The Application of Intelligent Techniques for Predicting Daily Flow at Telvar Basin River

River flow, which temporarily and spatially changes, is a major hydrological variable in water resource planning. In research, on water resources, the perdition of the river flow based on historical data is a main step for watershed management. In this study three intelligent techniques including wavelet artificial neural networks, gene expression programming, and support vector machine (SVM- LS) were compared in river daily flow perdition at the Telvar basin. Daily recorded data from 2002-2012 were used in the modeling procedure. Data were divided to train (75%) and test (25%) groups. Results indicated that all three modeling methods have high performance in predicting daily flow using the two-day lag data. The correlation coefficients for wavelet artificial neural networks, gene expression programming, and support vector machine (SVM- LS) were 0.9, 0.94, and 0.92 respectively. Therefore, it can be concluded that the gene expression programming has slightly better results in comparison with the other two modeling methods. The accuracy of the gene expression programming model increased due to the increasing of the lag data from 2 to 4 and five days.