Climate sensitivity – Could we be lucky?

One of potentially the most important climate change scientific papers for a long time has just been published in Science Express, the top science journal’s rapid-publication service. Unlike most, this does not deepen the general global warming gloom by suggesting things are ‘worse than we thought’ – instead it suggests that very high climate sensitivities (the kind that make it already ‘too late’, or turn us into Venus) are vanishingly unlikely. And more, that the most likely climate sensitivity could be slightly less than previous studies concluded.

This is good news because there is no sign of carbon emissions being curtailed any time soon, and indeed the current global picture is that we are above the worst-case A1F1 emissions trajectory traced by the IPCC in its original ’emissions scenarios’ projections. Even the IPCC, in all its wisdom, couldn’t imagine we would be as hopeless at cutting CO2 as has so far proven to be the case. Just as well then that the planet could be somewhat less sensitive to the warming effect of carbon dioxide and other greenhouse gases than has long been thought.

For those that have ‘sub req’d’ access to Science, the paper and its supporting online material are worth perusing in detail. For those that don’t, read on – I have no doubt that many will use these latest results to support their pre-existing political agendas, whatever these might happen to be. Ignore that stuff – because the science itself is what is interesting, and what should inform any policy implications that arise from this work.

So what is new? Well, what authors Andreas Schmittner et al have done is to use improved reconstructions of the planet’s ice-age temperature to drive a set of model runs using different climate sensitivities to see which can best reproduce what we think we know about the conditions of the ‘last glacial maximum’ around 20,000 years ago. This was a time when the Earth was in relative energy equilibrium (unlike now, when it has a lot of catching up to do with increased GHGs and other forcings) and when CO2 was about 100ppm lower than pre-industrial (about 185ppm).

And the conclusions? That the most climate-sensitive models, those which project a warming of over 6 degrees for doubled CO2, cannot reproduce the conditions of the last ice age – in fact, because they are so sensitive to climate change feedbacks, they experience runaway cooling and produce a near-completely ice-covered ‘snowball Earth’. This suggests that the models with the highest climate sensitivities (remember the ‘fat tail’ that could not be ruled out?) are not representing the physical properties of the climate system accurately and should, in the words of Schmittner et al, “be assigned near-zero probability”.

But what of the most-likely climate sensitivity, commonly put at or around 3C (plus or minus 1.5 degrees) for doubled CO2? The paper estimates a slightly lower median of 2.3C, but also usefully reduces the uncertainty on either side, giving a 66% probability of a 1.7-2.6C climate sensitivity. Since it seems rather likely at the moment that we will see a doubling of CO2 – and in the not too distant future either – this is somewhat reassuring.

There are caveats, as always: this is just one paper, based on one model, and previous work has come to quite different conclusions (but using less accurate data on LGM temperatures), as an accompanying commentary piece by Gabriele Hegerl and Tom Russon points out. Also, the actual ‘data’ on the last ice age temperatures is itself dependent on various proxies and therefore somewhat uncertain. As always, it would be sensible to wait for follow-up studies which may confirm or undermine these conclusions before reading too much into them.

Having said all that, this is still a very important piece of work because it suggests already-unlikely worst-case scenarios to be even less likely – and that we therefore should not be blinded by pessimistic paralysis about the situation we are in, especially with the ongoing rise in global emissions. That our situation may be slightly less hopeless than we thought yesterday should not lead to any complacency, but it should lead at least to renewed optimism and hope as the Durban climate talks get underway.


  1. Barry Woods

    And the climate policy implications are …….

  2. Robin Curtis

    need to be careful – it is only suggesting that climate sensitivity may lie towards the lower end of the IPCC range. It is still bad news!

    Lots of folk are going to take this as a reason to ease up on the pathetically small CO2 reduction that is taking place so far.

    Comment from a(nother) seasoned Climate Scientist: “Yes, interesting. Goes against lots of other assessments from Holocene and from earlier geological times (eg Pliocene)….will see what other scientists make of it.”

  3. Dan

    I can’t see a way in which this (too) will not be used by those who want to stick with fossil fuels for as long as they last. (Which is too long.) And particularly this coming straight after “Climategate 2” and (by the look of things) right before GFC 2. The world has “inaction” written all over it.

    It’s a good thing that models are getting tighter, but otherwise I don’t think much will change in the immediate future, and I certainly don’t expect anything apart from tokenism from Durban. Can’t really share your optimism.

  4. Oliver

    I think its worth taking on board James Annan’s posts on this.
    especially the second. The estimate for land only and for ocean only are markedly different, and the way the method works systematically down plays the larger results for the land dataset. It’s more evidence for sensitivity not being over four. Reading James it seems to me not very good evidence for sensitivity being less than three.


    Mark Lynas : “this is just one paper, based on one model, and previous work has come to quite different conclusions”

    That’s the point to remember. There are dozens of climate sensitivity studies, with different methods and different periods. So, one result is to be taken with caution, as each other of course (that’s true for high climate sensitivity). Here, from a larger proxy data, Schmittner et al 2011 find set a lower difference in temperature between Last Glacial Maximum and Holocene. So logically, it implies a lower sensitivity. But the quality of proxies (for LGM or past millenias) is subject to strong debates in the paleoclimate community. No hope for rapid change and robust conclusions in this domain.

    By the way, Schmittner et al results are coherent with 3 out of 17 models of IPCC 2007 AR4, whose equilibrium sensitivity for 2xCO2 ranges between 2,1 K and 2,3K. So, some climate models do reach a moderate sensitivity, when they project from present climate (rather than past climates) but it is not possible (at least in AR4, we’ll see in AR5) to classify all models for their realism (many others find a high sensitivity, over 4 K). We are obliged to deal with uncertainty in our energy and economic choices, and to to keep on the 2-4,5K sensitivity range for the moment. It would be a mistake to pick the 2,3 K of Schmittner 2011 as the most robust or more realist result.

  6. NewYorkJ

    Another way of looking at it: Schmittner’s study uses data suggesting only about a 3 C global mean temperature change from the LGM, rather than 5 C is other studies indicate. So predictably they get lower climate sensitivity. But if it only took 3 C of change to result in such dramatic climate change of the last glaciation, that’s not something to feel good about. Annan, however, indicates that Schmittner’s best model doesn’t fit the data as well as others, and suggests Schmittner’s climate sensitivity estimates would be bumped up a little as a result – closer to that 3 C. I’d be careful about calling their reconstructions “improved”.

    Also see a Q&A with one of the authors. Urban seems to be more cautious about the study than many commentators.

    1. Mark Lynas (Post author)

      Thanks – the Nathan Urban interview in the link above is particularly good, and I would strongly urge others to read it. I think the ‘fat tail’ issue here is crucial – whether or not this paper effectively eliminates (in a provisional sense; it must of course be checked by future work) the very high climate sensitivities some model runs come up with.

    2. John Mason

      Good point indeed – if Schmittner et al are correct (and there are some problems with their work) a small temperature rise led to 120m of sea-level rise!

      That ain’t so good!


  7. NewYorkJ
  8. NewYorkJ

    Over at RC, one of the study co-authors (Urban) mentions that he has a study in revision that estimates a 95% range for climate sensitivity at 1.8 C to 4.9 C, with best estimate of 2.8 C, very much in line with IPCC estimates (although note the lower bound is 20% higher).

    This goes to show you how absurd some of the headlines have been.

    The “fat tail” issue as I understand it is the possibility (very low chance) of some extreme values for CS (say 10 C). Since the IPCC 95% range extends “just” to 4.5 C (still catastrophically high), we’re talking very low probabilities to begin with, and I know very few scientists who currently entertain the possibility of a Venus runaway event, to which I don’t think even a 10 C CS would imply. The study in discussion perhaps is arguing that it’s 0% rather than 1%.

  9. Zeeshan Hasan

    This is the best news I’ve heard on the climate science side in ages. Quite a relief.

  10. Jon Flatley

    I’d be cautious at breathing a sigh of relief over the conclusions of one paper. I’ve read the book, “Storms of My Grandchildren” by James Hansen (arguably the most famous climate scientist in the world) and he’s used paleloclimatological data to point out just how sensitive the climate system is.

    For example, there was 1-2 miles of ice covering much of what now is New York City at the height of the last Ice Age. The average global temperature of earth was only 5 C cooler than today. The ocean was 350 feet lower than today. All with just 5 degrees! The opposite is an “ice free” condition quite possible and an extra 250 feet of ocean if the CO2 concentration reaches 450 parts-per-million (ppm). Right now it is 390 ppm and rising at 2 ppm/year.

    According to the IPCC, the global temperature could be anywhere from 1.1 to 6.4 C (depending on emissions) higher by 2100. Not something I can relax with!


Leave a Comment

Your email address will not be published. Required fields are marked *