May 4, 2004

Assault From Above

Filed under: Satellite/Balloons, Surface

Nature authors make an assumption that defies the laws of physics. But that doesn’t stop them from concluding that the satellite-based temperature record is dramatically cooled by the atmospheric layer just above it.

It’s common knowledge that the satellite-based temperature record of the earth’s lower atmosphere shows much less warming than surface observations or climate model predictions for the past 25 years. The big question, though, is why. A new study in Nature magazine claims the satellite measurements are in error because they include a cooling effect from the atmospheric layer just above it.

But in formulating their case, the authors assume the impossible.

The idea behind what Fu et al. reported to Nature seems simple enough. The satellite-based temperature measurements of the earth’s atmosphere are not collected from a single altitude, but instead represent the weighted average temperature of a rather large atmospheric layer. So large, in fact, that the measurements that are supposed to represent the middle to upper troposphere (from about 5 to 8 miles up) actually include part of the lower stratosphere (from about 8 to 12 miles up), as Figure 1 shows.

That extra few miles is potentially problematic because ozone loss in the stratosphere has led to a cooling trend there. Ozone absorbs incoming solar radiation, thus warming the air, so less ozone means less warming. That cooling trend in the stratosphere contaminates to some degree the temperature trend in the mid- to upper troposphere and perhaps masks the warming from an enhanced greenhouse effect.

Weighting Functions

Figure 1. Atmospheric contribution to the satellite-based temperature observations of the stratosphere (red line) and mid- to upper troposphere (blue line). Notice the region of overlap (shaded pink) that indicates that some portion of the stratosphere contributes to the observations of the tropospheric temperatures.

What happens when you remove the stratospheric cooling component from the tropospheric temperature record? That’s what Fu and colleagues’ attempted to do. Their procedure was to use the satellite temperature measurements of the stratosphere as an indicator of how much cooling had worked its way into the history of the tropospheric temperatures. By combining the two datasets in such a way as to remove the stratospheric cooling contamination, Fu concluded that the true temperatures in the troposphere have been warming at a greater rate than that which has been reported to date, and one which brings the satellites more in line with surface observations and climate model projections.

They therefore believe that they were able to solve the riddle as to why a large discrepancy had existed between the trends in surface temperatures and those of the troposphere—a discrepancy that climate models could not replicate.

Not so fast. It turns out that the recognition that the stratospheric temperatures were contaminating the satellite measurements of the middle and upper troposphere was made long ago by the co-founders of the original satellite-based temperature history, University of Alabama-Huntsville scientists John Christy and Roy Spencer.

Spurred by a desire to produce a “true” tropospheric-only temperature dataset, Spencer and Christy extensively examined various methodologies, including one like Fu et al. describe, in the late 1980s and early 1990s. Spencer and Christy quickly realized that such a technique was infeasible because it produced a situation that violated some basic laws of physics—namely, that energy must be positive (Figure 2).

Weighting Functions

Figure 2. The stratospheric contribution to the satellite-based measurements of tropospheric temperatures. Solutions that don’t violate the basic laws of physics must have a weighting that is greater than or equal to zero. The stratospheric contribution to the raw tropospheric temperature observations is given by the red line (top). The best, most physically realistic adjustment to the raw observations using data in the stratosphere to adjust tropospheric temperatures is given by the green line (middle). The solution used by Fu et al. is shown by the blue line (bottom). The physically impossible negative contribution (shaded region with hatching) equals the amount of positive contribution (blue shaded region), such that the total contribution from the stratosphere equals zero.

Figure 2 shows three curves. The top one represents the contribution that the stratosphere makes to the temperature measurements of the troposphere; ideally, that value would be zero, which would indicate no contribution. By combining the tropospheric measurements with actual measurements of stratospheric temperatures, it is possible to “subtract out” some of the stratospheric impact. But that must be done carefully, or else a non-physical result is produced—after all, you can’t remove more than you have. The zero line in Figure 2 shows the limit of removal; you can’t produce a result that goes below this line (or else you rely on negative energy—and break the rules of physics). The middle line of Figure 2 shows the best physically possible solution that can be produced using the data from the stratosphere to adjust temperatures in the troposphere—but one that still makes a sizeable stratospheric contribution. Figure 2’s blue line (bottom) represents Fu et al.’s solution. They succeed in setting the stratospheric contribution roughly equal to zero (the average of the area above and below zero). But they unfortunately created an unrealistic representation of the real atmosphere.

Though the portions in the stratosphere appear to cancel each other, here is the problem: the atmosphere in the negative portion is cooling rapidly (because of ozone depletion) while the atmosphere in the positive portion is changing very little. Fu et al., in terms of net effect, multiply the rapid-cooling-trend layer by a too-large negative factor (making it appear to warm up), while multiplying a near-zero-trend layer by a positive factor (which will have a very minor impact). Rather than canceling the stratospheric influence, therefore, Fu’s method adds a spurious warming trend to the net result.

Simply put, Fu et al. overcorrected for the stratospheric cooling.

In the name of good science, Spencer and Christy abandoned that technique years ago because even the best acceptable solution (the middle curve of Figure 2) still had too much stratospheric influence. Instead, they developed a clever solution that made use of different viewing angles by a single channel from the satellite to measure the temperatures occurring lower in the atmosphere. By doing so, they were able to produce the now famous lower tropospheric temperature history, one that is essentially free of stratospheric effects, and one that shows only about half as much global warming during the past 25 years as does the network of surface thermometers scattered across the globe. Just a few weeks ago, the latest in a long line of publications again demonstrated the confidence we can all have in that dataset (for more details see

Yet more than a dozen years later, a technique that was discarded as being physically implausible by the original developers of satellite-based temperature histories rears its head again. That Fu et al.’s physically impossible result made it through the review process at Nature magazine is unfathomable. If it proves one thing, it is that yet again the reviewers Nature magazine relies on to ensure scientifically sound results are failing them. Are Nature’s editors are intentionally allowing articles of certain subjects to be published despite their lack of scientific worthiness? Are the editors themselves underqualified to be judging the scientific merit of the research or assigning competent reviewers to do so? Whichever the case, the result is the same: When it comes to climate change, the reputation of Nature magazine is becoming severely degraded.


Christy, J. R., and W. B. Norris, 2004: What may we conclude about global tropospheric temperature trends? Geophysical Research Letters, 31, L06211, doi:10.1029/2003GL019361.

Fu, Q., et al., 2004. Contribution of stratospheric cooling to satellite-inferred tropospheric temperature trends. Nature, 429, 55-58.

Spencer, R.W., and J.R. Christy, 1990. Precise monitoring of global temperature trends from satellites. Science, 247, 1558-1562.

No Comments

No comments yet.

RSS feed for comments on this post. TrackBack URI

Sorry, the comment form is closed at this time.

Powered by WordPress