The Irrelevance of Nonlinear Effects In the Climate System

Posted by William M. Connolley on December 18, 2006
(25)
More »

RP Sr has yet another post The Relevance of Nonlinear Effects In the Climate System pushing the usual stuff: Thus if we accept that small perturbations can result in significant changes in the climate system through nonlinear interactions, then all of the human- and natural climate forcings need to be assessed in this context. I would sigh and move on (I did, earlier today) but now I feel moved to find an old post of mine: Climate is stable in the absence of external perturbation. See, its so old its on the old blog…

Anyway, the point is that people vastly overplay the importance of this non-linear / chaos stuff, because you can get exciting words like “tipping point” out of it, and it plays well in the press. Where would we be without the possibility of “climate surprises”, after all. There is a good RC post on this.

Yes, the weather is definitely chaotic. And yes the climate is in principle non-linear. But in many important aspects this all averages out into linearity; and there is no good evidence for important “tipping points” *in climates close to the present*. RP quotes Tas saying once the Greenland ice sheet collapse passes a certain point, it is unlikely to regrow in the current regime. Which is true. But not very interesting. The timescale for Greenland melt is so long that we (now) don’t need to care a great deal about it reforming, since GHG’s are going to be high enought to prevent that for rather a long time. And also, this isn’t a “small perturbation”: its a big one. Where is the evidence that small perturbations lead to large climate changes on a global scale? Holocene global climate appears to be quite stable and reacts linearly to forcing.

(25)
More »

Comments

  1. #1 bob koepp
    2006/12/18

    a couple questions…
    _How_ stable is the climate?
    Is the stability of the climate simply an observed phenomenon, or is there some sort of dynamic equilibrium?

    [Observably, fairly stable over the last 10kyr in the global mean. Observably, quite unstable during the last glacial, because there are big swings - Dansgaard-Oeschger oscillations (which people get very excited about). And maybe there are exciting instabilities ahead. But there is precious little evidence for them.

    There is clearly a dynamic instability. The weather varies wildly, but always returns to the same patterns - W]

  2. #2 Lubos Motl
    2006/12/18

    I agree with Bob that all these things are cliches unless you provide some quantitative model what you mean by the climate being linear and stable.

    The simplest models of its “stability” are, in my opinion, easily falsified by observations. If the climate is stable in the absence of external perturbations, how do you explain that the temperatures dropped by 0.05 C from 2005 to 2006?

    [You don't mean the same things by the same words. There is year to year variation, of course. So what? -W]

    But more generally, don’t get me wrong: I also think that most people especially outside science tend to overestimate various chaotic things and underestimate our ability to get – more or less – accurate predictions.

  3. #3 Lubos Motl
    2006/12/18

    William, you posted your answer to Bob before I sent my comment above. Are you joking when you say that the climate is stable at the 10kyr timescale? Have you ever heard of Milankovitch cycles?

    http://en.wikipedia.org/wiki/Milankovitch_cycles

    [I said: its been fairly stable over the last 10kyr. I didn't say absolutely constant -W]

    These are effects that significantly influence the climate whose typical period is exactly a few times 10kyr up to 100kyr.

    [Yeeesss. So what? -W]

    At more or less every timescale, one can find similar patterns. Some of them are determined by known factors, such as the precession, and some of them have unknown causes.

    The statement “climate is stable in the absence of external perturbations”, if you really mean things like “there are no signals at the 10kyr timescale”, sound like statements from someone who has seen the climate science for 7 minutes so far.

  4. #4 Lab Lemming
    2006/12/19

    But aside from the holocene, when else has climate been stable? More specifically, if we exceed the holocene climactic stability band (whatever the fuck that is), will we end up in a stabe or unstable post-holocene regime?

    [As far as we know, the last interglacial was stable too. The mechanisms for glacial instability only work in glacial times. Of course there may be others -W]

  5. #5 Eli
    2006/12/19

    I know what you’re thinking Lemming. You’re thinking did the mixing ratio increase six or only five times. And to tell you the truth I forgot myself in all this excitement. But being this is a the only planet you got, the most powerful climate perturbation in the world and will blow your head clean off, you’ve got to ask yourself a question. Do you feel lucky? Well do you, Lemming? (C Eastwood, with some liberties taken)

    [Very good... but as you say: "the most powerful climate perturbation in the world" - undoubtedly, the climate system responds to forcing. But the interesting question (which is what RP Sr was originally on) is do subtle non-linearities act to amplify tiny perturbations? Well obviously not, or there would be no "climate" -W]

  6. #6 Eli Rabett
    2006/12/19

    I had a long interchange over there with the usual characters about whether climate was a boundary value problem and weather an initial values problem. Basically they (and that includes Prof. Pielke) do not accept that distinction.

    However, a someone better answer might go like this. Weather forecasts are done as ensembles, where the track is found to be confined to a small area of the parameter space. The range of the forecast is the time at which the members of the ensemble begin to diverge from each other wildly. This is somewhere between one and two weeks at the present time.

    One can look at climate hindcasts/forecasts in the same way. The fact that they all agree withing limits with each other and do not chaotically diverge is evidence that there are significant bottlenecks in the parameter space which limit divergence of the current climate from the model tracks. Unless you can make an argument as to why and where those bottlenecks will be reached, Pielke;s argument is papp.

  7. #7 bob koepp
    2006/12/19

    Eli – Lemming can, of course, speak for himself. But I’d turn your eastwoodesque question around — Do you feel unlucky? Well, do you?

    Whether we are lucky or unlucky, however, is irrelevant to the truth or falsity of claims about how the climate system will change. The epistemic status of claims about what the future holds needs to be clearly separated from questions about whether the future in question would be good or bad for whomever. That’s just basic scientific method, no?

  8. #8 Eli Rabett
    2006/12/19

    Lucky or unlucky has everything to do with policy choices. As was pointed out a nuber of years ago by Michael Tobis, the characteristic of the problem is that there are a range of possible outcomes (climate sensitivity et al.). If one looks at the most probable range (2-4.5 C or something like that by 2100) the low end is not very concerning with a few costs. The most probable value has serious costs and the high end has catastrophic costs. Given the asymmetry of outcomes a great deal of caution appears to be logical.

  9. #9 bob koepp
    2006/12/19

    Feelings of being lucky or not (actually, attitudes about risk taking) do, indeed, have a lot (but not everything) to do with policy choices. That doesn’t make policy choices, or attitudes toward risks, into scientific problems. How willing someone is to risk being wrong about climate change is, I repeat, irrelevant to the truth or falsity of claims about climate change. Whether I am deathly afraid of a warming world, or whether I think it would be highly desirable, or a toss-up, has no bearing at all on the validity of factual claims regarding climate. In other words, “scariness” is not a proper epistemic criterion for assessing factual claims.

  10. #10 Eli Rabett
    2006/12/19

    Unfortunately bob, science will not get you the certainty you want in the time we have to take action given our current scientific understanding of the issue. Waiting is a policy choice as much as acting, with just as many, if not more consequences. And yes, risk is a proper epistemic criterion for assessing actions, which is what this is about no matter how much you complain. Moreover, since the risk is to everyone, your wanting to wait and see is irrelevant.

    Dirty Harry was offering a 50-50 chance, but as with climate change the risk on the downside, to get blown away, was a lot higher than the risk on the other, to get arrested. Even there, the odds of a neutral outcome if we continue increasing GHGs is a lot less than 50% (that’s the science).

  11. #11 bob koepp
    2006/12/19

    Eli -
    I don’t quite grasp the nature of my misfortune, but perhaps you were trying for irony. Certainty? Who said anything about certainty?

    I don’t deny that policy issues loom large. I do deny that the policy implications of climate change have any relevance whatever to the truth of climate change.

    You speak confusedly of risk and the epistemic assessment of actions. The only kinds of actions where epistemic assessment can get traction are cognitive acts like believing and doubting. Now believing and doubting certainly do expose us to risks. And it even makes sense to speak of epistemic risks; the epistemic consequences of error. But the notion that risk (especially risk relative to non-epistemic goods) is any sort of “proper epistemic criterion” is nonsense.

  12. #12 James Annan
    2006/12/19

    In other words, “scariness” is not a proper epistemic criterion for assessing factual claims.

    Maybe, but it’s a proper criterion for decision making (even though IMO the scariness is overblown in some climate research).

  13. #13 CapitalistImperialistPig
    2006/12/19

    William or anyone,

    I have some comments and questions prompted by your post at My Place. My basic question is what light, if any, do climate models shed on the stability question?

  14. #14 Eli Rabett
    2006/12/20

    Without disagreeing with James (whom I agree with), allow me to point out that bob has substituted his version of what I wrote

    “scariness” is not a proper epistemic criterion for assessing factual claims”

    for my version of what I wrote:

    “risk is a proper epistemic criterion for assessing actions”

    Very different things, but you hardly saw him take the pea out from under the pod, and if you don’t factor risk into your actions, well, I think I know some folk who would like to bet with you.

  15. #15 Eli Rabett
    2006/12/20

    Perhaps I should have made it clearer in this context. Both the range of climate change and the nature of the risks associated are epistemic (associated with knowledge), although they both have aleatory (associated with randomness) components and also components which are beyond our current knowledge. Therefore, our knowledge of both the range of climate change and the associated risks that depend on that range are separately and together epistemic criteria for assessing actions.

    The hard part is that current knowledge shows that it is necessary to take immediate action to avoid the probability of the most serious risks, moreover, given our epistemic knowledge of the system, it is strongly possible that there will be little significant improvement in our ability to describe and predict future climate changes beyond the current range.

  16. #16 bob koepp
    2006/12/20

    Eli – I really don’t want to get into a pissing contest with you, especially in a digression from the topic of stability; but I made no substitutions of the sort you claim. My only point in our exchange has been that risk is not an _epistemic_ criterion — there is no systematic relation between risk and “truthiness”. That’s all.

  17. #17 Eli Rabett
    2006/12/20

    There are three points. The first is that knowledge of the system allows you to evaluate the risks associated with the behavior of the system. The second is that evaluation of possible risks informs the direction that further research should take and is informative and contains important information about the system. Thus the study and understanding of risk is epistemic.

    Simply stating that something is risky may not be _epistemic_ but a careful evaluation of the probability of the risk occuring IS and is a necessary part of any study. Perhaps the best example is that an understanding of safety risks is both _epistemic_ and vital to eliminating them. Points of failure must be first identified before they can be mitigated. Such studies have two axes, frequency and severity. A frequent, but trivial risk, or a very rare, but very serious risk might both be tolerated, but you have to understand the risk before you make that decision.

    The third is that you did change what I wrote, in a way as to make it appear less valid. Risk is not scary, ignoring risk IS scary.

  18. #18 bob koepp
    2006/12/20

    Eli – Piss away. After you’ve relieved yourself, please substantiate your claim that I’ve changed any of your words. Oh, and please note the temporal order of the postings in question (it would be quite a trick for an earlier comment of mine to involve a substitution for a later comment by you…)

  19. #19 Eli Rabett
    2006/12/20

    Risk does not equate with scariness in the context of this exchange. Now if you think perception and belief are all that are needed for _epistemic_ understanding, then you can have the irrational _epistemic_ understanding and since I don’t I will take the rational _epistemic_ understanding. YMMV Let’s get that one off the table.

    The perception of risk is a cognitive issue subject to rational investigation. At some signicant level of understanding belief/perception) of the risk becomes rational. Have you looked under your bed?

  20. #20 Lab Lemming
    2006/12/21

    Stoat:
    I generally don’t deem isotopes lighter than Sr worthy of peering down my hard-rock nose at, but I seem to recall that the last interglacial d18O was a sharp increase followed by a rapid drop-off, and not the plateau that characterizes the holocene.

    [There is a pic here: http://en.wikipedia.org/wiki/Image:Vostok-ice-core-petit.png we don't have such good records for the Eemian: the Greenland early cores are corrupt then and NGRIP doesn't cover it all: http://en.wikipedia.org/wiki/Image:Ngrip-epica-do18.png But the D-O spikes are missing -W]

    Lubos & Rabbit:
    My suspicion of the happy, stable holocene comes from the cliamte records that the softrock people who somehow gain access to by lab generate. At least on the regional scale, nature seems to be perfectly happy at dishing out inconvenient variations all on her own.

    Everyone:
    How does the natural variability of the past have any bearing on what to do about greenhouse gasses?

    Who feels safer by the demonstration that city X would have been up shit creek had it existed at its present size 2600, 4000, or 7500 years ago? How is that possibly a recipe for not building the capacity to adapt to change?

    [Because one way to make GHGs scary is to invoke the chance of rapid climate change -W]

  21. #21 Eli Rabett
    2006/12/21

    Lemming, interesting what you say about the geo analysis. I thought most of the temperature data for the recent Holocene came from 18O analysis out of ice cores, which is a very soft rock indeed. Of course, my focus may simply be a time scale difference between geology and gas phase chemical dynamics.

  22. #22 Tas
    2006/12/22

    Hi William,

    Sorry you found RPs picking up my comment an issue to “sigh and move on” over. For the record, when I made the comment *I* was not seeking to overplay the importance of non-linear stuff and climate surprises. I was rather trying to point to a fallacy in the argument that tiny incremental changes must produce tiny incremental effects that can always be ignored. In a system that has thresholds (and the Greenland example is one that is in fact noted in the good RC article you linked to) there comes a point when you do indeed pass a point of no return (however we might despise cliches).

    Regarding Greenland, I’m not sure why you class it as “not very interesting”. Sure, the outcome plays out over centuries but the irreversabilty ensures that the consequences have to be lived with for ever. Can we adapt to 6m extra sea-level in the time frame of melt? I suspect yes, in general, with some losers. Do we want to adapt to such a world though? And are we confident that other knock-on effects (eg W. Antarctica) will not make it significantly more extreme?

    If the attitude is that you might rather avoid any of these reasonable scenarios then it *is* interesting *now* because lags in the system mean that our decisions over drivers must occur on a human timescale, not the glacial response timescale of the Greenland ice sheet.

    [My sighing was at RP not you; he seems to pick up things and then run off in an all too predictable direction. Anyway: my point was that small changes almost always do have small (climate) effects. You can find some counter examples (you could look at Greenland that way; but since we're not there yet its also perfectly possible to argue that *from the current climate* there are no such perturbations).

    There will come a point where we push Greenland into irreversible melting. How interesting (from a global point of view) is this? Maybe "not very interesting" is going too far; but it is certainly not as wildly exciting as often portrayed. Gr will probably contribute 50 cm/century as it melts; though I admit those estimates are uncertain -W]

  23. #23 Eli Rabett
    2006/12/24

    Eli has been pointing out over on RPS that it does not matter how wrong or chaotic the hypersurface is if the trajectory never reaches the places where it is wrong, or chaotic. Small excursions, especially if random, don’t get you to those places if you are not there already, and we are not and have not been for tens, if not hundreds of thousands of years (fortunately). There are undoubtedly bottlenecks to very bad (for us) places. The danger of a directed walk is that it can get you there incrementally.

  24. #24 Roger A. Pielke Sr.
    2006/12/24

    William – I am pleased that you linked to Climate Science, but you (and Eli and Tas) continue to ignore the significance of the nonlinearity of the climate system. While I agree with Tas that

    “Anyway: my point was that small changes almost always do have small (climate) effects”,

    the key qualifier here is “almost always”. The exceptions represent situations where sudden threshold changes in some important aspect of the climate system would occur.

    The failure of the multi-decadal climate models to predict such threshold changes certainly conflicts with the observed climate record for a variety of space and time scale, as we reported, for example, in

    Rial, J., R.A. Pielke Sr., M. Beniston, M. Claussen, J. Canadell, P. Cox, H. Held, N. de Noblet-Ducoudre, R. Prinn, J. Reynolds, and J.D. Salas, 2004: Nonlinearities, feedbacks and critical thresholds within the Earth’s climate system. Climatic Change, 65, 11-38.
    http://blue.atmos.colostate.edu/publications/pdf/R-260.pdf

    In a broader question, I invite you to respond to the conclusions summarized at Climate Science [http://climatesci.atmos.colostate.edu/main-conclusions/].

    Best wishes for the Holidays to the readers and contributors of STOAT.

  25. #25 Chris O'Neill
    2006/12/28

    Regarding inspecting the ice-core record for climate stability, I think it’s worth getting hold of the data files and plotting them yourselves to look in detail at whatever time frame you choose. e.g. Vostock, Epica. Looking at a 400,000+ year graph isn’t clear enough. There are periods at least as long as the Holocene that are at least as stable, e.g. the last 14,000 years of the last ice age up to 17,000 years ago. The vast majority of the time the cores show stability with a slowly changing long-term trend. Rapid sustained change, indicative of global strong positive feedback, i.e. instability, happens only a small fraction of the time.