Satellite temp data

Mark S

Has anyone looked at the satellite temperature data? http://www.drroyspencer.com/latest-global-temperatures/

It seems to me that much of the reconstructed historical temperature data has been called into question.

I haven't found much debate centered around the satellite data and it seems less prone to manipulation. Granted we only have data since 1979 - though at this point it seems like the only reliable data to work from. Any thoughts or history on the satellite data is appreciated.

Xnn

There has actually been a lot of work with satellite data. There are 3 global efforts and it takes a lot of work to compile the data correctly. A number of corrections need to be made with satellite data to account for decay of the orbits, differant satellites, differant times of day the observations are made as well as cloud cover and what not.

Two of the global efforts are in good agreement, while the third (Spencer) is something of an outlier. This doesn't mean Spencer is necessarily wrong, but he has not been able to show the other 2 groups why his methods/corrections are more accurate.

Andre

Both satelite temperature reconstructions from UAH and RSS are in good agreement despite the difference in methods, which should say something about their robustness.

sylas

We seem to have drifted a bit from the question actually posed in the thread title. Here's my attempt to comment on that....

I'm going to stick exclusively to the matter of satellite data, on which there has been a considerable amount of debate between the scientists working with this data. I like this particular problem, because the debate has generally been so constructive, with different groups working together even as they find problems in each others' work. Perhaps because it is less in the public eye, this has not had the same level of angst as various other matters.

Background

The raw data from satellites is microwave brightness, in different frequencies, coming up from the Earth. One can estimate temperatures for different altitudes in the atmosphere, or for temperatures at the surface, by appropriate weighting of the different microwave frequencies. It is also necessary to correct for satellite orbital decay, satellite drift for the time of day at which points are sampled, merging and calibration of data from different satellites, effects of cloud, and a lot more. No one satellite has been in orbit this whole time; the satellites have a limited a lifetime, and data from successive satellites has to be calibrated and combined to a single time series.

There are a number of researchers who work at processing this raw satellite data into a temperature record. Two groups in particular are very influential, and their datasets are both widely used. The two groups are Remote Sensing Systems (RSS) and University of Alabama Huntsville (UAH). Both groups use exactly the same raw data, but process it independently; and there are some significant differences in their results. The differences seem to be greatest over the tropics.

An introduction to the microwave data and the temperature extraction processes is available at the RSS (remote sensing systems) website. See Description of MSU and AMSU Data Products. The differences between RSS and UAH (University of Alabama Huntsville) are entirely due to the mathematical procedures for homogenizing the raw data. It is not possible at this stage to identify plainly which one is correct, and so frequently people who use satellite data will try both; with the differences considered as a part of the error uncertainty with this data.

It is important to recognize that when you look at temperature data from satellites, you are not looking at a direct measurement as if with a thermometer; but at the end result of a very complicated series of calculations.

The data you have shown at your link is for the lower troposphere, from the UAH group; an estimate averaged over the entire globe.

Comparisons

One method of attempting a check on the satellite calculations is to compare with atmospheric temperatures from radiosondes (weather balloons). These measurements, however, also have major problems and systematic errors, and as with the satellites there are a number of different groups who obtain different results. Measuring the atmosphere is a lot harder than measuring the surface, where you can put an instrument exactly at the point you are measuring, and expect it to stay there; and where you can examine it closely while it is in operation. Although satellites sound like they ought to be really accurate, the issues with an atmospheric measure such as you have cited mean it is still substantially less reliable than the direct surface measurements.

The RSS and the UAH groups are in constant contact with each other, and appear to have a strong working relationship, even though there are unresolved questions concerning which group is doing more reliable calculations. It is a cordial scientific rivalry.

The RSS group has an interactive data visualization page, where you can quickly compare numbers from RSS, UAH and four different radiosonde groups, over the whole globe or three alternative latitude bands, and over four distinct atmospheric layers. See Comparing RSS and UAH Atmospheric Temperatures to Adjusted Radiosonde Data.

I'll quickly summarize the differences, by comparing the regression trend from 1979 to the present, for RSS and UAH, over the three latitude bands. I'm only looking at lower troposphere, which is what you have linked to. The numbers are all a measure of a warming trend in degrees Celcius per decade.
The RSS group tends to find smaller trends at mid to high latitudes, and larger trends in the tropics, by comparison with UAH. Both groups get a roughly similar global average, which shows a slightly stronger warming trend than the surface measurements. Thus the UAH group has a much larger difference in the trend between tropics and mid latitudes than the RSS group.

Puzzles and solutions
The history of this satellite data is very interesting. Around 2000 and onwards there was a very significant discrepancy between the two groups, with the UAH showing an even weaker tropical trend, quite contrary to what was expected from atmospheric physicists. In 2005 the problem was solved very dramatically, when the RSS group discovered an algebraic error in the UAH calculations, pretty much a case of using a negative sign rather than a positive for a correction associated with the diurnal day/night cycle. The UAH group immediately acknowledged the correction, with thanks, and greater part of the problem was solved.

• Reference: Mears, A.C., and Wentz, F.J. (2005) The Effect of Diurnal Correction on Satellite-Derived Lower Tropospheric Temperature, in Science, 2 Sept 2005, Vol. 309. no. 5740, pp. 1548 – 1551. DOI: 10.1126/science.1114772

The UAH group has likewise helped located some errors in the RSS processing; not quite so dramatic, but still a good indication of how well these two groups work together, despite being rivals.

It is actually quite an exciting story of hard science, solved puzzles and ongoing unknowns. To get an idea of the significance of the satellite data and the resolution of the problem back in 2005, you can read the two different versions of "Global Warming: The Balance of Evidence and Its Policy Implications", from 2003 and 2007, which can be found in the sticky Earth forum thread: Reviews on Global Warming.

Cheers -- sylas

PS. The trends I have quoted above are for a "sub-sampled" comparison, which is required to facilitate a proper comparison with the radiosonde data. More about how this works is explained on the page. The trend numbers I have given are actually for data sub-sampled to fit with the coverage of the HadAT radiosonde data.