Akademik Veri Yönetim Sistemi
2nd International Conference on Multidisciplinary Sciences and Technological Developments, Bayburt, Türkiye, 12 - 15 Aralık 2025, ss.193-199, (Tam Metin Bildiri)
This study focuses the predictability of river discharge using artificial neural networks method using hydrometeorological data. Daily precipitation and air temperature data (1986-2022) from five meteorological stations were processed via the Thiessen polygon method, and the average values were obtained for the Enoree River Basin, South Carolina, USA. To determine the input parameters, correlation analyses were performed between the discharge data provided by a monitoring station in the basin and average precipitation and temperature values, as well as cumulative precipitations. Four models were developed using different input combinations. The performance of the models was evaluated with the root mean square error (RMSE), mean absolute error (MAE), and Nash-Sutcliffe efficiency coefficient (NSEC) statistics. The models including cumulative precipitation as an input parameter had higher NSEC values, ranging from 36.70 to 83.44%, compared to other models. It was concluded that including cumulative precipitation significantly improved model performance. For the best model, which have non-lagged temperature, two-day lagged precipitation, and cumulative precipitation of 15 days, NSEC values range from 0.565 to 0.677. These values are acceptable in the literature. This study highlights the potential of data-driven approaches for discharge predicting, particularly in data-scarce or hydrologically complex regions. Additionally, the study provides a basis for climate change impact assessments in the basin by offering insights into the discharge predictability.