Estimating Daily Streamflow Values Using Artificial Neural Networks, Support Vector Regression and Multiple Linear Regression Models for Ceyhan River Basin

Abstract

Listen

Translate article

Streamflow data are very important for effective planning and management of water resources in basins. In this study, Artificial Neural Networks (ANN), Support Vector Regression (SVR) and Multiple Linear Regression (MLR) models were developed to estimate the daily streamflow of three different rivers in the Ceyhan River Basin. Daily precipitation and temperature data obtained from The Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2) re-analysis data were used as predictor variables in the models. The estimation performances of the models were evaluated with different statistical performance measures. According to the evaluation results, the SVR model demonstrated the best performance in daily streamflow estimation for the Ceyhan River, achieving R² = 0.95 and RMSE = 28.20 m³ s-1. Additionally, for Söğütlü Creek, the results were R² = 0.82 and RMSE = 6.57 m³ s-1, while for Keşiş Creek, R² = 0.93 and RMSE = 1.45 m³ s-1 were obtained. The findings indicate that the SVR model predicts daily streamflow more successfully than the other models. Furthermore, it was found that the performance of the models developed using machine learning algorithms was superior to that of the linear regression model.