Prediction of daily streamflow using ANN, WNN and ANFIS models

The management and use of water resources, which have become increasingly important over time, has become one of the significant news today. In this study, Artificial Neural Networks (ANN), Wavelet Neural Networks (WNN) and Adaptive Neural Fuzzy Inference System (ANFIS), which provide feasible and reliable results, were used for daily streamflow estimation. For streamflow estimation, there are many parameters that affect the results and therefore, it is difficult to conclude exactly the same as the observed data. However, realistic results were obtained with the methods and models developed in the study. To verify the performance of the models, 70% of the data (1996-2007) were used to train and 30% of the data (2008-2011) were used for testing. Although the results of the models were close to each other, WNN model showed the best performance among ANN and ANFIS models. This shows that the decomposition of the original data into sub-series, identified and cleaning the noises significantly affect the results.