to perform as well as the extended Kalman filter.
Interrelation between the two methods is discussed theoretically and indicates that the simple method corresponds to the extended Kalman filter without noise. The interrelation is confirmed by comparing updated state variables.
