Prediction of monthly average discharge using the hybrid model of artificial neural network and wavelet transforms (Case study: Kor River- Pol-e-Khan Station)

Awareness of flow rate data in rivers is essential for management of water resources, flood forecasting, engineering design and environmental management. The presented models for flow rate predicting in rivers, such as rainfall-runoff and time series are not consistent with the observed data in many cases due to the lack of accuracy and complexity of the factors affecting the discharge. Wavelet is one of the methods that has been considered in recent years in the field of hydrology. Wavelet method is also very effective in the field of signals analysis and time series. This paper presents a hybrid intelligent model based on artificial neural network and wavelet transforms is used to simulate monthly average discharge in Kor River and Pol-e-Khan Station. Performance of prediction models were evaluated using the criteria of Root Mean Square Error (RMSE) and determination coefficient. The results showed that the hybrid model of artificial neural network and wavelet transform with 2 degrees of decomposition offers the best results for the most suitable structure. In this structure, the output discharge for flow rate in the following month is calculated based on discharge in 4, 3, 2 and 1 month ago and current month and the values of RMSE and obtained 7.14 and 0.941 respectively.