Akademik Veri Yönetim Sistemi
Theoretical and Applied Climatology, cilt.157, sa.4, 2026 (SCI-Expanded, Scopus)
The correct evaluation of evapotranspiration is very important for hydrological cycle comprehension, agricultural water resource management, and climate change impact assessment, especially in areas with limited water supply where occasional rainfall and rising temperatures are posing a threat to water security. The research focuses on hydro-climatic variability and evapotranspiration dynamics in the Dam Region of Türkiye, a key hydrological area in Southeastern Anatolia, which includes the Keban, Karakaya, and Atatürk reservoirs. Daily weather data were collected from the ERA5-Land reanalysis dataset for the period of 1960–2023 for six provinces (Malatya, Adıyaman, Şanlıurfa, Diyarbakır, Elazığ, and Tunceli). Computational intelligence models, including Multiple Linear Regression (MLR), Random Forest (RF), Support Vector Regression (SVR), Artificial Neural Networks (ANN, specifically a Multilayer Perceptron (MLP) architecture), Bidirectional Long Short-Term Memory (Bi-LSTM), and a Triple Hybrid (RF–ANN–Bi-LSTM) approach, were applied to evaluate the predictive skill and variable sensitivity. Performance of models revealed that the provinces could be grouped into three clusters defined via K‑means clustering based on predictive‑skill metrics and runoff‑regime characteristics. The SHAP-based sensitivity analysis confirmed that runoff is the most important mediatory factor in maintaining hydro-climatic stability of semi-arid and transitional environments. The Hybrid model, which merges the selected features of Random Forest, the nonlinear mapping of ANN, and the temporal dynamics of Bi‑LSTM into an attention-enhanced architecture, was assessed with a thorough array of metrics and showed an excellent ability to capture nonlinear hydro‑climatic interactions, thus increasing the predictive capacity in data-scarce, dam-regulated basins.