Abstract

Floods are among the most destructive natural hazards, making it essential to assess how climate change may influence their frequency and intensity. Yet, future flood behavior remains uncertain and varies regionally. This study uses continuous hydrological simulation with a calibrated and validated Hydrologic Engineering Center's Hydrologic Modeling System model to analyze historical and future peak flows in the urbanized Marga Marga Creek basin in coastal Chile. Historical daily meteorological data were stochastically disaggregated to hourly resolution to model multiple discharge series and account for internal climate variability. Future simulations were based on projections from eight downscaled and bias-corrected general circulation models under the SSP5-8.5 scenario. Results show that most historical peak flows were below 50 m(3)/s, driven by daily rainfall under 17 mm. Toward the end of the century, peak discharges are projected to decline relative to the 1980-2014 baseline. A nonstationary frequency analysis was conducted, but high inter-model variability limits its reliability for this basin. Overall, continuous simulation captures interactions among climate, hydrology, and land use, offering a valuable approach to address uncertainty. The findings support the design of early warning systems based on meteorological forecasting and the integration of flood risk into urban planning and multihazard strategies, especially in regions with limited observational data.