Many of U.S. Bureau of Reclamation's surface water storage facilities in the Western U.S. are managed during summer and autumn to support downstream aquatic habitat objectives related to stream temperature. For example, the Central Valley Operations Office (CVOO) in Sacramento, California, develops management plans using reservoirs and river temperature models to assess the feasibility of proposed management objectives for the warm season (July through September). These objectives are defined by downstream temperature thresholds and compliance locations, and they are constrained by other priority water allocations and an initial stock of stored "cold" water resource upstream (i.e. storage having temperature less than 52 degrees Fahrenheit).

Current management plan development involves forcing these reservoir and river models with a daily meteorological sequence representing expected plan-horizon weather conditions, which CVOO currently defines to be expected monthly median climate conditions based on historical records. A proposed alternative approach is explored in this study, where Climate Prediction Center (CPC) Local Three-Month Temperature Outlook (L3MTO) products are used to develop forecast-based distributions of monthly climate outcomes and, from these distributions, produce adjusted expectation for median-month climate conditions.

The study scope includes assessment of forecast availability, reliability, and decision-impact. Focus was placed on lead-2 L3MTO products issued in May, June and July for Sacramento Valley stations, generated from 1995-2005 data (no retrospective Summer issues are available prior to 1995). Forecast skill and reliability assessment was used to determine best station products for informing CVOO's stream temperature planning process on the Sacramento River below Lake Shasta. Forecast value then was evaluated through a retrospective analysis, where two CVOO plan developers were given scenarios of unlabeled meteorological boundary conditions based on either no-forecast (i.e. current approach), L3MTO-based, or perfect foresight daily meteorological sequences from any 2001-2006 warm season. Results from these planning exercises then were used to evaluate how decisions on stream temperature thresholds and compliance locations were affected by perfect foresight potential and current L3MTO forecasting capability. The presentation will summarize methods and key findings.