Development of Water Rating Curve in Shatt Al-Arab River

Abstract

This study investigates the abilities of artificial neural networks (ANN) to improve the accuracy of stream flow-water rating curve in Shatt Al-Arab River. Development of stage-discharge relationships for the daily stream flow to Shatt Al-Arab is a challenging task. In this study, the hydrological data was used as a tool for the identification of critical (information) segments in a river, to covers all study area in Shatt Al-Arab over a period of seven years started from January /2009 to January/2015from water resources office in Basra Province. Data from different gauging sites were used to compare the performance of ANN trained on the whole data set. The neural network toolbox available in MATLAB was used to develop several ANN models. Five layers feed- forward network with Log-sigmoid transfer function was used. The networks were trained using Levenberg-Marquradt (LM) back-propagation The Levenberg-Marquradt (LM) back-propagation was found to be the best ANN model with minimum Mean Squared Error (MSE) and maximum correlation coefficient (R) 0.9, and MSE 1.05*10-7, respectively. The optimum neuron number in the two hidden layers of (LM) was 8 neurons with R greater than 0.9, and MSE 1.05*10-7, respectively