Bishop Hill New study finds low climate sensitivity
Jan 25, 2013 Climate: sensitivity Via Leo Hickman, more evidence that aerosols have a small impact and that climate sensivity is low.
When the researchers at CICERO and the Norwegian Computing Center applied their model and statistics to analyse temperature readings from the air and ocean for the period ending in 2000, they found that climate sensitivity to a doubling of atmospheric CO2 concentration will most likely be 3.7°C, which is somewhat higher than the IPCC prognosis.
But the researchers were surprised when they entered temperatures and other data from the decade 2000-2010 into the model; climate sensitivity was greatly reduced to a "mere" 1.9°C.
Professor Berntsen explains the changed predictions:
"The Earth's mean temperature rose sharply during the 1990s. This may have caused us to overestimate climate sensitivity.
"We are most likely witnessing natural fluctuations in the climate system – changes that can occur over several decades – and which are coming on top of a long-term warming. The natural changes resulted in a rapid global temperature rise in the 1990s, whereas the natural variations between 2000 and 2010 may have resulted in the levelling off we are observing now."
I wonder how much more evidence we need of low climate sensitivity before policymakers are forced to take notice?
Reader Comments (82)
I wonder how much more evidence we need of low climate sensitivity before policymakers are forced to take notice?
To get their noses out of the green money trough, "force" may indeed be necessary.
It has been known for some considerable time that there is a ~60-year cycle in the temperature data - for example, if the 30-year trends (necessary to define "climate") of the 150-year temperature data are plotted on a continuous running basis, this cycle is brought out very clearly.
The rising portion of this cycle contributed a significant component of the temperature increase over the period 1980-1998, with a subsequent levelling out. Only now is this starting to be more widely acknowledged as a possible explanation for the rapid rise and post 1998 levelling-off of temperatures. Yet it did not require complex models, or supercomputing capability to establish this underlying (“natural”) characteristic.
Expect a whole new generation of scientists in 20 years saying "we have 10 years of cooling data, and the trend is downwards... it's worse that we thought... we're all going to freeze unless we do something about it... we have 50 days to save the planet"
Then around the turn of 2100, a new generation of climate sscientists will say the cooling has levelled off, and now there's a warming trend and we're all going to die...*
*I'll be long dead by then
Odd really: no warming from 2000-2010, yet significant rise in CO2 level - but still they suggest quite a high climate sensitivity!
The idea that computers can model the climate is completely preposterous.
Yes, climate 'scientists' will write as if they can, but they can't. All they are doing is providing the best estimate they can. That is valid science.
What isn't valid is completely uneducated lifestyle correspondents like Leo Hickman commenting on it.
That goes for the other side too. Science based on politics is utterly witless. Anyone who does so should get a new hobby.
My angle is that the media promote AGW because they are promoting carbon trading for their advertisers. In the Guardian's case, Shell. That is the fundamental issue and why we read such drivel from Bob Ward who is sponsored by a $100 billion hedge fund owned by his sugar daddy, Jeremy Grantham.
"I wonder how much more evidence we need of low climate sensitivity before policymakers are forced to take notice?"
30 ft of ice covering London in July before my MP, Nick Clegg, will even admit there is another side to the argument
I don't really get on with the idea that models can't model climate, an oft repeated refrainin here.
I think they CAN, but I don't think any of them DO yet.
When modellers say that a model is "modelling the climate" what they mean is the model expresses their own current theories and ideas about how the climate works. They still have to feed in these rules into the model. So a model models a theory about climate, not the climate itself.
It sounds nitpicking, but it is very important to make the distinction.
When the model approximates reality in the past (hindcasts) then they use this as validation that the theory is correct, but that's tautological - the theories were built from past observations, so it's not unusual or surprising (or even clever) that any model built using those theories then manage to calculate the past observations. As a validation tool, it's dodgy at best.
So when we criticise models, be careful not to criticise Modelling as a technique, which is a valid way of expressing theories, especially those which don't fit into static algabraic equations.
There is a paper from this group a year ago that says that S is around 2.0. So it's not clear why this is news. It uses uniform priors (eek!) but they do investigate the effect of varying this (see sec 4.6 where they try a prior uniform in 1/s).
Interestingly, this low-sensitivity paper is not cited in IPCC AR5. [Correction, it is in the figure, page 12-153, but they forgot to put it in the list of refs. And it is cited in Ch 10].
That paper doesnt have the bit about results changing when you add 10 yrs of data. I asked Berntsen and he said the news release was based on a new paper they are still working on so the numbers may change.
TheBigYinJames
Climate sensitivity was greatly reduced to a "mere" 1.9°C.
Ha ha ha ha !!
Was there a WMP ? Don't know.
Was there a Roman warm period ? Don't know.
What is the effect of clouds ? Don't know.
Is water vapour a positive or negative feedback ? Don't know.
Does temp follow CO2 or reverse ? Don't know.
How accurate are proxies ? Don't know.
What is the temperature above the oceans in the past ? Don't know.
How much do cosmic rays affect the climate ? Don't know.
The CERN experiments show a perfect correlation of cosmic rays and temperature. Jasper Kirkby is a real scientist, a physicist, not some jumped up idiot of a environmentally obsessed climate modeller .
Watch the video
http://cds.cern.ch/record/1181073
Can computers model the deeply interacting and feed backing behaviour of the above ?
No.
Can computers predict future temperatures ?
No.
Freeman Dyson says it's all bollocks
Ivar Gyvar (nobel prize, physics) says it's all bollocks
Even James Lovelock says it's all bollocks.
Every physicist on earth knows it's bollocks.
Here is a commentary of one climate scientist on another's work. Pielke Senior on Gavin Schpidt.
Let me translate it.
"Let's meet in Cambridge." "Great. That's in France, isn't it." "Not really" or
This is a ball - no it isn't, it's a corner flag or
This is a car - not to me it isn't. It's a horse.
http://pielkeclimatesci.wordpress.com/2012/01/30/comment-on-gavin-schmidts-post-on-his-weblog-real-climate-regarding-the-dominate-role-of-anthropogenic-greenhouse-gas-concentrations-on-the-global-average-temperature-trends/
"The idea that computers can model the climate is completely preposterous."
I prefer chicken entrails to computer entrails for telling the future.
When the Perfesser said
the obvious question is, How can he be sure that this very, very slightly different statement is not true?
* I'm sure we all know what the issue is with his use of "long term" to describe 20-odd years so I'll refrain.
@ TBYJ
I dunno. Can you build a computer model of, say, how 5-year-old children will behave in 100 years' time?
A child seems like a fairly simple system to model to me, not complex at all compared to the climate in 100 years' time. I'd think 5-year-olds in 100 years' time will be much like those now. But it still doesn't seem possible to me.
Paul Matthews (4:12 PM) -
You cite (and link to) Aldrin et al. (2012), mentioning that its mean estimate of climate sensitivity is around 2.0 K. Aldrin is one of those mentioned by Nic (see prior post) as using a uniform prior; in fact their uniform prior is wider than most, being flat from 0 to 20 K. (!)
.
Their figure 6a shows the posterior distribution with this uniform prior: mean of 2.0 K, 90% interval 1.2 to 3.5 K. Figure 6f shows the result with a 1/S prior (which Nic says is the Jeffreys' prior): mean of 1.5K, 90% interval 1.1 to 2.5K. [by eye] The mode doesn't change much -- from ~1.6K to ~1.4K (again, by eye).
There's almost no discussion of Chaos Theory in the blogosphere. Read what Benoit Mandelbrot writes about the mathematical absurdity of seeking trends in a fractal phenomenon.
It staggers me that gravy-train science has become so unethical that it claims to be able to "moddoo" the future of climate. They MUST know they're doing it. Scoundrels. Latter-day soothsayers, curse their black hearts.
It's about time that the hard sciences dissociated themselves from the climate charlatans. The longer they remain silent the greater the likelihood that the profession of "scientist" will attract the same scorn and derision as bankers and double glazing salesmen.
Standard modelling practice in the natural sciences is to calibrate a model with recorded data and then validate the model with data that was not used for calibration. So hindcasting is used as part of the calibration process and forecasting provides an opportunity for validation.
So while it appears that climate models are calibrated based on their ability to hindcast, they have clearly not been validated. If a model does not meet the validation test it means that the calibration is spurious.
This is deeply suspicious.
CERN 'gags' physicists in cosmic ray climate experiment
The chief of the world's leading physics lab at CERN in Geneva has prohibited scientists from drawing conclusions from a major experiment. The CLOUD ("Cosmics Leaving Outdoor Droplets") experiment examines the role that energetic particles from deep space play in cloud formation. CLOUD uses CERN's proton synchrotron to examine nucleation.
CERN Director General Rolf-Dieter Heuer told Welt Online that the scientists should refrain from drawing conclusions from the latest experiment.
"I have asked the colleagues to present the results clearly, but not to interpret them," reports veteran science editor Nigel Calder on his blog. Why?
Because, Heuer says, "That would go immediately into the highly political arena of the climate change debate. One has to make clear that cosmic radiation is only one of many parameters."
http://www.theregister.co.uk/2011/07/18/cern_cosmic_ray_gag/
Odd that by adding a further ~10 years of data that calculated sensitivity drops. Does that not tell them that either they are not calculating sensitivity at all or that it is a variable?
Validation of a model by hindcasting is no validation at all just as a mimic may be able to do a good impression of the voice of another but the mimic cannot predict what that voice will say next.
Brent Hargreaves
as a double glazing salesman I take great exception to being compared to a climatologit - but otherwise you have hit the nail on the button!
The climate does not know what it is going to do in 6 months, 6 years or 60 years so quite frankly it is laughing its sock off at those who think they can predict its future.
TBYJ:
"I don't really get on with the idea that models can't model climate, an oft repeated refrainin here.
I think they CAN, but I don't think any of them DO yet."
'Fraid you've got that one wrong TBYJ, the climate is chaotic, by which I am led to understand that extremely small changes in the initial input to the system can result in massive changes in the output. The butterfly effect from chaos theory. Even the IPCC admits it, it's just never put it into a SPM as far as I know, which is why assorted politicians believe that the IPCC is giving them accurate predictions made by scientists, so they must be right, when all they're getting is a set of scenarios prepared by scientists who believe that CO2 is the root of all evil, and worse yet, if they were to say it wasn't their funding would dry up.
” … In climate research and modelling, we should recognise that we are dealing
with a coupled non-linear chaotic system, and therefore that the
long-term prediction of future climate states is not possible.”
IPCC TAR, Section 14.2 “The Climate System”, page 774.
I'll bet they're sorry they ever wrote this now, because it will come back to bite them when the politicos realise they've been duped and the mob are at the gates begging for energy to heat them in the winter and baying for blood.
What do people mean by low or high here?
I'd say the range -1C to +1C spans low sensitivity.
Absolute values above 1 and up to say 2.5 would be medium .
Above 2.5 would be high.
Just my subjective impression. If I were to posit a subjective prior for S, it would span some negative values. My ignorance is large though.
geronimo,
There's a myth you can't model chaotic systems. Of course you can, you use multi-run distributions and monte carlo. Also, just because a system has chaotic elements, doesn't mean any particular metric, e.g. temperature, behaves chaotically. If it was truly chaotic then the temperature would be moving from 0K to infinity and back again on a random walk. It doesn't because even chaotic systems have bounds.
Take an example - drop a ball from a tall building and plot where it ends up. Because of many chaotic aspects, e.g. wind, exact trajectory, bounce surface, etc that you can never model absolutely, the ball never falls in the same place twice. But it does fall in a range of locations with a particular probability curve, and this has finite bounds. While this won't tell you where any individual ball will land, you can say how many balls land in which areas for a large number of balls.
Not arguing that current models are right (I was arguing the contrary), but chaos in climate is similar - just because some aspects are chaotic doesn't make the system as a whole not able to be modelled - it just means you have to use probability to express outcomes. If many chaotic elements cancel each other out in the macro, you can get behaviours.
Yeah but..nobody can do it now, and there is no guarantee that anybody can do it ever. And wouldn't it need about a thousand years of validation? Or if not, how many years? Must be more than the seventeen years they are saying is not long enough for a trend.
Chief Scientist, Met Office
This only means that climate sensitivity is not a good predictor of climate behaviour and that we are missing the point regarding the main drivers of climate variability.
If we look at the amount of attention climate sensitivity gets in relation to other physical parameters, using climate sensitivity after the results of this experiment, is really like predicting planetary orbits using a seasonally changing gravitational constant.
Some time ago Lubos Motl in an almost throwaway suggestion on his blog pointed out that if the GHE was amplified by some simple multiplicative sensitivity factor, that factor applies to water vapour as much as to CO2. The more or less stable 30 degrees in the real world is attributed to water vapour. Clearly empirical observation has ruled out a GHE adding 100 to 150 degrees to this (i.e. x 3 or more). Hence any feedback coeficient is probnably low. Can somebody explain what's wrong with this argument (and I apologise in advance if I have misrepresented Motl's argument, which I am quoting from memory.)
I sometimes wonder whether climate sensitivity itself is not a mere conjecture. It is intuitively obvious that it has a value in local immediate terms, but not that there is some magic way in which all the local sensitivities, influenced as they are by latitude, time of day, time of year, ocean cycle, weather and a host of other things, can add up to an average climate sensitivity that is good for anything except self-deception.
Millikan effect, anybody?
'Expect a whole new generation of scientists in 20 years saying "we have 10 years of cooling data, and the trend is downwards... it's worse that we thought... we're all going to freeze unless we do something about it... we have 50 days to save the planet"
If we run the windmills as fans will that do the trick?
"just because some aspects are chaotic doesn't make the system as a whole not able to be modelled - it just means you have to use probability to express outcomes. If many chaotic elements cancel each other out in the macro, you can get behaviours."
I would tend to agree, and note that this makes the habit of "throwing out" certain runs as "unrealistic", based on pure subjective criteria, problematic - throwing away "low" runs (where we have "stalled" or even dropping temps) and keeping "high"runs (where the temp goes up by 10+C) - skews (distorts) the probability curve towards disaster. Given the current "stalling", it would certainly be interesting to see what putting those runs back in does to the probability density curve.
There's a myth you can't model chaotic systems.
They don't want to hear you. They want "chaos" to be a magic bullet that rubs out any possible modelling.
Thanks for pointing out that even "chaotic" systems can be modelled. In my world the seasons roll around in a predictable pattern, suggesting that any underpinning chaos is not dominant over the main driving forces. Winter is always significantly colder than summer.
(I agree that the climate can't be modelled yet, but not because of – unproved – allegations that is chaotic at heart..)
"There's a myth you can't model chaotic systems.
They don't want to hear you. They want "chaos" to be a magic bullet that rubs out any possible modelling."
Not sure if your point Mooloo. Are you saying we can model chaotic systems now? Or do you have a peer reviewed article that shows us how we can do it? Or us the IPCC wrong?
Moreover the notion that you can foretell the future because you can tell that summers are hotter than winters is.... Well daft.
Just to get it straight, I don't think any of the current models actually model the climate, but I'm also acutely scared of the idea that we throw out all good ideas with the bathwater. Modelling is a fantastic way of expressing theories. Just because the current theories are incorrect, doesn't mean we should throw out the idea of modelling theories. Models are the betes noires of climate skeptics, but they are just the physical embodiments of theories, and it's those we have the beef with, not the models.
I think TheBigYinJames has a valid point --especially what he has said at 10.57 PM.
But somehow TBYJ you / we have to convince the MSM , policymakers and Joe Public of what you're saying because are right that use of models is a valuable tool if used correctly ( or more importantly the results are used properly --to further the development of the theory NOT to make policy costing billions of dollars before the theory is anywhere near proven to have some substance)
There is nothing wrong with computer models. Scientists in the same field understand their limitations, but the average Beano or Monbiot reader doesn't.
The admirable Chris Snowdon has written about the phenomenon of policy based evidence making extensively here
Expect much more along the lines of "well, we should be reducing the use of fossil fuels anyway, because of peak oil / energy security / fill in as required..." intention, of course, to deflect from the flawed SCIENCE (capitals - policy wonks' own)
BigYin
Yes...as someone who has built a lot of models, it is incumbent on the modeller to state the limits of the models, what the risks and unknowns are. Then someone with greaster knowledge can decide what to do. If my model says that the best way to get from London to Manchester is to go via York, because of the data fed into the model, the modeller should point out that the geographical straight line is different, and the great circle route too. A model is dependent on the data and assumptons that feed it, plus the intelligence of someone who might want to correct its short-comings.
Alan Kennedy
A lot of the mechanism of climate sensitivity is the effect of CO2 induced warming on water vapour. The resulting higher atmospheric water vapour amplifies the warming. The change is driven by CO2 rather than water because water vapourhas a short dwell time and water content responds rapidly to short term weather conditions. CO2 has a long dwell time, and has a correspondingly long term effect on atmospheric average water content. A widely quoted figure is a 4% rise in water vapour content per 1C of atmospheric warming.
http://www2.cgd.ucar.edu/hurricanes
Not yet pinned down are the exact effect of this on low and high altitude cloud cover, and their warming or cooling effects. Low altitude cloud has a net cooling effect in daylight by reflection and warming by night as insulation. Cirrus has a net warming effect. Both are presently under study to quantify their effects.
The climate is complex both in the number of variables and its non-linear behaviour. It is not beyond understanding.
"I wonder how much more evidence we need of low climate sensitivity before policymakers are forced to take notice?"
That is subject to changes in the public opinions of swing voters that various political parties are targeting. It was never much about evidence.
Perhaps this might clarify the way climate models operate.
Consider a pendulum. A simple pendulum's behviour can accurately be predicted using t=(2pi * square root L ) g . L is the length and g is the strength of gravity.
Given the starting position you can directly predict its position at any future time by a simple calculation.
Now consider a double pendulum, with two weights and hinges top and centre.
http://www.myphysicslab.com/dbl_pendulum.html
At small angles of swing this pendulum behaves classically and its future position can be predicted by simple calculation.
At larger angles of swing the pendulum behaves chaotically and its future position can no longer be predicted by simple calculation. Instead it must be modelled. Input the equation describing the physical behaviour of the pendulum and the starting position. Starting with the initial position you calculate the position after a short time. That is then used for a second calculation, whose answer provides the input for another calculation. You have to do this by repeated steps until you reach your future time.
Climate models work like a chaotic pendulum model. You put in the physical equations governing the system's behaviour, add a set of starting conditions and let it run. Repeated runs with different starting values and values for other variables allow you to gain understanding of how the model, and by extension the actual climate behave. It also makes clear the shortcomings of your model. The process is not easy, but it is not impossible and it is not meaningless.
Here we go again with the natural fluctuations excuse.
Climate science is itself following the First Law of Climate: "it's worse than we thought".
I feel like I'm watching rank amateurs here.
Did they just work out that here _are_ natural fluctuations?
Entropic Man...
I see you have accepted the proposition from Hansen and co. that water vapor completely precipitates out of the atmosphere whenever it rains. This is of course nonsense since water vapor is never completely removed from the air due to rain. Warm moist air will produce rain as it cools.The relative humidity level varies with temperature mostly (and some other factors). It never goes to zero and is always at a concentration level several times that of CO2. Often times in regions such as the tropics, water vapor has a concentration level a hundred times (or more) that of CO2. It stops raining when the equilibrium value of humidity is reached for the present temperature of the air. There is no time that air has a higher concentration of CO2 than water vapor, ever. Not even the polar regions have zero levels of atmospheric water vapor. Scientists that propose this theory should present a time history grid of the atmosphere stating the level of water vapor in the air at all altitudes and locations to show that water vapor is not present sometimes. This should be easy as the aviation field makes continuous measurements of all of the factors that concern the issue. Weather balloons and such you know. I pilot a plane often, and the weather reports I have seen have never had a humidity reading of zero (or even close to zero) at any altitude at any time for any place.
On a further note, I believe that even if water vapor did completely precipitate out, it would not have the effect that is proposed. Water vapor would be still be present most of the time. Since it accounts for the vast majority of the GHE, the time that it would be in residence would still cause it to contribute most of the GHE. To me an analogy would be if there were two heaters in a cold room, one very large and the other small. If the small one was running continuously and the large one only ran 90% of the time, the large one would still produce most of the warming. The idea that CO2 is some sort of "control knob" is nonsense. Heck, the other GHGs (methane, ozone, etc.) are in the air constantly too. No one seems to think they are the climate "control knobs".
TBYJ I've heard the arguments for the dropped ball before, also that you can't tell what the weather will be like in the summer but you can tell it's going to be warmer. Note that both are capable of being implied because there is empirical evidence from experience/experimentation. No such evidence/experimentation is available in models, which themselves have to be flawed because of our imperfect understanding of the climate. So even multiple model runs only give us what they call "scenarios".
Entropic Man
Thanks for the helpful explanation. I now have another problem. If your account is correct, and given water vapour is a powerful GHG, the setup you propose would obviously be very unstable indeed. It doesn't seem that way in reality. Isn't that an issue?
TBYJ: First off let's agree you believe that chaotic systems can be modelled, and that modellng is a valuable tool. Not sure about the chaotic systems bit, but agree 100% that provided you have all the real world data to hand modelling, say the design of aeroplanes, has reduced the development costs massively. The problem with your statement about modelling the climate, for me at least, is that it gives rise to the sort of comment we have had from mooloo.
"They don't want to hear you. They want "chaos" to be a magic bullet that rubs out any possible modelling."
Implying that "they" are somehow or other wishful thinkers who don't want to believe the outputs from perfectly good models foretelling the future. Here's what Kevin Trenberth had to say about the models on his nature blog on 4th June 2007, keep in mind people like mooloo want to bet the ranch on the outputs of these models:
"... In fact, since the last report it is also often stated that the science is settled or done and now is the time for action.
In fact there are no predictions by IPCC at all. And there never have been. The IPCC instead proffers “what if” projections of future climate that correspond to certain emissions scenarios. There are a number of assumptions that go into these emissions scenarios. They are intended to cover a range of possible self consistent “story lines” that then provide decision makers with information about which paths might be more desirable. But they do not consider many things like the recovery of the ozone layer, for instance, or observed trends in forcing agents. There is no estimate, even probabilistically, as to the likelihood of any emissions scenario and no best guess.
Even if there were, the projections are based on model results that provide differences of the future climate relative to that today. None of the models used by IPCC are initialized to the observed state and none of the climate states in the models correspond even remotely to the current observed climate. In particular, the state of the oceans, sea ice, and soil moisture has no relationship to the observed state at any recent time in any of the IPCC models. There is neither an El Niño sequence nor any Pacific Decadal Oscillation that replicates the recent past; yet these are critical modes of variability that affect Pacific rim countries and beyond. The Atlantic Multidecadal Oscillation, that may depend on the thermohaline circulation and thus ocean currents in the Atlantic, is not set up to match today’s state, but it is a critical component of the Atlantic hurricanes and it undoubtedly affects forecasts for the next decade from Brazil to Europe. Moreover, the starting climate state in several of the models may depart significantly from the real climate owing to model errors. I postulate that regional climate change is impossible to deal with properly unless the models are initialized.
The current projection method works to the extent it does because it utilizes differences from one time to another and the main model bias and systematic errors are thereby subtracted out. This assumes linearity. It works for global forced variations, but it can not work for many aspects of climate, especially those related to the water cycle. For instance, if the current state is one of drought then it is unlikely to get drier, but unrealistic model states and model biases can easily violate such constraints and project drier conditions. Of course one can initialize a climate model, but a biased model will immediately drift back to the model climate and the predicted trends will then be wrong. Therefore the problem of overcoming this shortcoming, and facing up to initializing climate models means not only obtaining sufficient reliable observations of all aspects of the climate system, but also overcoming model biases. So this is a major challenge."
So it isn't me who's saying they're bunkum, it's Trenberth, he is agreeing that they don't/can't get enough observations to build an accurate model, I believe he's asking for more funding.
Oh and mooloo, "next summer will be warmer than it is today on average" doesn't prove I can foretell the future state of a chaotic system unless I can give a day by day forecast of the weather, temperatur is only one parameter and easily forecast. I really don't understand how the alarmists got to believe that they could kill the discussion on chaos and climate with such a fatuous analogy.
It is impossible to predict whether the climate will be warmer cooler or just the same in 2100. Models can and do produce scenarios all based on the notion that CO2 is some sort of driver and ignoring the fact that there are vast areas of climate science that are simply not understood. If anyone is in doubt about this check the levels of scientific understanding in AR4. And that is before we get to the processes that we don't even know about yet. Add to that the difficulty (impossibility) of capturing the starting conditions and it just becomes a big and very expensive guessing game. I know some claim that even models that are always wrong are useful. I would like to see a cost benefit analysis to support such claims. They are clearly useful at modelling weather, but some argue that Piers does it better on his Sinclair Spectrum.
Geronimo I had not seen your post of 8.21 am when I crafted my own meagre rejoinder which is rendered otiose.
Chaotic Systems
There has been some Can! Can't! discussion on chaotic systems. Isn't the reality as follows?
1. Chaotic systems can be modelled in general. A proper model of a chaotic system will itself display chaotic behaviour.
2. A model of a chaotic system is useless for predicting its long term future states.
Jan 25, 2013 at 9:48 PM | Kneel:
"I ... note that this makes the habit of "throwing out" certain runs as "unrealistic", based on pure subjective criteria, problematic - throwing away "low" runs (where we have "stalled" or even dropping temps) and keeping "high"runs (where the temp goes up by 10+C) - skews (distorts) the probability curve towards disaster"
I agree. Also, the models use low pass filters between integration steps in order to preserve "stability", and even stop/reset/restart techniques when physical laws (conservation of mass, momentum and energy) are violated. Furthermore, errors (deviations from reality) accumulate exponentially in a time step integration.
I agree with the IPCC:
"In climate research and modelling, we should recognise that we are dealing with a coupled non-linear chaotic system, and therefore that the long-term prediction of future climate states is not possible.” From the 3rd IPCC report, Section 14.2.2.2 “The Climate System”, page 774