Has Antarctica been getting warmer? To the frustration of many environmentalists, it’s not an easy question. Manned weather stations have existed there for over 50 years, unmanned stations from 1980 onwards. Coverage is patchy both in space and time, with weather stations clustered in a few spots and records full of gaps from when sensors got buried in snow, breakdowns occurred or stations were closed. From 1982, satellites have measured the temperature over the continent but inaccurately, because of clouds and instrument inconsistencies.
So taking Antarctica’s temperature involves considerable mathematical skill and leaves much room for statistical disagreement. The disagreement has just became a lot less polite, as several scientists exchanged the blogosphere equivalents of declarations of war. What started out as a difference over methods and results has descended into character assassination and accusations of duplicity. Underlying all this, one side claims that Antarctica — a continent which holds 90 per cent of the world’s ice — is warming significantly. The other side claims this is not so. The battle has implications far beyond Antarctica. It has exposed a real problem of bias in scientific journals.
The story starts in 2009 when the prestigious journal Nature carried a cover story with an image of Antarctica glowing a warm pink. The paper inside used a relatively novel statistical reconstruction method to combine the detailed satellite data with the more accurate, but sparse, weather station data. It concluded that most of Antarctica — and particularly West Antarctica — had warmed significantly since 1957. This was worthy of a cover story because it contradicted the previous understanding: that only the most northerly part of this ice continent, the slender Antarctic peninsula, was warming.
The paper’s senior author was Eric Steig of the University of Washington. His work duly captured world headlines. Four days later, the editor of the famously pugnacious RealClimate blog, Gavin Schmidt, boasted that the sceptics were surprisingly silent. Mission, it seemed, accomplished: Antarctica was no longer an embarrassment to the global warming narrative.
He spoke too soon. The indefatigable Canadian climate analyst Steve McIntyre had already started to post analyses of the Nature paper at his blog, Climate Audit. He was joined by the engineer Jeff Condon, whose blog the Air Vent had, along with Climate Audit, played a crucial role in the release of the Climategate emails in 2009. Within a few weeks, McIntyre and Condon, together with engineer Ryan O’Donnell, statistician Roman Mureika, economist Huston McCulloch and others, deconstructed Steig’s results.
McCulloch soon discovered that an important factor — serial correlation — had been overlooked, which had the effect of making the warming trend look more statistically significant than it was. After this appeared at Climate Audit, Steig sent a correction to Nature, claiming he had found the error independently.
Mureika found that Steig’s reconstructions were, mathematically, derived from only three underlying temperature sequences — too few properly to represent an entire continent. McIntyre then showed that these sequences, claimed by Steig to be meaningfully related to important physical climate processes, resembled instead off-the-shelf mathematical patterns known as Chladni vibration patterns, and could be expected to arise irrespective of the claimed causes.
The blog workers were by now convinced that Steig’s significant warming trend was an artefact of a mathematical method employed inappropriately. It smeared a trend from an area containing many recording stations over an area with few stations, specifically from the strongly warming Antarctic peninsula to the much larger West Antarctica — a region containing only two short-record manned stations.
At this point O’Donnell began to go through the paper’s maths in detail with the help of one of us (Nicholas Lewis, who has a background in maths), Condon and McIntyre, and worked out how to correct Steig’s method. Together with these three, O’Donnell drafted a paper, which argued that Steig had implemented his reconstruction method in an inappropriate way (with misleading results), and proposed a corrected method that fitted the data better. The new method also indicated much lower continental temperature trends.
Pause here to note that Nature’s original peer-review process had let through an obviously flawed paper, and no professional climate scientist then disputed it, perhaps because of fear that doing so might harm their careers. As the title of Richard Bean’s new play — The Heretic — at the Royal Court hints, young scientists going into climate studies these days are a bit like young theologians in Elizabethan England: they quickly learn that funding and promotion dries up if you express heterodox views, or doubt the scripture. (The scripture, in this case, being the reports of the Intergovernmental Panel on Climate Change.) That is why in the Antarctic case, it took a group of amateurs, independent from conformist institutions and funding, to spot and correct the flaws. None of them deny that man is warming the globe to some extent; but they have grown increasingly frustrated by the distortion, exaggeration and sloppy methods being used to support a preconceived conclusion.
O’Donnell’s team submitted its paper to the Journal of Climate, which sent it to three reviewers. They duly came back with various criticisms and comments, the normal give and take of peer review. Two reviewers had minor points and recommended publication subject to these being resolved. But one, known as Reviewer A — all reviewers are anonymous — demanded major changes before publication. O’Donnell responded and made minor changes, but disputed Reviewer A’s major points and questioned his objectivity.
The journal editor then appointed a fourth reviewer of the revised draft, who recommended publication with minor changes. But Reviewer A demanded further major changes — insisting the ‘most likely’ West Antarctic trends be shown in the main results and suggesting that these could use a mathematical procedure called ‘iridge’ to fill in missing data. (The algorithm employed by both Steig and O’Donnell offers a choice of ‘iridge’ or ‘TTLS’ methods to fill in missing data.) Although the vote was now 3-1 against Reviewer A, the journal editor accepted his demand. Now, O’Donnell’s revised draft had not used iridge for its main results, but mentioned a West Antarctica trend using iridge as a way of checking, indicating that the ‘most likely’ regional trend was in line with the (warmer) iridge results. The O’Donnell team therefore concluded that in order to forestall further criticisms from Reviewer A and achieve publication, switching method for all their main results was the only sensible option.
The third submission appears to have been reviewed only by Reviewer A. Through the use of iridge, it sidestepped his main previous criticisms, and showed a higher temperature trend in West Antarctica. But after saying that ‘The use of the “iridge” procedure makes sense to me, and I suspect it really does give the best results’, he went on to criticise one aspect of the method and demand further changes, including another round of review. By this time, even the journal editor appears to have become exasperated with Reviewer A and asked O’Donnell’s team to respond to the review to his satisfaction, which they did. The paper duly appeared in December 2010.
By the end, Reviewer A’s criticisms — and the responses to it — had generated 88 pages of corre
spondence. Climate change sceptics are often accused of sniping from the sidelines but failing to submit alternative conclusions to the peer-reviewed literature. The obvious response is that the burden of proof should be on those wishing to prove their case, not on the challengers for failing to prove it wrong. Sceptics tend not to have the taxpayer-funded time to do studies. They also suspect that the scientific publishing process is biased against them. The war of attrition that O’Donnell’s paper had been through reinforced this impression.
Hard evidence of such obstruction first came to light in the now infamous ‘Climategate’ emails from the University of East Anglia, which included this gem from climatologist Phil Jones: `The other paper by [Ross] M[cKitrick] and [Pat] M[ichaels] is just garbage… I can’t see either of these papers being in the next IPCC report. Kevin [Trenberth] and I will keep them out somehow — even if we have to redefine what the peer-review literature is!’ The paper in question pointed out that regional temperature rises correlated with economic development, implying they were contaminated by local urban warming unrelated to greenhouse gases.
Of course, O’Donnell and his co-authors had long suspected that Reviewer A was associated with Steig’s team. When O’Donnell put this to Steig in early December, Steig admitted that it was him. For an up-and-coming scientist like Steig to get a major paper published in a journal like Nature was a career-enhancing moment. He therefore had every interest in opposing a paper that refuted its main conclusions — which hardly made him an impartial choice of reviewer.
At Steig’s request, O’Donnell kept the secret and did not reveal his identity as Reviewer A. But he changed his mind following a spat which broke out earlier this month. On 1 February, Steig published an essay disparaging the O’Donnell paper at the RealClimate blog. What really stuck in O’Donnell’s craw was criticism for using the ‘iridge’ method. Steig was now criticising the methodology which he himself had put forward a few months before. The blog battle — still ongoing — stemmed from this.
So has Antarctica been warming? Mostly not — at least measurably. The peninsula (2 per cent of the continent) shows substantial warming. The rest is patchy: some parts are warming slightly, others cooling slightly. Over the continent as a whole, since 1957, O’Donnell et al. found no statistically significant warming trend. The retreat of floating Antarctic ice shelves is a favourite story for the media. But, except in a very few peripheral parts, Antarctica is far too cold to lose ice by surface melting — even if gets much warmer. Primarily, ice loss is believed instead to be caused by currents of deep ocean water, itself warmed naturally a long time ago. A new study concludes that the melting process has been going on for too long to be blamed on industrial increases in greenhouse gases. (This study, of course, attracted far less coverage than alarming first reports.) Antarctic sea ice, unlike Arctic, has shown no signs of retreating: it recorded a maximum in 2007.
All this fits a disturbing new trend in the scientific literature on climate. Papers that come to lukewarm or sceptical conclusions are published, if at all, only after the insertion of catechistic sentences to assert their adherence to orthodoxy. Last year, a paper in Nature Geosciences concluded heretically that ‘it is at present impossible to accurately determine climate sensitivity to carbon dioxide’ (high sensitivity underpins the entire IPCC argument), yet presaged this with the absurd remark: ‘Earth’s climate can only be stabilised by bringing carbon dioxide emissions under control in the 21st century.’
Likewise, a paper in Science this month linking periods of migration in European history with cooler weather stated: ‘Such historical data may provide a basis for counteracting the recent political and fiscal reluctance to mitigate projected climate change.’ Sceptical climatologist Pat Michaels pointed out that the sentence would make more sense with ‘counteracting’ removed.
Science as a philosophy is a powerful but fragile thing. Science as a philosophy is a powerful but fragile thing; in the case of climate it is now in conflict with science as an institution.
For details of The Spectator’s forthcoming debate on climate change, see p. 38 or go to new.spectator.co.uk/events.