Using Empirical Relationships and Neural Network in GIS for Developing Rainfall-Runoff Model

SUMMARY The estimation of runoff volume from a catchment is required for planning and design of water resources projects such as the design of storage facilities, assessment of water available for municipal, agricultural or industrial purposes, etc. but proper planning is possible when the accuracy of hydrological process output would be sure. Simulation and prediction of runoff is very important because of two reasons: the first one is the nonlinear interaction between rainfall and runoff on a catchment and the second one is recorded runoff data are not available for many catchments. So, in the present study were used artificial neural networks as a model which has ability of extracting nonlinear relationships from the analysis of available information in GIS. The Method that is presented in this research identifies that using of empirical equations in the form of neural network models causes not only to add spatiotemporal effective parameters in model but also to increase accuracy of prediction and generalization of model for homogeneous catchment effectively.