Climate models have proven extremely skillful in predicting the warming that has already been observed…

15874730_1415047378519039_3981464146765466730_o It am Michael Mann, saying

Climate models have proven extremely skillful in predicting the warming that has already been observed and, by many measures (e.g. Arctic sea ice loss, melting of the major ice sheets) it is proceeding faster than climate models predicted…

Notice any problems with that quote? Well yes: model inaccuracy is taken as proof of model skill. You cannot assert simultaneously that models are extremely skillful, and that change is occurring faster than they predicted. You can, of course, plausibly say If anything, uncertainty is breaking against us, not with us but that’s rather different. Can you then add belies Tillerson’s implications to the contrary? Not on the available evidence, because it is a response to Tillerson’s The increase in the greenhouse gas concentrations in the atmosphere are having an effect, our ability to predict that effect is very limited. Read completely straight, as you are obliged to, our ability to predict that effect is very limited neither under nor overplays the predictions as compared to reality.

FWIW, I think that calling models “extremely skillful” isn’t right, and nor is “our ability to predict that effect is very limited”, except in the trivial sense that neither phrase has defined it’s terms clearly enough and so either could trivially be true by shifting and stretching definitions in plausible ways. The truth is somewhere in between.

This is all against the backdrop of the confirmation hearing on Wednesday for Tillerson. Predictably enough, the hearings are dominated by stupid politics. So the lead question we’re invited to consider is Democratic Senator Tim Kaine of Virginia asked Tillerson about Exxon’s decision to fund organizations that put out misleading information about climate science. This is pure politicking. No-one gives a toss what the actual answer is, because anyone who cares already has their own reality and won’t believe anything that contradicts it. Being asked about his own views on GW makes sense, and he managed a vaguely sensible answer: The risk of climate change does exist and the consequences of it could be serious enough that action should be taken. The obvious follow-up questions are: what kind of action should be taken? Would you support a carbon tax? Although to be actually effective that question has to be “would you support a carbon tax, as I do?” Since hardly any US pols can actually ask that question, it falls apart a bit.

Ultimately, the article is disappointing because, like so many others, it is more interested in conflict than in reporting what people said. See-also On the referendum #21: Branching histories of the 2016 referendum and ‘the frogs before the storm’.

15972497_1415113388512438_116255927821281674_o

Refs

* The Buzzfeed leaked document – Rich Puchalsky
* U.S. Needs a Robust Carbon Tax, not an Exxon Carbon Tax?
* Climate Change Deniers: Even Exxon’s CEO Has Abandoned You – make up your minds, people
* Fuckwittery from Bill McKibben – about the kind of warmed-over drivel you’d expected.
* Yet more drivel this time from mediamatters.org/ – get a grip people and stop hyperventilating.
* Don’t fall into the trap of restarting last decade’s ‘climate wars’ – Myles Allen

40 thoughts on “Climate models have proven extremely skillful in predicting the warming that has already been observed…”

  1. It seems that, cute as they are:

    Stoats (Mustela erminea), along with their mustelid cousins weasels and ferrets, were introduced in the 1880s in an attempt to curb invasive populations of rabbits and rats, which had grown out of control. But stoats don’t just eat rabbits and rodents: they also prey on native birds such as kiwis, as well as on insects and fish, and have been implicated in the extinction of the bush wren and laughing owl

    Like

  2. I think you’re leaping before looking there. Mann specifically says “extremely skillful in predicting the warming“. Skill in predicting warming doesn’t necessarily imply skill in predicting Arctic sea ice etc.

    [Depends what you mean. Sea ice retreat implies warming where it retreats, so the models can’t be skillful at predicting that area. And, whilst I don’t believe the septics on “warming is out of (below) the model range” I don’t really think you could call the predictions-to-now “extremely skillful” -W]

    Like

  3. As I understand it, model “skill” does have a definition and is something like a model providing a more reliable estimate that a naive projection from past data would produce. This Hargreaves & Annan paper describes James Hansen’s 1980s model as showing significant skill when it came to warming, so maybe there is some indication that it’s not unreasonable to describe models as being extremely skillful when it comes to warming.

    [Is that not to confuse words? In common English, “significant” and “extremely” could be interchangeable or closely related. But A+H have used “significant”, I imagine, in the technical statistical sense. “extremely” is being used in a totally different sense. Take, for example, the “standard” pic. Is that “extremely skillful”? Not really. Is it “significantly skillful”? Maybe. But straight linear interpolation might be better, or not much worse. On short time scales, climate prediction is hard -W]

    Like

  4. Yes, I agree that significant and extremely could be seen as different, with one having a formal meaning and the other being taken to mean “really, really, really good”.

    Like

  5. CIP –

    Simple solution, find something that eats stoats.. (Next question, can you find some thing the eats Wolves? And if not, can someone get down to the genetic engineering lab and make something? This problem just keeps getting worse..)

    Anyway.. yes, models should be conservative WRT catastrophic of unexpected events, because the presence of these events in your model makes it look wrong or over sensitive. Would a model that perfectly predicted the Younger Dryas have been discarded as clearly erroneous 13,000 years ago?

    Like

  6. How about “the performance of the CIMP5 climate models have been deemed acceptable, however several discrepancies were noted”

    In certain wessels, that is high praise. About as high as it gets.

    Like

  7. It’s not one of Mann’s best quotations. It’s too easy to pick holes in, not just ‘extremely’ but also ‘predicting’ and ‘proved’. It also seems odd (inconsistent) to claim accuracy then claim inaccuracy.
    I’m wondering if this comes from one of those ‘just a quick word’ phone call-type interviews in search of a sound byte. It doesn’t sound a considered response, whatever the question was.

    [Yeah, might be. But I think I’ve seen similar quotes before by others, which was why I picked it up. Need more examples -W]

    By definition, if that’s Stroke, aren’t the others all late?

    [Yes, that was the point. I’m stroke, in case you didn’t realise (that one is div 4; the one of me coxing was div 2, if that too was confusing). Try https://www.facebook.com/AlCraigiePhotographer/photos/a.1415112538512523.1073742553.197284073628715/1415116058512171/?type=3&theater, though you can’t see our blades -W]

    Like

  8. You don’t really think we’re going to let you get away with “…I think I’ve seen similar quotes before by others…” , given the way you smack us down when we’re a bit vague!
    Note this is the Winter League. Nice to see Proper British Attitudes to the weather/cold being demonstrated with such panache. Shame I missed the joke 😦 .
    On model skill, JDA/JH are definitely the ones to go to, IMO. But I’m biased. If H&A say ‘shows significant skill’ to me reads as ‘is excellent’ (don’t take this too seriously). IFRC, the take-out was that models often show skill with the big picture stuff, but struggle when it comes to the details.

    Like

  9. I saw a good definition of “extremely” somewhere:

    So far off the normal distribution that if he dropped all his marbles, they wouldn’t roll.

    Like

  10. Obviously a ‘no global warming’ scenario would just be natural variability wrapped around a straight line. So, if we are comparing to ‘no global warming’ then current models are definitely ‘skillful’ and probably ‘extremely skillful.’

    I don’t believe anyone reasonably expects current models to reproduce the ENSO variability in phase with the real world – yet. If we allow for that or select model runs that most closely match actual ENSO, then ‘extremely skillfull’ again would seem to be a fair description.

    Sticking with more mathematical descriptions woud leave less room for interpretation, but also be less informative for lay listeners. So we’re back to communicating science. Queue up some Schneider quotes.

    Like

  11. If my work emails had been stolen, taken out of context, broadcast to the world in a slanderous fashion, and then used as ammunition against me by vicious fossil fuel supported wingnut groups, I suspect I might have a hard time writing clearly and rationally while thinking about the CEO of a company historically at the heart of vicious fossil fuel wingnut denialism.

    But what I am really wondering about now, though, is how Tillerson Rex is going to reconcile his own rather mild position on climate change with the highly un-nuanced, balls-to-the-wall denialism of his boss, our new world leader, Kim Don Trump. That should be an interesting battle of the scorpions.

    And here is a mechanical engineering puzzle for you. If you superglue the speed regulator’s or governor’s balls to the cylinder wall, what will happen to the speed of the machine? I don’t really care, it’s just an interesting concept.

    Like

  12. SteveP — If glued so the mechanism cannot go up and down then it cannot adjust the steam escaping; won’t regulate.

    Like

  13. Somewhat related (meaning I don’t recall and don’t feel like searching for the most recent appropriate post where this comment could/should be left):

    Valuing Climate Damages: Updating Estimation of the Social Cost of Carbon Dioxide

    NAS report, link is to the’Discounting Module’ chapter. Pretty much what you can read elsewhere, though they seem to push for the low end (2.5%) discount rate. Biggest hole still seems to be the lack of accounting for possible catastrophic damages.

    Like

  14. Nothing unclear about this, when you add the rest of the quote:

    “… it is proceeding faster than climate models predicted,” he added.

    “If anything, uncertainty is breaking against us, not with us, which belies Tillerson’s implications to the contrary.”

    Anyone remember Pete Seeger”

    “waist deep in the Big Muddy, the big fool says to move on”?

    Like

  15. Any model predicting watming needs to be measured against the rough predictions that any lay person, with knowledge of the past 10,000 years of warming and the warming properties of CO2, could make.

    JD

    Like

  16. JD Ohio,

    Why should a model be measured against something so silly?

    [It made sense to me. Or rather, it would have made sense, if you could have gone back in time to make the prediction. I think “lay person” means someone who has paid attention, read the IPCC SPMs, and so on -W]

    Based on the past 10,000 years of warming, any lay person would be a complete loss what to predict, although the most likely scenario would be cooling (being past the HCO and based on the general downward trend of interglacials.
    And any lay person would also need to know quite a bit more than “the warming properties of CO2”, like a good estimate of what those warming properties actually are.

    With that solid knowledge, the lay person is no longer a lay person, but someone building a model to predict warming…

    Like

  17. Why should a model be measured against something so silly?

    In my world if temperatures have been rising for 10,000 years and CO2 has temperature increasing properties it is reasonable to forecast that it may rise some more in the next 100 years. One could do back of the envelope calculations based on temperature average rise over the last 100 years, 200 years, 500 years, 1000 years or 10,000 years and make a projection/prediction. One could then add some more to this to account for CO2. Then for instance, my quick calculation would be that I would expect temperatures to rise between .5 centigrade to 1.75 centigrade over the next 100 years. My best guess (not doing much work–(others could put in more time with this type of back of the envelope calculation) would be that temperatures would rise 1.1 degrees centigrade over the next 100 years. Then if I was exercising my rights as a voter, I would compare the accuracy of various models to my (in my view) reasonable first estimate and see how they fare.

    I would ask you what you would compare models to in an effort to ascertain their accuracy?

    JD

    Like

  18. Oopsie:
    http://www.bbc.com/news/science-environment-38643420

    “The British Antarctic Survey is to pull all staff out of its space-age Halley base in March for safety reasons.

    The highly unusual move is necessary because the Brunt Ice Shelf on which the research station sits has developed a big new crack.

    BAS officials say neither staff nor the base are in any immediate danger ….”

    Like

  19. The Brunt megasplinter is just the Southern Coneheads getting even for Kissinger calling Chile :

    ” A dagger aimed at the heart of Antarctica.”

    Like

  20. http://www.bbc.com/news/science-environment-38640413

    “It is clear that they [Trump] actually accept a great deal more of the science of human influence on climate than they are prepared to let on,” said Prof Myles Allen from the University of Oxford.

    “They are acknowledging there is a link, there is a potential problem and that’s already more than enough to justify continuing the relatively modest goals of both the Paris agreement and Clean Power Plan.”
    At odds with the base

    Prof Allen believes that the statements of the transition team to date are far removed from the views expressed by their grassroots supporters.

    “We’re in a situation where the foot soldiers of denial are well behind their generals,” he told reporters.

    Like

  21. An example of what preliminarily looks like good weather modeling appears here. http://www.sanluisobispo.com/news/weather/weather-watch/article126639939.html In late December of this year the Fleet Numerical Meteorology and Oceanography Center in Monterey projected the potential for the jet stream to split which would lead to rain in California. The rain came pretty much as predicted. It has been a long time since California has had this much rain, and the fact that the predictions were spot on (although made in short-term time frame) is an initial indication, of a reliable, useful model.

    JD

    Like

  22. #12Kevin O’Neill
    “Obviously a ‘no global warming’ scenario would just be natural variability wrapped around a straight line. So, if we are comparing to ‘no global warming’ then current models are definitely ‘skillful’ and probably ‘extremely skillful.’
    I don’t believe anyone reasonably expects current models to reproduce the ENSO variability in phase with the real world – yet. If we allow for that or select model runs that most closely match actual ENSO, then ‘extremely skillfull’ again would seem to be a fair description.”

    Have to comment on this,Kevin.
    Models have to be able to predict ENSO .
    If they cannot do this actually and accurately they have no skill.
    ENSO patterns have an important role in actual heat loss to space and failure to predict it properly means all calculating skill is out the window.-

    [This is nonsense, I’m afraid. Climate models need to have an ENSO – exactly how accurate is unclear, since the long term trend (AFAIK) isn’t desperately sensitive to it. But they certainly don’t need to predict individual ENSO phases, which s just as well, since its impossible -W]

    The other comment about natural variability wrapped around a straight line is also a throwaway line best thrown away. No one claims that the world has a flat line trend. There is a basal global warming with various cycles of warming and cooling overlaid. Definitely not a straight line.

    Like

  23. JD Ohio,

    You can only compare to the actual observations. Sometimes a wild guess can be correct, completely and totally by accident.

    Take the hypothetical case in which the models project a 2-degree increase, and your back-of-the-envelope calculation says 1. Due to a few very large volcanic eruptions, your guess turns out to be more correct. It wasn’t because your approach was better, it was because your calculation was saved by volcanic eruptions. Now also suppose that adding those volcanic eruptions into the models result in an equally correct estimate for the T increase. In that case, the models are clearly more accurate, because they are able to correctly determine the T increase when all factors are taken into account, while your own model (because that *is* what it is) turns out to be accurate completely and totally by accident.

    Of course, we should also remember that GCMs do quite a bit more than just model the T evolution as a function of changes in forcing. I doubt your own guestimates would include precipitation changes, evolution of oceanic currents, etc. etc.

    Like

  24. Marco: “You can only compare to the actual observations. Sometimes a wild guess can be correct, completely and totally by accident.

    Take the hypothetical case in which the models project a 2-degree increase, and your back-of-the-envelope calculation says 1. Due to a few very large volcanic eruptions, your guess turns out to be more correct. It wasn’t because your approach was better, it was because your calculation was saved by volcanic eruptions.”

    You are mischaracterizing my argument. I am not saying that my back of the envelope calculations would be more accurate or scientific than any particular model. I am saying that you have to start off somewhere as to what would be used to validate a model that is designed to predict 100 years into the future because the time frame of the prediction (as opposed to 2 weeks for instance) is so long that you can’t have multiple completed runs to test the model’s efficacy. I would start off with past temperature trends and others more sophisticated than me could tweak it or add more sophisticated or accurate components.

    My main point however, is that just because a model from 20 years ago predicted warming, and it did warm, does not give you much good evidence that the model is particularly skillful. Based on the last 10,000 years virtually anyone could predict warming.

    I would again ask you the basic framework you would use to test the efficacy of models such as Mann’s or Hansen’s.

    JD

    Like

  25. JD Ohio, you claim once again that based on the last 10,000 years we would expect warming. This is, however, simply not true. Any naïve lay person would, based on historical reconstructions, expect cooling:

    If told CO2 is expected to cause warming, the lay person would still not be able to predict warming, unless given a number for the expected warming per doubling of GHG forcing *and* the expected GHG increase.

    With that information you are already getting really close to an actual, albeit still simplistic, model.

    Now that layperson would also need to know how forcings have changed over the last 50-100 years to be able to reproduce to some extent how temperatures have *actually* changed over the last 50-100 years. Your simplistic model has now already become a pretty advanced model…

    A small BTW: Mann is, AFAIK, not an author of any GCM.

    Like

  26. Marco, I was going by this study which post dates the Marcott study and shows warming over the past 10,000 years (which is consistent with the receding of glaciers) http://www.dailymail.co.uk/sciencetech/article-2723861/Earth-getting-hotter-past-10-000-YEARS-contradicting-studies-humans-began-global-warming-trend.html

    Instead of saying what people would need to know, please give the basis, either simplistic or detailed, for your conclusions.

    In any event, I am not interested in arguing over which work is correct. For the third time, I am asking you as a lay person, who presumably believes that models discussed by Mann are skillful: what basis do you have to believe that they are [or, potentially,are not] I have given you my grounds for my evaluation. What are yours?

    JD

    Like

  27. JD Ohio, you linked to the Daily Mail. That newspaper has various nicknames, including the “Fail”. I wouldn’t trust it if it said the Sun would rise in the East.

    Like

  28. So isn’t hansen saying something akin to the observation that CO2 is responsible for (handwaving here) a hundred fifth percent of the warming observed so far (because we’re not accounting for the cooling from sulfates)?

    Like

  29. Major part of the problem is that global surface temperature is a lousy metric to judge GCM skill. Patterns of warming and especially circulation are much better. Steve Easterbrook had a great comparison of global circulation as measured and calculated. The model was scary good

    Like

Leave a reply to cstoudt Cancel reply