A Few Things Ill Considered

A layman's take on the science of Global Warming featuring a guide on How to Talk to a Climate Sceptic.

Thursday, March 02, 2006

send this to... Digg it! | Technorati | Del.icio.us | Reddit | Furl | Spurl

We Can't Even Predict the Weather Next Week

(Part of the How to Talk to a Climate Sceptic guide)

This article has moved to ScienceBlogs

It has also been updated and this page is still here only to preserve the original comment thread. Please visit A Few Things Ill Considered there. You may also like to view Painting With Water, Coby Beck's original fine art photography.

Labels:

posted by coby @ 3/02/2006 03:56:00 PM  

17 Comments:

  • At March 03, 2006 11:11 PM, Anonymous Anonymous said…

    You say: "Climate is defined as weather averaged over a period of time, generally around 30 years. This averaging over time removes the random and unpredictable behaviour of weather."

    What if we apply the same logic to a less ambitious problem like a container of gas sitting on a warm stove? Let's see:

    Gas is formed from millions of molecules that collide with each other 10^9 times per second. Their motion is chaotic and unpredictable. However, define a small volume of gas, say, 1mm3, which would still have 10^16 molecules, and define averages over, say, 30 milliseconds, which will give you an avearage of about 10^23 collisions. That's a lot of events, so this averaging would remove all that chaos of scattering, and each volume will have a well-defined energy and temperature. Obviously, the gas container is just a substantial collection of such volumes. Therefore, the overall gas behavior cannot be random and must be always easily predictable. Right?

    -aap

     

  • At March 04, 2006 10:55 AM, Blogger coby said…

    Hi Alexi, thanks for the comment!

    Before I allow you to educate me on Brownian motion (did I guess your point correctly?) I will try to cut you off at the pass, so to speak. Analogies are useful tools if you treat them right. They are of course limited, as is any model or approximation, because let's face it, every thing is unique when looked at closely enough. Climate is, in fact, defined as average weather. Does that mean that every macro behaviour can be similarly describe as an average of micro behaviours? No. Does the fact that this is not a universal principle mean that it is inappropriate when applied to the climate system? Again, no.

    I know that quite a few people think that climate is a chaotic system, and maybe on some large scale level, it is. But it is not, in my opinion, chaotic on anything approaching human time scales. Frankly, I have never heard any argument supporting that notion, only arguments that take that as a given. Certainly the march of the seasons is nice and regular, and determined directly by the orbital inclination of the earth. If a large volcanic eruption occurs, the global temperature drops for a few years quite predictably. Look at the glacial/interglacial cycles, not perfectly regular but far from random, also determined by orbital variations. I will grant you that the data is quite chaotic on the multi-century time scale even as it follows a 120Kyr cycle, but who is to say that had we enough data and understanding, these spikes and dips could not be thoroughly explained by solar infuences, volcanic eruptions, ice sheet dynamics etc.

    While I will grant you it is a complex system and is capable of some surprising behaviours, I don't see it as chaotic and I see no problem with speaking in a meaningful way about average weather. Did I understand you point?

     

  • At March 05, 2006 12:58 AM, Anonymous Anonymous said…

    You are correct that I started the problem description from first principles, from Brownian motion. However, your leap to my point is incorrect. Unfortunately, your opinion that the analogy is incomplete and inappropriate is also wrong - the fact that the macroscopic volume of gas heated from below can be inherently unstable is related to the omissions we made when we neglect small leftovers after we apply averages of those microscopic volumes. Same thing occurs when you use an artificial 30-year average of daily-monthly-yearly oscillations but consider all small deviations as a result of "external forcing". When you do this, you lose all internal dynamics of the system. So, the difficulties associated with arbitrary-selected scale for averaging is quite universal.

    You seem to admit that the weather system is chaotic and cannot be predicted for more than 3-5 days. Now you also admit that on multi-millennial time scale the glacial cycles are "not perfectly regular but far from random", and even has concerns about the 120ky cycle. What makes it so difficult for you to bridge the gap and start thinking about climate dynamics as being inherently chaotic on all time scales?

    I think the problem is in misunderstanding of what does it really mean for a system to be chaotic. It does not mean that the 30-year averages would look completely random. First of all, at least the three-frequency solar forcing does exist, and subjects all climate fields and interface coefficients to corresponding oscillations. It immediately places the initial phase space of the whole climate system on a three-dimensional torus. If recall some old memories correctly, it immediately makes the system even more than chaotic, it makes it "structurally unstable". In essence, the system follows the forcing on average, but exhibits internal dynamics on the top of it, and may have substantial deviations on certain original frequencies, like the unexplained magnification of the 120ky cycle.

    There are different degrees in how a dynamic chaos manifests itself. Look at Figure 8 of this research:
    http://arxiv.org/PS_cache/nlin/pdf/0111/0111022.pdf
    Although the regime belongs to a mathematically proven stochastic attractor, the behavior of observables in time is clearly "far from random".

    More, most known strange attractors have another very pronounced feature - the variables would exhibit oscillations with a broad spectrum of time scales, theoretically having infinitely long cycles. It means that if the weather is a real strange attractor (and many models, e.g. Lorenz-84, show that it is), then atmospheric variables would have very long variations including the same time scale as the oceanic CO2 pump cycle. Granted, this would be a weak coupling, but it would be also a coupling between oceans, and systems with weak coupling are known to produce incomprehensibly long intermittent behavior. This broad spectrum of time scales also implies that there is no such thing as "sufficient averaging time", be it 30-years, or another "human time scale".

    The assumption that the 30-year averages remove all unpredictable climate variations leads to a couple of problems that cannot be resolved within "equilibrium" model with "forcing". For example, climatologists still cannot account for about 2 Gigatonn of CO2 that is sunk somewhere in the cycle. As I found, all estimations of CO2 fluxes are based on "general consensus" of global equilibrium. If the dynamical chaos model is true instead, the whole system never has an equilibrium, it is always in some dynamic transition, so the methods of CO2 flux estimations based on equilibrium may be completely wrong.

    Another blunder occurs when someone tries to find a "feedback mechanism" responsible for glacial terminations, and for termination of those terminations. The problem is that they are looking for some understandable physical fields, interfaces, or processes that can be expressed in simple linear feedback equations. Unfortunately, the nature of instability in spatially-distributed systems is more complex than that. Typically, a spatially-extended hydrodynamic system becomes unstable not with regard of some particular concentration or boundary interface, it becomes unstable with regard to some eigenfunction of this concentration field associated with a critical eigenvalue of a linearized system. Those critical eigenfunctions are some complex wave modes of, say, the whole salinity field, and not just a field itself. In most cases the unstable mode cannot be identified apriori from intuition, and is identified only after a complete stability analysis.

    As you can see, failure to consider true nature of the climate leads to a few problems for equilibrium-based models with external forcing. That is my point.

    Regards,

    -aap

     

  • At March 06, 2006 12:29 PM, Blogger coby said…

    You seem to admit that the weather system is chaotic and cannot be predicted for more than 3-5 days.Now you also admit that on multi-millennial time scale the glacial cycles are "not perfectly regular but far from random", and even has concerns about the 120ky cycle. What makes it so difficult for you to bridge the gap and start thinking about climate dynamics as being inherently chaotic on all time scales?

    Alexi, it is the "far from random" part that makes me reject climate as inherently chaotic, especially if you are going to go as far as say "on all time scales". What about seasons? What about diurnal cycles? And what about at least 6 approximately 120Kyr glacial interglacial cycles?

    It is also the broadly deterministic response to forcings that a chaotic system would not exhibit that seperates climate from a chaotic "on all scales" system. Clearly, if you turn down the sun, the temperature drops. Clearly, if you turn the surface completely white, the temperature drops. And clearly, if you double the amount of an important GHG in the atmosphere, the temperature rises.

    Now, as I re-read your comment I see this:
    In essence, the system follows the forcing on average, but exhibits internal dynamics on the top of it, and may have substantial deviations on certain original frequencies, like the unexplained magnification of the 120ky cycle

    This is an agreeable description for me. (I don't agree, however, that the magnification of the 120Kyr cycles is all that unexplained). So, if chaotic behaviour can in fact be deterministic on a broader scale then we may in fact not have any disagreement, save some quibbling about what "substantial" means.

    The assumption that the 30-year averages remove all unpredictable climate variations leads to a couple of problems that cannot be resolved within "equilibrium" model with "forcing".

    I have not made any such assumption. There are many degrees of seperation between "removes all unpredictable variation" and "like predicting the weather 100 years in the future". I think broadly deterministic with chaotic flucuations is a fine mental model.

    Thanks again for you lengthy comments, I hope it makes this page a more interesting read for any future visitors.

    Coby

     

  • At March 09, 2006 12:26 PM, Blogger coby said…

    I have an entry specifically about the Chaotic Climate Gambit here

     

  • At June 21, 2006 8:08 PM, Anonymous Anonymous said…

    Just a suggestion here perhaps a mention and brief explanation of chaos theory. A pendulum will swing steadily until the forces become smaller and less predictable. Everyone can relate to that.

    A nice ironic touch would be to include a reference to Jurassic Park as well.

     

  • At June 21, 2006 8:10 PM, Anonymous Anonymous said…

    Oops.. sorry Coby. I didn't see your previous post about chaotic systems there. Please ignore previous. Nice work really!

    - Geoff

     

  • At June 21, 2006 8:17 PM, Blogger coby said…

    Thanks Geoff! and for the suggestions...

     

  • At November 13, 2006 7:40 AM, Anonymous Anonymous said…

    so is predicting future hurricane cycles considered climate-prediction or weather-prediction? it seems to be in the middle time-wise, so i suppose it is a hybrid of the two.

    and the predictions are little better than random chance it seems, just as with weather predictions more than a few days out.

    personally i don't buy your argument; i think it is easily disputed logically (as alexi has done) yet you present it as being bullet-proof.

    i believe global warming is occurring and that human activity is a contributing factor; i do not believe that climate predictions as to how warming will proceed hold any more water than do last year's predictions as to how busy this year's hurricane season was going to be.

     

  • At November 13, 2006 8:24 AM, Blogger coby said…

    so is predicting future hurricane cycles considered climate-prediction or weather-prediction? it seems to be in the middle time-wise, so i suppose it is a hybrid of the two.

    Cycles would be climate, a single season would be weather.

    personally i don't buy your argument; i think it is easily disputed logically (as alexi has done) yet you present it as being bullet-proof.

    Alexi's argument does not dispute the logical argument I make about the fallacy of equating weather prediction and climate prediction. We were talking at cross purposes but I thought we arrived at an agreement that the climate is "broadly deterministic with chaotic flucuations".

    His main argument, which we go into in much greater depth on the "Climate is Chaotic" page is simply that climate is in fact chaotic. But observations of its behaviour do not support his very mathematical point of view.

    i believe global warming is occurring and that human activity is a contributing factor; i do not believe that climate predictions as to how warming will proceed hold any more water than do last year's predictions as to how busy this year's hurricane season was going to be.

    Well now, that was a weather prediction. FWIW, the Pacific season was very active and saw many broken records.

     

  • At November 21, 2006 6:07 PM, Anonymous Anonymous said…

    coby I've been reading this whole collection and find it great. I do just have to comment on one thing: chaotic systems are still deterministic. Check out the wikipedia article for a good explanation. Really the common parlance chaos/randomness is better described as stochastic.

     

  • At November 21, 2006 7:56 PM, Blogger coby said…

    Hi Mikkel, thanks for the feedback and correction. I do let Alexi draw me into the deep waters, ie over my head a bit!

    Gavin Schmidt from GISS chimed in one one thread, I think the Climate is Chaotic one and knows what he is talking about better than I do.

     

  • At December 07, 2006 4:48 PM, Anonymous Anonymous said…

    rUnlike Dorithy in "The Wizard of Oz" who found out that she didn't have to look further than her own back yard to find happiness, one Must look beyond their own back yard to understand Global Warming.

     

  • At December 12, 2006 4:42 AM, Anonymous Anonymous said…

    It's really very simple: take a global climate model; input data from, say, 1950; run it on a supercomputer; and see if it predicts the climate in 2000.

    If it is able to do so completely on its own, without humans manipulating any of the parameters mid-run, then high confidence in the model's predictions for the future would be warranted. Though, I should note, there would still be no guarantee of correct predictions.

    If it cannot do this, then it has bugs, and the scientists who created the model and programmers who coded it need to go back to the drawing board.

    To my knowledge, none of the current models pass this test. I don't want to hear what a computer projects temperatures will be 100 years from now unless, in the same article, the authors describe how the computer successfully passed the test I've just given.

    Minus that it's just another example of human operators predetermining the results they expect, or "garbage in, garbage out."

     

  • At December 12, 2006 9:15 AM, Blogger coby said…

    I agree with you, it is that simple and this is a completely reasonable test. So I am pleased to inform you that the major climate models do indeed succeed at hindcasting not just the last 50 years but the last century, mid-century cooling, late century accelerated warming and all.

    Please see this article for references to that and other examples of model successes:
    http://illconsidered.blogspot.com/2006/03/models-are-unproven.html

    Let me know if you change your opinion in light of new information or not.

     

  • At February 20, 2007 4:50 PM, Anonymous Anonymous said…

    I agree with you, it is that simple and this is a completely reasonable test. So I am pleased to inform you that the major climate models do indeed succeed at hindcasting not just the last 50 years but the last century, mid-century cooling, late century accelerated warming and all.

    Please see this article for references to that and other examples of model successes:
    http://illconsidered.blogspot.com/2006/03/models-are-unproven.html

    Let me know if you change your opinion in light of new information or not.

    ____________________________________

    where has 'hind-casting' been successful? Don't spam me with useless links, all I need is one good link.

     

  • At July 04, 2008 6:14 PM, Anonymous Anonymous said…

    All these articles read like propaganda and spin. They say, Yes, you were correct, but from a certain angle and a certain way of thinking AGW is still happening. I still haven’t found the link that explains away the relationship between observed sun-spots and the emanated radiation which results interrelating with cosmic rays to affect cloud cover and so follows Earth’s service temp. The graph of the before mentioned much more closely correlate with observed temp change. I noticed the proposed “carbon tax” is carefully avoided, and the push for world government is not mentioned. This web-site is clearly one-sided and political. I see a lot of extra words that make the sentences convoluted. I do not sense straight talk. This web-site stinks of wrong-doing.

     

Post a Comment

<< Home