Monthly river flow forecasting by an adaptive neuro-fuzzy inference system

Abstract

In this study, the applicability of an adaptive neuro-fuzzy inference system (ANFIS) to forecast for monthly river flows is investigated. For this, the Göksu river in the Seyhan catchment located in southern Turkey was chosen as a case study. The river flow forecasting models having various input structures are trained and tested by the ANFIS method. The results of ANFIS models for both training and testing are evaluated and the best-fit forecasting model is determined. The best-fit model is also trained and tested by feed forward neural networks (FFNN) and traditional autoregressive (AR) methods, and the performances of the models are compared. Moreover, ANFIS and FFNN models are verified by a validation data set including river flow data records during the time period 1997–2000. The results demonstrate that ANFIS can be applied successfully and provides high accuracy and reliability for monthly river flow forecasting.