2008 年 52 巻 p. 49-54
Distributed hydrological models are capable of predicting the internal behavior at any point within a study basin. However, most cases, models are calibrated only at the catchment outlet and prediction accuracy at ungauged points inside is not assured. In this study, we investigate the capability of a physically based distributed hydrological model for predicting internal behavior by calibrating only using discharge observations at the basin outlet. It is found that the prediction ability for internal discharges is different depending on the initial condition; and especially the prediction accuracy of the floods with dry initial condition is low. This is due to a spatially uniform parameter setting for satisfying the calibration at the catchment outlet. Incorporating spatial heterogeneity of the parameter value for easily generating surface flow along the channel network improves the discharge prediction inside for different initial conditions.