Balancing Water Supply and Flood Risk for FIRO by Combining Optimal Hedging and Risk Tolerance-Based Operating Rules

This study integrates an analytical method with realistic heuristic rules for more effective realization of forecast-informed reservoir operation (FIRO). It presents dFLWL-EFO, a new FIRO decision-support framework that combines optimal hedging and risk tolerance-based operating rules, considering flood control and water conservation for the postflood season, which are often inconsistent, especially under increasing inflow variability. Effective use of weather and hydrological forecasts via FIRO is essential in addressing this challenge. dFLWL-EFO adapts ensemble inflow forecasts via the Bayesian model averaging method to quantify forecast uncertainty. It includes a two-stage stochastic optimization model for dynamic control of the flood-limited water level (dFLWL), explicitly incorporating hedging policies to balance flood mitigation and water shortage reduction under low to moderate flood risk; meanwhile, the framework integrates risk tolerance-based release rules from ensemble forecast operations (EFO), an approach that has been validated for the operation of real-world reservoirs, to ensure dam safety during high-risk flood events. Thus, the integrated framework holds the complementariness of dFLWL and EFO. Moreover, dFLWL-EFO provides reservoir operators with the flexibility to incorporate operational priorities, legal constraints, and institutional guidelines into the implementation of FIRO. This FIRO framework is tested on Folsom Lake in California. Results show significant increase in water conservation in the postflood season without increasing flood risks. By introducing operational flexibility enabled by forecast, the proposed dFLWL-EFO presents a realistic, effective, and generalized approach to move research to real-world reservoir operation practices.