PREDICTION OF RIVER DISCHARGE USING ARTIFICIAL NEURAL NETWORKS: AN EXAMPLE OF GHARRAF RIVER, SOUTH OF IRAQ

Abstract

The applicability and performance of the artificial neural networks are investigated by predicting river discharge one and two days ahead for Gharraf River, south of Iraq. Gharraf River system is located at southeast Iraq within Mesopotamian Plain. A multilayerd percpetron artificial neural net is selected to achieve experiments which trained by using back-propagation algorithm. Three models are presented firstly to explore the affect of the previous discharge on the specified discharge. The ANN generated results are evaluated using statistical parameters: squared correlation coefficient R2 and root mean squared error RMSE. The results of this study indicate that ANNs are capable of producing very good results for both one and two days ahead predictions. Correlation between observed and simulated discharge values of both high and low is estimated with a good accuracy.