More chaos than you can shake a stick at.
Guest post by Andi Cockroft
(Anyone familiar with the Moody Blues should recognise the title – from “Higher and Higher”)
As some will have learned by now, I do not possess the scientific skills of the regular WUWT contributors to engage in in-depth evidential-based posts. Rather I like to think that just like the great unwashed masses, I have an intellect and an enquiring mind, and want here to share my musings and seek feedback to better help me (and hopefully other readers) understand some of the more complex subject matter. For my shortcomings I apologise. For raising questions requiring answers I do not!
Way back in January 2011, Phil Salmon posted here on WUWT “Is the ENSO a nonlinear oscillator of the Belousov-Zhabotinsky reaction type?” – His alternate title “Standing on the shoulders of Giant Bob” Perhaps my alternate title should be the same. I have read and re-read Phil’s definitive article, and as Isaac Newton was believed to have complained of the three-body problem:-“his head never ached but with his study on the moon”. So in my own way, I want to raise issues associated with what I see as true chaos, and whilst following (albeit very slowly) Phil’s work, that was nonetheless focused towards ENSO and away from the general Climate Models that I want to investigate.
Why a butterfly? Most students of chaos theory will know the “butterfly effect”, whereby it is asserted that in a chaotic non-linear system, a butterfly beating its wings in say Brazil could inject such feedback as to disrupt the airflow sufficiently that many years later a tornado would form over Texas.
Similarly, travelling back in time and moving the butterfly a few centimetres would cause the tornado not to form, or form elsewhere.
Also note that the Quantum version of the Butterfly Effect (the Quantum Butterfly), is a different beast altogether – although some parallels are drawn.
Ascribed to Edward Lorenz, the US mathematician and meteorologist who sadly passed away a few years ago. Initially Lorenz referred to a Seagull flapping its wings forever changing the weather, but later in a 1972 speech entitled “Does the flap of a butterfly’s wings in Brazil set off a tornado in Texas”, the butterfly analogy was born.
It is Lorenz’s work on meteorological chaos that I want to look at specifically.
In or around 1961, vacuum-tube computers arrived at Lorenz’ disposal, and not surprisingly the first weather model was born as a set of a dozen or so differential equations involving such things as temperature, pressure, wind velocity etc.
During a re-run of this early model, Lorenz is believed to have restarted the program in the middle of its run by entering a variable to 3 decimal places – to his surprise the results were completely at odds with what was achieved earlier.
Restarting and re-entering the variable to its full 6 decimal places produced a repeat of the initial results – from this Lorenzo drew the inevitable conclusion that with his dozen or so equations, even a miniscule variation on input is capable of creating massive change in output.
But why such a radically different outcome for such a miniscule difference in input?
It transpires that the equations were non-linear, with a so-called “great dependence on initial conditions” – change the initial conditions even slightly and a large change in output is observed in a later state.
The conclusion that Lorenz drew, was that given that such small variations can create such massive variation in output, it was impossible to “model” a weather system.
This “sensitive dependence on initial conditions” was destined for higher things however.
Around this time, Lorenz published perhaps his most important paper “Deterministic Nonperiodic Flow”, and with it was born Chaos Theory, and the concept of the Lorenz Attractor.
(For those wanting to investigate the Lorenz Attractor further, I would recommend reading Phil’s original article here, or Wikipedia here )
I’m not sure if there is any relationship here to Peter, but James Gleick’s bestseller, ‘Chaos: Making a New Science’ built on Lorenz’ musings, became the mantra for many to follow in all walks of life:- finance, science and even time-travel effects in science-fiction. It is now bandied around in insurance, marketing and business boardrooms throughout the world.
Being quite a reserved individual, Lorenz was taken aback by the devotees to his speech. ‘I was just trying to determine why we didn’t have better luck with our weather forecasts, I never reached a point where I believed the butterfly was a scientific fact. At most, it’s a hypothesis. I never expected it to become so huge outside meteorology.”
So my question now is, given that we cannot even begin to measure things such as SST to anywhere near 3 decimal places, how can we expect low-accuracy and highly volatile data, entered into chaotic models of even more chaotic systems to produce anything of significance?
If Lorenz gave up when he discovered weather to be totally chaotic and unpredictable beyond a few days, what makes the Climate Modellers believe they can do better forecasting years or even decades ahead?
Andi
They really don’t care about most of their forecasts because they will be gone before they are proven right or wrong. It’s the MONEY now they care about and the blessings of their peers and politicians.
faith.
Ray Bradbury had a different take on butterflies, namely causing one to die. This story was published in Collier’s magazine in 1952.
http://www.lasalle.edu/~didio/courses/hon462/hon462_assets/sound_of_thunder.htm
If Wiki is accurate, James Gleick is Peter’s brother. They are brothers of Elizabeth.
http://www.people.com/people/static/mediakit/editorial-gleick.html
I believe James is Peter’s brother
Surely the origin of the butterfly theory is Kipling’s Just So story, “The butterfly that stamped”.
…and so all of the CAGW hype is undone by a single butterfly.
And what exactly makes policymakers so enamoured with the pronouncements of these modelmakers? It seems such a suicidal adherence. And the likes of Hansen and Mann, whose condescending machinations about nothing other than a total fantasy make such great cannon-fodder for even a street-level sceptic? (or great grist for the imagined legions of consensus science, some of whom populate the troll gallery at WUWT). All of this over a nonsensically-small delta-T, of which an even tinier amount “may” be caused by CO2. Totally wacky, I say.
Weather is not climate. To conflate the initial condition dependency problem of weather forecasting with the boundary condition problem of climate simulation is fundamentally erroneous.
How is it conceivable that climate could be forecast years ahead?
It could be that climate and weather are something like Russian dolls from the dynamical systems viewpoint. That is, Lorenz says to expect weather to be a chaotic dynamical system on the time scale of days, which has a predictability horizon of a few weeks at the very best.
On a time scale of months, if you average the weather, your inner Russian doll begins to look like a single object because you can no longer follow the wanderings over the doll at that level. But on the next larger doll (say, ENSO) you may discern wanderings associated with another dynamical system. And it may be that, if you give up on weather forecasting, you could still find predictability of ENSO beyond the weather forecast horizon.
My understanding is that ENSO models show predictability beyond the weather horizon, but at the time scale of more than one ENSO cycle we are currently stumped.
But even that would not necessarily be the end of the story. It could be there is another doll (perhaps PDO) that can be understood as yet another dynamical system, after weather and ENSO are smoothed out by multimonth or multiyear averaging. What the forecast horizon of that might be is probably at the research frontier.
In order to claim climate forecasting skill, not only would these dolls need to be well understood, with a substantial portion of the decadal variability captured in the largest doll; you would also need to prove that GCMs can simulate them correctly. My understanding of Roger Pielke Sr’s and Judith Curry’s views is that we are very far from achieving that.
So skillful decadal forecasts are conceivable, but at this point it’s just as conceivable that, say, PDO dominates decadal variability, that PDO dominates the variability over the last 100 years; that PDO can be described as a dynamical system, but its major shifts cannot be forecast more that a few years before they happen. (Not saying that’s true, just that it can’t be ruled out with our present understanding.)
“If Lorenz gave up when he discovered weather to be totally chaotic and unpredictable beyond a few days, what makes the Climate Modellers believe they can do better forecasting years or even decades ahead?”
__________________
The thing with these models and attractors is, just looking at the famous image (e.g. http://en.wikipedia.org/wiki/File:Lorenz.png) you can see that you cannot predict the exact point where will your system get after just a few circles, but you can predict pretty well the rough volume in which it will reside.
What models are used for is not to determine whether it will rain 100 years from now, but rather to determine the attractors around which the weather is oscillating and where will they move in the years to come.
Their skill in figuring that out is a completely different story but that does not have that much to do with chaos.
jhultquist, Ray Bradbury’s book is the first thing I think of when ever someone mention’s the “Butterfly Effect”
….
…If Lorenz gave up when he discovered weather to be totally chaotic and unpredictable beyond a few days, what makes the Climate Modellers believe they can do better forecasting years or even decades ahead?….
As jack morrow said it is MONEY. But I would also add that it is POWER Dr. Evans explains it very nicely on JoNova’s Website.
See: Climate Coup — The Politics: How the regulating class is using bogus claims about climate change to entrench and extend their economic privileges and political control. http://joannenova.com.au/2012/03/climate-coup-the-politics/#comment-1014902
As evidence a commenter posted:
I would add:
Obama’s Green Jobs Program “Infused with Politics at Every Level: http://blog.heritage.org/2011/12/26/obamas-green-jobs-program-infused-with-politics-at-every-level/
Obama make a remark at one point that working for the government was the best job a person could have or something along those lines but I can not find a link.
Andi,I believe they have a valid answer to that.
Consider a different famous chaotic system, a swinging pendulum dangling from a swinging pendulum.
Although it’s chaotic nature means we cannot model where the pendulum will be beyond the very near future, we are still able to say a number of valid things about it. For example, we can determine the region it may be found within, and the region it can never be found within. We can expilicitly determine it’s location in the third dimension. We can determine the probabilities of it being found within particular bands of the regions that it may occur within.
In the same fashion, climate models can’t determine future weather. Not only does chaos mean they can’t determine snow won’t form in the Sahara desert, it means they can be sure that at some point it will. This is analgous to the bands that the pendulum may be found within. However, they can still say some extremely likely things about the temperature within the Sahara. They can say even more impressive things about the likely average temperature there over time. All this even though chaos does indeed mean they can’t tell for a particular day next decade whether that day will be a scorcher or a snowstorm.
So on this issue, I have to conclude ‘even a Climate Modeller can’t be wrong all the time’.
It’s not the chaotic nature of their models that troubles me. It’s not even my programmers distrust of the accuracy of very large programs. It’s the whole idea that the future can be foretold from the entrails of a computer.
All their models will be confounded if any of a number of possible things happens. Among the many possibilities are nuclear war between Israel and Iran, India and Pakistan, Russia and Georgia, China and India, or the U.S., or France, or England and whoever they deem themselves to be at risk from. Or a cometary impact. A supervolcano exploding. A new method of electricity production being developed. Natural or artificial changes in the biosphere that make it more carbon absorbant. Vastly improved battery storage would have the same effect as developing a new energy generation system. Changes in the economics of renewables. Changes caused by bioengineering or nanotechnology. Any of the unknowns that I haven’t mentioned. We don’t know what the future will be. But we do know the models do not foreshadow it.
Any idiot can forecast disaster with a computer and logarithmic growth, and many already have. Consider the forecasts of Ehrlich or the Club of Rome. Reality has always confounded them. And for the same reasons those earlier models were wrong despite all their claimed science and accuracy, these models too wiill be found wrong.
Of course, by then we’ll have moved on to the next prophesy of disaster. The late, respected G. Harry Stine, engineer and writer, anticipated it would be a fresh water shortage disaster.
I’d suggest enrolling in some advanced math courses at a good university. There’s far more to this than just sensitive dependence on initial conditions. And Lorenz’ work has been greatly expanded on. Univ. of Texas has a good curriculum. Blogs (even this one) cannot provide the depth and breadth of knowledge this subject requires.
As another meteorological novice, I understand your intuitive objectives to AGW “science” (really psudo-science).
Simple observations pointed out the fallacies inherent in making CO2 the sole determinate of future climate.
– It was obvious from my experience with western US evenings (vs. Southeastern US evenings) that water vapor, not CO2 is the real “greenhouse” gas. (Not the CO2 isn’t a greenhouse gas, it is just not very potent compared to the effects of water vapor and clouds.) Even my country bumpkin farmer relatives (with little or no education) knew that clear weather made for colder nights. However, we can model climate (to the 3rd decimal place) without really understanding cloud formation (as CERN experiments made clear) using the concentration of a trace atmospheric gas.
– Ignoring any possible effects of solar variability as being irrelevant to climate seemed ludicrous even though I no inkling of the real mechanisms by which the Sun affects climate.
– Assuming positive feedback mechanisms in any complex system that does not have a history of constantly going to extremes seemed stupid. Without negative feedback mechanisms in place it seemed obvious that any small variability would be amplified to the point of climate catastrophe every week, not just when man got involved.
I could go on but (hopefully) you get the idea.
CAGW is not just incorrect, it is based on so many obvious falsehoods that it should have been laughed at by people with any common sense long ago…except for the fact that it fits the prejudices (and political goals) of a large group of humanity (lets call them the “ruling elite” just for a handy nickname.) This ruling elite hates humanity (with the curious exception of their own group) and wants to punish (other) humans for destroying the perfect world that used to exist (sounds kind of like a religious story, except they HATE religion, go figure). Also, the ruling elite wants to control the lives of others and, oddly enough, CAGW gives them the perfect reason (either submit to our rule or you will be destroyed…Baw Ha Ha Ha!).
I salute those that work so hard to better understand the climate hoping that real science will discredit the pseudo-science of CAGW. However, I think that if facts would settle the argument, it would never have gotten going. Simply thinking about the climate, without really studying it should have made skepticism of CAGW the default position of all thinking people unless there were reasons of self interest for our beloved leaders to accept irrational arguments for taking our freedom (and our money). We will not win this argument with science because it was never about science (IMHO). That was obvious to this “casual” observer from the start.
“If Lorenz gave up when he discovered weather to be totally chaotic and unpredictable beyond a few days, what makes the Climate Modellers believe they can do better forecasting years or even decades ahead?”
=======================================================
I don’t think they actually believe they can. What they believe is the cultist leftist economic and social dogma. They have to come up with something which fits their belief system….. thus climate modeling showing how rational humans, adapting and progressing through history, is evil.
This is an interesting and relevant post for “Climate Change”. Ten or more years ago I read an excellent book on Chaos Theory, which inevitably covered the story of Edward Lorenz and his work on the subject. He was truly a brilliant man who, but for WW2, would probably have become one of the greatest mathematicians of all time, instead of ending up as a meteorologist. (Sorry, Anthony!) I have always thought it unfortunate that the term “Chaos” became attached this theory. “Complexity” or “Complex” Theory seem to me to be more appropriate. In the same way, “Climate” is inappropriate when discussing the Earth’s changing ATMOSPHERIC conditions. In astronomy, for example, scientists never referred to the “climate” of a planet. “Climate” is only appropriate as a general description of regional annual weather conditions, viz “Mediterranean”, “Tropical”, “Temperate”, etc. And in each of these, considerable variations can occur throughout the year and by slight differences in geographical location. I reside in a small region of this planet called the British Isles, and we have great difficulty in keeping track of the variations in our “Temperate Climate” North to South, East to West, hour by hour. If the Earth has a climate, would someone kindly define it for me.
I’ve often thought it would be interesting to run one of these models with 128 bit floating point numbers and then repeat the exercise with 64 bit and 32 bit numbers and then compare the outputs. Any differences would highlight the futility in attempting to use models to predict a chaotic system.
I doubt that weather is completely chaotic in the long term. There are too many cyclic forces at work that are clearly observable. For example, we are in a 60 year ENSO cycle currently. The sun has an 11 year cycle. The duration of the natural cycles are not precise so may vary a bit.
My take is that mathematical equations of gases in the critical region may have zones of instability under certain process conditions. These zones of instability prevent the computer programs from converging. I recall one time when my department overhead shot up sharply one month. We were trying to resolve systems containing carbon dioxide. The Cray computers could not ever converge from one direction so would go the other way, but still could not converge. This back and forth was taking extremely high priced time on the Cray. I hired a retiree to come back and in short order he did the calculations near the zones of instability by hand and created convergence. Mathematics do not always reflect real world processes, especially when using computers to resolve. Computer models do not have as much judgment as knowledgeable humans.
The AGW models are so simplex that all driving forces except carbon dioxide do not have much impact. With carbon dioxide being a second order driver to water vapor, the models are essentially meaningless in the short run and doomed to complete failure in the long run. All they can do is to predict higher temperatures as the carbon dioxide content in the atmosphere increases. Actual observations of measured temperatures over the last 15 year prove that the models are incorrect as far as meaningful exactitude is concerned.
The UK Met Office is on record as saying it was very good at forecasting 24 hours ahead and 30 years into the future, but not very good a fortnight from now.
I guess that’s one of the best descriptions of the supposed validity of climate science I have ever come across.
While the weather may jitterbug around chaotically, the temperatures are still limited by the conservation of energy. If the solar heat input doesn’t change, and the earth is really losing less heat to outer space (possibly because of GHGs), then the earth should get warmer. No telling for sure how the heat is distributed, but there is a tendency for things to mix.
Of course, this is an energy balance and an energy limitation; temperature is not conserved, energy is. If GHGs are adding energy to the earth, then it would not violate the conservation of energy for that to result in lifting the atmosphere further from the surface, without raising the tropospheric temperature. Indeed, energy could still be conserved if the additional energy went to increasing the speed of the earth’s rotation without raising temperatures at all. Theoretically.
I suspect that the Team have simply enlisted large numbers of volunteers to track the movements of all butterflies in Brazil and feed this data into the models.
Simples.
Of course we could model things, but then the sun will decide to have two or three quiet cycles and suddenly everything is covered in ice, or a medium distant star will go nova and all the radiation balance assumptions of the model go right out the window. We live in a dangerous universe, but for sanity’s sake, we don’t dwell on the what-ifs of Alpha Centauri going nova on us and pretty much ending habitability in this region of space (probability ~nil, but the precautionary principle dictates we should take drastic action to see that such a catastrophic event does not occur /snark).
fredb says:
March 15, 2012 at 7:35 am
“Weather is not climate. To conflate the initial condition dependency problem of weather forecasting with the boundary condition problem of climate simulation is fundamentally erroneous.”
True, weather is not climate. However, climate modeling IS an initial value problem and the modeled climate WILL depend on initial conditions. This would be an excellent opportunity to discuss numerical stability and discretizations for system of differential equations, but no one from the warmist side is usually interested in those details…
Constrained infinities and constrained attractors make for good reading. The tip of a bull whip can crack at an infinite number of places but never farther away than the length of the whip. The reference frame can be moved from place to place such that it cracks in Bolivia one day and Tristan da Cunha the next. What actually happens inside the framework (weather) is not so important as what shapes the framework (climate) and what trends are possible. We have billions of years of weather history to ponder but we don’t have a clue as to what drives the climate to change and what it will soon become. We are trying to divine the characteristics and trend of the climate from the evidence given in the weather – learning the cause from observing the effect. We ignore Lorenz’s chaos and lessons of precision of initial starting points at our peril.