Improvement of streamflow simulation for gauge site of hydrological model

Abstract

The paper presents an improvement procedure for streamflow simulation at gauged site of a semi-distributed river basin model. In addition to streamflow and precipitation, meteorological observations that are not employed in the HEC-HMS model calibration are used as inputs in the procedure. Some of the available meteorological variables may be of limited values in calibrating a large range of streamflow hydrographs for obtaining the optimum state variables and parameters of a river basin model. This study presents the integration of the Bayesian regularization neural network with the HEC-HMS model to provide most accurate streamflow simulations at gauged site, for a wide range of streamflow hydrographs pertinent to the hydrometeorological conditions. The artificial neural network is capable of generating a good generalization with given hydrometeorological patterns.