The use of Artificial Intelligence methods is becoming increasingly commonin the modeling and forecasting of hydrological and water resourceprocesses. In this study, applicability of Adaptive Neuro Fuzzy InferenceSystem (ANFIS) and Artificial Neural Network (ANN) methods, GeneralizedRegression Neural Networks (GRNN) and Feed Forward Neural Networks (FFNN),and Auto-Regressive (AR) models for forecasting of daily river flowis investigated and Seyhan River and Cine River was chosen as case studyarea. For the Seyhan River, the forecasting models are established usingcombinations of antecedent daily river flow records. On the other hand, forthe Cine River, daily river flow and rainfall records are used in inputlayer. For both stations, the data sets are divided into three subsets,training, testing and verification data set. The river flow forecastingmodels having various input structures are trained and tested to investigatethe applicability of ANFIS and ANN and AR methods. The results of all modelsfor both training and testing are evaluated and the best fit inputstructures and methods for both stations are determined according tocriteria of performance evaluation. Moreover the best fit forecasting modelsare also verified by verification set which was not used in training andtesting processes and compared according to criteria. The resultsdemonstrate that ANFIS model is superior to the GRNN and FFNN forecastingmodels, and ANFIS can be successfully applied and provide high accuracy andreliability for daily river flow forecasting.
展开▼
