TCS Daily

When Is Global Warming Really a Cooling?

By Roy Spencer - May 5, 2004 12:00 AM

Much media attention is focusing on the forthcoming big-budget climate disaster movie "The Day After Tomorrow" and how much scrutiny the "science" on which it is based deserves. But there are some developments in the world of serious climate science that certainly deserve greater scrutiny.

A new paper1 appearing today in the journal Nature purports to solve the long standing "problem" of the satellite-based global temperature record not showing much warming over the last 25 years (only +0.085 deg C/decade -- about a third of what is expected from climate models for the troposphere). Instead, all it does is help answer the question: "is the quality of peer review in the popular science journals getting worse?" (The answer is "yes.")

By way of background, the Microwave Sounding Units (MSU) on the NOAA polar orbiting satellites measure deep layers of the atmosphere, with each instrument channel measuring the average temperature of a different layer (see Fig. 1). John Christy, a professor of atmospheric science at the University of Alabama, Huntsville, and I discovered in 1990 that these instruments were so stable in their calibration that we have been using them ever since for climate monitoring of tropospheric and lower stratospheric temperatures since the satellite record began in 1978.

The lowest layer (the troposphere) is measured by channel 2, and this is where global warming is supposed to occur. The lower stratospheric layer is measured by MSU channel 4. Christy and I have measured substantial cooling (-0.47 deg. C/decade) in this layer over the same 25 year period.

But because the layers measured by the satellite are so thick, there is a intermingling of the warming and cooling signals. This means that warming in the tropospheric channel would be partly cancelled by stratospheric cooling occurring in the upper portion of this layer. Because of this problem, we devised a lower tropospheric retrieval2 based upon different Earth viewing angles from the MSU tropospheric channel. As can be seen in the accompanying figure, this removes the stratospheric influence and so allows us to monitor the lower and middle troposphere for signs of global warming.

Fig. 1 Microwave Sounding Unit weighting functions for channels 2 and 4, along with derived weighting functions meant to remove the influence of lower stratospheric cooling on MSU channel 2 by Spencer & Christy (1992, "TLT") and Fu et al. (2004). The Fu et al. weighting function shows substantial negative weight above 100 hPa, a pressure altitude above which strong cooling has been observed by weather balloon data. This leads to a misinterpretation of stratospheric cooling as tropospheric warming.

Enter the new Nature study. The authors, noticing that channel 4 measures the extreme upper portion of the layer that channel 2 measures (see Fig. 1), decided to use the MSU channel 4 to remove the stratospheric influence on MSU channel 2. At first, this sounds like a reasonable approach. We also tried this thirteen years ago. But we quickly realized that in order for two channels to be combined in a physically meaningful way, they must have a large percentage of overlap. As can be seen in Fig. 1, there is very little overlap between these two channels. When a weighted difference is computed between the two channels in an attempt to measure just the tropospheric temperature, an unavoidable problem surfaces: a large amount of negative weight appears in the stratosphere. What this means physically is that any attempt to correct the tropospheric channel in this fashion leads to a misinterpretation of stratospheric cooling as tropospheric warming. It would be possible for their method to work (through serendipity) if the temperature trends from the upper troposphere to the lower stratosphere were constant with height, but they are not. In this instance, the negative (shaded) area for the Fu et al. weighting function in Fig. 1 would be cancelled out by its positive area above about 200 millibars. Unfortunately, weather balloon evidence suggests the trends change from warming to strong cooling over this altitude range.

This kind of mistake would not get published with adequate peer review of manuscripts submitted for publication. But in recent years, a curious thing has happened. The popular science magazines, Science and Nature, have seemingly stopped sending John Christy and me papers whose conclusions differ from our satellite data analysis. This is in spite of the fact that we are (arguably) the most qualified people in the field to review them. This is the second time in nine months that these journals have let papers be published in the satellite temperature monitoring field that had easily identifiable errors in their methodology.

I will admit to being uneasy about airing scientific dirty laundry in an op-ed. But as long as these popular science journals insist on putting news value ahead of science, then I have little choice. The damage has already been done. A paper claiming to falsify our satellite temperature record has been published in the "peer reviewed" literature, and the resulting news reports will never be taken back. This is one reason increasing numbers of scientists regard Science and Nature as "gray" scientific literature.

1. Fu, Q., C.M. Johanson, S.G. Warren, and D.J. Seidel, 2004: Contribution of stratospheric cooling to satellite-inferred tropospheric temperature trends. Nature, 429, 55-58.

2. Spencer, R.W., and J.R. Christy, 1992: Precision and radiosonde validation of satellite gridpoint temperature anomalies, Part II: A tropospheric retrieval and trends during 1979-90. J. Climate, 5, 858-866.


TCS Daily Archives