Özet:

Flow forecasting is necessary to ensure sustainable water resources. Machine learning (ML), which is among the popular programs of recent times, has become more and more important in predicting river flows. Machine learning has been used in this field for a long time, but the quality of prediction is still improving. In this study, the performance of Göksu Himmetli and Körkün Hacılı flow measurement stations located on the Seyhan Basin were compared with the hybrid model benchmark model (GRU) and Liner Regression model created with GWO (Grey Wolf Algorithm) integrated Gated Repetitive Unit (GRU). In the study, the timespan of Göksu Himmetli station is between 2002-2011. The timespan of the Körkün station is between 2003-2012. The data sets consist of daily flow values. To validate the performance of the model, the first 80% of the data was used for training and the remaining 20% of the data was used for testing. Observed and predicted flow data were compared and it was examined which model gave promise results. RMSE, MAE, MAPE, R2 and STD were used as statistical evaluation criteria. When all evaluation criteria were examined, it was seen that the performance of the hybrid model was successful.

Anahtar Kelimeler: