Un-cherry-picking and the Singer Event

By Christopher Monckton of Brenchley

The dwindling number of paid or unpaid trolls commenting here – dwindling because the less dishonest and less lavishly-funded ones realize the game is up – do not like The Pause. They whine that in order to demonstrate a long period without global warming I have cherry-picked my start date.

No, I have calculated it. I have not, as they suggest, naïvely cherry-picked 1998 as my starting-point so as to take unfair advantage of the Great El Niño of that year. I have not picked 1998 at all.

Instead have determined by iterative calculation the earliest month in the record that shows no global warming at all as far as the present. On the RSS dataset, which I shall use for the analysis to follow, that month is September 1996, giving 17 years 6 months without any global warming.

This graph has become famous. Patrick Moore, one of the founders of Greenpeace (which is now, with frantic mendacity, trying to disown him), displayed it recently on Fox News, followed by weeping and gnashing of Medicaid dentures right across the leftosphere. Marc Morano has the graph as a lead indicator at the inestimable ClimateDepot.com. It is popping up all over the blogs.

The true-believers wring their tentacles and moan about cherry-picking. So let us deal with that allegation, and discover something fascinating on the way.

To put an end to the allegation that we have cherry-picked our start-date, let us start at the beginning of the satellite record. So here you go, from January 1979.

Oo-er! Instead of a zero trend, we now have a terrifying increase in global temperature, at a rate equivalent to a shockingly sizzling – er – 1.24 Cº per century. That is below the 1.7 Cº/century near-term warming predicted in IPeCaC’s 2013 Fifth Assessment Report. Well below the 2 Cº/century predicted in the 2007 Fourth Assessment Report. Scarcely more than a third of the 3.5 Cº/century mid-range estimate in the 1990 First Assessment Report. A quarter of the 5 Cº/century predicted by the over-excitable James Hansen in front of Congress in 1988.

Evidently, the predictions are becoming less and less extreme as time passes. But they are still well over the top compared with reality. Let us illustrate IPeCaC’s latest version of the “consensus” prediction, comparing it with the RSS real-world observation:

Whichever dataset one chooses, whichever starting point one picks, the rate of global warming predicted by the models has been and remains grossly, flagrantly, egregiously in excess of what is actually happening in the real world. It is this central truth that every attempt to explain away the Pause fails to get to grips with.

One of the slyest ways to pretend the Pause does not exist is offered by “Tamino”, one of the Druids’ white-robed, snaggle-bearded, bushy-eyebrowed archpriests.

“Tamino”, taking time off from his sun-worshiping duties at Stonehenge, invites us to suppose that in 1997 we had predicted that the previously-observed warming rate would either continue as is or turn horizontal. Where would the subsequent annual data points lie?

Tamino’s excited conclusion is that “fourteen of sixteen years were hotter than expected even according to the still-warming prediction, and all sixteen were above the no-warming prediction (although one is just barely so)”.

Phew! Global warming has not Paused after all!! The Earth still has a fever!!! Global temperature is continuing to rise – and at a faster rate than before!!!! What a relief!!!!! We said we were more certain about future global warming than about anything in the whole wide world before, evaah – and we were right!!!!!!

Up to a point, Lord Copper. The truth, if one goes looking for it rather than bashing and mashing the data till they fit the desired outcome, is much more interesting.

First, it is not appropriate to use annual data points when monthly data are available. Using the monthly data multiplies 12-fold the degrees of freedom in the analysis, and makes the picture clearer.

Secondly – and this cannot be said too often – trend lines on observed data, particularly where the data are known to be stochastic and the behavior of the underlying object chaotic, are not, repeat not, repeat not a prediction.

Let us play Tamino’s game of breaking the RSS dataset into pieces. But let us break it into three pieces, not two. Let us look at three periods: January 1979 to January 1993, January 1993 to January 1999, and January 1999 to February 2014.

It was Fred Singer who first pointed out this most startling characteristic of the post-1979 temperature record to me. We were sitting in front of the big computer screen and playing with the temperature datasets using my graphing engine in the library at Rannoch, overlooking the loch and the misty mountains beyond.

An otter was flip-flopping past the window, and the ospreys (a ménage a trois that year, with two adult males, a female and two chicks in the nest) were swooping down to catch the occasional trout.

Fred told me there had been very little trend in global temperature until the Great el Niño of 1998, and very little trend thereafter. But, he said, there had been a remarkable step-change in the data in the short period culminating in the Great el Niño.

The trend till January 1993 and the trend from January 1999 are indeed near-identical at just over a quarter of a Celsius degree per century. Not a lot to worry about there. But the trend over the six years January 1993 to January 1999 was a lulu. It was equivalent to a spectacular 9.4 Cº per century.

So, what caused that sudden upward lurch in global temperature? Since absence of correlation necessarily implies absence of causation, we know it was not CO2: for CO2 concentration has been rising monotonically, with no sudden leaps and bounds.

Indeed, no phenomenon in the atmosphere could really have caused this lurch. True, there had been a naturally-occurring reduction in cloud cover over the 18 years 1983-2001, and that had caused a forcing of 2.9 Watts per square meter (Pinker et al., 2005), greater by 25% than the entire 2.3 Watts per square meter of manmade forcing from all sources in the 263 years since 1750.

However, as Dr Pinker’s graph shows, the trend in cloud cover forcing was relatively steady, and there was actually a drop in the cloud-cover forcing in 1993-5.

Since solar activity did not change enough over the period to cause the sudden hike in global temperature, there does not appear to have been any external reason why, for six years, temperature should suddenly have risen at a mean rate equivalent to almost 1 Cº per decade.

Nor was there any more land-based volcanic activity than usual: and, if there had been, it would if anything have caused a temporary cooling.

When the impossible has been eliminated, whatever remains, however improbable, must be the truth. The culprit is plainly not atmospheric, not extra-terrestrial, not land-based, but oceanic.

To kick atmospheric temperature permanently upward by almost 0.3 Cº, and to boot the centennial trend northward by almost a full Celsius degree, something drastic must have happened beneath the waves.

Perhaps the still poorly-monitored pattern of overturning that takes warmer water from the mixed upper stratum of the ocean to the much colder benthic strata temporarily slowed.

Perhaps we shall never know. Our ability to monitor changes in ocean heat content, negligible today, was non-existent then, for the ARGO buoys were not yet on duty: and even today, as Willis Eschenbach has pointed out, the coverage is so sparse that it is the equivalent of taking a single temperature and salinity profile of the whole of Lake Superior less than once a year.

Perhaps there was massive subsea volcanic activity in the equatorial eastern Pacific, beginning in 1993/4 and peaking in 1998. Again, we shall never know, for we do not monitor the 3.5 million subsea volcanoes on Earth (the largest of which, occupying a footprint greater than any other in the entire solar system, was only discovered last year).

But the curious thing about this sudden and remarkable warming over just six years – let us call it the Singer Event – is that so very little consideration is given to it in the ramblings of IPeCaC.

The Singer Event accounts for fully four-fifths of the global warming trend across the entire satellite era. Without it, no one would still be wailing about global warming.

But will you find the Singer Event mentioned or discussed in any of IPeCaC’s Summaries for Policymakers from 2001 onward? Er, no. It doesn’t fit the story-line.

Bluntly, unless and until the cause of the Singer Event is nailed down there is nothing whatsoever in the global temperature record during the satellite era to suggest that CO2 is having any detectable effect on global temperature at all.

Finally, just in case any of the trolls want to whinge about why I have confined the analysis to RSS, I prefer RSS because, alone of the five datasets, it correctly represents the 1998 Great el Niño as being significantly bigger than any other el Niño in the instrumental record.

There have been two previous Great el Niños in the past 300 years. Each of those, like the 1998 event and unlike any other in the global instrumental record, caused corals to bleach worldwide on a large scale. Bleaching is a natural defense mechanism against sudden ocean warming, and the corals have recovered from it just fine, as they have evolved to do.

But the corals are a testament to the fact that the Singer Event is something atypical. And it is the RSS dataset that best reflects how exceptional the Singer Event was.

But here, just to please the trolls, is the entire global temperature record since January 1979, taken as the mean of all five global temperature datasets (it is pardonable to take the mean, because although the coverages and measurement methods vary the discrepancies are small enough that they tend to cancel each other over a long enough period).

The bottom line is that the warming trend since 1979 on the RSS dataset was 0.44 Cº, while on the mean of all the datasets it was not vastly greater at 0.51 Cº. So, when the trolls argue that I am cherry-picking, they are arguing pettily about hundredths of a degree.

The Singer Event is highly visible in the five-datasets graph, and still more so in the earlier RSS graph.

I should be most interested in readers’ comments on the Singer Event, because I have been invited to contribute a paper to the reviewed literature about it. All help, even from the trolls, would be very much appreciated.