METHOD COMPARISON FOR REDUCTION OF THE FLOOD PREDICTION UNCERTAINTY USING REAL-TIME HYDROLOGICAL OBSERVATION DATA

Abstract

In flood prediction, reduction of the uncertainty is one of the biggest issues. There are several methods to reduce the prediction uncertainty by using real-time hydrological observation data, but the difference of the methods was poorly understood. In this study, we developed one distributed hydrological model and two statistical models. The distributed model was composed of 2D surface- subsurface flow model, 1D saturated-unsaturated infiltration model and 1D dynamic wave model. As the assimilation method, particle filter, simple error-compensating, and multi data error-compensating methods were applied. Two set of statistical models are the artificial neural network model and the multiple regression model. Applied to OYODO-river, difference of these methods was discussed.