Bishop Hill Dessler rebuts
May 8, 2014 Climate: sensitivity Is it just me that reads every paper by Andrew Dessler as an attempted rebuttal of some sceptic position or another? His latest paper was submitted to Geophysical Research Letters just three weeks after the publication of the Lewis and Crok report on climate sensitivity and reads as though it was written in direct response to it. Here's the abstract:
Estimates of the Earth's equilibrium climate sensitivity (ECS) from 20th-century observations predict a lower ECS than estimates from climate models, paleoclimate data, and interannual variability. Here we show that estimates of ECS from 20th-century observations are sensitive to the assumed efficacy of aerosol and ozone forcing (efficacy for a forcer is the amount of warming per unit global average forcing divided by the warming per unit forcing from CO2). Previous estimates of ECS based on 20th-century observations have assumed that the efficacy is unity, which in our study yields an ECS of 2.3 K (5%-95%-confidence range of 1.6-4.1 K), near the bottom of the IPCC's likely range of 1.5-4.5 K. Increasing the aerosol and ozone efficacy to 1.33 increases the ECS to 3.0 K (1.9-6.8 K), a value in excellent agreement with other estimates. Forcing efficacy therefore provides a way to bridge the gap between the different estimates of ECS.
Efficacy is not a measure with which I am familiar, and a cursory scan of Google search results suggests that it has not exactly been a research priority in the past, although there is a Hansen paper from 2005 and a section in AR4 on the subject. Hansen seems to be determining efficacy from climate models, which seems a bit back-to-front to me.
Hockey Schtick, who wrote about the paper yesterday, describes it as a "fudge factor" and I have to say that I too detect a whiff of sugar and butter on reading the abstract. If someone can send me the paper then perhaps we can see how much physics is involved.
Bishop Hill
I've now got hold of the paper. Here is the discussion of efficacy:
It has long been expected that forcings with the same global average magnitude but different spatial patterns could evoke different responses in global surface temperature [Hansen et al., 1997; 2005; Shindell and Faluvegi, 2009; Shindell et al., 2010]. For example, forcing concentrated at high latitudes, which are less strongly restored by infrared radiation to space, will lead to more warming than well-mixed forcing agents. Certain forcing agents, particularly aerosols and tropospheric ozone, are indeed not uniformly distributed and impact the climate system differently than well-mixed constituents [Shindell et al., 2003; Feichter et al., 2004; Chung and Seinfeld, 2005; Crook et al., 2011].
Different formalisms have been adopted in the literature to account for this process [e.g. Hansen et al., 2005; Winton et al., 2010; Armour et al., 2013]. One is to account for the effect using a so-called forcing “efficacy” [Hansen et al., 1997; 2005], which is the amount of warming per unit of global average forcing divided by the amount of warming per unit of forcing from carbon dioxide. Most calculations of ECS based on 20th-century observations assumed that the efficacy of different forcers is one, so this effect was ignored.
Recently, Shindell [2014] analyzed transient model simulations to show that the combined ozone and aerosol efficacy is about 1.5. At least some of the high efficacy of aerosols and ozone was due to nature of the transient runs he analyzed, but his analysis nevertheless clearly showed that, in the models at least, the efficacy of aerosol and ozone forcing was significantly greater than 1.0.
To test the impact of efficacy on the inferred λ and ECS in our calculations, we multiply the aerosol and ozone forcing time series by an efficacy factor in the calculation of the total forcing. We find that increasing the efficacy shifts the PDF of λ to lower values (Fig. 1), corresponding to increased climate sensitivity.
Using Shindell’s [2014] estimate of efficacy of 1.5 decreases λ to 1.1 …, corresponding to an ECS of 3.5 K … We can reasonably simulate the IPCC’s climate sensitivity range using an efficacy of 1.33, which gives an ECS of 3.0 K…
This doesn't help much, since the physics seems to be in Shindell 2014. However, the latter paper is available online.
Here I analyse results from recent climate modelling intercomparison projects to demonstrate that transient climate sensitivity to historical aerosols and ozone is substantially greater than the transient climate sensitivity to CO2. This enhanced sensitivity is primarily caused by more of the forcing being located at Northern Hemisphere middle to high latitudes where it triggers more rapid land responses and stronger feedbacks.
Ah ha. It's because the model says so.
Reader Comments (33)
Does the paper give any empirical or theoretical reason as to why the correct value for efficacy should be 1.33? From reading this, it would appear the sole reason for choosing this figure is that it produces "a value in excellent agreement with other estimates".
Which no doubt strikes Dr Dessler a s a very good reason indeed. Can anyone with the patience to read the whole thing advise if he has any actual evidence
He appears to be saying that just as there is assumed to be some unproven thing amplifying the warming effects of CO2, there is also some unknown mechanism amplifying the cooling effects of aerosols and/or ozone. Which then means that the sensitivity to CO2 has to be EVEN MOAR in order to give the results that we see.
"Does the paper give any empirical or theoretical reason as to why the correct value for efficacy should be 1.33? From reading this, it would appear the sole reason for choosing this figure is that it produces "a value in excellent agreement with other estimates"."
Indirectly the reason is political and based only on the UNFCCC.
epicycles
We already knew they only get a high CO2 sensitivity by adding aerosols and ignoring natural variation. The big problem is that observations do not match expectations when that assumption is made. If you were to increase the component of internal variation (which is more in line with reality and the new voodoo science of deep ocean heat trapping plus pause-producing pdo fluctuation) while ignoring aerosols then you get a very small value for CO2 sensitivity. The conclusions in climate science are always dictated by the initial assumptions but if the results don't agree with the observations then the assumptions are wrong.
Dessler by the way is fond of trying to fit a trend line into a scattergraph and declaring positive feedback. Using more or less data or using a different cause-effect assumption gives the opposite trend but nevertheless you could fit a picture of a dancing hippo in the same scattergraph with equal validity. I don't really think anyone takes him seriously.
His apparent Damascene 'gut-feeling' moment for belief in catastrophic global warming despite all the apparent evidence to the contrary is that Gerald North once produced ice-age style cooling in a model by removing CO2 so CO2 must therefore be the driver of global temperature. He didn't bother to wonder precisely where this sudden, massive CO2 sink might have come from nor did he notice the circular reasoning that if you build a model with the assumption of CO2 as a climate driver then that is precisely what you get in the output.
I read this as:
"Observations give a less disastrous conclusion than models. Therefore we need to put in a multiplier to the observations to make them fit with the models. Because consensus."
Perhaps not fair basing that on the extract and my poor understanding of gobbledegook.
Why can't these people be happy that Gaia seems to be repairing herself by providing unexpected negative feedbacks that are currently unknown.
Add adjustments to taste or until you reach the figure you first thought of.
Efficacy?
Wasn’t that something to do with Lily the Pink’s Medicinal Compound? We were assured by Scaffold that it was most efficacious in every way, hence the need to drink a drink a drink a drink.
Models. Aren't. Real.
The basic idea doesn't seem hard to understand (unless I have missed it).
1. To get climate sensitivity you need to know how a given radiative forcing (that is, extra watts / square meter of heat) translates in to a given rise in surface temperature.
2. Up to now, calculations have assumed that radiative forcing is global. For a gas such as CO2, that should not be too bad an approximation, because CO2 is well mixed.
3. But forcing from aerosols and tropospheric ozone are bigger in the Northern hemisphere than the Southern hemisphere, because emissions are larger there, they have a shorter lifetime in which to mix, etc.
4. Because water is harder to heat than land and there is more ocean in the Southern hemisphere, the same watts / square meter will have a bigger effect on surface temperatures if it is concentrated in the north than in the south.
5. If this logic holds, the same forcing from aerosols and tropospheric ozone concentrated in the north should have a bigger impact on surface temperature that the same forcing from CO2 distributed globally.
6. The Dessler paper is an attempt to account for this, parameterising the difference between forcing by 'efficacy'.
I'm open to the idea that the whole scheme is wrong, or that the implementation is wrong. I probably won't make up my mind for a few years until we see many more studies and reflective responses. My judgement now is that, like most scientific ideas that don't completely fail, it most likely contains some partial truths although they will be integrated in to a bigger picture we can't yet see.
In the comments above I don't see even a glimmer of interest in helping me see what could be wrong or right with this. Even the OP, to be honest, with it's dismissive 'because the model says so' isn't doing anything for me. Of course theory has to be continually checked against experiment. But if any result of arithmetic and reasoning is dismissed as not worth following up because it's only a model then we won't get very far.
Dr. Dessler is a politican cooling his heels at one of the best engineering schools in the US from where he practices politics dressed up as science.
JK (2.52)
I think people are dismissing this because having built a theory around a set of conjectures (which may have some logic behind them, but sadly no experimental support), the author does no more to test the theory than fiddle with the values in order to get the presumed level of sensitivity which agrees with other (unproven) models of catastrophic effects of CO2. If this were decent science then there would be some proposed experimental model which could be tested. This does not appear to have been done in this case and therefore it can (and should be) dismissed until there is some evidence. The onus is not on the reader to refute this theory - it is be on the proposer to test the theory and provide some level evidence that the logical steps do in fact add up.
Kummer & Dessler seem to have performed their basic calculations improperly.
They claim that the AR5 forcing time series are referenced to the late 19th century, and therefore deduct from the temperature time series the 1880-1900 average temperature. But the AR5 forcing time series are, as stated in Table 8.6 (which they cite), referenced to 1750. They should therefore have likewise deducted the 1880-1900 average forcing from the forcing time series.
Upon making this correction, I calculate that Kummer & Dessler's basic estimate of climate sensitivity should be (taking the median) 1.5 C, not 2.3 C.
I dispute their/Shindell's contention that aerosols and ozone have an efficacy substantially above one. But even taking their high case of an efficacy for them of 1.5 would only increase the climate sensitivity median estimate to 1.75 C, not to 3.5 C as they calculate.
This appears to be an unscientific attempt to undermine the robust energy budget evidence for climate sensitivity being low. It is appalling that the paper has been accepted for publication by GRL.
Has Roy Spencer checked this for errors yet? You would think after last time GRL would call him as a reviewer.
Thanks for the response Nic.
We can wait to see how Kummer and Dessler respond to your point about the calculations.
In the mean time, a couple of questions if you have the time / inclination:
First, you write
Is that because you think there is no good evidence that efficacy differs from one (a negative statement) or are you saying there is positive evidence that it is close to one?
Second, what do you think of their methodology (done right) as a means for assessing the impact of an efficacy greater than one? Do you have an estimate of what efficacy would be "substantially above one" in the sense of, say, making a difference of 0.5 degrees or more to an estimate of climate sensitivity?
Third, Kummer and Dessler mention that different formalisms than an efficacy scaling factor have been used to account for non-uniform forcings. Do you think that any of these are more enlightening than efficacy for thinking about this in relation to climate sensitivity?
They cite the Hansen for the term "forcing efficacy," but do they attempt to rebut his actual findings?
Per 3.1.4 and 3.2, Hansen modeled ozone efficiency at .82, volcanic aerosol efficiencies at .91, black carbon aerosol at .52, and sulfate aerosols at 1.09. None of those are close to 1.33.
RR
My thought exactly :-)
And as with probability, how can you have an 'efficacy' greater than 1?
Nic Lewis,
OK, you are apparently an advocate for low sensitivity, energy balance methods do not become robust just because you want to or just say so, you have to do the science first.
Even before recent papers on the energy balance calculations, 3/4 methods agree about 3C sensitivity with some likely interval 2-4.5 or so: GCMs, paleo and interannual variations. Just denying these does not make them less relevant in the rational world.
Now the outlier method, by a number of papers, is shown to be biased low, like one would expect from the phenomena regression to the mean. So everything is very logical, including that Hansen, as usual, understood it years before everybody else. The efficacy factor is physically well based and could be understood without climate models: Aerosol forcing is rather local/regional and almost all such pollution is from particular latitudes over land in the northern hemisphere.
The fundamental problem for your case is that the Shindell calculation is based on same aerosol-data that was used in Otto et al, he just adds an obvious additional physical factor. If you say Shindell is very uncertain then your own work is at least as uncertain, so you have no way out (at least if you keep on advocating and still want to appear as an objective scientist).
Regarding the only substance in your comment, Shindell has already responded to that on realclimate.org:
"Alternatively, one could assume we can estimate forcing during this early period realistically enough to remove if from the longer 1750-2000 estimates, and so compare forcing and response over 1850-2000. In this case, this must be done for all forcings, not just for the aerosols......If this is done consistently, the denominator of the climate sensitivity calculation containing total forcing barely changes and hence the TCR results are essentially the same (a change of only 0.03°C). Lewis’ claim that the my TCR results are mistaken because they did not account for 1750-1850 aerosol forcing is incorrect because he fails to use consistent time periods for all forcing agents. The results are in fact quite robust to either analysis option provided they are done consistently." - See more at: http://www.realclimate.org/index.php/archives/2014/04/shindell-on-constraining-the-transient-climate-response/#sthash.JFGn0aWY.dpuf
Whenever I read these " retro-fit" papers, I think of Garth Paltridge's comment, " There is a fair amount of science behind the global warming debate , but in general, and give or take a religion or two,never has quite so much rubbish been espoused by so many on so little evidence."
Dessler's specialty is arm waving after the fact excuses to help climate hype believers stay true to the faith.
That Dessler's excuses don't hold up under scrutiny is ironically his job security: The more climate predictions fail, the more the climate obsessed demand sciencey sounding excuses as to why cliamte apocalypse is still on track. This is like Paul Ehrlich and his schtick of population doom: He has never been right, but for the fools that believe him, it just means there is a final denouement coming at the end of the narrative to show that he was right after all.
@John L
"OK, you are apparently an advocate for low sensitivity, energy balance methods do not become robust just because you want to or just say so, you have to do the science first."
You know when a comment gets off to a bad start when the commenter is an anonymous person admonishing a published researcher in his area of specialty, to go do his homework...
"Even before recent papers on the energy balance calculations, 3/4 methods agree about 3C sensitivity with some likely interval 2-4.5 or so: GCMs, paleo and interannual variations. Just denying these does not make them less relevant in the rational world."
I think what you're saying here is that GCM's (theories), paleo (speculation combined with theory) and very short term climate responses (interannual variations) -- are as important as, or more important than, actual long term empirical observations...
Hmm... that may be a conclusion, but it doesn't sound much like one found in a rational world.
Will Nitschke,
"You know when a comment gets off to a bad start when the commenter is an anonymous person admonishing a published researcher in his area of specialty, to go do his homework..."
So you dismiss the majority of comments on this thread, this blog and other "skeptical" blogs, which is all about this.
But what really matters is arguments, look at my comment, I do show with explicit arguments why Nic Lewis is wrong.
@John L
You do realize you are an anonymous commentator with no credibility, no expertise, no record of scientific publishing, etc., and you're lecturing an academic who is a specialist in this area of scientific research that you're right and he's wrong? I just find the lack of total self awareness of activists endlessly fascinating.
Will, John
Let's focus on the scientific arguments. Please watch tone of your comments so that the thread doesn't get derailed.
John
As Nic noted in his GWPF report, the IPCC suggests in AR5 that apart from estimates based on observational record and GCMs, other methods are unreliable. Do you disagree? (Worth pointing out that Forster and Gregory, based on satellite obs, also found low ECS.)
Bish that's cruel. If the list is reduced to just the observational record and GCMs, and the GCMs are as Robert G. Brown described earlier this week, Nic Lewis has a slam-dunk.
Well, paleo is ruled out even before you get out of the starting gate. Academics are struggling to quantify what is happening right now and what has happened as little as 100 years ago, yet paleo purports to be evidence based on fanciful reconstructions of what may have happened hundreds of thousands or millions of years ago. If it's evidence of anything it certainly should be given very little weight.
Will: Yes, I've been amazed by the rise of arguments from such sparse paleo data. But the Bish as ever has his finger on the zeitgeist, pointing out that the IPCC agrees with us. Its remaining blindspot is the complete inadequacy of the GCMs but that too, one feels, is being taken care of by the best of the blogosphere. And these intellectual foundations for CAGW do matter. Increasing strength to Nic's elbow.
Bishop Hill,
IPCC cites PALAEOSENS which was a summary of all paleo-studies. They gave a combined 68% interval of 2.2–4.8 K. So the experts who studied the issue have quantified the uncertainty. "are unreliable" doesn't really mean anything unless you quantify it and review every method with the same fair and critical eye. Similarily, Will Nitschke's opinions on the issue has no meaning to me. Especially as he disregards arguments against greater experts, which then hits back onto himself as well at Nic Lewis which has much less merits than for example Dessler.
So paleo-studies agrees very well with complex climate model calculations, which have a similar range. Dessler mentions the "interannual period studies" which is based on ENSO and satellite observations of feedbacks (water , clouds, albedo etc.).
It is not all clear why the energy balance methods applied to the 20th century should be regarded more reliable, it looks rather the opposite:
For example, aside Shindell:
Rose et al 2014: "Results imply that global and regional warming rates depend sensitively on regional ocean processes setting the OHU pattern, and that equilibrium climate sensitivity cannot be reliably estimated from transient observations."
Armour et al 2013:
" While the global climate feedback is often assumed to be constant, its value—diagnosed from global climate models—shows substantial time variation under transient warming."
Olsen et al 2013:
"The results open the possibility that recent climate sensitivity estimates from global observations and EMICs are systematically considerably lower or higher than the truth, since they are typically based on the same realization of climate variability."
The simplest answer would be that energy balance methods (based on long term observations and physical measurements) while containing many approximations, still contain vastly fewer assumptions than other methods, such as the huge number of assumptions contained in paleo studies, or the problematical issue of GCM's running consistently too hot. A method that agreed with GCM's might be an argument for consistency but also an argument against physical reality.
Models are not running too hot, or rather, you don't know yet if the models eventually will run too hot or to cold.
Current global average land temperature are still within normal error bars, you cannot say they run too hot just because there is a current deviation to the hot side, that is not how to do statistical interferences.
When evaluating models you need to take into account uncertainty and bias in assumed forcings, uncertainty in temperature measurements and natural variability like ENSO-oscillations that are not synced between models and the real world, even if the long term trend might be the same.
A 100% perfect physical model down to individual quarks would still show some more or less random deviation to the real world, perhaps like the current average, due too measurements errors and small differences in the initialization state, causing out of synch natural variability.
You can speculate as much as you want about uncertainties of different methods, it is still only amateur opinions which has little correlation with the real evidence. And there is a stiff challenge for Nic Lewis is too back up his opinions in a form that will pass expert review. Currently that looks very unlikely, judging from his comment above.
According to Santer's latest paper models run too hot everywhere except the North polar region, and are outside 95% certainty bounds.
Will Nitschke,
Santer et al 2014 points to that negative forcing from volcanos probably has been underestimated. Best guess explains about 15% of the difference with large uncertainties.
One of the authors has written another paper http://www.nature.com/ngeo/journal/v7/n3/full/ngeo2105.html which concludes that
"errors in volcanic and solar inputs, representations of aerosols, and El Niño evolution, may explain most of the discrepancy."
But it is the nature of things that you don't get an exact answer or a definite answer from these studies.
In any case it is inconsistent to claim large uncertainties, for example in future climate projections, and then say that GCMs definitely are biased high.
"The multimodel average tropospheric temperature trends are outside the 5–95 percentile range of RSS results at most latitudes."
"The likely causes of these biases include forcing errors in the historical simulations (40–42), model response errors (43), remaining errors in satellite temperature estimates (26, 44), and an unusual manifestation of internal variability in the observations (35, 45). These explanations are not mutually exclusive"
Identifying human influences on atmospheric temperature
- Contributed by Benjamin D. Santer, June 22, 2012
http://www.pnas.org/content/early/2012/11/28/1210514109.full.pdf
(Nearly 2 years later, models are doing even worse as there has been no measurable warming since then either.)