Reconstructing Western North American Temperatures Over the Past 500 Years

NCDC scientist, Dr. Eugene Wahl together with Dr. Jason Smerdon from Columbia University published a new tree-ring based reconstruction of western North American annual temperatures, which they used to evaluate the performance of paleoclimate reconstructions derived from climate proxies versus those derived from noise-only predictors.

 Climate proxies are naturally preserved records like tree rings or the characteristics of sediments in lakes and oceans that can indicate past climatic conditions through their biological, physical, or chemical connections to environmental conditions. To test the ability of the proxy data to reconstruct the past climate, the scientists also performed reconstructions with random noise, or non-informative predictors. These predictors are randomly drawn and have properties like real proxy data but are not directly related to actual climate information. Using these predictors in place of real proxy data with the same methodology offers a way to evaluate reconstruction performance that can happen by chance.

This derived paleoclimate reconstruction provides 5° x 5° gridded surface temperature coverage primarily over the western United States from 1500 to 1980, and it uses a large statistical ensemble to characterize uncertainty. A key result based on this ensemble methodology is that the most recent decades are, at the level of the 20-year average, very likely the warmest across midlatitude western North America during the past 500 years.

Temperature reconstructions, like this one, have played a significant role in understanding the climate prior to the beginning of the instrumental record by quantitatively extending the record back in time in an objective way. Dr. Wahl and Dr. Smerdon’s new research was published in an article, “Comparative performance of paleoclimate field and index reconstructions derived from climate proxies and noise-only predictors,” in Geophysical Research Letters. In this article, the real proxy data are shown to strongly outperform the non-informative predictors when both are used to derive otherwise parallel reconstructions of western North America annual temperatures.

This record's data can be accessed from NCDC at