Quantifying uncertainty contributions of hydrologic modeling process to hydrologic drought projection in South Korea

The Andong, Hapcheon, and Seonjingang basins in South Korea are situated in the upper reaches of major rivers with reservoirs. The increasing severity of climate change leads to problems in reliable water supply, especially during extreme droughts. Unexpected climate change highlights deep uncertainty in projecting and monitoring future hydrologic droughts, emphasizing the critical importance of understanding the sources of uncertainty. The uncertainty contributions of the climate projection and the hydrologic processes to two hydrologic drought projections (low flows and characteristics of drought events) were quantified by the ANOVA method in South Korea. For the climate projection process, the selection of GCMs was the dominant source of uncertainty in both runoff and hydrologic drought projections, while the emission scenarios especially contributed to the projected drought events. For the hydrologic modeling process, the selection of hydrologic models exhibited significant contributions to projected low flows, mainly because of the complexity in estimating low flows, accounting for 51.9 %, 21.9 %, and 50.5 % for Andong, Hapcheon, and Seomjingang basins, which demonstrated site-specific effects. The contributions to the projected drought events were relatively weaker, showing an increasing trend for the durations, accounting for 5.6 %, 8.9 %, and 12.4 % for future periods. The findings emphasize the importance of quantifying uncertainty contributions of the hydrologic modeling process to the hydrologic drought projection regionally. • GCMs highly contributed to the uncertainty in runoff projections. • Interactions of emission scenarios and GCMs contributed to projected drought events. • Contributions of hydrologic model selection were significant in projected low flows. • Contributions to projected low flows were site-dependent but independent for periods.