Towards continental-scale SWOT-based modeling of reservoir operations

The advent of reservoir models has enabled the monitoring and prediction of reservoir operations (inflow, storage, and release); however, the in situ operation data used for model parameterization is spatially scarce. In this study, we evaluate reservoir monitoring capabilities by leveraging observations from multiple satellite missions to reconstruct reservoir operations, optimize reservoir model parameters, and simulate operations. For reservoirs in the US and Brazil, we design a set of experiments relying on different combinations of in situ , satellite, and hydrologically modeled data sources as reservoir model inputs to characterize the error of satellite- and model-informed reservoir operations. We also make the first attempt to evaluate the use of the Surface Water and Ocean Topography (SWOT) satellite mission data record for reservoir optimization and simulation. Our results show that simulated storages (median nRMSE ≤ 0.30–0.50; represents maximum across experiments for each country and altimetry dataset) perform significantly better across all experiments than simulated releases (median nRMSE ≤ 0.68–0.94). Accuracy of simulated releases is strongly dependent on the quality of the inflow time series: using in situ inflow (median KGE = 0.56–0.68) for reservoir modeling has greater accuracy than hydrologically modeled inflow (median KGE = −0.19–0.15). However, when simulating operations based on altimetry and modeled inflow, 42–67% of reservoirs improve in release KGE compared to simulations without reservoir operations. Overall, our study demonstrates the potential of SWOT and other satellite altimeters for continental-scale reservoir storage monitoring, and that improving inflow simulations is crucial for release monitoring.