Tallbloke's Talkshop

Andrew Cooper: IPCC using differing graph versions

Guest post from Andrew Cooper, AKA ‘Scute’. It’s a bit hard to tell if the baseline for temperature on video version from the IPCC press conference runs from 1861 or 1851, so we’ll let the talkshop readership decide. They certainly did switch the cumulative emissions baseline. It all makes a difference to presentational impact, especially the misleading slope  in temperature from 2000-2010.

graph from page 36 of the Summary for Policy Makers (SPM) with the version shown at the IPCC press conference below :


The graph includes an unabashed cherry-picking of temperature data points (2000, low and 2010, high) so as to fit the linear relationship the IPCC are promulgating between total accumulated CO2 and temperature since the late 19th century. It is true they can say they were just plotting in tens of years, to show a long term trend from 1860 but the current 15-year standstill is now turned into a soaring rise. They, of course, know this is a complete misrepresentation of the truth but are hiding behind the fig leaf argument that the decimal system dictates that you choose years with a ’0′ on the end.

This is made worse by the fact that the graph was touted in the live streaming of the SPM press presentation but is languishing on the last page of the 36-page SPM document. And you had to be quick to see the problem in the few seconds that they zoomed in on it at the conference.

Here is a clip from the BBC where they parade the offending graph (at 1:30):


Notice that Thomas Stocker, the co-chair of Working Group 1, says it was finalised “after many hours of deliberation and preparation by the scientists”.

There is so much wrong with this graph. For a start, there are in fact two versions, one presented at the conference and the other on page 36 of the SPM. You can tell because the 2010 temperature plot is at 0.91 C (above 1861 to 1880 average) in the press conference graph and at 1.05 C in the printed summary. The 2000 temperature has crept up between the press conference video and the print version too. It’s at 0.59 C in the video and at 0.74 C in the print version. That’s a 0.14-0.15 C discrepancy between two sets of historical (i.e. set in stone) data. The 1861-1880 baseline is the same for both graphs so the plots should be identical. The same thing is happening to the 1950 and 1980 plots.

In other words, the “preparation” done by the scientists was to take the page 36 SPM print version and shunt the entire historical plot down by 0.15 C- that’s a decade’s worth of warming. No wonder they “deliberated” for “many hours” over it and got not a wink of sleep.  The video version of the plot now sits a little lower and seems to show off the linear trend rather better. So they probably think their lucubrations paid off. Trouble is, in their haste they allowed two incompatible graphs to slip into the public domain.

It doesn’t end there. According to the 3 main temperature sets, averaged, there is a 0.253C difference between the 2000 and 2010 temperatures. Both the offending graphs show this difference to be  0.33 C. The combined datasets show around that trend from 1990 rather than from 2000!

The cherry-picking of these two dates is already a gross misrepresentation of the underlying data but the spurious stretching of the gap allows an even steeper upward kick where, as we know, there is a flatline. If that flatline had been shown, it would have drifted off horizontally to the right with never a hope of linking smoothly to the projected future trend that soars above it. An ugly 90 degree vertical hike would have been unthinkable so, indeed, “much deliberation” was needed to devise a way of avoiding it.

Source for global averaged temperature: Hadley Centre Global Temperature page (table at bottom).


Deliberation’ and ‘preparation’. Those two words sound like euphemisms for something rather stronger. Stocker uses them specifically and exclusively in relation to this graph.

This all smacks of massaging the plot data to fit a dubious narrative (force-fitting the linear trend emphasised in the video) and soaring temperatures (plus ca change). I think Thomas Stocker and his scientists should be called on it.


There is one legitimate difference between the two graphs. The press conference graph shows cumulative carbon emissions from 1850; the print version shows it from 1870. This difference shunts the plot along the x axis by a tiny amount for the press conference version. This has no bearing on the spurious shunt of the plot down the y axis for the temperature. As stated above, both graphs cite the 1861-1880 baseline.

The temperatures were measured carefully with a ruler with the plot zoomed. They are the most accurate readings I could get. I concede that they could be out by 0.02 to 0.03deg C. However, this is a small fraction of the discrepancies discussed and therefore doesn’t detract from the thrust of the article.