Title: Daily inflow forecasting for Dukan reservoir in Iraq using artificial neural networks
Authors: Rafa H. Al-Suhili; Rizgar A. Karim
Civil Engineering Department, University of Baghdad, Baghdad, Iraq; City College of New York, 160 Convent Avenue, 140th Street New York, NY, 10031, USA
Dam and Water Resources Engineering Department, Faculty of Engineering, University of Sulaimania, University Main Campus, Sulaimania, Iraq
Abstract: Five ANN model versions were developed for the daily inflow forecasting to Dukan reservoir in Iraq. These models are dependent on the preceding days' lags (1, 2, 3, 4, and 5), respectively. The model versions forecasting correlation coefficients were found to be (94.6%, 94.6%, 95.2%, 95%, and 73%), respectively. The third model was used for forecasting and found capable of forecasting long term daily inflow series of the Dukan reservoir. Moreover it was found also capable of preserving the high and low persistences of this series in addition to the perfect simulation of the recession part and time to peak of the hydrograph.
Keywords: ANN models; daily inflow forecasting; reservoir operations; time series analysis; Dukan reservoir; Iraq; artificial neural networks; simulation; water reservoirs.
Int. J. of Water, 2015 Vol.9, No.2, pp.194 - 208
Available online: 22 Apr 2015