Bishop Hill Trenberth et al in Remote Sensing
Sep 16, 2011 Climate: Models Kevin Trenberth has his own response to Spencer and Braswell in Remote Sensing. It's billed as a commentary rather than a proper paper.
The first thing you notice is this:
Received: 8 September 2011 / Accepted: 8 September 2011 / Published: 16 September 2011
so presumably it's fair to say that this is not a peer-reviewed contribution to the literature.
I've not had time to read it yet, but my eye alighted upon this sentence.
There are obvious differences among models. As noted by Dessler [10], it is important to sample all model results and not just select a few that may have certain specific deficiencies, as was done by SB11 [8].
Which is a surprising thing to say given that Spencer has already pointed out that he has picked models that span the range of sensitivities. I note that Spencer's blog post is dated 7 September. Presumably then Trenberth is guilty of being in such a hurry to get his rebuttal out that he managed to preempt Spencer's explanation.
However, given that the blog-post explanation merely reiterated the one in the Spencer's original paper, we might conclude that Trenberth was also in too much of a hurry to actually read that either.
Bishop Hill
Steve M has some rather more considered thoughts, which really make Trenberth look rather grubby. Then there's this:
Clouds were the major source of uncertainty in climate models in Charney 1979 and remained so in IPCC AR4 (2007). If Dessler 2011 did in fact show that "cloud effects are small", this would be an epochal achievement in climate science. Given that a preprint of Dessler 2011 only became available on Sept 2, 2011, there has been little opportunity to analyse its results so far. Whether Dessler 2011 really proves that "cloud effects are small" remains to be seen. If, like Dessler 2010, it makes such assertions based on r2 of ~0.01, I think people could reasonably disagree on whether such far reaching claims had been firmly established.
One has to appreciate the achievement of Trenberth in giving us so much to laugh about.
Reader Comments (39)
My if the time between Received and Published gets any shorter they will have to bend the laws of physics again !!
Is there a video?
The WG1 chapter has been written. It would be a pain to change it. Better to throw up a lot of muck here and keep doing what's been done.
I wonder why he failed to mention that Spencer is a Creationist.
Trenberth point (which is clear from various other rebuttals already out there) is that the models that Spencer picked all share the same deficiency in not preoperly simulating the effects of ENSO.
So rather than picking models that span the range of sensitivities, it would presumably have been better to pick models that span the range of ENSO simulation. Better yet, all.
Spencer's explanation of having chosen the highest and lowest sensitivity models is not responsive to Trenberth's point that ENSO is more important in explaining the model-obs match than sensitivity is.
My eye alighted on this sentence:
which is remarkably similar to Wagner's statement to IOP. Did Trenberth tell Wagner what he should say?
(It is also ironic since this is exactly what sceptics have been saying for years re CO2 and warming)
Bish
my reading of the blogosphere (ok I know it is not peer reviewed) is that there is an issue with the models chosen by Dr Spencer. I think he has addressed this on his blog
where would we be without tinternet. dead good. init
D'head
Trenberth like Hansen is wriggling on the hook of having put their faith in false science ['non-existent back radiation and cloud albedo effect cooling].
Not only is the gravy train stopping, their leadership of research is being challenged as outside scientists have identified their mistakes.
Bart--
"Better yet, all [models]"
Spencer responded to this in his blog by providing a graph with all models. The conclusion was unaffected.
It is a "travesty" that Kevin Trenberth hasn't been sacked.
Posted and accepted in one day?
Will Dr Spencer get the same, swift acceptance of a comment, if forthcoming?
From a quick scan, I get the impression that Trenberth's critique is a general 'we don't like it', and not engaging the reasons laid out in SB11.
But good to see that the US taxpayer can shell out the publication fee at such short notice ...
Bart
Your point is in inverse.
For the point that Spencer's making, any set of models, ought to be enough.
BH:"...presumably it's fair to say that this is not a peer-reviewed contribution to the literature."
Dr Phil Jones: "Kevin [Trenberth] and I will keep them out somehow — even if we have to redefine what the peer-review literature is!”
OK, that's a cheap shot, I admit it.
Why would anyone take Trendberk seriously after all he is a member of the Fiddlestick Team?
It's all fishier than Billingsgate's market. We have: Abraham showing up for Amateur Hour; wagner's words repeated almost verbatim; a discussion on the importance of using all models followed by a figure showing that only the "best" model truly resembles ENSO; Dessler's upcoming-or-maybe-not paper referenced at light-speed using the past tense...
I'm sure there's more for those who have time for this stuff.
We also have the magic anti-Mannian words: the description of their method was incomplete, making it impossible to fully reproduce their analysis. Such reproducibility and openness should be a benchmark of any serious study
And the best one: the interpretation of causality between clouds and temperature is often a major challenge. Yeah, right.
I've seen enough 'articles of faith' in Ken's scientology to fill a life time of ennui, why does he bother?
It's all arguments over deckchair placement, on a ship fast disappearing into the briny.
Steve Mc now has a post on this: More hypocrisy from the team, mainly making the point that Trenberth's complaints about S&B apply to many papers by 'The Team'.
Much of HSI was about the behind the scences wrangling to keep the right stuff in, and the wrong stuff out of AR4.
The Hockey Team playing hokey cokey pokey with science, seems to be recurring
McI just posted a blog...noticing how strange is for Dear Kev to claim that Dessler's invisible paper "quantifies the magnitude and role of clouds". One wonders if Dear Kev has read Dessler's latest work in full.
<I>Bart Verheggen said:
"So rather than picking models that span the range of sensitivities, it would presumably have been better to pick models that span the range of ENSO simulation. Better yet, all."
------------------------------
The models that try to simulate ENSO do a godawful job. For example - ECHAM (which Trenberth's RC post brought up in response to S&B) produces science-fiction style swings in global temperature because its ENSO spews out record breaking temperatures right across the equatorial Pacific. Almost constantly. Hot and cold.
Trenberth's paper has a graphic which compares the response of two models to SB11's observations: the first panel shows a model which does not match well, but the model of the second panel matches reasonably well. On his website, Spencer displayed the response of all models (which he would have done well to include in SB11), and indeed one of the curves in that spaghetti graph isn't too far off. [Can one say that Trenberth "pasta-picked" his second panel?]
But if we return to Trenberth's graphic, it is the third panel which is most telling -- it shows the mean and spread of all models' responses. The mean response reinforces SB11's conclusion that (in general) the models do not agree with the observations in this particular regard. But look at the spread! It's so large that it encompasses the observations. It seems that Trenberth's argument runs something like: "The models are all over the place. So one can never say that observations contradict models, because there's always at least one model which will come close to the observations." I note also that the third panel appears to show the *range* of responses. However, most analyses (e.g. IPCC) report on the mean and *confidence interval* of the models' results. The confidence interval would be significantly less than the range, as it is typically assumed that the average of several models' results, produces an estimate with less uncertainty than any one model's results.
Trenberth has show repeatedly he does not need to read someone to know its wrong , such are the powers of the man .
Fergal posted while I was composing the above. The model which Trenberth put forward as matching SB11's observations well, was indeed the echam model which Fergal mentions. Trenberth describes it (see caption to Figure 1) as "a model which reproduces ENSO fairly well."
From Trenberth's paper: "...20 years or longer is needed to begin to reolve a significant global warming signal in the context of natural variations". How much longer, given that some think they have detected a 0.8 degree C rise in the last 100 years?
50 years (since anthropogenic atmospheric CO2 is thought to have greatly increaed)?
300 years (since 2-3 degrees C warming following the LIA)?
800 years (ocean CO2 outgassing from the MWP which was 2-3 degrees C warmer than today)?
A global warming signal has not been detected and it's a travesty that it hasn't.
The speed of this reaction and its rapid acceptance tells me that the Forces of Thermageddon are nolonger thinkimg ratonally or strategically, but are panicking. Victims of thefearful reaction of the weak and clueless 'I don't know what to do, but must be seen to be doing something'
Can anyone explain why they have been so spooked so quickly by the turn of events? Is there a deadline approcahing that I don't appreciate? Or a further instalment in teh works?
Very very curious
You have overlooked the post-Modern concept of post-facto peer review. It's jolly efficient--cuts days off the publication cycle. Only certain people are eligible for this, of course, but it saves quite a bit in postage, since it's done with a phone call or two.
The really sad thing about this is that it is not really about science but solely about the corruption of the science journals carried out by the "Gang".
Trenberth, and it really is a travesty, seems to have taken over from Mann when it comes to sticking both feet in it. I guess Mann is to busy trying to use lawyers to, once again, hide the truth/data! I simply cannot fathom that they are so dumb to think no one is watching/checking their every move!
Robert Conquest's 3d law of politics: 3.The simplest way to explain the behavior of any bureaucratic organization is to assume that it is controlled by a cabal of its enemies.
Clearly -- global warming has become a religion of entrenched political interests which, as embodied by the IPCC and the national academies, has become a sclerotic bureaucracy with all the incompetence, corruption, and cronyism one would expect from such a bureaucracy.
Evil corporate fossil fuel intersts could not pay enough cash to secure the kind of benefits that Trenberth, Algore, and company are providing. Kev and the boys couldn't do more damage to their cause if they executed a circular firing squad.
Wow ! they are realy spooked. Roy and Co. must really be on to something. Go get them Roy !.
I love the idea that a model is valid if it can hide within confidence intervals, which are defined as the MAXIMUM and MINIMUM points found in 14 models.
Why bother going to all the trouble of examining the residuals of the fit vs. actual data when you only need have large confidence levels to 'prove' that they are one and the same?
The max-min span also means that when one foretells the future using these models the 95% confidence levels have to be in the order of a slope 40 to-40 W/m2 per year.
The lower panel of their figure 1 puts me in mind of the O.J. Simpson trial; the glove didn't fit. It is absolutely clear that the models have a very different line-shape from the real data.
Trenberth is claiming that because he has a credit card and a password he can take money out of anyone's account, because all you need to take money out of a bank account is a credit card and a password.
If I had refereed a paper attempting to claim a good fit on the basis of the figure I would have bounced it; where are the residuals? what are the confidence intervals of the n=14 models?
"Moreover, in examining the relationship of the regression’s strength among models to
climate sensitivity we find a weak positive correlation—that is the models that do a better job of
replicating the observed relationship are the higher sensitivity models, though it should also be
remarked that the correlation is of marginally statistical significance, the precise value of which
depends on the degrees of freedom attributed to the model ensemble."
'the correlation is of marginally statistical significance'. Like being a little bit pregnant then. Statistical significance, like virginity, is absolute.
Sixteen (16) years and no warming:
http://www.woodfortrees.org/plot/hadcrut3vgl/from:1995/to:2011/plot/hadcrut3vgl/from:1995/to:2011
Bart Verheggen, FergalR and HaroldW
I commented at JuithCurry about D11 and the models selected in it for comparison with the observations and how well they modelled ENSO. To repeat some key points from that comment by way of background to Trenberth et al:
D11 said “.. the models that do a good job simulating the observations (GFDL CM 2.1, MPI ECHAM5, and MRI CGCM 2.3.2A) are among those that have been identified as realistically reproducing ENSO [Lin, 2007].”
First off Lin 2007 does not include MRI CGCM 2.3.2A as one of its models that realistically reproduce ENSO. It is in the group that “.shows an oscillation with a constant period shorter than the observed ENSO period, sometimes also with a constant amplitude.” . . . .
[Further] “ENSO Feedbacks and Associated Time Scales of Variability in a Multimodel Ensemble” Belmadi et al (2010) offers a closer look at the GCMs classifying them “into three groups that account for the dominant feedback process of the ENSO cycle: horizontal advection (mainly in the western Pacific), vertical advection (mainly in the eastern Pacific), and the combination of both mechanisms.”
Belmadi et al do an initial assessment of the models based on absence of any significant peak in the interannual ENSO frequency band. . . . .
Belmadi then classifies the various models according to the dominant feedback mechanism, and in so doing eliminates GFDL CM2.1 and MRI ECHAM5 from their preferred group (INM-CM3.0, IPSL CM4, UKMO HadCM3, and UKMO HadGEM1). Both are eliminated on the basis that they are overly dominated by the thermocline feedback. MPI CGCM 2.3.2A is also eliminated because it’s dominated by zonal advection feedback.
So D11’s claim that the three “best fit” models are the best to model ENSO is in part not supported by his reference, and also not supported, full stop, in the more recent literature. This claim must therefore be seen as at best controversial in the literature.
Trenberth et al rather curiously doesn’t actually given any direct evidence that the models that fit well to the observations are the ones that fit ENSO well. It is all done by proof by example and assertion. They start by asserting evidence for a general principle:
. . . here we have taken the CERES EBAF product [25] to explore the methods employed by SB11. . . . Our results suggest instead that it is merely an indicator of a model’s ability to replicate the global-scale TOA response to ENSO. [my emphasis]
So we wait with baited breath to see these results.
First we are told The MPI-Echam5 model replicates the observations very well.
There then follows a critique of SB11’s model’s inability to model ENSO and then the comment but rather than stratifying [the climate models] by climate sensitivity as done without basis by SB11, one should stratify them by their ability to simulate ENSO.. Why “should”?
Trenberth et al then shows that the ECHAM5 runs over each of the last 10 decades encompasses the observations, asserts ECHAM5 reproduces ENSO reasonably well and finally reaches the weak conclusion that the relationship with ENSO and the model ability to replicate ENSO, as well as climate sensitivity could be explored further.
How did Trenberth et al get to assert: Our results suggest instead that [S&B’s result] is merely an indicator of a model’s ability to replicate the global-scale TOA response to ENSO. from finding one mode (ECHAM5) that purportedly replicated the observations well, coupled with an unreferenced and controversial in the literature claim that ECHAM5 is a good at modelling ENSO?
Spencer is a Creationist.
Oops!!
You should learn the difference between intelligent design and creationism.
Source: Trenberth et al Sept. 2, 2011
IOW, Remote Sensing is an "off topic" journal, regarding publication in which Bob <fast fingers> Ward is reported to have opined, circa Sept. 2):
So, I'm sure we'll see Bob Ward rapping Trenberth's knuckles for using such a "thoroughly disreputable approach" any day now, won't we?!
Mind you, if anyone is capable of "finding weaknesses in the peer review system", it would seem that Trenberth is an expert (particularly when it comes to his own papers). Ol' King Kev is obviously so damn good at finding such "weaknesses", that he managed to avoid any vetting - let alone an "honest vetting" - for his "commentary" in this "off topic" journal.
The double-standards and hypocrisy are almost beyond belief! And we're supposed to trust them?!
Trenberth and the Team: The planet is going to catch fire because of Global Warming, our totally reliable peer-reviewed models say so.
Sceptics like Spencer and Braswell: But the models don't match the observations. See?
T&tT:There are obvious differences among models. As noted by Dessler, it is important to sample all model results and not just select a few that may have certain specific deficiencies, as was done by SB11.
SlS&B: Hang on a sec. Deficiencies?
T&tT: Yes. Some of the models are a bit... well... rubbish.
SlS&B: You didn't mention this before.
T&tT: Well obviously not. It would have been giving sceptics the ammunition to shoot our IPCC science down. Given the stakes, we couldn't allow that to happen.
SlS&B: I see. So why use them at all?
T&tT: Because that's how climate science works - you try to model the climate as a means of research.
SlS&B: No, I mean why use them to draw conclusions from in the IPCC reports, if you know they're duff? Why not just use the ones that don't have deficiencies where they don't match observations?
T&tT: Because the errors don't matter, as the same conclusions are reproduced in many other studies.
SlS&B: But wouldn't it strengthen your argument to be able to say your models didn't have any errors? Why waste time and space continuing to use the wrong ones if you've got this long list of good ones lined up? Why not drop all the ones that don't match reality and show us only the ones that do?
T&tT: Because the errors don't matter, as the same conclusions are reproduced in many other studies.
SlS&B: I see. Could it be that *none* of the models actually matches reality, but they all have different faults, so whatever objection is raised to one can be countered by pointing out others that don't have that particular fault?
T&tT: As we said, the errors don't matter, as the same *conclusions* are reproduced in many other studies. The methods don't matter if all the conclusions come out the same.
SlS&B: So can we take it that since you haven't actually weeded out the models that don't match reality, that this is because you don't have any that do? Can we take it that since these models that don't even reproduce short-term phenomena like ENSO got past the IPCC's scrutiny, that they are not applying any checks on models to be sure they are fit for purpose? Why should we think *any* of the models used are reliable enough to be able to model future climate outside the bounds of our experience, when we already know they don't even model our current climate correctly?
T&tT: Because the errors in individual studies don't matter, as the same conclusions are reproduced in many studies.
SlS&B: I see. Thank you.
[Exeunt. Curtain falls.]
Bart,
"Trenberth point.....is that the models that Spencer picked all share the same deficiency in not preoperly simulating the effects of ENSO".
Then why are these models used at all ...by anyone....ever?
If the science is settled, why are there so many different models? Why more than one?
I don't understand the content. It’s a complete hatchet Job on SB11, the likes I have never experienced in my 25 years as a scientist/statistician. Thank heavens I do not work in such a nasty field.
I have a question for Kevin Trenberth. Aren't you even a little embarrassed you boarded the AGW gravy train?