4:30 PM - 4:45 PM
[AHW29-05] Impacts of runoff data and reservoir management on flood dynamics
Keywords:Flood, Simulation, Runoff, Global, Uncertainty
Fluvial flooding remains a critical global challenge, causing significant socioeconomic losses and complicating disaster risk reduction efforts. While the CaMa-Flood model is widely used for large-scale flood hazard assessment, its performance hinges on the accuracy of input runoff data. Two prominent datasets—ERA5-Land (0.1° resolution) and runoff from global hydrological models (GHMs) in ISIMIP3a (0.5° resolution)—offer distinct trade-offs: ERA5-Land captures fine-scale hydrological patterns but may introduce biases in long-term hydroclimatic trends, whereas runoff from the GHMs involved in ISIMIP3a are forced by bias-corrected climate data, prioritizing reliable long-term trends over spatial detail.
This study evaluates how differences in input forcing spatial resolution and bias corrections propagate into flood hazard simulations by systematically comparing CaMa-Flood outputs (discharge, water level, and inundation extent) at 0.1° resolution when forced with each dataset and with observations (Fig. 1: Kling–Gupta efficiency (KGE) scores from the naturalized CaMa-Flood simulation for reproducing daily streamflow using ERA5-Land during 2010–2019). We further examine the role of human interventions by conducting paired simulations—naturalized (no reservoirs) and regulated (with reservoir operations)—validated against global observational datasets.
Our analysis quantifies the impacts of runoff source selection on flood dynamics, revealing how resolution and bias correction influence model performance across diverse basins. Additionally, we assess the extent to which reservoir management modulates flood hazards in simulations. The results provide actionable insights for optimizing runoff data selection in flood modelling, balancing spatial detail with long-term reliability, and clarifying the importance of anthropogenic factors in hazard assessment.
This study evaluates how differences in input forcing spatial resolution and bias corrections propagate into flood hazard simulations by systematically comparing CaMa-Flood outputs (discharge, water level, and inundation extent) at 0.1° resolution when forced with each dataset and with observations (Fig. 1: Kling–Gupta efficiency (KGE) scores from the naturalized CaMa-Flood simulation for reproducing daily streamflow using ERA5-Land during 2010–2019). We further examine the role of human interventions by conducting paired simulations—naturalized (no reservoirs) and regulated (with reservoir operations)—validated against global observational datasets.
Our analysis quantifies the impacts of runoff source selection on flood dynamics, revealing how resolution and bias correction influence model performance across diverse basins. Additionally, we assess the extent to which reservoir management modulates flood hazards in simulations. The results provide actionable insights for optimizing runoff data selection in flood modelling, balancing spatial detail with long-term reliability, and clarifying the importance of anthropogenic factors in hazard assessment.