Bishop Hill Myles out of line
The reverberations from the Lewis/Crok report are still playing out in the blogosphere. In particular there are some interesting comments at Ed Hawkins' blog.
One of the memes that is being pushed by our climatological friends is the idea that the Lewis/Crok range for transient climate response (i.e. short-term warming) is similar to that of the IPCC models. Myles Allen was the first to promote this idea, in his comments at the Science Media Centre.
Their 5-95% range of uncertainty in TCR (kindly provided by Nic Lewis) is 0.9-2.5 degrees C, almost exactly in line with the range of the models shown in their figure (1.1-2.6 degrees C).
This idea has now been repeated by Ed Hawkins:
Focussing on TCR, Lewis & Crok give a ‘likely’ range of 1.0-2.0K. The same IPCC AR5 range is 1.0-2.5K, taking into account other lines of evidence. The 5-95% ranges from Lewis & Crok are almost the same as the 5-95% ranges from the GCMs (Nic Lewis confirmed this to Myles Allen).
Well up to a point, Lord Copper. As Troy Masters - another independent scientist with publications in the area - has pointed out in the comments at Ed Hawkins' blog, a 5-95% range, being so wide, can hide a multitude of sins, including for example vastly different skews from one distribution to another.
...these ranges can be similar while at the same time the specifics of the actual distributions (median, mode, etc.) can be very different and have largely different policy/impact implications. You are right that the similarity in these ranges between Lewis and Crok and the IPCC report might make the distinction insignificant (although estimated impacts/costs are disproportionately affected by the upper bound) IF the IPCC did not imply knowledge of the distribution within that range of TCRs. However, while the report does not explicitly give a “most likely” value for TCR, it implicitly suggests one with the statement in the SPM that global surface temperatures are “more likely than not to exceed 2C [above pre-industial by 2100] for RCP4.5 (high confidence)”. This is based on the GCMs with an average TCR of 1.8 K, significantly higher than the estimated “most likely” value used in Lewis and Crok. Similarly, I think most people would agree that if TCR is 1.0 K and ECS is 1.5 K (both within the “likely” range of the reports), we are unlikely to hit that 2K target under RCP 6.0 by 2100, but again AR5 says “warming is likely to exceed 2C for RCP6.0…(high confidence)”, thereby implicitly assigning a low probability to the lower end of that range.
I don’t think the IPCC report can have it both ways. Either the SPM must say “we don’t know” about whether we are likely to exceed 2K under RCP4.5 and RCP6.0, because the likely range of TCRs include those that exceed or fall below this mark, or it must stop saying that estimates of TCR are consistent simply because the ranges are consistent (as those statements rely on more specific aspects of the distribution). Given that RCP4.5 includes increasing emissions up to 2040, and RCP6.0 increasing emissions up to 2060 (albeit neither at the rate of RCP8.5), consider the difference of the implications when using the IPCC implicitly assumed distribution of TCR estimates vs. the explicit most likely values of Lewis and Crok:
IPCC AR5 using implictly assumed TCR distribution: we are more likely than not to exceed 2K by 2100 under the scenario where emissions continue to increase up to 2040, and likely to exceed 2K under the scenario where emissions continue to increase up to 2060 (high confidence).
Lewis and Crok: we are unlikely to exceed 2K by 2100 under the scenario where emissions continue to increase up to 2040, and more likely than not to stay below 2K by 2100 under the scenario where emissions continue to increase up to 2060.
In both cases we might say the TCR ranges are the same, but they convey largely different messages to policy makers.
In fact it's worse than that, because although the IPCC's CMIP5 models have an average TCR of the order of 1.8°C, as Lewis and Crok point out in their report the amount of warming that the models predict is even higher than this number implies.
It is actually possible to show the very significant discrepancy between what the IPCC is telling policymakers and Nic Lewis's work, based on the AR5 forcing and heat uptake data and the observational studies. In Figure 12.8 of the Fifth Assessment, the IPCC reports the range of projected warming under the various RCP scenarios, including the 5-95% and 17-83% ranges as well as the medians. Nic Lewis has kindly sent me the equivalent projected figures, calculated using a two-box model and using the various key percentiles from the TCR and ECS ranges in his report. I have drawn these (in blue) next to the IPCC's figures (in red - main warming projections 17-83% ranges and means, based on the ensemble of CMIP5 models; in yellow - 5-95% and 17-83% ranges and medians from AR5 Figure 12.8, based on Rogelj et al 2012).*
;)
The difference, as I'm sure you will agree, is pretty stark. But Lewis's results are, in climatological speak, "almost the same" as the IPCC's.
One last thing. I wondered how it was that Nic Lewis could have provided Myles Allen with the extra data about the range of his TCR estimate without the details of the skewness getting passed on too. Did he not pass on the likely range too?
I asked Nic about this. It turns out he provided Myles with the 'likely' range first, and Myles then asked for the 5-95% range. Nic was worried that quoting just the 5-95% range would mislead people. So he sent Myles the 5-95% range, stating that he was happy for Myles to quote it as long as he also gave his (Nic's) median and likely range.
Oh dear.
*Nic asks me to note the following caveats: The best estimates and likely ranges are not identical to those given in the Lewis and Crok report (although the TCR best estimate is virtually the same): the values given in the report have been made more conservative. The percentiles that these projections represent will only approximate the percentiles of the TCR and ECS values used, because some uncertainty is not common to TCR and ECS.
Reader Comments (71)
Nic
1. Yes we agree
2. Regarding the red boxes, you say:
The Bish's post actually says:
So I'm not sure why you disagree with my point here, but to be honest I'm not bothered enough to argue it further!
3. OK, just a coincidence then. As you say, it's not surprising I thought there was an issue here. I guess this is the difficulty of blog posts as opposed to papers - general the technical detail needed to understand exactly what was done is not included. Which set of numbers do you regard as your most reliable ones?
4. I think all information that helps describe the distribution is important, so means as well as medians (as long as one compares like with like). Aldrin et al showed a skewed distribution and reported the mean. What are your mean values?
So to summarise, your projection for global warming by 2081-2100 under RCP8.5, relative to 1985-2005, is:
5-95% range: warming of 1.5-4.0 degrees C
17-83% range: warming of 1.7-3.0 degrees C
median: warming of 2.1 degrees C
(this is reading off the Bish's figure by eye - happy to have the numbers corrected if necessary)
Richard,
Or in other words, Lewis' figures for RCP8.5 actually give predictions almost the same as the next IPCC scenario down which is RCP6.0. Perhaps you can confirm whether you agree with this statement?
Perhaps you might also agree that if Lewis' figures are correct this would be great news?
thinkingscientist: That would be agreement indeed. Great news shared is a problem blown to pieces?
Richard Betts @ 5:11pm
//
"4. I think all information that helps describe the distribution is important, so means as well as medians (as long as one compares like with like). Aldrin et al showed a skewed distribution and reported the mean. What are your mean values?"
//
What will you learn from the mean values?
A project for Geoff Chambers, and others.
In time for the Christmas season 2014, and perhaps in the spirit of Gilbert & Sullivan, perhaps as a pantomime in our great tradition of same, create an entertainment to afford us some enlightenment and amusement.
Words and music. Actors and musicians. A YouTube product to tell future generations how we saw things. Any volunteers?
Guest spots. Dame Julia as herself. Richard as Buttons, her helper but yet with an impish rapport with the audience and a hint that he might help, in the end, defeat the bad guys and see that all turns out for the best. Michael Mann as a sort of decidedly uncongenial uncle up to all sorts of legal actions to protect his interests. Jim Hansen as a twisted version of Homer, lacking in charm and obsessed with some peculiar vision with which he wants to darken everyone's life. And so on.
Lewis as the gentle prince searching for truth and goodness and who in the end finds them, living out his days in contentment, admired by one and all. The Bish as the Bish, rogue prelate on a mission to expose wrongdoers, brainless bletherers, and worse, and show such benighted ones the errors of their ways, ever believing that redemption is possible for even the worst of them.
Maybe in the midst of the misery and destruction and intellectual and moral degradation brought to us by the CO2 campaigners, there is yet some scope for a chuckle? Or maybe not. It is all so grim and wasteful.
thinkingscientist
Yes, it's pretty obvious that Lewis's RCP8.5 warming seems reasonably similar to the IPCC's RCP6.0, but with Lewis's upper end of the range being higher.
And yes, if this were correct then it would indeed be great news.
Emphasis on the "if" of course…. ;-)
John Shade
Audience participation should be fun.
The cast:
3% of the audience:
97% of the audience:
*grins impishly*
Hopefully it'd play outside the academic circuit too.
In front of an impartial BBC invited audience.
Richard Betts
"Emphasis on the "if" of course…. ;-) "
If there's one person who couldn't answer that, it is you.
Thanks to all for an interesting comment thread -- and, of course, many thanks to Our Host for starting the discussion.
Special thanks to Thinking Scientist, who appears to be taking something of a Steve McIntyre approach to auditing published work. It is especially noteworthy that we often-derided "industry shills" take a more responsible -- and certainly more accountable! -- approach to reporting our results. It does seem that our publicly-funded scientists should (at least) meet the minimum stnadardsof reporting prescribed for oil & gas and mining-company resource-reserve claims.
Cheers -- Pete Tillman
Mining geologist (retired)
--
"It is difficult to get a man to understand something when his job depends on not understanding it."
-- Upton Sinclair
Can I by any chance be the Unidentified Friend of the Master of the Horse? Or, failing that, Kublai Snoo? (Though, I must confess, I can't wiggle my ears.)
Peter Tillman, your comments are much appreciated, although comparing my hastily penned and lightly researched contributions to this thread to the detailed and exhaustive investigations of Steve McIntyre is rather more than a stretch.
But thanks away.
RB, your imagined audience is a peculiar one, made up it would seem of a non-random sample from a frame that held an ill-defined group of people, most of whom sensibly declined to take part in the sloppy survey that first led to the 97%. I've had a notion for a post on that rather malodorous statistic for a while now. I'll try to get it done tomorrow, and if I do, I'll put a link to it here. In the meantime, I appreciate your attempt to add some humour. The spin of the 97% is a fitting item given the Spin of the Ranges which is our proper topic here. The97% one has had a lot of mileage, but I think the Range one may be being nipped in the bud.
Jonathan Jones - I can and so can my children. I do not intend to play the ass though.
Richard
2. OK, we'll agree to disagree; as you say the point is not important.
3. Agree re blog posts often having less detail - although technical ones of my own probably have as much detail as you find in many papers. But the good thing about blog posts is that you can ask about missing detail and get a quick answer (on sceptics' blogs at least), so the lack of detail doesn't matter so much. I've asked a colleague of yours about missing details and inconsistencies on a published paper and never got an answer (not his own fault, I think).
The best estimate and likely range in our report were chosen to be more conservative than those derived purely from the AR5 forcing and heat uptake and HadCRUT4 distributions (+ internal variability). Which one regards as more reliable is a difficult call. The median warming projections are almost identical on both bases; the likely range from the report, being set more conservatively, gives projections extending further out on each side of the median.
4. I entirely disagree re means vs medians. And neither do I normally quote modes, which are lower than medians for ECS, TCR and projected warming. If you look at Figs. 10.20 and Fig. 1 Box 12.2 in AR5, you'll see that they do the same as me. I of course agree that one needs more information about a distribution than just the median, but the appropriate data are the values at different percentiles - 5%, 17%, 50%, 83% and 95% (likely and very likely ranges, plus the median) generally provide adequate information. They are all I have calculated. Means and modes aren't invariant under reparameterisation (e.g. from ECS to its reciprocal the climate feedback parameter); medians are. I can't help it that Magne Aldrin quoted means – he's about the only author who has. The AR5 scientists, quite rightly, ignored his means, obtained the full distributions and showed only the medians as the central values.
You have my AR5 forcing etc data based projections for global warming by 2081-2100 under RCP8.5, relative to 1985-2005, almost right. They are:
5-95% range: warming of 1.5-4.0 degrees C
17-83% range: warming of 1.7-2.9 degrees C
median: warming of 2.1 degrees C
rounded to the nearest 0.1 C.
But it is not appropriate just to look at RCP8.5 - that seems to be nearer a worst case than a business-as-usual scenario (see my comment at http://www.economist.com/comment/2322688#comment-2322688). On RCP6.0 I make the corresponding figures:
5-95% range: warming of 0.9-2.4 degrees C
17-83% range: warming of 1.0-1.7 degrees C
median: warming of 1.3 degrees C
We have already experienced 0.15 C of the above rises, of course.
Alas, I have nearly run out of time today to complete that post I mentioned at 9:37 PM yesterday. It will be Sunday before I can get back to (hopefully) complete it. It is a bit marginal for this thread, and so a further delay is fine by me since my conscience will be less troubled then about the risk of diverting the comments off topic ( since the most active phase of discussion here may well be over by then).
Back on topic, Nic Lewis is doing an admirable job of clarification and confirmation of his take on it, and the civil exchanges and comments from Richard Betts about it are also helpful for that end.
An excellent thread. I'd nominate it for inclusion in the Key Post archive.
Bumping Richard Betts at 5:11pm 13 March - please can you tell us what you would learn from comparing mean values in this context? Nic Lewis's answer at 10:38pm suggests it would be a pointless exercise - presumably you disagree?
"Occam's Broom and the Stink of '97% of Climate Scientists'"
That's my post on the 97% statistic, a statistic which I think stinks to high heaven. I got it done at last on Friday, just before heading for the hills for a few days. It can be found here: http://climatelessons.blogspot.co.uk/2014/03/occams-broom-and-stink-of-97-of-climate.html
If anyone has been waiting up for this, sorry to take so long about it after my promise here on 13th March.