Hybrid Optimization Rainfall-Runoff Simulation Based on Xinanjiang Model and Artificial Neural Network

Abstract

A hybrid rainfall-runoff model that integrates artificial neural networks (ANNs) with Xinanjiang (XAJ) model was proposed in this study. The writers extracted the digital drainage network and subcatchments from digital elevation model (DEM) data considering the spatial distribution of rain-gauge stations. Then the semidistributed XAJ model was established based on DEM. Considering the runoff routing cannot be calculated by the linear superposition of the route runoff from all subcatchments, artificial neural networks as effective tools in nonlinear mapping are employed to explore nonlinear transformations of the runoff generated from the individual subcatchments into the total runoff at the entire watershed outlet. The integrated approach has been demonstrated as feasible and was applied successfully in the Yanduhe watershed, the upper tributary of Yangtze River Basin. The results indicated that the approach of integrating back-propagation ANN with semidistributed XAJ model may achieve the promising results with acceptable accuracy for flood events simulation and forecast.