How data revisionism hypes global warming

By Christopher Monckton of Brenchley

I have now had the opportunity to study SteveF’s remarkable essay at Lucia’s Blackboard, to which Anthony kindly draws attention in his footnote to my earlier posting on the absence of statistically-significant global warming for 17 years 4 months.

SteveF’s conclusion is that once allowance has been made for three naturally-occurring influences – volcanic aerosols, the ~11-year solar cycle and the el Niño/la Niña cycles – the HadCRUt4 warming rate from 1979-1996 was six times faster than from 1997-2012. In the abstract, to allow for uncertainties, he cautiously reduces this to three times faster.

Even if one were to take the unadjusted HadCRUt4 data, the rate of warming from 1979-1996 was more than twice as fast as the rate from 1997-2012.

I decided to look not only at HadCRUt4, as SteveF did, but also at the two satellite datasets, RSS and UAH. RSS showed warming at 0.7 Cº/century from 1979-1996 and cooling at almost 0.1 Cº/century from 1997-2012.

UAH, however, in contrast to both HadCRUt4 and RSS, showed warming in the later period, 1997-2012, that was thrice as fast as the warming of the earlier period, 1979-1996.

SteveF’s essay takes no account of the most substantial medium-term natural cycle that seems to influence global temperatures: the Pacific Decadal Oscillation. The cycles of that great Oscillation tend to exercise a warming influence for about 30 years followed by a cooling influence for about 30 years. This cyclical influence is visible throughout the HadCRUt4 global temperature record since 1850.

There was a remarkably sharp transition from the “cooling” to the “warming” phase of the PDO at the beginning of 1976 and a transition back to “cooling” late in 2001.

The HadCRUt4 warming rate from 1976-2001 was equivalent to almost 1.8 Cº/century (compared with warming at just 1.1 Cº/century from 1979-1996), but from 2002 to the present HadCRUt4 shows cooling at a rate equivalent to almost 0.5 Cº/century (compared with warming at almost 0.5 Cº/century from 1997-2012).

Much of the fall in the warming rate identified by SteveF, therefore, appears to be attributable to the PDO. It would be interesting to adjust the global instrumental temperature anomaly record not only for volcanic aerosols, solar cycles and el Niños but also for the cycles of the PDO, but that is above my present pay-grade.

What is far from clear is the influence, if any, from CO2. Its influence must be very small, for it seems easily overwhelmed by natural influences such as the PDO and the three phenomena studied by SteveF.

During the three “warming” phases of the PDO that are visible in the HadCRUt4 instrumental record since 1850, the warming rates were as follows: 1860-1880 less than 1.0 Cº/century; 1910-1940 1.4 Cº/century; and 1976-2001 1.8 Cº/century.

Superficially, there appears to be an inexorable and strikingly near-linear increase in the warming rates during successive “warming” phases of the PDO. Might this increase be attributable to the monotonic increase in CO2 over recent decades?

If the increase in warming rates were to continue, perhaps as a result of the growing warming influence from CO2, the warming from about 2040-2070 might be equivalent to 2.2 Cº/century; and from 2100-2130 2.6 Cº/century.

It would not be until around 2160-2190 that the warming rate would reach the IPCC’s currently-projected central estimate of 3.0 Cº/century. And, even then, the mean centennial rate after allowing for the “cooling” phases of the PDO would be considerably less.

However, the apparently tidy 1.0 to 1.4 to 1.8 Cº/century-equivalent increase in the rates of global warming during the “warming” phases of the PDO may not be attributable to CO2 at all. The true cause may be another and more sinister man-made phenomenon: Orwellian data revisionism.

Late in 2009, after the first Climategate emails had been sprung on a naively unsuspecting world, Roger Harrabin of the BBC, an acquiescent true-believer in the global-warming Party Line, was told by his superiors that for the sake of what little is left of the BBC’s reputation he should – just for once – ask Professor Jones of the University of East Anglia some critical questions about the temperature record.

Harrabin had never before stopped to think about whether the Party Line was true. That is the trouble with the Party Line: as Orwell points out in 1984, it is intended as a substitute for independent thought – or for any thought.

So he did not know what questions to ask. He asked me for help in framing suitable critical questions.

I told him to ask Jones whether there had been any statistically-significant global warming over the previous 15 years. He thought that was an absurd question. The Party Line said warming was occurring at a rate unprecedented in human history.

I told him to ask the question anyway. To his astonishment, Jones – albeit testily – admitted there had been no warming statistically distinguishable from zero for 15 years.

I also told Harrabin to ask Jones whether the rates of warming during the three “warming” phases of the PDO in the instrumental record since 1850 were statistically distinguishable from one another.

Harrabin got a further surprise when Jones told him that the three rates could not be distinguished from one another, statistically speaking. On the then HadCRUt3 version of the global dataset, the rates of warming were equivalent to 1.0, 1.6 and 1.7 Cº/century respectively. The uncertainties in the data during the first of the three periods, 1860-1880, were so large that the rate could not be distinguished from that of the later two periods.

Our CO2 emissions could not have influenced the second period of PDO-driven warming, but we could in theory have influenced the third. Yet in the HadCRUt3 dataset the two periods showed warming within 0.1 Cº/century of one another: far too little an increase to be statistically significant.

At a climate conference in Cambridge a few years ago, I asked Jones whether, given that the global warming rates in the three “warming” phases of the PDO could not be distinguished from one another statistically speaking, any anthropogenic influence was yet discernible in the temperature record. He said there was a discernible influence, but did not say where or how large it was.

Not long afterwards, and perhaps not coincidentally, he produced HadCRUt4. Suddenly the rates of warming during the second and third PDO “warming” phases were changed from 1.6 and 1.7 Cº/century respectively to 1.4 and 1.8 Cº/century respectively.

As with other such instances of data revisionism in the terrestrial datasets, the later period was changed very little because the satellites were watching and prevented cheating. But the record in the earlier period was pushed downwards, artificially steepening the apparent warming over the 20^th century. It is as though we knew better than those who took the earlier measurements what measurements they ought to have recorded, all over the world.

Disentangling the true contribution of CO2 to warming from not only the numerous natural influences but also from the effects of data revisionism is near impossible. We shall have to wait and see. The one fact that is already clear, however, is that the warming rate predicted by the models on whose output the climate scare is founded is proving to be a hefty exaggeration.