Simulation of rainfall-runoff process by ANNs and HEC-HMS model

The rainfall-runoff relation is nonlinear process and it is not possible to adequately represent the response of the catchment's system with the simple models. In the present research, simulation of the rainfall-runoff process was carried out by Artificial Neural Networks (ANNs) and HEC-HMS model. The ANN models of Multi Layer Perceptron (MLP) with two structures of one and two hidden layer, and Radial Basis Function (RBF), was used for simulation of this process. It has been applied to the Zard river basin in Khuzestan province using daily rainfall and runoff data, during the period of 1991-2000. In this period, 14 flood events were selected for simulation of the HEC-HMS model. The obtained results of the above models were compared with the observed data from Zard river basin. This comparison shows that RBF model has much more power than MLP and HEC-HMS models for simulating of the rainfall-runoff process in Zard river basin.