Real Time Flood Forecasting Using Variable Infiltration Capacity Model

Abstract

The variable infiltration capacity (VIC) model had been widely used for rainfall-runoff analysis since early 1970s. In this study, the VIC model was modified for the purpose of flood runoff analysis and real-time flood forecasting. The modified model has another non-linear tank model into which the direct discharge in the original VIC model is inputted as effective rainfall. Besides a percolation hole was added to the original one in order to preserve water balance during flood periods. The accuracy in calculated flood discharge by the modified model was confirmed by flood runoff analysis of the Kuroki Dam Basin in Okayama. In addition, the real-time flood forecasting system based on the modified VIC model combined with sequential state updating technique as extended Kalman filtering technique or inverse estimation technique was developed. The result of the application of the developed system to flood forecasting at the Kuroki Dam showed satisfactory performance that is comparable to that of the real-time forecasting system based on the non-linear tank model previously developed, though the result also showed the dependency of the flood forecasting accuracy on the rainfall forecasting.