Efficiency of River Flow Prediction in River Using Wavelet-CANFIS: A Case Study

Application of coactive neuro-fuzzy inference system (CANFIS) and wavelet coactive neuro-fuzzy inference system (WCANFIS) models to predict river flow time series is investigated in the present study. Monthly river flow time series for a period of 1989–2011 of Ganga River, India were used. To obtain the best input–output mapping, different input combinations of antecedent monthly river flow and a time index were evaluated. Both model outcomes were contrasted using mean absolute error (MAE) and coefficient of determination (R2). Assessment of models signifies that WCANFIS model predicts more accurately than CANFIS model for monthly river flow time series. In addition, outcomes revealed that inclusion of surface runoff and evapotranspiration loss parameters to input of models enhances the accuracy of prediction more appreciably.