This study evaluated the applicability and reliability of various baseflow separation methods in the Yeongsan and Seomjin watersheds, and investigated the relationship between watershed characteristics and the baseflow index (BFI). Both graphical approaches (HYSEP-FIM, SIM, LMM, UKIH-Min, PART) and digital filter methods (Lyne-Hollick 1·2-Pass, Chapman, CM, Eckhardt, EWMA) were applied to estimate BFI, and their performance was assessed using statistical indicators including NSE, KGE, PBIAS, and RSR. Sensitivity analysis revealed that BFI results were highly dependent on parameter settings, particularly for digital filters, with the Eckhardt method demonstrating the most stable and reliable performance, thus identified as the optimal method. Correlation analyses between watershed characteristics (e.g., area, elevation, slope, land use) and BFI showed statistically significant relationships in some main river reaches, while tributary basins exhibited complex interactions requiring multivariate analysis. These findings provide a methodological foundation for future baseflow estimation, and serve as essential baseline data for watershed-scale hydrologic modeling, ecological flow assessment, and integrated water resources planning.

Keywords: Baseflow, Baseflow index (BFI), Digital filter, Graphical method, Parameter sensitivity, Watershed characteristics.