2011-11-26 06:28:51Schmittner et al. (2011) on Climate Sensitivity - the Good, the Bad, and the Ugly
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

My take on Schmittner et al.:

Schmittner et al. (2011) on Climate Sensitivity - the Good, the Bad, and the Ugly

2011-11-26 06:54:24
Agnostic

mikepope_9@hotmail...
118.208.124.229

Dana - an excellent article.  But what about the way Schmittner deals with (or doesn't deal with) slow feedbacks - or isn't my earlier comment relevent?

2011-11-26 07:01:28
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

I sort of addressed that by specifying "fast feedbacks" a couple of times, and by adding the BBC quote that cold-to-warm sensitivity isn't necessarily the same as warm-to-hot sensitivity.  But I'll add a bit to be more explicit about that.

I don't think his low sensitivity estimate is based on excluding slow feedbacks, because like Hansen's 3°C estimate, this is fast feedbacks only.

2011-11-26 07:18:12
Agnostic

mikepope_9@hotmail...
118.208.124.229

Hmm.  But paleoclimate evidence from the Eemian maxima has slow feedback effects (albedo, methane, sea level) "built-in" as do present confitions whereas glacial maxima excludes them and because their effect is unquantified, they can not be be added in or excluded can they?  Sorry if I have got this wrong.

2011-11-26 07:58:28
Daniel Bailey
Daniel Bailey
yooper49855@hotmail...
97.83.150.37

Schmittner quote @ RC (h/t to john byatt):

john byatt says:

After all the poor science journalism and opinion on Schmittner’s paper he has stated the following.

“While our statistical analysis calculates that high climate sensitivities have very low probabilities, you can see from the caveats in our paper, and my remarks in this interview, that we have not actually claimed to have disproven high climate sensitivities….Our study comes with a number of important caveats, which highlight simplifying assumptions and possible inconsistencies. These have to be tested further.”

Unsure where john got the quote from, however.

2011-11-26 08:05:26
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Agnostic - Hansen has also given sensitivity estimates when including slower feedbacks (see James' post, which is linked within my post in a few places, including the 'fast feedback' text).  But his fast feedback only estimate is 3°C.  That includes his estimate based on the LGM (see the figure in my post).

2011-11-26 08:30:57
Agnostic

mikepope_9@hotmail...
118.208.124.229

Dana - I repeat ... an excellent article

2011-11-26 08:31:07
Albatross
Julian Brimelow
stomatalaperture@gmail...
198.53.65.169

Hi Dana,

Looks good.  Just gave is a perusal though.

Given that deniers are trying to use this study for inaction on AGW or for not drastically reducing emissions, I think it would be prudent to highlight this comment made by the lead author when interveiwed by ScienceDaily:

"Hence, drastic changes over land can be expected," he said. "However, our study implies that we still have time to prevent that from happening, if we make a concerted effort to change course soon".

2011-11-26 09:10:16
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Agnostic - thanks.

Albatross - have added that quote.  Frankly I think the study means we have less time to prevent drastic changes from happening, for reasons outlined in the post.  If sensitivity were 2.4°C and glacial-interglacial transitions were 5°C, then he'd have a point.  But if those transitions are just 2.2°C, and we're already at 0.8°C, it seems to me that means we're pretty much out of time.  But then again I really don't think they're right about glacial-interglacial transitions just being a 2.2°C change, so I don't put much stock in their results in general.  Frankly I'm a little surprised they got published in Science with that low estimate, and only using UVic.

2011-11-26 09:42:03
Albatross
Julian Brimelow
stomatalaperture@gmail...
198.53.65.169

Dana,

Thanks. I hear what you are saying about how much time we have. To bew candid, (and I am not a climate scienetist) I was expecting the kind of extrme weather we are seeing of late to only occur circa 2020.  Everything seems to be happening quite fast IMHO, certainly faster than I expected.

From what I understand and have read, possibly their  biggest contribution to the science is in their methodology.  James Annan liked their method of analyzing the data and he , as you know, it pretty picky.  So I'm hoping that they have established a framework upon which others can build upon using models like the GISS-E.

Yes, I am dubious of the 2.2 difference, but I'm pretty sure that if not robust, someone will fix that very soon.

As I have said before it is a paradoxical paper, arguing for somehwat lower sensitivity bringing about huge changes..greater minds than mine are needed to unravel this.  This is by no means the last paper on climate sensitivity, so the skeptics were too zealous to try and spin it as such.

2011-11-26 10:33:57
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

I added some info from tb's interview with Nate Urban.  He clarified that while the proxies they used showed 2.2°C LGM cooling, globally their models actually estimated the cooling at 3.3°C.  Their global coverage from their proxies is rather sparse.  Urban also explained that most of their lower sensitivity boils down to that lower LGM cooling value.   So if it's determined that the LGM value is too warm, that would bring their sensitivity estimates up as well.  But the LGM reconstruction is a new one, so it remains to be seen whether it will stand the test of time.

2011-11-26 10:59:54
Albatross
Julian Brimelow
stomatalaperture@gmail...
198.53.65.169

Thanks Dana, hmmm, interesting. 

Someone just emailed me asking about this after seeing the denialist spin.  So I will forward them to your post when it goes live :)

2011-11-26 11:28:24Dammit, Dana, I needed this post 24 hours ago
John Cook

john@skepticalscience...
143.238.130.246

Did a public talk last night and a skeptic in the audience asked about Schmittner!

I suggest adding a new skeptic argument 'Schmittner finds low climate sensitivity' as a sub-argument to 'climate sensitivity is low' and adding this as a rebuttal.

2011-11-26 11:33:44
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Heh yeah just saw your email, sorry John! :-)  Will create that rebuttal.

I hope you were able to come up with a decent answer for that skeptic in the audience.

2011-11-26 11:52:54
skywatcher

andycasely@hotmail...
122.107.164.176

Good article - the BBC is already mangling this somewhat, at least in the headline (what's new), so a good explanation is necessary.  Is it possible to estimate what the Schmittner EQS is likely to be on the assumption they had used a 5C LGM-Holocene change rather than a 3.3C LGM-Holocene change?  I don't suppose it linearly scales up, but I'm guessing that it would place the sensitivity estimate right in the ballpark of teh Knutti and Hegerl estimates.  Thus, when you account for the alterations of the experiment, Schmittner ends up telling us a similar story as before, but because the numbers are lower and our absolute rate of warming remains unchanged, the effects are actually worse.  A stange thing for the deniers to be happy about.

Regardless of that, thumbs up from me, I don't see any obvious errors either.

2011-11-26 12:00:19
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Thanks skywatcher.  Unfortunately it's not so simple as a linear scaling up, but I believe I mentioned (in a recent revision to the post) that Urban explained (in his interview with thingsbreak on Planet 3.0) that almost all of the discrepancy boils down to their warmer LGM estimate.  Urban ran the numbers with the simple linear scaling (noting that it's a big oversimplification) and showed that the warmer LGM accounts for basically the entire difference between their sensitivity and Hegerl/IPCC.

I suspect their LGM is too warm because the Hegerl/IPCC estimates are based on a whole bunch of different lines of evidence, not just the LGM.  And when Hansen used 5°C for the LGM, his estimate lined up with Hegerl/IPCC.  But the Schmittner LGM is based on a lot of proxy data, so it should be a pretty good estimate.  Then again, it's heavily weighted towards the oceans, which is why their proxy-based estimate (2.2°C) was lower than a global average estimate (3.3°C).

Bottom line, it's just one study, and remains to be seen whether their reconstruction withstands the test of time.  And it's also one more study ruling out low sensitivity, so it's funny that the denialists are so happy about it.

2011-11-26 12:53:24
Albatross
Julian Brimelow
stomatalaperture@gmail...
198.53.65.169

Dana have you seen Hegerl and Russon's perspective piece?

2011-11-26 14:15:45
skywatcher

andycasely@hotmail...
122.107.164.176

Ha! Tom (Russon)'s a friend of mine, nice he's got something in Science!

2011-11-26 14:33:39
Albatross
Julian Brimelow
stomatalaperture@gmail...
198.53.65.169

You know what is incredibly ironic about this folks?  Skeptics like Lindzen, Watts, Spencer , Michaels etc. do not trust the paleo data and models--they say paleo data is a "grey area" and we all know what they think of those ridiculous/useless models.  Intriguing now then to see "skeptics" endorsing this work, getting all excited and spinning its findings as proof that climate sensitivity is low, when it relies heavily on both paleo data and models (and not a particulalry good model at that)....

Maybe that should/could be noted prominently?  It just goes to show how they will latch onto anyhting that supports their belief/ideology, even if their argument is internally inconsistent and/or contradictory.

2011-11-26 16:09:54
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

I did Alby - John emailed it to me.  They didn't really say anything that I hadn't already said in the post, so I didn't add anything from their piece.  It was a good summary though.

You also reminded me of the fact that when we tried to pin Pielke on the low sensitivity-paleoclimate contradiction, he said there was too much uncertainty about paleoclimate conditions to compare to today.  I might add a bit about that, but not too much, because the post has already gotten rather long.

I see on his blog Pielke has stooped to obsessing over the stolen emails as well.  Not surprising, but I had hoped he was better than that.

2011-11-26 16:47:28
Andy S

skucea@telus...
66.183.179.88

Mostly very good, especially the second half. Some points:

A new paper in Science from Schmittner et al. (2011) attempts to constrain climate sensitivitybased on temperature reconstructions of the Last Glacial Maximum (LGM): [suggest you add 19-23,000 years ago]

uses a relativley new global (typo)

 Marginal posterior probability distributions for equilibrium climate sensitivity to doubled atmospheric CO2 (ECS2xC) from Schmittner et al. (2011) 

I have no clue why these distributions are so lumpy, can anyone explain why?

Further, James Annan says:

remember, they are not estimates of "the pdf of sensitivity" but rather, probabilistic estimates of the sensitivity - but they do need to overlap in order to be taken seriously

I'm not clear what Annan is getting at here (the distinction is lost on me).

 just 2.2°C cooler than interglacial temperatures Is that interglacial or post glacial? (given that the Eemian interglacial was a degree or so warmer than the Holocene, if my memory of Steve Brown's articles serves).

A point I read somewhere (perhaps in Gavin Schmidt's comments) is that higher sensitivity figures are needed to explain the Pliocene; perhaps different sensitivities apply in different environments. I would argue that we're now closer to the Pliocene (400ppm, but 2-3 degrees hotter than now) than we are to the conditions of the LGM and we're headed, via the Miocene, for the Oligocene (>600 ppm, no ice caps). For me, that's the biggest stumbling block that prevents me feeling relieved about this study. Perhaps it's because I'm more of a geologist than a physicist. The Earth at the last glacial maximum was a different planet.

2011-11-26 17:00:53
Andy S

skucea@telus...
66.183.179.88

I think maybe you go a little too far into "meta" territory with sentences like this

Somehow the climate denialists glossed over this aspect in their reporting on the paper, which is further proof that they are not true skeptics.

I fully agree, of course, but when you say it like that  it looks a bit like point-scoring and detracts from the rest of the article.

2011-11-26 19:56:14
John Mason

johntherock@btopenworld...
86.134.155.184

Good piece - one thing that could be changed is the footnote below the first paragraph below fig.4 - on my screen it is very difficult to make out.

 

Cheers - John

2011-11-26 22:41:17
Tom Curtis

t.r.curtis@gmail...
112.213.154.102

1)  You should consider eliminating the lower half of figure 1.  Although interesting in itself, it is complex and does not feature in your discussion, so its inclusion is likely to shed more confusion than light.

 

2)  The 3.3 degree global cooling figure is in error.  At line 138 it is indicated that the cooling in Surface Air Temperature is -3.3 degrees K, while for Sea Surface Temperature, the cooling is -2.0 degrees K.  Area weighted, that indicates a reduction in global mean surface temperature of around 2.4 degrees K.  For comparison, at lines 157 and 158 we learn the reduction averaged over gridpoints with data is 2.2 degrees K, while it would be 2.6 degrees K globally averaged.  There appears to be an inconsitency, however, with the claim at line 146 that their reconstructed temperature reductions are only 30 to 40% less than other reconstructions.

Although I believe the post to be excellent as it stands, and highly desirable that it be posted, we should probably clarrify this apparent inconsistency.  If we are unable to do so, then the increase in estimated climate sensitivity due to the low effective estimate of global temperatures is between 50 to 100% rather than  40 to 70%.

As a side note, in the Urban interview, he correctly describes the 3.3 degree reduction as a reduction in SAT, but incorrectly compares it to the 5 degree reduction in GMST commonly found.

 

3)  Where you say that the study uses a 3.3 degree reduction globally, that is incorrect in that the study methodology makes no use of the global mean data.   The phrase:

"For example, the study uses a relativley new global mean surface temperature reconstruction for the LGM of just 2.2°C cooler than interglacial temperatures in the locations where they have proxy data, or 3.3°C globally. "

should be rephrased as:

"For example, the study uses proxies which indicate much less cooling than has been found in previous studies.  They find the LGM was just 2.2oC cooler in the locations where they have proxy data, which is approximately the equivalent of 2.4oC globally."

Obviously the change in global equivalent is based on point 2 above.  You may prefer 2.6 degrees C as having more justification from the text, or to stick with 3.3 degrees C and let me add complexity in the comments ;) 

3)  Given an extensive discussion of this paper, the following is a crucial issue:

"The second reason is limited spatial data coverage. A sensitivity test excluding data from the North Atlantic leads to more than 0.5 K lower ECS2xC estimates (SOM section 7, Figs. S14, 154 S15). This shows that systematic biases can result from ignoring data outside selected regions as done in previous studies (22, 23) and implies that global data coverage is important for estimating ECS2xC. Averaging over all grid points in our model leads to a higher global mean temperature (SST over ocean, SAT over land) change (–2.6 K) than using only grid points where paleo data are available (–2.2 K), suggesting that the existing dataset is still spatially biased towards low latitudes and/or oceans.  Increased spatial  coverage of climate reconstructions is therefore necessary in order to improve ECS2xC estimates."

This lack of spatial coverage, given the method they use is an obvious source of uncertainty in the results, uncertainty which is not obviously included in their PDF.  Given the significant spatial discrepancies in their best fit model, inadequate proxy coverage represents an even greater problem.

However, to discuss this problem will require a proper discussion of their method, which will add substantially to the complexity of the post.  Possibly detailed discussion of the method and additional related problems could be left to a follow up post.

2011-11-27 01:29:16
Kevin C

cowtan@ysbl.york.ac...
144.32.196.109

Does Tamino prefer a lower case t?

tamino found that the CCCMA models were consistent outliers:" could be strengthened by adding ", failing to reproduce 20th Century temperatures".

I think "There are some unusual aspects about this study which may raise some red flags" is not detached enough. How about "There are some unusual aspects about this study which require further investigation before the conclusions of the study can be accepted, as the authors themselves point out."

Otherwise excellent work.

2011-11-27 03:48:07
Albatross
Julian Brimelow
stomatalaperture@gmail...
198.53.65.169

Looks good to me.

2011-11-27 04:20:24
Riccardo

riccardoreitano@tiscali...
2.33.129.69

I find the "ugly news" section a bit confusing.
Fig. 4 does not show that the LGM temperature change will be reached sooner; true it's nearer, but the slopes are different. You need to prolong the curves untill they (or at least the blu one) reach the dashed lines.
I understand the message but I think it's not obvious. You probably need to spend a few more words on the meaning of the dashed lines, i.e. possible dramatic changes like those from glacials to interglacials (like you wrote in the "take-home" section).

In the introduction you could add that they used "Bayesian inference, using the likelihood of the observations DTobs given the model DTmod(ECS2xC) at the locations of the observations." This is important because any regional bias of the model, analysed by the authors at length, will be reflected into the determinantion of the sensitivity.

Did anyone note that both the model and the reconstruction show a warmer than now high latitude (> 75° N) northern hemisphere? It surprises me, I have no explanation.

2011-11-27 04:37:04
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Good comments guys.  Tom, good catch on SAT vs. GMST.  I think it should be good to publish now.

2011-11-27 04:59:09
Rob Painting
Rob
paintingskeri@vodafone.co...
118.92.41.145

Late to this, but I agree with Riccardo on figure 4. It's one of those many things that is poorly understood by most - that lower climate sensitivity implies a quicker response to forcing. Are you able to show the first line in figure 4 intersecting? Maybe insert the extended graph image into the current image? 

The temperature data most certainly cannot be right - quite a few of the ocean proxies around Greenland (in Figure 3 of the paper) have the LGM warmer than today! As Riccardo points out, Figure 4 in the paper has the Artctic warmer than today too!

Nice work Dana, and wicked turnaround time too.

2011-11-27 05:40:27
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Well I had to take an educated guess to extend the data, since IPCC emissions scenarios only go out to 2100, but I revised the figure to show the blue lines intersecting close to 2120.  I'll see if I can clarify the Ugly section a bit.

2011-11-27 13:20:29
thingsbreak

things.break@gmail...
98.204.66.145

Dana and others,

 

1. Where did the "CCCMA = UVic" idea come from? I don't think that's right. Is it actually the case that UVic (vs. CCCMA's models) has a problem repoducing the 20th century instrumental record?

2. As I mentioned before, the surface temp difference was not 2.6°C but rather 3.3°C (this includes the ocean surface) colder. You're combining surface and ocean temps to get 2.6°C, no?

2011-11-28 16:57:14
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Somebody in another thread mentioned CCCMA and UVic.  I looked at a CCCMA page and it mentioned UVic - I thought UVic might have been one of the CCCMA models, but I may have misinterpreted.

Schmittner didn't mention this point in his comment on the post, interestingly.  I think UVic is his model, so I'm surprised he didn't say anything if we got it wrong.  But it's not a very important part of the post, and I can easily remove it.  Another commenter said that part was wrong too.

Schmittner questioned the 2.6°C.  It just depends whether the Hansen and Sato 5°C and IPCC 4 to 7°C LGM cooling estimate is for just surface air, or for both air and sea surface temperatures.  I'll have to see if the IPCC is specific about that, because I don't think Hansen and Sato were.

2011-11-28 20:44:19
Tom Curtis

t.r.curtis@gmail...
112.213.154.102

Dana, here is the paper which describes the UVic Earth System Climate Model (Ref 14 from Schmittner et al).

 

From a quick skim, the UVic model is not the CCCma model.  They are closely related models, however.  The UVic sea ice module was included in CCCMa, and the UVic model is extensively used in developing and testing the CCCma model.

2011-11-28 22:46:52
Riccardo

riccardoreitano@tiscali...
192.84.150.209

A somewhat related paper :

J. C. Hargreaves1 , A. Paul2 , R. Ohgaito1 , A. Abe-Ouchi1,3 , and J. D. Annan1

"Are paleoclimate model ensembles consistent with the MARGO data synthesis?

Here they correctly compare apples to apples, i.e. MARGO SST with model SST. Schmitter et al., instead, use MARGO SST as surface temperature throughout. This is the reason why they find warmer than present high northern latitude. See fig.  2a. Note that the effect of North Atlantic on sensitivity is large, 0.5 K.

2011-11-29 02:12:10
thingsbreak

things.break@gmail...
66.7.151.194

Dana,

 

Nate brought it up to me. It looks wrong to me, but I haven't exhaustively tried to document the model's pedigree. It also should be easy enough to determine whether the claims being made about the model's ability to simulate 20th C temps are correct. From what I remember (this was a while back), they aren't.

2011-11-29 03:43:37
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Romm has a good post on Schmittner too, which quotes Richard Alley agreeing that their LGM reconstruction is too warm, due to the MARGO data.  I'll probably revise the post/rebuttal to include an Alley quote and some MARGO-specific discussion.

The IPCC talks a lot about SSTs, including MARGO, and concludes

"It is thus very likely that the global warming of 4°C to 7°C since the LGM occurred"

This sounds to me like both sea and air surface temps, which would mean that 2.6°C is the correct comparison value, but it's not entirely clear either.

I'll update the UVic discussion to note that as Tom found, it's related to CCCMA, but not the same.

2011-11-29 03:52:13
Daniel Bailey
Daniel Bailey
yooper49855@hotmail...
24.213.18.68

RC just posted on it.  Nate Urban weighs in.

2011-11-30 03:34:51
Tom Curtis

t.r.curtis@gmail...
112.213.154.102

dana, I disagree with the claim that Romm's post is good.  He is excessively, and IMO unwarrantedly critical of the paper.  We need to distinguish between papers which use a reasonable method and a justifiable selection of data and come up with wrong results, and those which use poor methods, or unjustifiable data choices.  The latter type are flawed, and need to be criticized as such, but Schmitter et al does not fall in that category.  The former category includes such illustrious company as MBH98.  That is the sort of thing that happens in emperical sciences because they are emperical sciences.  Such papers can represent major advances both by demonstrating new and valuable techniques, and because correcting them will hone in on the truth.

 

IMO RC and you have adopted the correct perspective.  Schmittner et al is a valuable paper, but it should not be over interpreted.  There are reasonable, but not certain grounds to think its climate sensitivity estimate is too low, and more research is required to show whether it or the host of other studies showing a higher climate sensitivity is more accurate.

2011-11-30 04:18:34
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Well I agree with you on Romm's tone, Tom.  But his post did  have a lot of useful information, like the comments from Richard Alley.  That's the part I liked.

2011-11-30 04:25:22
Alex C

coultera@umich...
67.194.32.234

I still feel that Schmittner's complaint about the 2.6˚C figure comes from you saying the figure is from the data, which it is not.  It is the global model average.