The endpoint fallacy: A dishonest statistical abuse by the IPCC
 
Lies, damned lies, & statistics: The IPCC’s 2007 report, cited in a science lecture by Rajendra Pachauri, chairman of the IPCC’s science working group, and also about to be cited with approval in a “Technical Support Document” in justification of the EPA’s imminent finding that CO2 and five other gases are jointly or severally “dangerous” in terms of the Clean Air Act, contains the above graph purporting to show that the rate at which the world is warming is inexorably increasing. The graph is an egregious instance of the endpoint fallacy, a dishonest abuse of statistics by which false trends are demonstrated by careful selection of endpoints or (in the present instance) startpoints when evaluating data trends.

Removal of the IPCC’s false trend-lines from the data reveals the true position –

The truth: 1860-1880 and 1910-1940 warmed just as fast as 1975-1998
 
No anthropogenic signal: The world warmed at the same rate from 1860-1880 and from 1910-1940 as it did from 1975-1998 (see the three parallel magenta trend-lines). The former two periods occurred before humankind can possibly have had any significant influence on temperature. Therefore there is no anthropogenic signal in the global temperature record, and no basis for the IPCC’s assertion that the warming rate is accelerating.

To demonstrate why the endpoint fallacy is a shoddy statistical abuse, we can use the IPCC’s own global temperature data to deliver a result precisely the opposite of that which the IPCC tries to draw –

Heading for a new Ice Age? “Global warming” becomes cooling
 
Any result you want: Beginning in 1993 (top left) and advancing the start-date successively by 4 years at a time, the IPCC’s own data show the world heading for an Ice Age. Using the same data as the IPCC, we reach a diametrically opposite (and equally unjustifiable) conclusion, proving the IPCC’s abuse of statistical method.

No reliance can be placed upon purported temperature trends that depend arbitrarily upon a careful selection of start-dates and end-dates. The IPCC and Dr. Pachauri were wrong, and the EPA will be wrong, to rely upon the endpoint fallacy as the basis for their erroneous conclusion that warming rates that are far from unprecedented are accelerating when they are doing nothing of the kind.

Plainly, a longer perspective is desirable. Let us go back 600 million years –

Lack of correlation implies lack of causation
 
Sub specie aeternitatis, the correlation between atmospheric CO2 concentration and global mean surface temperature is non-existent: above 915 ppmv the logarithmic relation between CO2 and temperature fails (Myrhe et al., 1998), and the addition of further CO2 has little further influence over global atmospheric temperatures.

Throughout the past 600 million years, the mode of temperature has been 12.5 Fahrenheit degrees warmer than the present, but atmospheric CO2 concentration peaked at 7000 parts per million by volume in the Cambrian era.

It was at this time that the calcite corals originated. There was also a very high CO2 concentration compared with today’s during the Triassic era, when the delicate aragonite corals were first created by algal symbiosis. Congressman Inslee suggested that corals were no longer adapted to high CO2 concentrations: he felt that acidification of the oceans would harm them.

However, measurements of ocean pH over time are few, and are not adequate to demonstrate any acidification of the oceans: the generally-quoted reduction of 0.l pH units is derived chiefly from modeling. Today’s CO2 concentration is almost the lowest in the geological record, endorsing Will Happer’s testimony before the Senate earlier this month that the planet is currently starved of CO2, and has been so starved for several million years.

The Vostok ice cores provide a detailed record for the past 650,000 years –
650,000 years’ methane, temperature, and CO2
 
Which came first? Methane concentration (red), temperature proxy (black) and CO2 concentration (blue) from the present (left) to 650,000 years before present (right).

Red flag 10: The Vostok ice cores show that in the past 650,000 years the correlation between greenhouse gases and temperature was close. Al Gore said in his movie that whenever CO2 changed, temperature changed.

However, it was temperature that changed first, and CO2 that followed 800-2800 years later. The latter change cannot have caused the former.

It is also worth considering the temperature record during the Holocene – the 10,000-year period following the last Ice Age –

It was warmer than the present for 10,000 years
 
“Global warming” in perspective: The recent 300-year period of “global warming”, nearly all of which cannot have been anthropogenic, is insignificant in comparison with the Holocene climate record. Throughout much of the past 10,000 years, including the Minoan, Roman (R), and Medieval (M) warm periods, global temperatures were up to 5 Fahrenheit degrees warmer than the present. Today’s temperatures are not unprecedented.

Red flag 11: Unfortunately, the IPCC has made a determined effort artificially to abolish the medieval warm period, apparently with the intention of making it appear, falsely, that today’s global mean surface temperatures are unprecedented in recent history.

Now you see it (IPCC, 1990) ...
 
Medieval warm period? Yes. This drawing of a graph in the IPCC’s 1990 report shows it clearly.

... now you don’t (IPCC, 2001)
 
Medieval warm period? Not any more: the UN purported to abolish it in its 2001 assessment report. The above graph appeared six times, in full color, and at large scale, in the 2001 report, the only graph to be so favored.

The IPCC notoriously abolished the medieval warm period in its 2001 report, having explicitly acknowledged its existence in its 1990 report. Its justification for the purported abolition was highly questionable. The prominence that it accorded to the 2001 graph – the only one in the entire report to be reproduced six times, in full color, and at large scale – suggests a political rather than a scientific motive.

Red flag 12: The UN’s report relied upon a paper in Nature that contained a number of abuses of sound statistical practice. The paper, (Mann et al., 1998-1999), relied heavily upon bristlecone-pine proxies for pre-instrumental temperature change, even though a previous UN report had explicitly recommended against the use of such proxies on the ground that the width of the tree-rings is influenced not only by temperature change but also by changes in precipitation, and most notably by changes in atmospheric CO2 concentration. Recent attempts by Mann et al. to revive the unsound graph regrettably suffer from the same central defect as the original: removing the bristlecone proxies and a further defective outlier (the Tiljander proxy) from among the proxy datasets clearly shows that the medieval warm period was real, and appreciably warmer than the present day.

The unsatisfactory statistical methods in Mann et al. were thoroughly exposed by McIntyre & McKitrick (2003, 2005). In all material respects, the findings of McIntyre & McKitrick were powerfully endorsed by a detailed investigative study by three statisticians at the instigation of the House (Wegman, 2005).

Red flag 13: It is of particular concern that the compilers of the now-discredited graph upon which the UN unwisely placed such undue weight in its 2001 report were extremely reluctant to release their computer programs and data. Nature failed to require them to produce the data; and it was only after numerous requests by McIntyre and McKitrick that Mann et al. eventually parted with the information necessary to allow a proper, independent, academic review of the graph that the UN had been so willing to accept without any real peer review.

Red flag 14: It is worth demonstrating one or two of the statistical abuses that led to the false abolition of the medieval warm period. One startling abuse was the disproportionate weight given to temperature proxies that provided Mann et al. with the “hockey-stick” profile they desired, in comparison with the lesser weight given to proxies that demonstrated the presence of the medieval warm period. An instance –

The upper panel was given 390 times the weighting of the lower panel
 
A false balance is abomination to the Lord: In the compilation of the UN graph purporting to abolish the medieval warm period, the upper data, showing the present day to be warmer than the previous 600 years, was given 390 times the weight of the lower data, showing the Middle Ages as warmer than the present.

Red flag 15: The computer model which was used to generate the defective UN graph was tuned to generate data curves showing the present day to be warmer than at any time over the past 600 years, regardless of whether the graph were based on genuine temperature proxies or on random red noise –

Genuine proxy data (top) and random red noise (bottom)
 
Whatever answer you want: The computer model that generated the UN’s graph that “abolished” the medieval warm period generates “hockey-sticks” that show today’s temperatures as warmer than for 600 years, with the post-1900 temperature increase serving as the blade of the hockey-stick. Remarkably, the model generates “hockey-sticks” even if, instead of the genuine temperature-proxy data (upper panel), random red noise (lower panel) is used.

Red flag 16: The EPA, in the Technical Support Document that it will pray in aid as justification for its endangerment finding in respect of CO2, will disregard the overwhelming majority of the papers in the scientific literature, and will also deny history by finding that there was no medieval warm period –

‘No medieval warm period’ (EPA, 2009, after NRC, 2006)
 
Notwithstanding official attempts either to eradicate the medieval warm period altogether or to show that it was not as warm as the present, in the past 25 years at least 670 scientists from 391 institutions in 40 countries have contributed to peer-reviewed papers in the learned literature establishing that the medieval warm period was real, global, and warmer than the present. Here are graphs from a few of these papers –
The medieval warm period graphically illustrated in the learned literature
           
Huang et al. (1998)    Dansgaard (1969); Schonweise (1995)    Tyson et al. (2000)
           
Kitagawa & Matsumoto (1995)    Noon et al. (2003)    Gupta et al. (2005)
           
Wilson et al. (1979)    Keigwin (1996)    Esper & Schweingruber (2004)
Medieval warm period? Yes, nine times: it is as well established in the scientific literature as it is in the historical record.

Red flag 17: It was only after the UN’s use of the defective graph had been challenged that a suspicious spate of papers supporting Mann et al. in their attempted abolition of the medieval warm period appeared in the scientific literature. However, the Wegman report showed that most of the authors of these papers had previously been co-authors with Mann himself. This incident illustrates a central difficulty. Many of the scientific journals have declared prejudices in favour of the “official” position on climate change: therefore, they are far more indulgent of authors who support the “official” position than of skeptics.

This declared bias among the journal editors allows supporters of the “official” position to knock down any skeptical paper by dashing off a quick rebuttal, which is eagerly printed after a minimum of scrutiny. Then the IPCC, which claims to operate by reviewing the literature, can concentrate on the rebuttals rather than the skeptical papers that question its position. At any rate, the IPCC, in its anxiety not to admit its mistake in attempting so prominently to abolish the medieval warm period in 2001, failed – and continues to fail – to take any account of the overwhelming majority of papers in the literature that demonstrate that the medieval warm period was real, global, and appreciably warmer than the present.
We conclude that today’s temperature is not exceptional. It was warmer than today in the medieval, Roman, and Minoan warm periods and throughout most of the Holocene; it was up to 7 Fº warmer than the present in each of the four previous interglacial warm periods; and 12.5 Fº warmer than the present throughout most of the past 600 million years. Yet Earth did not fry and the oceans did not acidify.
Is “global warming” happening faster than even the IPCC had thought?

In my testimony, I presented a graph indicating that the global atmospheric concentration of carbon dioxide was rising in a straight line at a rate well below the least of the IPCC’s exponential projections –

CO2 concentration, though rising, is well below the IPCC’s predictions
 
Observed and predicted CO2 concentration, 2000-2100: The pale-blue region, bounded by exponential curves, is the UN’s predicted path for CO2 concentration over the present century. The observed, deseasonalized CO2 concentration change calculated by NOAA from January 2000 to November 2008 (dark blue) is near-coincident with the least-squares linear-regression trend (solid, pale-blue line) on the data. CO2 concentration is no longer rising ever more rapidly, but only in a straight line, even though CO2 emissions are rising ever more rapidly.

Red flag 18: The derivation of the carbon dioxide graph is more fully explained in the technical paper annexed at Flag 1. Briefly, the data (the thick, dark-blue spline-curve) are taken from the NOAA’s deseasonalized global CO2 concentration anomaly dataset. The NOAA is the parent organization of the NCDC, of which Mr. Karl is the Director. The thick pale-blue line beneath the spline-curve is the least-squares linear-regression trend on the data. It is visible that the trend-line is almost coincident with the data, showing CO2 increasing not exponentially but merely linearly, well below the IPCC’s prediction.

The pale blue region, bounded by exponential curves that appear at the above resolution to be close to linear, is the IPCC’s projected range for CO2 concentration increase over the seven-year period, based on its “business-as-usual” scenario A2. Extrapolating the present trend to 2100 yields the following graph –