How do the properties of training scenarios influence the robustness of reservoir operating policies to climate uncertainty?

Reservoir control policies provide a flexible option to adapt the uncertain hydrologic impacts of climate change. This challenge requires robust capable navigating scenarios that are wetter, drier, or more variable than anticipated. While number prior studies have trained using large scenario ensembles, there remains need understand how properties training impact policy robustness. Specifically, this study investigates including annual runoff, snowpack, and baseline regret—the difference between perfect foresight performance in an individual scenario. Results indicate subsets with high regret outperform those generated other sets both wetter drier futures, largely by adopting intra-annual hedging strategy. The approach highlights potential improve efficiency robustness considering ensemble.