Randomized trial benchmarks reservoir routines for optimal performance in large-scale hydrological models, suggesting effective calibration methods.
Key Points
This research aims to benchmark different reservoir operation schemes within large-scale hydrological models to identify effective calibration methods.
Benchmarking four reservoir routines: LISFLOOD, CaMa-Flood, mHM, and STARFIT using the ResOpsUS dataset.
Evaluating performance across 164 reservoirs in the United States.
Testing the effectiveness of target variables for parameter estimation.
The mHM routine shows the highest performance but requires site-specific demand data, limiting its global applicability.
CaMa-Flood offers a robust compromise, outperforming LISFLOOD and matching STARFIT's performance.
Calibrating to reservoir storage is found to be more informative than calibrating to outflow.
Cite This Study
Casado-Rodríguez et al. (2026) studied this question.