UC Davis

The security and prosperity of California’s economy depend on a large, interconnected and highly engineered water supply system with vast surface and groundwater storage operated for over-year storage. Substantial public investment is needed to maintain and improve these systems with changing climate risks, increasing demands, and aging infrastructure. Hydro-economic optimization identifies promising opportunities for managing such systems, but scientific and engineering advances are needed to better incorporate hydrologic uncertainty and conjunctive management of streams and aquifers into such models. Over two decades of research the California Value Integrated Network (CALVIN) linear programming hydro-economic model has yielded insights for California’s inter-regional water system on many fronts, including climate change, groundwater sustainability, conjunctive management in southern California, water markets, and reservoir operations in Northern California. Yet there remains unrealized potential to improve the model’s formulation to better represent interannual hydrologic variability while concurrently developing reservoir control rules. To address this gap, this study implements linearized quadratic carryover storage value functions on 26 surface reservoirs and linear groundwater storage penalties on 32 groundwater reservoirs to optimize California’s inter-regional water system seasonal and inter-annual operations with limited seasonal foresight. A convex Pareto front tradeoff between total system cost and groundwater overdraft is developed through multi-objective evolutionary optimization of inter-annual carryover storage penalties. Resulting over-year storage valuations span a wide range of near-optimal reservoir operations, suggesting substantial system flexibility and adaptability when managed in an integrated way. Generalized relationships between carryover capacity, carryover use, and the marginal value of carryover storage are drawn for the multi-reservoir system. When compared to carryover operations based on California’s current simulation model, CalSim-II, the performance of these near-optimal reservoir operations suggests room to improve currently prescribed reservoir control rules, especially in supporting more sustainable conjunctive management with aquifers. Overall, this study demonstrates the practicability, feasibility, and utility for understanding large-scale integrated multi-reservoir conjunctive use systems using limited foresight with explicit carryover storage values.