Sunday
Jul212013
by
Bishop Hill
![Author Author](/universal/images/transparent.png)
Low-sensitivity model outperforms
![Date Date](/universal/images/transparent.png)
![Category Category](/universal/images/transparent.png)
![Category Category](/universal/images/transparent.png)
![Category Category](/universal/images/transparent.png)
Steve McIntyre has a must-read post about a low-sensitivity climate model which, when loaded up with actual greenhouse gas levels, completely outperforms the Met Office's HADGEM2.
Reader Comments (68)
Thanks for the pointer. Wow.
I am shocked by this post from Uncle Steve. He has not indicated that he knows how this model was created and yet he is praising it on the basis of the result. This attitude is the opposite of the way he investigated the Hockey Stick.
I am not at all surprised that toady Drake instantly says wow.
Bish
I find it hard to understand how after all the clinically accurate demolition jobs you have done on the CAGW agenda, you now swallow hook line and sinker the Climate Sensitivity models?
Let me reassure readers that I didn't merely receive these results and then praise them. I verified that the model was not tuned to recent data - I am aware of convincing evidence of this. I also verified that application of the model to observed GHG levels yielded the results shown.
I'm a tad puzzled. The low-sensitivity model output looks nothing like the temperature graph. Are we just supposed to be impressed that the graph goes up and up even with low sensitivity?
Steve
If you are still around, have you investigated how this model was put together and what assumptions and facts it includes? What about the effects we do not yet understand?
Steve
Shaking your hand and getting your signature in my copy of the HSI was one of the best moments of my life, even better than flying solo in a jet for the first time in the RAF, thats how important you are to me., However you are praising a model which so far you have not given any indication of having investigated?
Steve, as I see it, is currently having a rather effective go at Met Office salesmanship, as he amusingly calls it. This might have some further ramifications, for example for IPCC AR5 WG1. One never knows where these things are going to lead but the journey is the reward.
James,
In what way do you think the the model output doesn't look like the temperature? Do you mean that it's smoother? If so, that's probably because it's an ensemble average.
Dung,
Are you assuming that Steve is endorsing this model, or claiming that it is validated or credible? All he's saying is that it's possible for a low sensitivity climate model to match history. No more. Steve's generally very precise about what he says. Don't read more into it than is there.
Toady and Nullius
It is not normally Steve's method of proving a point to throw up bogus models and have a quiet snigger about it.
I agreed with what Martin A wrote three days ago:
What Steve is doing is demolishing the nonsense argument that the models prove high sensitivity. What fun. Believe me, it's those arguments that are going to croak.
Quite so. Also, I've written the post in a style that ought to alert regulars readers to a little surprise.
You folks just don't get it do you.
Producing a low sensitivity model that does as well as HADGEM2 is only of interest if that model can be validated otherwise it is totally meaningless. Until and unless Steve says he is happy that this model is well constructed and valid then this whole thing is a waste of time.
Nullius,
"In what way do you think the the model output doesn't look like the temperature?"
I mean it doesn't look like it. The temperature graph zig-zags up in a very regular way. The slope of the graph from 1910 to 1940 is almost exactly the same as 1970 to 2000. The model produces a graph that pretty much flat-lines (with a tiny wiggle) until 1960, and then sets off in a straight line towards the heavens.
Steve on CA:
I don't think we've been told everything about that last sentence. Best not to fall off too many trolleys until we have?
Most of us believe that high sensitivity models are bogus, I do not believe that a bogus low sensitivity model is the right response, therefore I need Steve to tell me it is a good model.
Dung,
Given that the model says we warmed up about 0.2C over the last 15 years, I don't think there are many senses in which it could be considered a good model.
Dung,
It seems to me that you are missing the logic of what Steve M. is doing here. It is (if I am understanding correctly, a big 'if') offering a comparison simply for the purpose of showing that "high sensitivity" models are not more accurate in hindcast than one example of a "low(er) sensitivity" model. This form of argument does not accept or affirm the ultimate soundness of any of the models, necessarily.
It is loosely analogous to when Steve M. offers comparative proxy curves to show that a "team" reconstruction is not statistically robust, and team members start shouting that Steve is offering his own "alternative reconstruction" which is not well validated.
As I see it, he is offering neither alternative models nor alternative proxy reconstructions as statistically and scientifically valid. He is showing how fragile, flaky, and/or non-robust the various papers are (whether model-based or proxy-based) in too many cases.
It is a common misperception to think he is providing an alternative reconstruction-model-claim etc., when his comparative methodological arguments are prior and more fundamental, showing why the field is not yet in a position to make such definite claims based upon the data being offered.
[corrections welcome if my layman's gloss proves inaccurate]
Dung,
This is the same sort of argument Steve faced when he produced versions of the paleoclimate reconstructions with various errors corrected, to show that the methods were not robust, and the warmists kept on interpreting them as serious proposals for corrected paleoclimate reconstructions, and saying they were no better when it came to validation, etc. The alternatives were shown only to demonstrate that changing the method in a certain way made a major difference to the conclusion, *not to say that the new conclusion was correct*. This is the same sort of situation. The model output is presented to make a precise and limited technical point that a low-sensitivity model can match the historical temperature record, not to say that the model is therefore 'good'/valid and can be trusted for making predictions.
We'll have to wait and see what conclusions Steve plans to draw from that - I expect he pointed out that it's an example of a low-sensitivity model matching history because that's just the point he needs to do so. But until we know what his intention is, we don't know if it's a waste of time or not.
Skiphil, that seems a fair description to me.
Putting the real physical world to one side for a moment, Met Office difficulties are compounded if old models can produce similar output to much-trumpeted 'state-of-the-art' models. It begs the question "Was it worth the expense?"
Michael: Yep, I have a feeling that's likely to be one of the questions at the end of this.
Since, coincidentally, I am looking at CMIP3 data, the curve looks like a CMIP3 one. They all have that shape
nullius, the green model does not seem to do well with the 1910-1940 warming. On the other hand, any model that doesn't do well on predicting known data, I'm ready to take that as a good sign...
More silly propaganda. "... completely outperforms ..."
SM's actual comment is: " ... one could argue that GCM-Q actually outperformed HadGEM2 in this respect [ie recreating 20th C temperature history]." I think this is an unobjectionable statement - clearly the claim is arguable - but I can see how some might argue the other way; and in any case hind-casting isn't all that's involved in GCM performance, surely.
I'm going to wait until the music stops before Steve pulls the mask from the partner he brought to the ball. I've a feeling some of the things some posters are sure of are wrong.
No surpise here. A model with a climate sensitivity and no aerosols must do well in the second half of Century 20. But it will still have trouble with the first half because the temperature increased just as fast as the second half but without much GHG warming. It needs a bigger kick from natural sources.
In my view, this is a potential humiliation for all the climate modellers who obstinately cling to the notion of strong positive water vapour feedback and high climate sensitivity.
It may also suggest that the scientists were closer to modelling reality some time ago and departed from this to pursue a path of more aggressive warming, ignoring the deviation from reality over nearly two decades. Perhaps they were driven by their belief system rather than observation; perhaps they underestimated temporary contributions to warming by natural factors.
It would be interesting if this model is more compatible with other observations such as the lack of a hot spot in the troposphere and the predicted relative humidity.
If this model is typical of early models then it really does raise questions about the current direction of climate science and the competence of those involved.
I will stick my neck out and make a guess about the origins of the model. It is probably one that was developed by some well-known climatologists about 20 years ago and since then has been modified out of existence as more and more bells and whistles are added.
If I am right, Steve McIntyre, or one of his contacts, has somehow got hold of the original version and has shown that it works better than any of the improved versions. In other words, all the advances made by climatologists in the past couple of decades have made the model worse!
As I said, the above is just a guess. I may be barking up the wrong tree but the mysterious nature of Steve McIntyre sparked my imagination and caused me to speculate wildly and probably wrongly.
Well, retirement does happen personally.
===========
Roy:
Knowing a bit of Steve McI's sense of humour, my guess would be that the model is even older. E.g. when Steve writes:
That could almost sound like Arrhenius (who had some kind of a "model" of CO2-induced GW), with the "personal reasons" being mostly that he is now dead. Though other parts of the post do not fit with that. But I would bet that this is a very old model.
Unusual in that it's a post at Climate Audit which itself has no audit trail, the green squiggle could be anything that went up in the 20th century with a lot of smoothing. Dow Jones, stamp prices, GDP etc.
I'm with dp@Jul 22, 2013 at 5:26 AM, there's some reveal coming. eg a very early paper proposing a simple relationship.
Maybe international man of mystery Gavin Q is behind the secret GCM-Q?
To clarify my position:
The model referred to by Steve is either a model he believes to be good and accurate or it is a model which he has no more respect for than any other.
I do not believe it is the former case but if it is then I would like Steve to say that.
If it is the latter case and almost all posters believe that it is, then Steve is using a bogus model to ridicule other bogus models and I do not agree with the strategy.
I do not think it is the same as Steve changing the data in the MBH Hockey Stick models to show what they would look like with correct data as opposed to tree rings and or upside down Tiljander etc. Accusations then that he was presenting his own new paleoclimate models were ridiculous.
I think it is totally out of character for Steve to play this game whatever surprise he has in store. If Climate Sensitivity models are wrong/inaccurate then he/we should argue the case and give evidence to support our arguments, not play their own stupid game with them.
The reason for 'playing their own stupid game with them' is that under the auspices of that game the most damaging policies imaginable may still be imposed on the human race. (We haven't seen anything yet, if we're not careful.) If there was a knockout proof of no danger from rising CO2 emissions fair dinkum, as my mother would say in earlier years. But I don't see that emerging from the "climate sensitivity is bunk" camp. So it's entirely right, from every real sceptic's point of view, for Steve to play 'em at their own game and show how stupid it is. As I said last night, I can't wait.
Let me play a stupid game with you Richard ^.^
Just answer clearly yes or no (that will be the day hehe):
Do you think that modelling Climate Sensitivity can (at this time) produce accurate results?
Do you think the phrase Climate Sensitivity has any meaning in the real world?
Do you believe that scientists fully understand even half the factors which affect temperature?
I feel I've said enough on this thread for now. Let's wait and see what Steve comes up with and whose reaction to his teaser last night, the Bishop or his critics, turns out to have been the most balanced.
" I don't see that emerging from the "climate sensitivity is bunk" camp"
I suppose I am a major contributor to that camp. Don't misunderstand the intent. As with all of my awkward questions, the intent is to get the other side to respond. To discover on what basis they make the claims they do. I have never got an answer. I do not see why we (if there is a we) should allow them any 'givens'. If now I have a lot of sceptics who do not reject the concept of CS, why do they not answer? If you believe in CS, if it seems to you to be a compelling hypothesis, whether you are a sceptic or a CAGWist, just jot down a few lines of justification. This is not a no-CS hypothesis except in the sense of a null. Why accept CS without question? Put up or shut up.
Unless you are EM, who in another thread accuses me of putting the planet at risk with my evil questions.
I am not asking here for reasons why we shouldn't contest it for sceptic tactical reasons. That is a different argument. I will accept any tactical discipline handed down by the well-funded denial industry as soon as my cheque arrives.
Read the post carefully.
"For the same personal reasons, I do not, at present, have permission to identify him, though I do not anticipate him objecting to my presenting today’s results on an anonymous basis."
I think that Jeremy is correct regarding the personal reasons - and that the model is much older than 20 years.
According to Wiki, Arrhenius (1906) has a sensitivity of 1.6C / CO2 doubling.
Rhoda:
You're a scary woman sometimes. :)
Yes, Paul, one of the more suspicious uses of anonymity, even by my standards. :)
Steve promises a "little surprise" -- taking my cue from Jeremy and Paul above, I suggest it is something like an assumption of 1.65C sensitivity from Arrhenius with an assumption of homeostasis..... I.e., low sensitivity plus a planetary atmosphere that tends toward equilibrium proves more accurate than all the bells and whistles of the IPCC's motley collection of super-expensive climate models???
Richard Drake is as slippery as Baroness Verma ^.^
Answer the questions you old waffler.
Why would it matter if the model was 100 years old and produced by Arrhenius?
Skiphil
You can not suggest that the model was accurate, simply that it produced a result that happened to be close to the truth. Does it not matter how it was put together or how much Arrhenius actually understood about CO2 and temperature?
It was your criticism of both Bishop Hill and Steve McIntyre in such florid terms that I thought was a little OTT earlier. We all know you've desperately wanted me banned from Bishop Hill for months now, which, just from a personal perspective, I've always felt is a little illiberal. But no need to answer on any of that for the moment. Let's see what Steve comes up with. And chin up. You may not have damaged your credibility as much as initially seemed likely. And we do need a variety of opinions here at all times.
Richard
What a reasonable response (even if I didn't get answers to perfectly reasonable questions). I have never said and never considered that you should be banned, you are far too knowledgeable and intelligent to make that suggestion. However there have been times when your actions have been er um somewhat unreasonable and at those times I speak out. You can be charming and witty when you want to be ^.^
I will always say what I think, I don't hold the Bish or Steve in any less regard because I suggest that they may have made one mistake, is anyone perfect?
It's like who shot JR all over again.
There are two ways a detective story can go. We see the outcome of the crime but not the perpetrator and follow the clues along with the detective, leading to a revelation at the end. Or we see the perpetrator and observe the detective finding clues and overturning alibis in order to catch the person we already know is guilty. The difference between an episode of Poirot and Columbo perhaps.
If Steve had done the latter we would know where the model comes from but be waiting to see what it can do. Steve has done the former - showing us what the model can do and waiting to reveal whodunnit.
Monsieur McIntyre has laid out the facts of the case for us and all that needs revealing is what shady character with a hidden past or wronged femme fatale did the deed.
Mosher at CA lends weight to suggestions that the anonymous climate model comes from someone in the more remote past of climate science:
Dung, I'm not able to improve upon my description above of what I think Steve M. may be up to (I wish I could say it better). I sense he's comparing methods, data, arguments, and assumptions, but NOT arguing (at this time) about which model is ultimately "right" or "wrong"..... it's a more tentative set of comparisons, as in "all these expensive climate models do not seem to be improving the 'skill' of modeling....."
Dung,
As many people - including Nullius in Verba most notably - have argued above, I think you may be misconstruing Steve. Academic arguments proceed in strange ways, and often involve provisionally accepting a point that one may not accept as correct in order to explore other issues. I don't think anyone thinks "climate sensitivity" is defined in any fundamental way such that its value could be discussed in a way at all similar e.g. to that of the mass of the electron - but it may turn out to have some approximate usefulness so it seems unnecessarily rigorous to exclude any discussion involving it. Also, even in serious debates, one can sometimes have a bit of fun, and introduce ideas in a slightly cryptic way.
Skiphil, thanks for your posts - it sounds as though I may be right! The extract cited by Paul Matthews at 2:09 was the one that initially made me consider that it probably was not Arrhenius:
(my emphasis). The "at present" I took to mean that Steve might one day aspire to have the permission from whoever had constructed the report to present it, and getting that permission in the case of Arrhenius would clearly be difficult. I guess if you read the quote literally, though, it does not imply that. Wait and see."Steve McIntyre has a must-read post about a low-sensitivity climate model which, when loaded up with actual greenhouse gas levels, completely outperforms the Met Office's HADGEM2."
"All he's saying is that it's possible for a low sensitivity climate model to match history."
"It may also suggest that the scientists were closer to modelling reality some time ago..."
"If I am right, Steve McIntyre, or one of his contacts, has somehow got hold of the original version and has shown that it works better than any of the improved versions."
"I.e., low sensitivity plus a planetary atmosphere that tends toward equilibrium proves more accurate than all the bells and whistles of the IPCC's motley collection of super-expensive climate models?"
"...it produced a result that happened to be close to the truth."
Just some of the comments that leave me as bemused as I was when this thread started. Are we looking at the same graphs? The model ouput is an utterly lousy match for actual temperatures. And I'd say it only looks slightly better than the MO model because of the arbitrary centering choice that Steve made.
If you people think the low-sensitivity model output and the temperature graph look similar, then perhaps I can introduce myself: George. George Clooney.
BTW, I agree that it's probably some sort of Arrhenius model.
FWIW - this thread came to mind:
http://climateaudit.org/2011/05/15/willis-on-giss-model-e/
Lest I am thought to be going mad, this is what Skiphil wrote on 9th January:
I at once agreed with this (and everything else that Skip wrote). I was very conscious that I never saw Dung agree with it. But I appreciate that he now says:
That was not what I understood at the time. But I have never had this clear a statement to go on before. And what's a mere six months to wait when we're dealing with climatic timescales.