Structure of scientific devolution

by William Yeatman on March 31, 2003

in Science

The history of science and politics explains a lot about the manifold distortions, exaggerations, and outright hokum that regularly appear in some of our major journals on the subject of global warming.

Specifically, Thomas Kuhn’s landmark book The Structure of Scientific Revolutions (first published in 1967 and reprinted about a zillion times) demonstrates that scientists have a herd mentality in which almost everyone believes in some reigning “paradigm” (in this case, disastrous global warming). Careers are spent proving that paradigm and discounting anything that disagrees with prevailing theory. That’s “science.”

As for “politics,” the two intersect, writes James Buchanan, when researchers find they can sell their ideas to policymakers for mutual gain: If a little distortion is required, so be it. That’s a powerful prism through which we must view so-called “facts”: a throng of scientists acting in concert with the political process.

Prism in hand, we can’t say we were surprised when the science community’s weekly lobbying newsletter, Science magazine, couldn’t stop gushing about recent research purportedly demonstrating that climate models of gloom and doom are right after all and the satellite temperature records reporting precious little warming were wrong.

Don’t misunderstand: Science does also report on scientific research. But anyone who thinks it isn’t lobbying hasn’t read the news or editorial sections and just about anything else in it that isn’t hard science. And any idea that Science is fair and balanced on the subject of global warming can be thrown into the trash along with previous issues.

At least that’s all we can take from their most recent handling of a new take on satellite-measured global temperature data. On May 1, Sciencexpress (a web page where the magazine posts “selected” paperschosen for their “timeliness and importance”before they actually appear in print) featured research led by Lawrence Livermore’s Ben Santer showing that climate models were in better agreement with a newly produced, and as yet unpublished, version of the satellite-measured temperature history that includes a warming trend about three times that of the oft-peer reviewed satellite-temperature history constructed by John Christy and Roy Spencer.

Although the paper itself does not come right out and say it, a between-the-lines read suggests that because the new satellite dataset is in closer agreement with model results, it is likely the more accurate reflection of the actual state of the lower atmosphere. And just in case you didn’t get that message from the paper itself, the accompanying Science press release makes it clear: “A stubborn argument against global warming may be discredited by a reanalysis of the raw data central to its claims.”

Santer used a climate model to discriminate between two versions of the satellite datathe one published by Christy that shows very little warming, and a modified, unpublished version by Frank Wentz that shows much more heating. That was certainly a novel approach because it wasn’t science. In science, we don’t bend data to fit models and then pronounce those models correct. Rather, we bend models to accommodate datathat is, reality.

If Science had any intent to be fair and balanced, it would have noted that John Christy and colleagues published a paper at the exact same time that checked the satellite data the old-fashioned (i.e., scientific) way. He and Roy Spencer compared their record with a totally independent measurement of lower-atmosphere temperature, taken from daily weather balloons, and found the two to be in strong agreement.

Weather balloons are launched twice daily from sites around the world; as they ascend through the atmosphere, they radio back observations of temperature, humidity, and pressure that are used to initialize models of daily weather forecasts. Balloon observations can be compiled into a record of lower-atmospheric temperatures that can then be compared with the satellite measurements.

It is important to realize that the way the weather-balloon data are collected is completely different from how we obtain satellite observations, and thus represents an independent measurement of the same quantity (atmospheric temperature). Of course, as is the case with any measurement, there are complications that must be considered when compiling the raw data.

For that reason, there are several different research groups that have released their own versions of the weather-balloon temperature history. To protect against any accusation of picking only the particular data set that best matches their satellite observations, Christy and Spencer compared the temperature trend during the past 24 years derived from their observations with the trend during the same period as calculated from four different manifestations of the global weather-balloon history.

Figure 1 shows what they found. The trend in their satellite record is 0.06C per decade. The trends from the various weather-balloon records range from 0.02C per decade to 0.05C per decade. In each case, the trend in the satellite record was slightly greater than that in the weather-balloon records, and the match with the two weather-balloon records with the most complete coverage of the globe was within 0.02C. That close correspondence is remarkable: Remember, these are independent observations.


Figure 1. A comparison of trends in satellite-measured temperatures (two columns on left) and weather balloonmeasured temperatures (four columns on right, including one equal to zero).

How does the Wentz and Schabel data set compare with the weather balloons? In a word, poorly. So poorly, in fact, that Santer and colleagues, instead of performing such a comparison themselves (as Christy and Spencer did), obviously anticipated the results and instead simply resorted to taking a swipe at the veracity of the weather-balloon data, saying that “As with [satellite] data, however, there is evidence that the choice of the ‘adjustment pathway’ for radiosonde data markedly influences the size and even the sign of the estimated global-scale trend.” The need to crash the satellite data is as obvious. As Tom Wigley stated in the press release from the U.S. National Center for Atmospheric research concerning the Santer paper: “The real issue is the trend in the satellite data from 1979 onward. If the original analysis of the satellite data were right, then something must be missing in the models.”

Obviously, the global warming community as a whole cannot take such a possibility lightly. Still, it is behaving in a much more predictable fashion than the climateprecisely, in fact, as Kuhn’s and Buchanan would predict. Faced with mounting evidence, it resorts to increasingly bizarre excuses as to why the models are still right, even to the point of disparaging actual real-world observations when necessary.

The question remains, and since they didn’t ask it, we will: Which do you believe, models or reality?

We’ll take reality every time.


Christy, J.R., et al., 2003. Error estimates of version 5.0 of MSUAMSU bulk atmospheric temperatures. Journal of Atmospheric and Oceanic Technology, 20, 613629.

Santer, B.D., et al., 2003. Influence of satellite data uncertainties on the detection of externally forced climate change, Sciencexpress,1 May.

Comments on this entry are closed.

Previous post:

Next post: