BOULDER--A new analysis of satellite data collected since the late 1970s from the lowest few miles of the atmosphere indicates a global temperature rise of about one-third of a degree Fahrenheit between 1979 and 1999. The results are at odds with previous analyses that show virtually no warming in the satellite record over the 20-year period. The findings will be published by the journal Science at its Science Express Web site (http://www.sciencexpress.org) on May 1. The team behind the study includes scientists Tom Wigley, Gerald Meehl, Caspar Ammann, Julie Arblaster, Thomas Bettge, and Warren Washington, all from the National Center for Atmospheric Research. The lead author is Ben Santer of Lawrence Livermore National Laboratory. "It's undeniable that the agreement with both global climate models and surface data is better for the new analysis than for the old one," says Wigley. Over the past 25 years, a series of instruments aboard 12 U.S. satellites has provided a unique temperature record extending as high as the lower stratosphere. Each sensor intercepts microwaves emitted by various parts of the atmosphere, with the emissions increasing as temperatures rise. These data are used to infer the temperature at key atmospheric layers. Since the 1990s, skeptics have pointed to the absence of a warming signal in the satellite-derived temperatures, which stood in contrast to a distinct warming trend in average air temperature at Earth's surface. A 2000 report from the National Research Council concluded that both trends might be correct--in other words, the global atmosphere might be warming more quickly near the ground than higher up. Although Wigley agreed, he felt there was more to be explained. "The real issue is the trend in the satellite data from 1979 onward," says Wigley. "If the original analysis of the satellite data were right, then something must be missing in the models. With the new data set, the agreement with the models is improved, and the agreement with the surface data is quite good." In order to glean temperatures from the raw satellite data, several adjustments and corrections must be made. Until now, only one group, based at the University of Alabama in Huntsville (UAH), had produced a complete set of global temperatures from the raw data. For the new study, a group based at Remote Sensing Systems in Santa Rosa, California, applied a revised set of corrections to the satellite data. These corrections accounted for the effects of heating on the radiation sensor itself--the first time this source of error had been addressed fully, according to the authors--as well as new adjustments for the drifting orbit of each satellite and other factors. The group found a warming trend of 0.16 degree F per decade in the layer between about 1.5 and 7.5 miles high, compared to a trend of 0.02 degree F in the previously published UAH analysis. Both estimates have a margin of error of nearly 0.2 degree F (plus or minus). According to the authors, the new results are a closer match with surface warming, as well as with four computer-model simulations of 20th-century climate produced by NCAR and Los Alamos National Laboratory. As a further check on the new satellite data set, the team examined regional patterns. Using a statistical technique, the group analyzed the 20th-century simulations and searched for an underlying "fingerprint" of climate change. For instance, the rates of warming in the satellite-monitored data vary by latitude from north to south. The authors found that the overall fingerprint of climate change in the models resembled this and other regional patterns found in the new satellite data set. The study was supported by the U.S. Department of Energy and the National Oceanic and Atmospheric Administration, with contributions from the National Science Foundation through its institutional support for NCAR.