Prediction of FDCs for ungauged basins: a comparative analysis of emerging and traditional regional approaches

ABSTRACT Long-term hydrological information is essential for planning and monitoring of water resources, especially given the challenges posed by the low density and insufficient spatial coverage of hydrometeorological networks. The goal of this study is to delineate homogeneous regions and predict the flow duration curve (FDC) at ungauged basins by applying regionalization approaches to assess streamflow. A current study introduced an application of the simple Tyler skill score (STSS) technique to ensemble the output of regionalization approaches, i.e., artificial neural network (ANN), inverse distance weightage (IDW) and stepwise regression (SWR) for better predication of hydrological information. The results indicate that the STSS ensemble approach outperforms individual methods, achieving the highest accuracy with a Nash–Sutcliffe Efficiency (NSE) of 0.78, R2 of 0.89, and PBIAS of −3.5%. The RMSE-observations Standard deviation Ratio (RSR) value of 0.40 confirms its superior predictive capability compared to ANN (RSR = 0.53), SWR (RSR = 0.67), and IDW (RSR = 0.80). Furthermore, the Volume Error (%) was lowest for STSS (−2.0%), demonstrating its effectiveness in streamflow volume accuracy. These results establish STSS as a robust, model-independent regionalization technique suitable for hydrological estimation in data-scarce regions. The findings are highly relevant for water resource management, flood prediction, and hydropower planning in ungauged basins.