Thursday, April 20, 2006

Hegerl et al on climate sensitivity

A new paper is out in Nature this week: "Climate sensitivity constrained by temperature reconstructions over the past seven centuries" Hegerl et al, Nature 440. What the authors have done is to, erm, use temperature reconstructions over the past seven centuries to constrain climate sensitivity. Just as it says in the title. In particular, they combine the analysis of the recent instrumental record (published by Frame et al in GRL) with some proxy records for global temperature over the previous few centuries. Both of these lines of evidence are rather weak, but jointly, they give a significantly better result than either one can by itself. Their result is a 5-95% CI of 1.4-6.2C.

So what is so newsworthy and significant about this? Well, I have to admit, I'm really not sure. Clearly, it vindicates the rather obvious point we made in our recent GRL paper, that using more information will generate lower uncertainty. And they have certainly done a more careful and precise calculation than our rather back-of-the-envelope approach. But it is not clear why they stopped after considering the past 7 centuries of globally-averaged temperature, and their bounds are not particularly exciting ones (plenty of people have previously published similar results elsewhere, eg Knutti et al and Piani et al recently). As we've discussed, there are numerous other data sets that all provide additional information (and what's more, which all point towards a sensitivity of close to 3C as having the highest likelihood), and any analysis that only uses a small subset of the data will necessarily give an result with unrealistically high uncertainty. Therefore, it's important to realise that this paper does not provide any support for the belief that S>6.2C, even at the 5% level. But still, if it's a sign that Nature is willing to row back from some of the more alarmist nonsense they have published recently, then that is certainly a step in the right direction.

Now for the long-overdue extended rant about Nature. We originally submitted our GRL manuscript to Nature, where an editor sat on it for a couple of week, before rejecting it with the boilerplate excuse:
we are unable to conclude that the paper provides the sort of advance in significantly constraining climate sensitivity relative to prior estimates that would be likely to excite the immediate interest of researchers in a broad range of other disciplines
Of course, I know they are snowed under with manuscripts and have to reject a large proportion without review. But we were very disappointed that they couldn't even be bothered to give a plausible excuse for their decision. If our estimate is not "likely to excite the immmediate interest of researchers" then it is very hard to see how Hegerl et al's rather weaker and less general result could have passed this test. And of course the editor's rejection gives us absolutely no clue as to what might be required to pass their test in the future, since what she wrote doesn't have a scrap of credibility about it. Note that the Hegerl et al manuscript was already under review at that time.

Of course, Hegerl et al might also have grounds to feel disgruntled that we leapfrogged and gazumped their result, which was submitted prior to our work (not that I knew about it at the time). That's the way the cookie crumbles - and they certainly got the better half of the deal. Note that the time they spent in review was nothing to do with me, cos no-one ever sends me papers on this topic to referee. And nothing I've said here about Nature should be taken as being in any way critical of the authors themselves - in fact, Gabi Hegerl was very friendly and encouraging when she heard of our work. But it's hard to avoid the cynical conclusion that the editor's judgement was not made so much on what was said, but who said it...


Oh, now I see that the "Editor's summary" claims that this is the "best guide yet", and that this work shows that there is "a small probability that climate sensitivity will exceed 6.2 °C". Well, that's pretty good work to fit two simple mistakes into one short paragraph. I guess they've got to put a brave face on it.


PixieBoy said...

When Nature says,

"It was thought that the upper limit of climate sensitivity (global mean temperature change due to CO2 doubling) was between 7.7 °C and above 9 °C."

Presumably, this was only Nature who thought so in the first place?

So, let me get this straight.

If last week I thought that there were pixies living at the bottom of my garden, this week I can write a press release revealing to the World that,

"it is now unlikely that there are pixies at the bottom of my garden".

I'll leave it to a future publication in Science to reveal that I don't have a garden.

Brad Fallon said...

What is meant by climate sensitivity parameter and global warming potential?

James Annan said...

The "best answer" here looks reasonable.