Ecology, Environment and Conservation Paper

COMPARISON OF ARTIFICIAL NEURAL NETWORKS AND HECHMS MODELS IN SIMULATING RAINFALL-RUNOFF HYDROGRAPH

Farshidsafshekan and Nader Pirmoradian

Abstract

The relationship between rainfall and runoff is a complex hydrologic phenomenon. Therefore, the modeling of this relationship is important in view of the many uses of water resources such as hydropower generation, irrigation, water supply, and flood control.In this study,it was derived an Artificial Neural Networks (ANNs) model tosimulate rainfall-runoff process. It was resulted a multilayer perceptron neural network model with 9-10-7 structure. To increase of the model stability and better training, the rainfall data due to occurrence time distribution were divided to four groups according to Huff rainfall distribution classification. The HEC-HMS model also used to simulate rainfall-runoff relationship and to compare with neural network model. The absolute relative error for hydrograph parameters simulation was lower in ANNs than HECHMS model. It was shown that the ANN method offers an accurate modeling of rainfall-runoff relationship.