UCAR > Communications > News Releases > 1998 Search

November 20, 1998
EMBARGOED UNTIL 4:00 PM EST, November 26, 1998

New Analysis of Global Temperature: Year-to-Year Correlations Reveal Human Fingerprint

David Hosansky
UCAR Communications
P.O. Box 3000
Boulder, CO 80307-3000
Telephone: (303) 497-8611
Fax: (303) 497-8610
E-mail: hosansky@ucar.edu

BOULDER -- A new statistical analysis of 115 years of global temperature data, compared with output from two leading computer models of climate, has strengthened the argument that human-caused emissions of greenhouse gases, such as carbon dioxide, are warming the earth's atmosphere. The new results by Tom Wigley, a senior scientist at the National Center for Atmospheric Research (NCAR); Richard Smith (University of North Carolina); and Benjamin Santer (Lawrence Livermore National Laboratory) appear in the November 27 issue of Science. NCAR's primary sponsor is the National Science Foundation.

The paper is the first one to examine each year's average temperatures for the Northern and Southern Hemispheres by correlating them with the readings taken up to 20 years earlier or later. If the numbers rise and fall randomly over time, then the correlations are weaker than if there is a consistent long-term trend. In their paper, "Anthropogenic Influence on the Autocorrelation Structure of Hemispheric-Mean Temperatures," Wigley and colleagues found that the correlations were far stronger for actual temperature data than for simulations taken from two global climate models. For this project, the models purposely omitted this century's increase in greenhouse gases, holding them constant, which means the models replicated only the natural year-to- year variability of the climate system. The implication is that this century's warming trend has overpowered the climate system's natural variability. Correlations were higher in the actual temperature data, where the thread of warming runs through the year-to-year record, than in the constant-greenhouse-gas simulations, where no such thread exists.

In examining the strong year-to-year correlations, Wigley and his coauthors considered volcanoes and changes in solar output. According to the scientists, volcanic eruptions are so infrequent and their effects so short-lived that "they may be rejected" as an explanation for the differences between the data and the models. However, variations in solar output over the last century could have been large enough to affect some long-term trends. Global temperature rose sharply from roughly 1900 to 1940, leveled off until the 1970s, and then began another period of warming that has accelerated in the 1990s. Could the sun be responsible for this century's entire warm-up of about 1 degree Fahrenheit (0.6 degree Celsius) across the globe?

The answer depends largely on how sensitive the earth's climate is. Each global computer model has a unique sensitivity level determined by the model's physics. This allows the model atmosphere to warm by a given amount for a given increase in either greenhouse-gas levels or solar input, or both. In separate experiments using a third, simpler model, Wigley and colleagues subtracted estimates of these possible influences from the actual temperature record. The scientists discovered that, in order to explain their results using solar effects alone, this model had to be about six times more sensitive to changes in solar input than is believed realistic. Thus, they write, "solar forcing alone is insufficient to explain the behavior of the observed temperature data."

In contrast, they found, when changes in greenhouse-gas levels are included with solar output, then a sensitivity in keeping with current understanding of the climate system was enough to reconcile the correlations in actual and simulated temperature data. The results, they note, "are noticeably degraded if a sensitivity outside this range is assumed."

The authors' work is based on their confidence--bolstered by other evidence presented in their Science paper--that the two simulations of climate with constant greenhouse-gas levels realistically simulate the year-to-year changes in hemispheric temperature that are distinct from long-term trends. These simulations were performed with climate models at the United Kingdom's Hadley Centre and at the Geophysical Fluid Dynamics Laboratory of the National Oceanic and Atmospheric Administration.

"Our results imply that both anthropogenic and solar forcing have significantly affected global climate," conclude the authors. According to Wigley, "These results provide another important piece in the jigsaw puzzle of climate change, strengthening yet further our confidence that there has been a discernible human influence on climate. Furthermore, they provide additional evidence that the models used to make projections of future climate change are realistic."

NCAR is managed by the University Corporation for Atmospheric Research, a consortium of more than 60 universities offering Ph.D.s in atmospheric and related sciences.

-The End-

Writer: Bob Henson

To receive UCAR and NCAR press releases by e-mail,
contact Milli Butterworth
telephone 303-497-8601 or by email butterwo@ucar.edu

UCAR news releases
UCAR news in brief


The National Center for Atmospheric Research and UCAR Office of Programs are operated by UCAR under the sponsorship of the National Science Foundation and other agencies. Opinions, findings, conclusions, or recommendations expressed in this publication do not necessarily reflect the views of any of UCARs sponsors. UCAR is an Equal Opportunity/Affirmative Action employer.
Prepared for the web by Jacque Marshall

Last revised: Fri Apr 7 15:38:50 MDT 2000
Last revised: Mon Nov 30 16:04:34 MST 1998