Affiliations: Geological Engineering Department, Hacettepe University, Beytepe, Ankara, Turkey | Computer Engineering Department, Hacettepe University, Beytepe, Ankara, Turkey
Note: [] Corresponding author. Naciye Nur Ozyurt, Geological Engineering Department, Hacettepe University, Beytepe 06800, Ankara, Turkey. E-mails: [email protected] (Naciye Nur Ozyurt), [email protected] (Ebru Akçapinar Sezer).
Abstract: Prediction of daily streamflow in mountainous karst basins by means of deterministic models requires vast amount of data which is often not available. Such predictions are critical for the optimal operation of hydropower plants utilizing streamflow. The powerful prediction tool adaptive neuro-fuzzy interference system (ANFIS) was used for prediction of downstream flow in the Zamanti River Basin. Model input parameters are upstream flow, precipitation and retrospective downstream flow. Four different models constructed and the model results were assessed by using the determination coefficient and the root mean square error. Applied models produced reliable prediction values. Model results reveal an acceptable deviation from observations at stream flows exceeding the average. The prediction success of the applied ANFIS model appears to be promising for streamflow prediction in similar hydrogeological settings.
