Date of Award

Summer 2025

Document Type

Open Access Thesis

Department

Civil and Environmental Engineering

First Advisor

Jasim Imran

Abstract

Effective reservoir management in hydrologically variable regions such as California’s Central Valley faces increasing pressure to balance flood control, water supply reliability, hydropower generation, and ecosystem health. Traditional rule-based operations, while operationally simple, often lack the adaptability required to respond to real-time hydrologic conditions and forecast uncertainty. This study presents an integration of a policy tree optimization model with ensemble streamflow forecasts to enhance Forecast-Informed Reservoir Operations (FIRO) at Folsom Reservoir. Building on historical data for inflow, storage, and release, an interpretable, threshold-based decision framework was developed and coupled with 14-day ensemble inflow forecasts from the California-Nevada River Forecast Center’s Hydrologic Ensemble Forecast Service. The resulting policy tree was evaluated against historical operations, with performance assessed across key metrics including flood risk reduction, water supply reliability, and robustness under forecast uncertainty. The optimized forecast-informed policy tree demonstrated improved operational flexibility and resilience compared to historical operations. These findings highlight the potential of integrating machine learning-based optimization and real-time forecasting to advance adaptive, data-driven reservoir management under changing climate and hydrologic conditions.

Rights

© 2025, Alessandra Estefania Perez Salas

Share

COinS