Using a new analysis of satellite temperature measurements, scientists from the Lawrence Livermore National Laboratory have determined that uncertainties in satellite data are a significant factor in studies attempting to detect human effects on climate.
Since 1979, Microwave Sounding Units (MSUs) have been flown on 12 different polar-orbiting weather satellites operated by the U.S. National Oceanic and Atmospheric Administration. MSU instruments measure the microwave emissions of oxygen molecules, which are related to atmospheric temperature. By monitoring microwave emissions at different frequencies, it has been possible to ‘back out’ information on temperature changes in various layers of the atmosphere.
Until recently, only one group — from the University of Alabama at Huntsville — had analyzed the raw MSU data. This analysis is complicated by such factors as the gradual decay and drift of satellite orbits (which affect the time of day at which MSU instruments measure atmospheric temperatures) and by problems related to the calibration of MSUs.
The pioneering Huntsville analysis of the MSU data suggested that the troposphere (the lowest layer of the atmosphere) had undergone little or no overall warming since 1979.
Some have used this finding to question both the reality of human-induced global warming and the reliability of computer climate models, which predict that the troposphere should have warmed in response to increases in greenhouse gases. The Huntsville results are also at odds with thermometer measurements indicating pronounced warming of the Earth’s surface during the satellite era. Now a second group has conducted an independent analysis of the same raw MSU data used by the University of Alabama scientists. This group, led by Carl Mears, Matthias Schabel, and Frank Wentz of Remote Sensing Systems in Santa Rosa, uses different methods to correct for satellite orbital drift and MSU calibration problems. They find that the troposphere probably warmed by roughly 0.1 degrees Celsius (0.18 degrees Fahrenheit) per decade from 1979 to 2001. This amounts to a total rise in tropospheric temperature of 0.4 degrees Fahrenheit over this period. The implications of these uncertainties for attempts to detect human effects on climate are explored by Livermore scientists Benjamin Santer, Karl Taylor, James Boyle and Charles Doutriaux, along with researchers from Remote Sensing Systems, the National Center for Atmospheric Research, Lawrence Berkeley National Laboratory and the University of Birmingham in England.
Their findings are reported in the May 1 online edition of Science Express in a paper titled, “Influence of Satellite Data Uncertainties on the Detection of Externally-Forced Climate Change.” The Lab scientists and their colleagues use results from a state-of-the-art computer climate model that was run with estimates of historical changes in greenhouse gases, sulfate aerosols, ozone, volcanic dust and the sun’s energy output.
These experiments were performed at the National Center for Atmospheric Research in Boulder, Colo., and the Department of Energy’s National Energy Research Scientific Computing Center in Berkeley, Calif. The model runs yield detailed patterns (or ‘fingerprints’) of tropospheric temperature change. These fingerprints are identifiable in the Santa Rosa satellite data showing a warming troposphere, but not in the University of Alabama MSU records. Model output from these and other simulations are freely distributed to the research community. The positive detection of model tropospheric temperature ‘fingerprints’ in the Santa Rosa satellite data is consistent with earlier research that has found human-induced signals in such climate variables as surface temperature, ocean heat content, tropopause height and Northern Hemisphere sea ice cover.