91ÃÛÌÒ¸ó

Western US water agencies can benefit from seasonal water supply forecasts to optimize their reservoir operations for water supply and flood protection. However, seasonal water supply in the western US is difficult to predict multiple months ahead, especially in the mid-latitudes of the western US where prediction varies widely by immediate location and derives little skill from the El Niño-Southern Oscillation (ENSO). We investigate how seasonal water supply prediction using sea surface temperatures (SST) varies between adjacent small to medium size mountain watersheds within several regions of the western US, and how this variation relates to differences in moisture pathways. Differences between rain-dominated and snow-dominated basins are also considered. Prediction centers, or areas of the oceans where SSTs have significant predictive skill for seasonal water supply, are identified separately for each watershed to further optimize prediction, since prominent crests can separate the influence of incoming moisture. Winter and spring precipitation, and co-occurring snow-water equivalent depth (SWE), are compared for their predictability by SSTs and usefulness in predicting streamflow. A particle tracking model is used to evaluate moisture pathways to watersheds and to compare their prevalence during periods characterized by positive and negative SST anomalies at identified prediction centers. Particle tracks and moisture sources are compared for storm events before and after peak SWE to understand physical mechanisms that explain seasonal differences in SST-water supply relationships. This work may improve prediction of water supply and understanding of moisture pathways for small watersheds, which can be masked via procedures to aggregate watersheds into larger regions.