What Drives Changes in the Timing of Snowmelt Runoff in the Western United States?

Future climate-change scenarios forecast widespread reductions in snowpack accumulation and resulting declines in snowmelt-derived streamflow in many mountainous watersheds around the world. These predicted changes could have important consequences for water resources supply and management by changing flood regimes and the seasonal availability of water resources in ways that current structures and policies may not be able to accommodate. The western United States, where precipitation and runoff are strongly seasonal, represent a very important laboratory for studying the impact of climatic changes on streamflow timing and water resources, as well as offering opportunities for identifying the immediate causes of such changes. It is a matter of considerable concern, then, that regionally coherent trends towards an earlier start of the snowmelt runoff season has been identified in the near-natural flow series of rivers throughout the western United States. The advance in the timing of the spring snowmelt pulse is associated with decreased April-July (AMJJ) fractional flows, and increasing fractions of the annual runoff occurring earlier in the water year, especially in March. Most of these trends began in the late 1940s and have continued through the 1990's. They are present in snowmelt-dominated streams throughout the western United States, except for an area in the southern Rocky Mountains/southwest. In order to clarify the patterns of year-to-year as well as trending changes in streamflow timing principal component analyses (PCAs) were carried out on three measures: (a) the day of onset of the spring snowmelt runoff pulses in each river, (b) the AMJJ fractional flows, and (c) the March fractional flows. The PCAs identify broad regions of common variability in the changes of streamflow timing by all three measures. The two regions that most consistently vary together in terms of the magnitude of trends and the redistribution of streamflow are the Sierra Nevada (California) and a region centered on western Idaho. Spatial patterns from the PCAs are consistent with large-scale winter and spring atmospheric circulation patterns suggesting that winter and spring atmospheric conditions could be used to predict the distribution of streamflow through the remainder of the water year. However, correlations between leading principal components (PCs) and regional-to-local scale winter and spring temperature and precipitation anomaly composites proved significant for only some cases. The results from this analysis indicate that an important part of the variability of streamflow timing derives from large-scale atmospheric-circulation changes over the western United States as a whole. Superimposed on this large-scale coherency, regional-to-local scale climatic conditions and basin-properties govern the amount and temporal distribution of runoff at the individual gauges.