Dodgy statistics and IPCC Assessment Reports

Guest essay by Christopher Monckton of Brenchley

In the 19^th century, British Prime Ministers used to say there were “lies, damned lies, and statistics”. In the 21^st century, we may say there are frauds, serious frauds, and IPCC Assessment Reports.

Recall, for instance, the notorious graph in the Fourth Assessment Report that falsely indicated that the rate of global warming is accelerating and we are to blame. Using the same statistical dodge, one can show that a sine-wave has a rising trend.

In the Fifth Assessment Report, the IPCC still cannot bring itself to behave. My expert review of an earlier draft of that report opened with these words:

“To restore some link between IPCC reports and observed reality, the report must address – but does not at present address – the now-pressing question why the key prediction of warming in earlier IPCC reports have proven to be significant exaggerations. The IPCC’s credibility has already been damaged by its premature adoption and subsequent hasty abandonment of the now-discredited “hockey-stick” graph as its logo; by its rewriting its Second Assessment Report after submission of the scientists’ final draft, to state the opposite of their finding that no discernible human influence on climate is detectable; by its declaration that all Himalayan ice would be gone in 25 years; and by its use of a dishonest statistical technique in 2007 falsely to suggest that the rate of global warming is accelerating. But the central damage to its credibility arises from the absence of anything like the warming it had predicted.”

The IPCC have indeed addressed The Pause. But they have addressed it by using statistical prestidigitation to air-brush it out. As Bob Tisdale has pointed out, the very first graphs the reader of the Summary for Policymakers will see are in Figure SPM.1, which consists of three panels. Each of these panels exploits bogus statistical techniques to vanish the pause.

Here is what They did and how They did it.

The first of the three panels shows the global instrumental surface temperature record since 1850:

And what is wrong with that? It looks innocuous enough, but a mathematician would take one look at it and sniff. He would see two things obviously wrong with drawing any conclusion about dangerously-rising 20^th-century temperatures from this graph.

First, there is the aspect-ratio dodge. For the x axis is in years and the y axis is in Celsius degrees of temperature change. One can choose any aspect ratio one wants. To make 20^th-century global warming look worse, just stretch the graph northwards.

Not all climate extremists know that. In a debate with me on Roy Green’s radio show in Canada a few years ago, one of the pointy-heads at TheSmugBlog asked the audience, with that earnest desperation in his voice that is mandatory, “But don’t you see how serious it is that global temperatures are rising at an angle of 45 degrees?”

I had to explain to the poor sap, as gently as I could, that degrees of arc and degrees of temperature change are clean different things.

But it is Dick Lindzen, whose vast experience and profound knowledge allows him to put the climate scare into perspective as no other can, who has best illustrated the insignificance of 20^th-century global warming.

His local paper, the Boston Globe, prints the previous month’s temperature movements in the city. He has superimposed on that record an orange band that shows the entire warming of 0.75 Cº over the 20^th century.

Even allowing for the fact that a global annual average will change less than a regional monthly one, it is difficult to look at Dick Lindzen’s orange band and draw the conclusion that 20^th century global warming was alarmingly beyond the bounds of natural variability.

The second statistical dodge in the IPCC’s first panel is the error-bars dodge. If you look carefully at the error-bars in the IPCC’s graph, you will see that they are absent. Let us remedy that absence:

Even today, the combined measurement, coverage, and bias uncertainties in the global terrestrial data are ±0.15 Cº. The uncertainties were far larger in the 19^th century. Notice also how much less drastic and exciting the graph looks once the 2 σ uncertainty bounds are plotted.

There is a third dodge that is not directly evident from looking at the graph itself. All around the world the record-keepers have been rewriting the temperatures in the early 20^th century to push them downward, so as to make the rate of warming over the century seem a great deal steeper than it was. Here, for instance, is New Zealand:

And Darwin Airport, Australia:

And the U.S. Historical Climate Network, before and after adjustment (this example and the next two are thanks to the vigilant Steven Goddard):

And the GISS record at Reykjavik, Iceland, before (left) and after adjustment (right):

And Santa Rosa, CA, this time with the trend-line added:

The effect of all these tamperings is to make it look as though there was more global warming in the 20^th century than there was. Fortunately, there is not so much scope for the compilers of the terrestrial temperature records to tamper with what has happened since 1979, because the watching satellites now provide an independent record of global temperature change.

So to the second of the three mendacious panels in Figure SPM.1:

This graph is an illustration of a meme that has become a favorite with the apologists for Apocalypse: the most recent decade was warmer than earlier decades, so global warming is still getting worse (for the theology of the New Religion, standing common sense on its head, is that warmer weather is worse than cooler).

The priceless advantage of taking decadal averages, if one wants to magic the Pause away, is that it wipes out the entire trend of the most recent decade. One can dock off a further two years if, as here, one uses the decades 1991-2000, 2001-2010 etc. rather than 1993-2002, 2003-2012 etc. Finally, using decades docks off all the months of the current year. So this statistical dodge neatly erases the past 12 years 8 months of the Pause.

And, by what is perhaps more than a coincidence, the length of the Pause, taken as the longest period exhibiting a zero least-squares linear-regression trend on the mean of the monthly global mean surface temperature anomalies on the three terrestrial datasets, is – er – precisely 12 years 8 months.

There is another and more subtle dodge here. As we saw in the earlier graph of the uncertainties in the HadCRUt4 global temperature dataset, the error bars narrow toward the present. The way the IPCC has presented the decadal blocks on the graph exploits this to make it seem that the blocks in the 1970s, 80s, 90s, and noughties are much further apart than those in the 1910s, 20s, 30s, and 40s, implying without quite saying so that the rate of warming over the four most recent decades on the graph was significantly greater than the warming earlier in the 20^th century.

Dick Lindzen, however, uses a graph that shows how little difference there is between the earlier and later periods of warming, even though it was only in the later period that we could have exercised much influence.

One panel shows the global temperature anomalies from 1895-1946. The other shows the anomalies from 1957-2008. Both cover 52 years. Both are plotted to an identical scale. Dick Lindzen asks his audiences whether they can tell which panel covers which period. It is not at all easy to tell.

Which brings us to the third panel. Here, the dodge is one of the newest in the arsenal of statistical shiftinesses on which the IPCC draws with such disfiguring frequency and relish. It is the use of colors, and bright ones at that, to try to suggest that the mild and beneficial global warming of the 20^th century was grievous and alarmingly damaging.

And here the IPCC will find that it has made a mistake. Previously it has chiefly used bright colors in the red scale to indicate predictions of future planetary overheating. However, most people, on looking about them, will see remarkably little change as a result of 100 years’ warming. The trees are greener; the deserts have shrunk by quite a bit (the Sahara by 300,000 sq. km in 30 years); sea level is 8 inches higher; and that’s it.

Recoloring the graph in neutral tones would have been more scientifically adult:

Does the Earth really look that much different as a result of 0.7 Cº global warming over 100 years? Not really. Let us end with a God’s-eye view of the planet He has given us. Really, our stewardship has not left it in too much of a mess.

Yet.