In 2004, spring snow melt began anomalously early across the Western United States. USGS-gaged streams draining the Sierra Nevada, Cascades, and Rocky Mountains all recorded a spring pulse of meltwater during the second week of March. However, data from streamgages monitoring nested streamgages along the Tuolumne and Merced Rivers in Yosemite National Park suggest that this early onset of melt did not occur at all basin scales and elevations. Gages are located at elevations from 1200 m to 3300 m in basins with various slopes and aspects. In 2002, spring melt began uniformly at all monitored elevations, but in 2004, spring melt was delayed at gages at elevations above 2800 m. Snowmelt at several locations above 3000 m did not begin until May. These differences were observed not only in streamflow records but also in the vegetation, as flowers bloomed anomalously early below 2800 m but bloomed at their average time in higher regions.

This paper seeks to answer the following questions: What are the primary factors controlling the observed differences between elevation in 2004? How did solar radiation and temperature inputs differ between 2002, with a uniform melt onset, and 2004? Was melt dominated by different processes at different locations? What are the implications for climate forecasts, which predict earlier spring onsets in a warmer climate? How can snowmelt and streamflow models better capture the different behavior observed at the highest altitudes?