Climatic Change

, Volume 62, Issue 1, pp 217–232

Changes in Snowmelt Runoff Timing in Western North America under a `Business as Usual' Climate Change Scenario


  • Iris T. Stewart
    • Scripps Institution of Oceanography
  • Daniel R. Cayan
    • Scripps Institution of Oceanography
    • U.S. Geological Survey
  • Michael D. Dettinger
    • U.S. Geological Survey
    • Scripps Institution of Oceanography

DOI: 10.1023/B:CLIM.0000013702.22656.e8

Cite this article as:
Stewart, I.T., Cayan, D.R. & Dettinger, M.D. Climatic Change (2004) 62: 217. doi:10.1023/B:CLIM.0000013702.22656.e8


Spring snowmelt is the most important contribution of many rivers in western North America. If climate changes, this contribution may change. A shift in the timing of springtime snowmelt towards earlier in the year already is observed during 1948–2000 in many western rivers. Streamflow timingchanges for the 1995–2099 period are projected using regression relationsbetween observed streamflow-timing responses in each river, measured by the temporal centroid of streamflow (CT) each year, and local temperature (TI) and precipitation (PI) indices. Under 21st century warming trends predicted by the Parallel Climate Model (PCM) under business-as-usual greenhouse-gas emissions, streamflow timing trends across much of western North America suggest even earlier springtime snowmelt than observed to date. Projected CT changes are consistent with observed rates and directions of change during the past five decades, and are strongest in the Pacific Northwest, Sierra Nevada, and Rocky Mountains, where many rivers eventually run 30–40 daysearlier. The modest PI changes projected by PCM yield minimal CT changes. The responses of CT to the simultaneous effects of projected TI and PI trends are dominated by the TI changes. Regression-based CT projections agree with those from physically-based simulations of rivers in the Pacific Northwest and Sierra Nevada.

Copyright information

© Kluwer Academic Publishers 2004