Akademik Veri Yönetim Sistemi
JOURNAL OF WATER AND CLIMATE CHANGE, cilt.16, sa.2, ss.736-756, 2025 (SCI-Expanded, Scopus)
Streamflow forecasts play an active role in hydrological planning and taking precautions against natural disasters. Streamflow prediction models are frequently used by scientists, especially in dam management, sustainable agriculture, flood control, and flood mitigation. Hence, streamflow prediction modeling was performed in this study, and six models were employed through four different machine learning (ML) algorithms, namely, the artificial neural network (ANN), random forest (RF), support vector machine (SVM), and decision tree (DT) that are well known in the literature, in order to predict the monthly streamflow of & Ccedil;ar & scedil;amba River, T & uuml;rkiye. To further enhance model performance, wavelet transform (WT) was applied to ML algorithms. In this study, monthly average streamflow and precipitation data between 1974 and 2015 was used, and the minimum redundancy maximum relevance method (MRMR) and the cross-correlation method were performed to determine model input data. Results of this study revealed that RF had superiority over the other models before WT, followed by the SVM model. The SVM after WT (W-SVM), M04 (r: 0.9846, NSE: 0.9695, and RMSE: 0.3536) gave the most effective performance results, while the W-ANN model (r: 0.9797, NSE: 0.9588, and RMSE: 0.4108) showed the second best performance.