Myles Allen's model and a controversial press release

This post is mainly for people who have arrived here from a link at Steve McIntyre’s ClimateAudit blog. If you haven’t come from there, you will need to look here and then here to understand what it’s about.

Transcript from ‘Overselling Climate Change?’, A BBC Radio4 programme presented by Simon Cox and broadcast in April 2006.

Continuity announcer: [quoting newspaper headlines]

‘Global warming is twice as bad as previously thought.’

‘Screen saver weather trial predicts 10 degrees Celsius rise in British temperatures.’

‘Global warming may be twice as bad as feared.’

Simon Cox: These are the broad sheet headlines from the first results of the world’s largest climate experiment. The first two headlines are simply wrong. The last one is just misleading. But how did all of these reputable newspapers get it so wrong. Our story starts with climateprediction.net, a group of British scientist who try and predict the effect of global warming using computer models running on thousands of ordinary home computers across the World. In 2005 they published their first results from the project in the well-known scientific journal Nature. The team had been testing what effect doubling the amount of carbon dioxide, or _CO2 , in the atmosphere would have on temperature; or as it’s known in the trade, climate sensitivity. They ran thousands and thousands of models, each with slightly different parameters, to try and get the widest range of responses. They were left with just over two thousand results. The vast majority of these, well over a thousand, showed that doubling CO₂ would lead to a temperature rise of around 3 degrees Celsius. This is in line with the findings of many other climate researchers. A tiny percentage of the models showed a lot of warming, the highest result was a startling 11 degrees. When it came to publicising their research the scientists chose to focus on this 11 degrees figure. In a press release headlined ‘Bleak Results from the World’s Largest Climate Change Experiment’ the only number mentioned is 11 degrees. There was no reference to the fact that most of the results were around 3 degrees.

Was this giving a good picture of the research? Scientific papers are shown to others for comment, so that they are accurate and not exaggerated, a process known as anonymous peer review. We decided to conduct our own review of both the press release and the paper published in Nature. We contacted several respected climate scientists and a statistician, and asked them to read both the Nature paper and the press release. All were critical. One of them wrote to us, ‘I agree the 11 degrees centigrade figure was unreasonably hyped. It’s a difficult line for all scientists to tread, as we need something exciting to have any chance publishing in places such as Nature, and to justify our funding. I do think that in this case they clearly overstepped the line in their presentation of what they had shown’.

I put this criticism to Dr. Myles Allen, the principle investigator at climateprediction.net.

Myles Allen: The press release was absolutely fine. The press release identified what was interesting about the paper. And the other thing which our, our, press advisers tell us to do is to make sure that a press release could be used by the sort of hard-working journalists in the Oxford Times who don’t have time to go and read the whole story, that they can essentially go and copy it out. And in that respect the press release was accurate as well. It said up to 11 degrees, and that was precisely the result that we got. There were problems with some journalists who decided to embroider on the press release, without actually going back to the paper. I have no sympathy for them really. If journalists decide to embroider on a press release without refering to the paper which the press release is about then that’s really the journalist’s problem. We can’t, as scientists, guard against that.

Simon Cox: I just wondered, I mean looking at this second parpagraph, where you say that the first results from climateprediction.net, a global experiment using computing time donated by the general public show that average temperatures could eventually rise by up to 11 degrees Celsius, even if carbon dioxide levels in the atmosphere are limited to twice those found before the Industrial Revolution. [That’s] doubling of CO₂ . Such levels are expected to be reached around the middle of this century unless deep cuts are made in greenhouse gas emissions. I mean, if you’re reading that, you know, you would think, you know, there’s no sense of context that, OK, it could be 11 degrees Celsius, but it’s much more likely to be around 3 degrees.

Myles Allen: I think that anyone reading the paragraph would understand what the phrase ‘could’ and ‘up to’ would mean. So I think that there is nothing inaccurate in that paragraph.

Simon Cox: You Don’t think that it was worth, in that, what, five hundred words, an A4 side, mentioning the three degree figure at all, just to give the context.

Myles Allen: Had we done so we could have been accused, conversely, of being misleading, because that would have implied that was a new scientific result we got out of this study. In fact the only reason we got the majority of models coming out with near 3 degrees was, as far as we could tell at the time, because the original model had a sensitivity of 3 degrees, so to have suggested in this press release that our results somehow showed new information to imply that the most likely climate sensitivity was around three degrees would have been very misleading. That would have been more misleading than drawing attention to the 11 degree figure.

Simon Cox: What’s your best guess on climate sensitivity?

Myles Allen: What’s my best guess? Well it depends what you mean by that. Do I think that there’s a 50 percent chance of a sensitivity greater than three or so? I’d say yeh! that’s probably about right – OK. So, if you had to give me 50/50 odds, on where the climate sensitivity was, I’d say it’d be somewhere, three to three and a half, maybe.

[end of interview]

Simon Cox: None of the reports in the press, or on the BBC, mention the fact that most of the models predicted a temperature rise of around 3 degrees, still in itself a significant change. All focused on the most extreme scenario.

Steve McIntyre discussed this matter in greater detail at ClimateAudit in May 2007 here and his post contains links too some very interesting background information.

A very lively debate about this programme between the producer, Richard Vadon and Myles Allen is raging at RealClimate: here.

Does anyone know when Myles Allen was appointed to his exalted post of Review Editor at the IPCC?

The sound file is here: rtsp://rmv8.bbc.net.uk/radio4/science/thu2000_20060420.ra?start=”0:22.5″ about 2:55 mins in, but the whole programme is worth listening to, particularly contributions from Hans von Storch. The BBC just doesn’t make them like this any more.

You can still find ClimatePrediction here: www.climateprediction.net