Paleoclimate Records
Paleoclimate refers to climate prior to the beginning of instrumental records—so in the Intermountain West, before the mid- to late 1800s. Various environmental indicators or “proxies” can be used to reconstruct paleoclimatic variability extending back hundreds or thousands of years. With this long window into the past, paleoclimate data give us a more complete understanding of the modes of natural climate variability, and a better context for interpreting recent climate events and projecting future change.
Tree-ring reconstruction of annual streamflows for the Colorado River at Lees Ferry, AZ, from 762–2005, with a 20-year running mean applied. The gaged period (green box) represents only 8% of the tree-ring record. The yellow box highlights the mid-1100s "megadrought". From Meko et al. (2007); see this page at the TreeFlow website for more information on this reconstruction.
In the Intermountain West, the most useful paleoclimate proxies for undertanding the past 2,000 years are tree rings. The annual growth of trees in many parts of closely reflects annual moisture variability, so tree-ring records can be used to reconstruct, or extend, observed records of precipitation, drought indices, and annual streamflow. These reconstructions indicate that more severe and sustained droughts occurred in the several centuries prior to 1900 than those that have occurred in more recent times.
Resources
The TreeFlow web resource developed and maintained by WWA and CLIMAS provides tree-ring reconstructions of annual streamflow for over 60 gages across the U.S., over 30 of which are within Colorado, Utah, and Wyoming. It also has links to tree-ring reconstructions of precipitation, drought indices, and snowpack.
The NOAA Paleoclimatology Program has the world's largest online archive of paleoclimate reconstructions, organized by climate variable and region. One particularly useful dataset is the North American Drought Atlas (gridded summer Palmer Drought Index reconstructions).
University of Colorado Boulder 
