Partitioning uncertainties of extreme flood estimates using long continuous simulations

• Modeling options can have a strong impact on of extreme floods estimates. • Uncertainty decomposition is a powerful tool for prioritizing critical sources of uncertainty in modeling chains. • Main sources of uncertainties depend on catchment characteristics and return period. • Hydrological model’s parametric uncertainty is critical for high-elevation catchments. • Stochastic variability is increasing with return period and is remarkable for rainfall-dominated catchments. Modeling options play a key role in the reliability of flood estimates, especially of extreme ones. Here, we investigate the influence of two components in a hydrometeorological modeling chain using long continuous simulations, under a stationary climate. We employed two parametrizations of the stochastic weather generator GWEX, and two model structures of the bucket-type hydrological model HBV that were configured, calibrated, and run for three representative model parameter sets. We analyze the impact of these modeling options on the magnitude and uncertainty of floods for return periods of 1 to 1000 years for selected large Swiss catchments with diverse physiographic characteristics. We found that uncertainty increases with return period, while the main source of uncertainty depends on catchment characteristics and return period. In higher elevation catchments, the hydrological model parameters were the dominant source of uncertainty, whereas in lower-elevation and rainfall-dominated catchments, the weather generator parameterizations and stochasticity were critical. Furthermore, we investigated the effect of the selected model structures on the identification of a threshold return period beyond which precipitation is the main driver of floods. No substantial differences were found between the threshold return periods for the two model structures. These findings highlight that physiographic characteristics affect the identification of the threshold return period and the contribution of the components to the uncertainty of flood estimates, challenging our ability to make a priori generalizations. Overall, this work underscores the importance of diversity in modeling options and uncertainty decomposition as a tool for informed decision-making in flood risk management.