Diminishing Returns From Multi-Decadal Global Climate Model Simulations
By Dr. Roger Pielke Sr.
I have posted that the NSF is funding grants which as part of (or all) of their focus is to provide multi-decadal global and regional climate model projections; i.e. see
The NSF is also perpetuating an erroneously narrow view of the climate system, as I posted in
These claims and projections are based on global climate models.
Judy Curry, on her weblog Climate Etc has very effectively summarized the diminshing scientific responses from the use of these models in her post
Decision making under climate uncertainty: Part I
where she wrote
“So it seems like we are gearing up for much more model development in terms of higher resolution and adding additional complexity. Yes, we will learn more about the climate models and possibly something new about how the climate system works. But it is not clear that any of this will provide useful information for decision makers on a time scale of less than 10 years to support decision making on stabilization targets, beyond the information presented in the AR4.”
I agree with this viewpoint. This culture of using models as the tool to communicate to policymakers is an inappropriate and misleading use of the scientific method. I also discussed this misuse of models in my post
Comments On Numerical Modeling As The New Climate Science Paradigm
where Dick Lindzen is quoted
“In brief, we have the new paradigm where simulation and programs have replaced theory and observation, where government largely determines the nature of scientific activity, and where the primary role of professional societies is the lobbying of the government for special advantage.”
Hopefully, the NSF (and other agencies) will soon realize that most of this funding is a waste of taxpayers money and could be better spent on other research uses in climate and elsewhere.
Max Hugoson says:
November 6, 2010 at 8:02 am
I do not doubt that human activity affects local climate as in cities and their environs or even that agriculture, irrigation etc. can affect climate on a regional scale: what I do doubt is that this any perceptible effect globally.
I have no use for these models, because unlike a weather model where, more or less, the forecasters get the result of their predictions within a day or so and can adjust their techniques accordingly, learning what works and what does not on the way as it were, climate models cannot be tested. And an ability to hindcast for which they are tweaked and twiddled is no guarantee of their predictive power which turns out to be zero. As somebody sagely observed if the models worked then you would only need ONE model.
Moreover few seem to understand that whichever way you care to look at it, from classical physics, to mechanistic assumptions to information theory the sheer scale of the repetitive data process handling in these models means that the degradation and increase in noise with every step forward they could not predict beyond the short term, years, if that.
Some scientists are fond of saying that due to the models we know more about the climate fifty years hence than a few years ahead, either they are woefully ignorant of physics and mathematics, or being deliberately disingenuous.
As to clouds, here in the hills over which I love to walk there is one, about 1300 ft which has a nearly sheer drop into the sea on one side. On a sunny. warm, windless day some years ago, with the sea like a mirror, I watched a cloud form on it’s seaward side from about two miles away. It started as tiny puff of white the size of a child’s hand halfway up the hillside and grew upwards and outwards at a terrific rate so that within half an hour or so it had become a massive cloud formation towering up to many thousands of feet. It was fascinating to watch.
Kindest Regards
Mike says:
November 6, 2010 at 8:41 am
Let’s cutoff funding for cancer and AIDS research. I mean what’s the point, we are going to die anyway.
Since you want us all to return to the Stone Age in a desperate attempt to make you feel good about being alive, Mike, why don’t you try it yourself first?
The graph implicitly assumes that effort is expended in an attempt to produce greater accuracy. This may have once been the case, but is not, now. Most latter-day modeling is focused on getting the politically correct answer, so accuracy has for some time been headed back down towards the abscissa in a “death spiral.”
JPeden says:
November 6, 2010 at 3:14 pm
Since you want us all to return to the Stone Age in a desperate attempt to make you feel good about being alive, Mike, why don’t you try it yourself first?
——————–
JPeden,
: )
John
It seems that there is a straw man argument implicit in this. Putting more funding in for bigger and better models does not mean that climate scientists are abandoning observation of the real world. The problem is that it will be hard to find funding for either sort of initiative.
After the fiasco of the Lindzen and Choi paper, where Lindzen ignored a lot of real world data which showed positive feedback of clouds, used the wrong models, ignored the horizontal flow of energy due to winds in his study of whether data and climate models agreed, Professor Lindzen hardly qualifies as an authority on the validity of climate models. In fact, the data that Lindzen had showed agreement with the commonly used climate models, contrary to the conclusion that he came to in his paper with Choi.
Lindzen claims he has written a correction to the paper which he is trying to publish.
http://climateprogress.org/2010/01/11/science-lindzen-debunked-again-positive-negative-feedbacks-clouds-tropics/
” The LC09 results are not robust….
LC09 misinterpret air-sea interactions in the tropics….
More robust methods show no discrepancies between models and observations….
LC09 have compared observations to models prescribed with incomplete forcings….
LC09 incorrectly compute the climate sensitivity.”
I’d hate to have these folks managing my investment portfolio.
Come to think of it, maybe they are managing it!
Ale Gorney says:
November 6, 2010 at 12:54 am
This stie is the defintieiotn of dimishing returns.. I watch it and observe, calculate and know.. damn all of you. Looking back it was easy.. agree? *SHAKING MY HEAD*
IT IS EASY
Here we are…. together, hands joined to fight the good fight but will you remember ? 3 months from now… will your thoughts be lucid of what you see and KNOW?
REMEMBER WHAT YOU SEE.
========================================================
The preceding public service announcement was brought to you by
The Friends of Prop 19
MartinGAtkins says:
November 6, 2010 at 5:00 am
The entire world would be immensely less confused if they understood that there can be no scientific predictions without reasonably well-confirmed hypotheses and statements of initial conditions. So, let us all recite as follows: I will not use the word ‘prediction’ unless I reference the hypothesis and initial conditions that were necessary to make the prediction. When you are hankering to say you have a prediction but no hypothesis or no set of initial conditions then use the word ‘hunch’. You can have good hunches. I am sure the climategaters have lots of good hunches. But no mountain of hunches attains the level of a single hypothesis.
Ellecent idea listed earlier, stop funding useless computer modelling and use funds to correct problems with the weather stations, especially the high arctic stations. Now the team would have us believe they care about data accuracy, so they would volunteer enmass to staff the reopenned stations, right? Or perhaps as court ordered community work these could be their new homes until they learn the scientific method.
Sorry post normal spelling is contagious, Excellent was the word..
You sure that curve isn’t supposed to be bell shaped? I bet anything that the returns become negative rather quickly after a certain amount of effort is lodged. I mean, misunderstanding the climate system is certainly a negative return, isn’t it?
MartinGAtkins
————–
How many times do we hear from the disciples of AGW that climate predictions are more accurate than weather predictions (or words to that effect)?
No one seems to challenge this concept. It’s just stated as fact, even though it can’t be tested.
————-
How do you know it can’t be tested when you have not even bothered to look into the matter.
Mike says:
November 6, 2010 at 8:41 am
Let’s cutoff funding for cancer and AIDS research. I mean what’s the point, we are going to die anyway.
———————————————–
I actually agree. I mean, with the we are going to die anyway mentality. When are we finally going to defeat death itself? Never, of course.
The major difference in this piss poor analogy is the fact that when you are trying to defeat death by curing diseases, these diseases don’t always prey on people that were going to die anyway. Cancer kills children, and so does AIDS. It would be humane to have a cure that would treat the young who are stricken with these diseases. When you try to crush an economy because of a perceived threat from climate change, the economy that funds the disease research Mike is ‘concerned’ with, then you are really just exacerbating the problem. So, Mike, if AGW was found to be an overblown scenario designed to secure nothing more than funding, look at your backhanded sarcasm and ask yourself, who does the money really get taken away from?
I read this and the background material and it did not make sense. It looks distinctly ambiguous and bad expressed or reasoned.
The models already capture most of the physics but still have problems with regional scale climate and decadal scale cycles. Getting the models to deal with cases like this with some skill is going to take a lot of effort.
In a sense these claims are stating the obvious. However it also shows a lack of imagination since the payoff of being able to deal with things like regional droughts which can have an enormous economic impact could be significant.
Theo Goodwin says:
November 6, 2010 at 6:37 pm
The entire world would be immensely less confused if they understood that there can be no scientific predictions without reasonably well-confirmed hypotheses and statements of initial conditions. So, let us all recite as follows: I will not use the word ‘prediction’ unless I reference the hypothesis and initial conditions that were necessary to make the prediction.
I don’t disagree with you as such, but we were talking about computer models and not a specific hypothesis.
A hypothesis must have predictive qualities as one of the conditions for it’s conformation. You don’t known if all aspects and components of the hypothesis are met until conventional experimentation or field research enhances the hypotheses.
A model can’t tell the researcher if a component or condition has been overlooked or wrongly parameterized.
Computers are fine for complex equations and visual enhancement but they can’t confirm a hypothesis.
LazyTeenager says:
November 6, 2010 at 10:22 pm
MartinGAtkins
No one seems to challenge this concept. It’s just stated as fact, even though it can’t be tested.
————-
How do you know it can’t be tested when you have not even bothered to look into the matter.
It’s not that that I can’t be bothered to look into it. It’s just the fact that we haven’t yet developed time machines.
LazyTeenager says:
November 6, 2010 at 10:46 pm says….
..” The models already capture most of the physics…”
Dear me why not all of the physics? Is there some physics you wish to leave out and if so why? And incidentally do they also incorporate the complex chemical and biological processes as well?
No doubt you, as an expert in these things, will be able to explain to us unenlightened souls how all this is incorporated into a model.
And further why the laws of thermodynamics, whether realised in terms of entropy or uncertainty somehow do not apply to these models.
And moreover, given your imaginative insight about how the results of these models might be used, can explain to us ignoramuses, just what we ought to do with the results of the model’s output, bearing in mind the limits of their predictive power, which we can demonstrate rather well is zero, quite why or what we should do about it?
Kindest Regards
“The models already capture most of the physics but still have problems with regional scale climate and decadal scale cycles. Getting the models to deal with cases like this with some skill is going to take a lot of effort.”
I suppose that depends on what you mean by “capture” and “most of the physics”. What nobody ever wants to discuss are questions like: (1) what partial differential equations are being solved? (2) What are their initial and boundary conditions? (3) what specific numerical methods are being used to time-march these equations to obtain a time-dependent “solution”?, (4) what is the order of accuracy of the numerical discretizations?, (5) what stability criteria are used to control the time step?, (6) how do you know (or ensure) that a numerical “solution” for a time 50 years from now isn’t corrupted with numerical noise (due to truncation errors and numerical stability issues) and is essentially garbage?, (7) how to change the surface boundary conditions as a function of time and how do you know your approximations are valid?, (8) do you have proper documentation related to questions (1) – (7)?
I find that many modeling organizations fail miserably on question (8)…
Everyone talks about computer models as if they are these mysterious black boxes, too complicated to analyze from an outsider’s POV. All numerical computer models are based on two things, and two things only: variables and constants. Variables are what you input. Constants “this number and/or this rate of change is always so” are what the variables work against to give you a result. No matter how large and complex you make the variable arrays, this variables vs. constants crunching is all a model can do. After all, computers can only do one thing; they can compare two numbers. And, they are really bad at subtraction!
Why no one has looked into the inner workings of these models is beyond me. I’d love to look at the constants :).
Not dead center on topic, but certainly collateral.
http://www.telegraph.co.uk/comment/columnists/charlesmoore/8116595/What-the-Green-Movement-Got-Wrong-Greens-come-to-see-the-error-of-their-ways
“”””” LazyTeenager says:
November 6, 2010 at 10:46 pm
I read this and the background material and it did not make sense. It looks distinctly ambiguous and bad expressed or reasoned.
The models already capture most of the physics but still have problems with regional scale climate and decadal scale cycles. Getting the models to deal with cases like this with some skill is going to take a lot of effort. “””””
Well I presume from your assertion that pretty much everything between “Big Bang” cosmology, and “Quantum Chromo-dynamics” is included as “most of the Physics”
I suspect they just dismissed all of “Archeo-Physics” which is all the really interesting stuff that happened in the first 10^-34 seconds after the BB.
I’d be really happy if they only included increments of time down to say one time zone; 15 degrees of Longitude or one hour of time. So that would be 24 different orientations of the planet each day throughout the year. Too much to expect that they would go down to say one degree of earth rotation; or how about one half of one degree; which is the angular diameter of the sun, as seen from earth.
Optical Mouse Navigation algorithms are easily able to do sub pixel interpolation; so these mostly all of the Physics models should be able to follow Mother Gaia’s solar blowtorch continuously as it encircles the globe endlessly.
Funny that you can include nearly all of the physics, and still get the clouds wrong. They would seem to be one of the more obvious parts of the Physics. Well if they can’t even get their weather stations properly located on land; who knows how flaky their models are.
I wonder if they keep track of each of the known possible Isotopic GHGs; such as H2O, HDO, D2O and of course each of those with 16O and 18O, and don’t forget all the different species of CO2 since we know which ones are natural, and therefore good; while the others are man made, and therefore bad.
Since the diffusion rates of molecules are presumably different for different isotopes; the atmospheric energy exchanges, must surely be different for each; different spectral lines and different line broadening effects etc.
And the beautiful thing about computers, is they really don’t car much how many variables you throw into the mix; it is just more simultaneous non-linear partial differential equations to solve.
The good thing about being stuck on a desert island with just a sandy beach and a stick to scratch with, is that you can dispense with a lot of that (Physics) and just concentrate on the important features; like Clouds ALWAYS reduce the total solar energy captured by the earth; and water vapor does too; and more of either of those means less solar energy for earth; pretty simple really, and you don’t have to compute and trend lines or R^2 values either.