Search

Estimation of stream flows is very important in basin and flood management. It is a very difficult topic to predict hydrological events that do not show linearity and depend on many parameters. In this study, streamflow modeling has been made and compared by using Artificial Neural Networks (ANN), Support Vector Machines (SVM) and Deep Learning (DL) optimizations using monthly average streamflow data of Mersin Lamas River. When the results are examined, it has been observed that the best method in solving the problem is ANN and SVM methods, which give more unsuccessful results. Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Nash Sutcliffe model Efficiency coefficient (NSE), Mean Absolute Relative Error (MARE) and Coefficient of Determination (R2) were used as comparison criteria. 1 input model of DL gave the best result (RMSE= 0.7647, MAE=0.7370, NSE=0.9724, MARE=27.9326, R²=0.9962). [ABSTRACT FROM AUTHOR]