H161-09
Value-Based Reservoir Operations in a Changing Climate
Abstract:
The Tuolumne River in California supplies municipal water to San Francisco and other cities and irrigation water to lands in California’s Central Valley. Don Pedro is the largest reservoir in the watershed with a storage of 2.03 MAF and is operated by Turlock Irrigation District (TID).
TID uses ensemble weather and hydrologic modeling with Hydrocomp Forecast and Analysis Method (HFAM) to forecast future inflows to Don Pedro, conditioned by the current watershed state. Historical hydro-meteorological data collection began in the Tuolumne River at a few stations in the 1890s. Comprehensive records for streamflow, precipitation and temperatures start in the 1920s. Climatic non-stationarity from 1975 to 2019 is indicated by increasing trends in minimum daily temperature and annual runoff volume variance. Trends are not found in mean long-term precipitation and long-term streamflow volumes.
TID is evaluating value-based algorithms that recommend weekly Don Pedro releases given ensemble future reservoir inflows. The algorithms calculate the marginal value of current reservoir release based on the future value of water for beneficial use and future damage caused by excess releases. Reservoir operators set beneficial release demands (instream flows, irrigation, municipal and industrial requirements) and corresponding value (beneficial value) as well as excess release limits and corresponding damage (flood damage). Release demands and limits can be time variable. The decision time frame for current releases can be from one year to five years in the future.
Analysis of Don Pedro operations over historic time shows how the assigned beneficial value and flood damage impact flood risk and beneficial releases. Non-stationary climate effects on flood risk and beneficial releases are quantified by dividing the 1930 to 2019 hydro-meteorological records into two periods: 1930 to 1974, and 1975 to 2019. Present day beneficial water demands are used for both periods.