There are five data sets of global average temperature in the troposphere (the part of the atmosphere where our weather occurs) based on satellite data, from the two main providers, RSS (Remote Sensing Systems) and UAH (University of Alabama at Huntsville). UAH provides two of them: TLT (temperature in the lower troposphere) and TMT (temperature in the mid-troposphere); while RSS provides three: TLT, TMT, and TTT (temperature in the total troposphere). They’re all different, and each has gone through a number of revisions since the satellites began collecting data in 1979.
The UAH data are overseen by John Christy and Roy Spencer. When Christy testifies in congress (which he has done several times) he invariably uses his own data product, usually the mid-troposphere version UAH TMT. Here’s the latest version, v6.0, together with a linear regression line (in red):
Determining the uncertainty in the trend according to linear regression is complicated by the strong autocorrelation of the data. We can eliminate much of that problem by using annual averages rather than monthly data, which look like this:
The estimated rate since 1979 is 0.85 +/- 0.42 deg.C/century (95% confidence). That’s the lowest rate of increase of all five troposphere data sets.
The highest rate of increase of the five data sets is for the TTT product from RSS:
For these data the estimated rate is 1.82 +/- 0.48 deg.C/century — more than twice as fast as for the UAH TMT data! If we compare the warming rates of all five, we get this:
Naturally one wonders, why are they so different?
The UAHTLT data are warming faster than the UAHTMT data simply because the lower troposphere seems to be warming faster than the mid-troposphere. But that raises the question, why is this reversed for the data from RSS? That’s because the RSS data for TMT are their latest version 4, but they haven’t yet published a version 4 for TLT, that’s still on the older version, v3.
In testimony before congress, retired admiral David Titley (formerly chief oceanographer of the U.S. Navy) mentioned three problems with satellite data sets: orbit drift, mismatch between different satellites, and stratospheric contamination. Addressing these problems is the reason RSS has revised their data products recently. Their newest version 4 addresses the problem of orbit drift and improves the match between different satellites. Doing so results in “substantially increased global-scale warming relative to the previous version”:
The impact is strong enough that their new TMT shows faster warming than their old TLT.
The new RSS product, TTT, is designed to reduce the effect of stratospheric contamination. The stratosphere is actually cooling (because of CO2 increase), and if it contaminates a troposphere estimate it will introduce false cooling, leading to estimates of warming which are too small. As RSS states, “In the simpler TMT product, about 10% of the weight is from the lower stratosphere. Because the lower stratosphere is cooling at most locations, this causes the decadal trends in TMT to be less than the trends in the mid and upper troposphere..” They also illustrate the effect of implementing their new version:
In short, the revisions to RSS products have directly addressed some of the known problems with satellite data, strongly suggesting that the new RSS data sets — especially TTT version 4 — are a more accurate measure of how tropospheric temperature is changing. If you use the UAH data, you still have to wonder about orbit drift and mismatch between different satellites, and if you use UAH or RSS data for TLT or TMT, stratospheric contamination is still a serious problem.
Another way to evaluate the satellite data is to compare is to the upper-air measurements from thermometers carried aloft by balloons. They radio their readings back to earth, which is why it’s sometimes referred to as “radiosonde” data. Satellites have the advantage of more global coverage, but radiosondes have the advantages of being from actual thermometers, and of dating back to 1958 (satellite data don’t start until 1979).
There are a number of different global estimates of tropospheric temperature from balloon-borne instruments; one which is specifically designed for climate study is RATPAC (Radiosonde Atmospheric Temperature Product for Assessing Climate). It’s easy to retrieve a global average because the data providers have already computed it. So too has the U.K. dataset HadAT2, but the version I find only extends through the end of 2012, while RATPAC has been kept up-to-date. The two versions are in excellent agreement:
Let’s compare the satellite data to the RATPAC data, since it extends to the present year.
RATPAC global estimates are seasonal, so I’ve computed seasonal averages from the monthly satellite data. The match between RATPAC and UAH TMT (the slowest-warming of the five satellite data sets) is rather poor:
The wiggles match pretty well, but not the trend — there’s a pronounced drift between them, with UAH TMT running quite a bit cooler than the balloon data. Also, the correlation coefficient of UAH TMT with RATPAC is poorer than that of any other satellite data set. This argues very strongly against the accuracy of the UAH TMT data.
The satellite data set with the best correlation to RATPAC data is the one warming fastest, RSS TTT:
Not only the wiggles, but the trends match. In fact the trend according to RATPAC is 1.91 +/- 0.51 deg.C/century, in excellent agreement with RSS TTT but flatly contradicting UAH TMT. It’s interesting to note that the correlation between RATPAC and satellite data sets is better for all three RSS data sets than for either UAH data set.
Bottom line: the reasons for the recent revisions to RSS data and direct comparison to balloon data both lead me to believe that the RSS data products are distinctly superior to the UAH data. In particular, the RSS TTT product, because it has a much better correction for stratospheric contamination, seems best of all, and matches the balloon data best. It’s also the satellite data set showing the fastest warming.
We don’t live in the upper atmosphere, we live on Earth’s surface, where temperature is measured by thermometers. Some people believe (and climate deniers want you to believe) that the satellite data sets are better than surface temperature data sets based on thermometer measurements. ‘Tain’t so. In fact, Carl Mears, lead scientist processing the satellite data for RSS, says outright that the surface temperature records are more accurate. You can hear him say so here (21 seconds in, at 0:21).
The reason deniers claim the satellite data is superior is that they want to discredit the surface temperature (thermometer) record, and that’s because the surface data show so much warming. But among the satellite data sets, there’s one which shows far less warming: UAH TMT. Odd that these days they make a habit of showing that one, the one satellite data set which shows the least warming and correlates least with balloon data. Come to think of it, it’s not odd at all…
This blog is made possible by readers like you; join others by donating at Peaseblossom’s Closet.