UPDATE: Steve McIntyre has posted on this in his own unique style.
And in case you missed the superbowl of blogs WUWT also has a post here.
After ten months of reviews and rewrites we have successfully published an improved version of Steig et al. 2009. While we cannot publish the paper here, we can discuss the detail. Personally I’ve never seen so much work put into a single paper as Ryan did and it’s wonderful to see it come to a successful conclusion. This is the initial post on the subject, in the coming weeks there will be more to follow.
Guest post by lead author Ryan O’Donnel.
DOING IT OURSELVES. . . a tongue-in-cheek reference to the RC post here:
Improved methods for PCA-based reconstructions: case study using the Steig et al. (2009) Antarctic temperature reconstruction
(Accepted 11/30/10, Journal of Climate)
A detailed analysis is presented of a recently published Antarctic temperature reconstruction that combines satellite and ground information using a regularized expectation-maximization algorithm. Though the general reconstruction concept has merit, it is susceptible to spurious results for both temperature trends and patterns. The deficiencies include: (a) improper calibration of satellite data; (b) improper determination of spatial structure during infilling; and (c) suboptimal determination of regularization parameters, particularly with respect to satellite principal component retention. We propose two methods to resolve these issues. One utilizes temporal relationships between the satellite and ground data; the other combines ground data with only the spatial component of the satellite data. Both improved methods yield similar results that disagree with the previous method in several aspects. Rather than finding warming concentrated in West Antarctica, we find warming over the period of 1957-2006 to be concentrated in the Peninsula (≈0.35oC decade-1). We also show average trends for the continent, East Antarctica, and West Antarctica that are half or less than that found using the unimproved method. Notably, though we find warming in West Antarctica to be smaller in magnitude, we find that statistically significant warming extends at least as far as Marie Byrd Land. We also find differences in the seasonal patterns of temperature change, with winter and fall showing the largest differences and spring and summer showing negligible differences outside of the Peninsula.
Copyright © 2010 American Meteorological Association
(early online release to be available on or around Dec. 7th)
Some of you remember that we intended to submit the analysis of the Steig Antarctic reconstruction for publication. That was quite some time ago . . . and then you heard nothing. We did, indeed, submit a paper to Journal of Climate in February. The review process unfortunately took longer than expected, primarily due to one reviewer in particular. The total number of pages dedicated by just that reviewer alone and our subsequent responses – was 88 single-spaced pages, or more than 10 times the length of the paper. Another contributor to the length of time from submission to acceptance was a hardware upgrade to the AMS servers that went horribly wrong, heaping a load of extra work on the Journal of Climate editorial staff.
With that being said, I am quite satisfied that the review process was fair and equitable, although I do believe excessive deference was paid to this one particular reviewer at the beginning of the process. While the other two reviews were positive (and contained many good suggestions for improvement of the manuscript), the other review was quite negative. As the situation progressed, however, the editor at Journal of Climate – Dr. Anthony Broccoli – added a fourth reviewer to obtain another opinion, which was also positive. My feeling is that Dr. Broccoli did a commendable job of sorting through a series of lengthy reviews and replies in order to ensure that the decision made was the correct one.
The results in the paper are generally similar to the in-process analysis that was posted at CA and here prior to the submission. Overall, we find that the Steig reconstruction overestimated the continental trends and underestimated the Peninsula – though our analysis found that the trend in West Antarctica was, indeed, statistically significant. I would hope that our paper is not seen as a repudiation of Steig’s results, but rather as an improvement.
In my opinion, the Steig reconstruction was quite clever, and the general concept was sound. A few of the choices made during implementation were incorrect; a few were suboptimal. Importantly, if those are corrected, some of the results change. Also importantly, some do not. Hopefully some of the cautions outlined in our paper are incorporated into other, future work. Time will tell!
Lastly, I’ll give a shout out to other folks whose comments helped shape the paper by their comments and analysis. In particular, Roman, Hu, and Carrick . . . thanks!
290 thoughts on “Doing it Ourselves”
Are you at liberty to post any of that? Sounds interesting…
Congrats Ryan, Nicholas, Steve, and Jeff…
Will we ever find out who the “one reviewer” was? Probably not, but it is not hard to speculate he is on the Team.
So any chance of publishing reviews both positive and negative ?
It would be interesting to look at the reasoning behind the comments. It seems only fair … after all we did get a front seat to see the paper unfold. It would be interesting to compare issues and comments about the paper generated at various blogs with what the actual reviewers commented on.
Great job and thanks to all the authors.
Will you get coverage at Real Climate??? Probably not; more likely badmouthing.
Congratulations Jeff and all
Thanks everyone but Ryan and Nic deserve most of the credit here. Steve and I got the ball rolling (along with a different Jeff C) and Ryan took off with it. Nic was also a big contributor but really you wouldn’t beleive the work they put into this. When the SI comes out, I don’t believe that will be behind a paywall but the code is simply massive.
I think that basically the same result was calculated with at least a hundred different minor and quite a few major variations.
The ‘abstract’ is unreadable from an IE 8 and Safari browser. The zoom needs to be 50% to see the line end, but it’s unreadable at that resolution. Could you fix it?
Best regards, Ray Dart.
REPLY: How’s that?
Maybe time to invest in a higher resolution monitor?
You success shows that on occasion the anonymous review process still works.
It could benefit from these minor changes:
a.) Reviewers who adamantly oppose publication or demand changes must be willing to identify themselves if the dispute cannot be resolved.
b.) Authors of disputed papers can override opposition by paying for the full expense of publishing their paper with the comments of the opposing reviewer – fully identified.
With these two minor modifications, abuse of the anonymous review system might vanish overnight and science could once again become an honorably profession.
With kind regards,
Oliver K. Manuel
Former NASA Principal
Investigator for Apollo
Congratulations on being published.
My recollection of the issues was such that there must be some massively large uncertainty levels. Very, very few stations on the interior of the continent (was it only 3?), quality control problems with the monitoring stations (sometimes buried under snow), local UHI issues with stations located in manned camps, etc. Any attempted reconstruction would be forced to extrapolate temperatures over extremely long distances.
What did you quantify the uncertainty levels to be?
I haven’t a clue, but if you’ve embedded the ‘abstract’ as a ‘graphic’, it doesn’t seem to fit well with the HTML page resolution for the browsers I’ve mentioned.
Best regards, Ray Dart.
Got it now I think.
Most of your paper will be way beyond my pea sized brain to be able to completely understand, but congrats on the accomplishment and thanks for all of the hard work done to produce your work.
Congratulations. Esp getting it into the Journal of Climate. Some of the remarks about Steve Mc not publishing enough will be trumped now. As a lover of irony, I thought it was great to have the abstract refer to “Fall”. AFAIK, there isnt (yet) one tree on (or should it be in?) Antartica!
Congratulations. A great personal achievment for the lot of you. Especially after all of the snark, the ad homs, the dismisivness, the comment censorship at RC, etc. Good job boys.
The dialogue for the ‘abstract’ is ‘repaired’, but parts of the rest of the page are difficult to read because of the resolution problem.
The ‘graph’ can be resolved by choosing ‘the expanded version’, but the dialogue that precedes this still encounters the same problems that the abstract encountered.
Best regards, Ray Dart.
Congratulations to all, particularly for having the patience to tolerate 88 pages of blovation by a reviewer gone amok.
My first thought when I read that was the reviewer was Mike Mann, since he writes like a firehose of prose.
I think I know who the reviewer was, and I honestly don’t think that Mann had a majority of the involvement. Perhaps as an advisor, or perhaps adding some supplementary comments . . . but I do not believe he was the primary author of the review.
Oh, and the 88 pages included our responses, too. 😀 Our responses were actually longer than the reviews, because the issues being discussed required a good deal of background information, lest they degenerate into a he-said-she-said orgy of arm-waving. 😀 Needless to say, though, nearly 40 pages of review was a bit daunting!
Though reposting the reviews was requested by at least one poster, I do not believe that posting them in their entirety is appropriate. I have a few nuggets that I can share, though. My favorite was one where the reviewer asserted that our seasonal patterns were not different than S09’s despite opposite behavior in some seasons – such as the South Pole showing maximum cooling in ours, but maximum warming in S09. Directly from the review:
Reviewer (quoting the S09 text but adding “[and fall]”):
Seeing as I am a moron and can’t use blockquotes, let’s retry this. 😀
I think I know who the reviewer was, and I honestly don’t think that Mann had a majority of the involvement. Perhaps as an advisor, or perhaps adding some supplementary comments . . . but I do not believe he was the primary author of the review.
Oh, and the 88 pages included our responses, too. 😀 Our responses were actually longer than the reviews, because the issues being discussed required a good deal of background information, lest they degenerate into a he-said-she-said orgy of arm-waving. 😀 Needless to say, though, nearly 40 pages of review was a bit daunting!
Though reposting the reviews was requested by at least one poster, I do not believe that posting them in their entirety is appropriate. I have a few nuggets that I can share, though. My favorite was one where the reviewer asserted that our seasonal patterns were not different than S09′s despite opposite behavior in some seasons – such as the South Pole showing maximum cooling in ours, but maximum warming in S09. Directly from the review:
Reviewer (quoting the S09 text but adding “[and fall]“):
I got a chuckle out of writing that paragraph. 😀
And Anthony . . . thanks for the repost!
#9 Jeff Id
Congratulations and thanks for remembering my small contribution back in early 2009. Before the satellite data was public and we had just managed to get RegEM working, we figured that Dr. Steig’s methodology smeared the peninsula warming around the continent and overstated continental warming by something like a factor of two. The very first alternate reconstruction came out at 0.07 deg C/decade; that value kept coming up over and over again. Hats off to all, and Ryan in particular, for continuing the hard work figuring out what was really going on.
I dropped out of the loop back in May 2009 when my 4 year old son was diagnosed as autistic. We eventually figured out the real problem was that he wasn’t digesting protein. Although he ate plenty of high protein foods, it was passing right through him. The neurological manifestations were the result of years of malnutrition and chronic inflammation brought on by his digestive problems. He responded immediately to treatment.
We had a teacher conference today, he’s in the upper half of his kindergarten class, has many friends, and is doing well. It was a great day. Reading about your paper being accepted just made it that much better.
Ryan, did the negative predictor weights of S09 reconstruction, which led to a classic discussion thread a tAV, get discussed in the paper?
I guess I hadn’t really known of the existence of Stieg 2009 but I had come across the idea that antarctic warming was focussed on the penninsula. In fact I’ve always assumed this was the mainstream position. Has Stieg et al 2009 had any impact in antarctic research because even without your analysis it seems to stand outside of the mainstream position?
Thank you, boys for your time and work. I can not believe you made it in the review process but the uea emails may have helped.
Well done, now I just need a JOC link.
“I would hope that our paper is not seen as a repudiation of Steig’s results, but rather as an improvement.”
Honorable to say but it will be seen as a refutation.
A little speculation here… given that this was J of C and the focus was on improving the PCA methodology of climate reconstructions, my guess is that Dr. Broccoli brought in Ian Jolliffe as the 4th reviewer.
Jeff, BTW I really would like your speculation on who you believe the one particular reviewer was, you had to have had some inclination.
Ryan, Nicholas Lewis, Steve and Jeff,
Congratulations! Thank you all for your fine work and tireless efforts to get to the truth.
Congratulations, you all are truly scholarly gentlemen, and a lot of it with humor! Thanks for all the blogging, its been some amazing reading.
#25- Jeff C.-congrats to you and your family, a very heartwarming story.
Congrats to the 4 (or 5) of you for doing it “your way”. A special treat for me to see it all from beginning to end. A real joy.
Jeff C. absolutely wonderful news about your son! Thanx for sharing.
So Jeff, would this constitute “citizen science” as defined by Dr. Judy?
Congratulations. Watching your efforts here was a truly educational experience. Well done.
Congradulations to all! Color me impressed.
#25 Jeff C.- Great news, warm wishes.
The lengths one reviewer went to to keep the paper out. Understand this is a ploy they have employed before.
1. A paper is written that improves the science.
A. they argue that it reaches the “same” conclusions
B. they are that it isnt novel enough.
Their point is not that they want to preserve errors in the record. The point is they want to keep skeptic names off the lists of published authors:
as mann wrote when skeptics first got published: “This was the danger of always criticising the skeptics for not publishing in the
I also wish to thank you all for your hard work. You deserve at least a lot of positive press for your contributions to improving statistical analysis in Climate Science!!
Congratulations on a wonderful achievement and for your persistence against the attempts to stifle academic freedom.
Re Steven Mosher #39.
It’s amazing how many hockey stick reconstructions that reach the “same” conclusions, but don’t improve the science, get published!
Even if it’s not seen globally, it will be seen by those who know. (it is sort of enhanged digital imagery)
Congratulation to all of you, a great synergy was at work there.
This is great to see. I take it to be unfunded, and a completely transparent process open for the world to observe and participate. May we see more like this published where it can be quoted to our friends and politicians.
Congratulations. Could we call it Blogs et al 2010?
Your results show that while the Antarctic Peninsula has warmed, the continent as a whole shows no statistical difference from temperature stasis.
My prediction? It won’t be reported in the MSM that trumpeted the Steig et al report. And Nature will attempt to rebut it by inviting the Hockey Team to take a free swing at it (with appropriate sighing and rolling of eyes) while denying equal opportunity for you to reply.
Dear Jeff, Ryan et al.,
congratulations for the paper!
I wonder if you could say a bit more about the disagreement of the two methods . .
Does this paper mean, that the Mannian took a beating it won’t recover from?
Thanks and cheers,
Look if you have the data to hand I’d be really interested in seeing the trends from the last decade or two.
A pity we can’t read it as yet, but could you elucidate some on:
(b) improper determination of spatial structure during infilling
As a geologist who has attempted many,many maps with poorly distributed data points, this aspect is of much interest to me
Well done guys! This is a great anniversary present after Climategate.
Now that you’re going to be published in a climate peer reviewed journal, does this make you all genuine climate scientists? Has the BBC or Guardian booked an interview or is the silence deafening?
Congrats on confirming Steig’s basic conclusion.
There are none so blind as those who will not see.
This has been a good week.
Or only seeing what they want to see
Congrats for the hard work. It was a pleasure to see it developing here in blogland. I would surmise that a greater sense of community and support developed from the participation of bloggers. I, myself, enjoyed the convolutions and evolution of thought. It is a shame that the review may not be seen in its entirety. I believe that it would help the general science community rather than hurt. Those looking for faults will find them whether they exist or not. But those who enjoy science, would consider the comments and responses as similar to a good detective story.
A press release always gets to make the headlines in the MSM.
December 2, 2010 at 7:53 am
Congrats on confirming Steig’s basic conclusion.”
Ah, not interested in any details, just the conclusion. Surprising for a “lib”.
I see that some have asked the identity of individual reviewers. Since reviewers are anonymous we don’t really know. However, it was fairly clear to me in the one case that the review was done by a group of people rather than an individual. Still, it wasn’t the worst review I have seen, just the longest and most verbose in my limited experience.
For some of the specific questions about what the paper means, there will be some follow-on posts that contain information we had to exclude from the paper for length reasons that will hopefully answer your questions. 😉
Ryan, Nic, Steve & Jeff
Nice achievement. An independent voice is encouraging to see.
Next step is to track that your paper is included in AR5 with proper assessment weighting of it.
I would hope Eric Steig would welcome this as a valuable addition to his own work. It fixes many of the defects I saw in his earlier effort, and does not (as far as I can read from the text available) appear to be an attack on him or his credibility.
Of course newer analyses should do a better job than older ones they were inspired by: That’s called “the advancement of science” in my neck of the woods.
Ryan, Jeff et al., a congrats are in order! Beautifully executed work.
Unfortunately for many advocates, preserving the answer is more important than improving the method used to achieve it (and thereby when the method gets refined, and the answer inevitably changes, the “great unwashed” get attacked for their hard work, as has happened to Steve many a time in the past).
Congratulations to all on publishing a paper that had its beings on the TAV and CA blogs. I personally obtained a lot of enjoyment and learning from those experiences and look forward to reading and reviewing this paper. The experience also encouraged me to go back and learn how to use PCA. I found the eventual authors, and particularly Ryan O, very forthcoming and patient in answering questions about the analysis that was discussed at the blogs.
Perhaps Eric Steig will be by the offer his views. Maybe even TCO could drop by to say congrats – and then, of course, leave without fanfare.
The warming that Steig et al found smeared from the Peninsula to West Antarctica was something that, I think, the authors of that paper hinted was a problem. The Nature cover, of course, showed what Steig et al produced in vivid color. The science, to be credible, needs to able to explain the much faster warming Peninsula rather than a luke-warming Peninsula with warming spread to West Antarctica.
I assume that the authors of this paper did not attempt to evaluate the accuracy and precision of the satellite and ground temperature measurements and would rather have assumed them as a given as the basis for their statistical analysis.
A case in point. SOP for an activist: Preserve the answer, dissemble and deflect when there are differences. The method (whether it is correct or in error) is unimportant to this group.
I was wondering about this a few weeks ago.
Bloody well done!
BCL, and others, please keep your eyes on those all important plus/minus degrees C per decade following the trend number.
One thing puzzles me. In your earlier posts, one of the issues you pointed out was that the Steig et al reconstruction “smeared” the trends over both area and time. The post 1980 trend (the AWS data) showed statistically significant cooling for everywhere but the peninsula. Is this subject also covered in the paper?
BCL, and others, please keep your eyes on those all important plus/minus degrees C per decade following the trend number.
I was looking more closely at the term “statistically significant”.
#68, In future posts, you will see that the differences between trends (which must be calculated from residuals when working with the same data) are statistically significant.
#67, The post 1982 trend was covered, however I think Ryan will describe the results in subsequent posts here.
Data should create light, not heat.
Steig et al said the warming was statistically significant everywhere. The new paper says the warming is dominated by the peninsula and the region of west Antarctica adjacent to the peninsula. That tells a somewhat different story. I looked myself at the Steig et al justifications for limiting the analysis to 3 PC’s, and it appeared to me they contradicted the “rules of thumb” normally used in selecting the best number of PC’s for this kind of reconstruction. I am pleased that RO et al got published, mainly because I think they address this very real shortcoming of Steig et al, and because their results focus attention on the question that should be answered: why has the peninsula region warmed so much and the rest of the continent very little, if at all?
BCL: Mmmm… I don’t really think you were looking all that closely. If you were, you would have observed the following:
RLS the continent as a whole isn’t warming… 0.06 C/Dec +/- 0.08
E-W the continent as a whole isn’t warming either… 0.04 C/Dec +/- 0.06
Yes, the trend shows some extremely modest warming over the time period studied… but with the error bars showing greater variability than the reported trend, it’s reasonably likely the trend is actually cooling rather than warming.
In S09, the continent as a whole was warming at 0.12 C/Dec +/- 0.09. That’s a very different finding… one that seemed so dramatic, it got published in Nature!
So, with RO10 generally using the same data and in some cases correctly adjusted satellite data, but using different (and improved) methods, there’s actually no net warming over the continent.
Not only that, but the authors in RO10 demonstrate the reasons why the results of S09 erroneously showed overall continent-wide warming… partly due to lack of adjustment in some of the satellite data, but also due to the method that was used in S09… the method “grabbed” warming from one area of the continent and transferred it across the continent to create the warming trend when one actually doesn’t exist.
Yep, pretty compelling. But, not in the way that you seem to think.
Congratulations guys, this is very good news.
For all the criticisms of the peer review process, this shows that it usually works – as long as you have a good fair editor, which is the case most of the time and clearly was the case here.
Obviously the editor sent the paper to one of the original authors, and naturally that person would try to mount a vigorous defence of his paper and pick holes in yours. He may have discussed the paper with co-authors. The editor then has to make a decision, in this case with the aid of a fourth reviewer.
The times-scale may seem long but it is not unusual.
The net result is that the paper is improved considerably by the comments – I expect by the negative reviewer’s comments also.
A point of etiquette is that you should not publish all the referee comments. They are supposed to be confidential.
“A point of etiquette is that you should not publish all the referee comments. They are supposed to be confidential.”
Makes perfect sense if the identity of the reviewer is disclosed. Makes no sense if the reviewer is anonymous, since it would seem difficult (impossible?) to offend an unknown person through a failure of etiquette.
BTW, I think the whole of the review process should be above board, with identities of reviewers and the content of the reviews disclosed in all cases. That would eliminate the temptations to influence the publication of papers when the reviewer has a personal vested interest in protecting their own work or the work of friends/associates from refutation. Having everything out in the open is a much better approach. Let the sun shine in on the process.
Congratulations. Based on his comments on other threads, it seems that you all have developed mutual respect with Dr. Steig. That’s a great accomplishment in itself.
So, will the press release headline say:
Antarctic Peninsula Warming: Worse than we Thought!
Congratulations. Looking forward to seeing the new and improved analysis.
And I notice that the new and improved graphic breathes life back into the volcanism conjecture for increasing Antarctica temps (that Steig 2009 quashed).
I agree. Complete openness is paramount.
The findings of S09 had serious implications for climate. Prior to S09 warming was thought to be confined to the peninsula with no continental trends discernable from decadal variability. S09 challenged this by showing that warming extended to the continent, and that there was much more of an underlying trend and less decadal variability (more in line with models and SH climate trends).
The area of Antarctica thought of as the most unstable and vulnerable to climate change – the west Antarctic ice sheet and the Ross ice shelf – was shown by S09 to have the greatest warming trend. This finding arguably had the greatest importance as the WAIS is obviously stable to the inherent decadal variability. A confirmed warming trend that could be attributable to AGW obviously increases the risks of the CAGW scenarios (which involve collapse of the WAIS and/or the RIS).
S09 had serious flaws which produced spurious reconstructed spatial and temporal climate patterns. When the flaws are addressed and the method enables a proper reconstruction, S09’s key findings – which distinguished it from previous research and therefore publishable in Nature – disappear (or are at least significantly diminished).
Congrats! And great work.
On disclosing reviewers comments: I agree that etiquette dictates they should not be posted. This is true even though reviewers are anonymous.
is actually a situation where etiquette is at odds with common sense, which should permit all discussion to be published, with the caveat that those who are anonymous remain anonymous provided they don’t out themselves. (Unfortunately, often people’s style of commenting might, in and of itself “out” them. But, to some extent, reviewers should be willing to reveal themselves. Quite often, they will– in private at meetings. )
Not sure why you think this. I just don’t see justification for not posting what reviewers say. What possible benefit to the progress of science does this have?
It’s not the sign of the trend, and whether it is positive or negative at any time, it’s the magnitude and whether the magnitude is consistent with model predictions. The new results certainly don’t suggest a dramatic inland heating of Antarctica, rather the only dramatic heating is confined to the northern peninsula.
So “no”, they don’t confirm Steig’s results in that sense.
Lucia, I’m with Steve F on this. There is no possible benefit from preventing the authors who work has been criticized from publicly airing those criticisms with the general science community. If others generally agree with the original critique, that information will get communicated to the authors, science advances.
In AGU’s case here the policy:
On the other hand, there are good reasons to protect the identity of the reviewers, even if you are an author whose work you have criticized.
Suppose I criticize a paper studying the efficacy of a new drug and my criticisms prevent the publication of that study, effectively blocking FDA approval. Am I then not exposed to legal liability from the drug company and its investors who have been harmed by my criticisms?
So anonymity is a good thing, and it is the responsibility of the authors if they disclose portions of a peer reviewed publication to protect that anonymity. They certainly should be careful about speculating about who the critics are (in this case, it’s obvious which cabal is responsible for the more negative review, the identifies of the individual players hardly matters).
I am in agreement with Lucia. It is quite possible that others may know who the referees were, but were unaware of what was written. This could expose people to discomfort from someone who does not like how the reviews were done. This can cut both ways. A referee who accepted the content of this paper could suffer abuse at the hands of other colleagues for contributing to the “skeptic cause”.
In an ideal world, you would be right, but under the current circumstances, there is an expectation that such communications will not be bandied about unnecessarily. Abusing that expectation would lead to fewer people agreeing to referee in order to protect their own careers.
“Abusing that expectation would lead to fewer people agreeing to referee in order to protect their own careers.”
I guess I hold the (perhaps naive) hope that people think their personal integrity and the goal of scientific progress are more important than the displeasure of colleagues. Maybe I am wrong about this. I hope not.
Great stuff Jeff- congratulations.
Hopefully this will be the catalyst for more blogosphere (skeptic) papers
questioning the “consensus science”.
I wonder if Nature will print an addendum?
Not sure why this paper needed to be published…my old analysis of 0.51 +- 0.52 C/Dec for the continent of Antarctica is entirely consistent with these new results. I can hardly be blamed if certain reporters claimed Antarctica might have experienced as much as 5 degrees of warming in the past 50 years years.
Seriously guys, congrats!
I guess I can see the point about anonymity of the reviewers, but I am more interested in the following anyway:
1. What were the strongest objections to the paper and how did you account for them ?
2. What were the weak and humorous objections to the paper ?
Surely you can “paraphrase” the objections to preserve anonymity, but still give us a feel for what various reviewers took issue with. I guess if an entire goofy/silly line of reasoning can be traced directly to the thought patterns of a specific reviewer we are sort of stuck, but if you avoid words like bizarre, robust and debunked you are already 90% of the way to making it anonymous 😉
Congrats on the paper and on your perseverance.
I think your paper will be of interest in the discussion of Arctic warming anomalies – I recall discussion of how circum-Arctic temperature stations are being used to smear the values over the whole polar region. Your paper, presumably is an example of why this needs to be carefully considered.
#90 Yes Steve reveals more.
Again I agree. Roman’s arguments actual support my position that it’s the responsibility of the authors if they disclose any of the criticisms to protect the identify of the reviewers as much as possible (wholesale publication of reviews in general should not be done without the approval of the original writer in any case), but excepts seem to fit in the “fair use” clause, as long as the purpose isn’t to egregiously expose the identity of the reviewer, or to unfairly represent his or her opinion.
The novelty of Steig et al was their purported claim that West Antarctic (home of the important WAIce Shelf) was a particular locus of warming and that the warming on the Peninsula was not distinctive. Their abstract started:
See CA for more on this.
I’ve been wondering for some time about the basis for supposing that authors have a duty of confidentiality to reviewers. I understand that reviewers have a duty of confidentiality to authors, but not all duties of confidentiality are symmetric (a client can waive privilege, but a lawyer can’t.)
I had a dispute with Nature about this in 2004. Mann was bruiting that our submission had been rejected. We posted up the review comments to show that they had been favorable, up to the addition of a third reviewer (who Climategate indicates to me was Phil Jones.) Ross originally posted them up at his U of Guelph website. Nature wrote the university, asserting that the reviews were confidential and asked them to take down the reviews.
I replied with a very sharply worded letter asserting that Nature had breached various implied terms of any agreement in how they conducted their review and could not selectively attempt to enforce terms while violating others. As a way of avoiding a showdown, Nature editors (probably Karl Ziemelis who seems very sharp) offered to get consent of the reviewers; this was obtained and I posted the review comments online at my old website and they are now online at climateaudit.info.
I haven’t taken this issue on though I’ve thought about it from time to time. The antagonism and obstructionism of Team reviews needs to be exposed. But before taking on this cause, I would like to understand precisely why Lucia, Roman and others think that authors have a duty of confidentiality to reviewers.
Re #83, Carrick’s quote of AGU policy:
Note that this language creates a duty of confidentiality on the editor and editorial staff. It does not state that authors have such an obligation.
94.Steve McIntyre said
December 2, 2010 at 4:09 pm
My read on RomanM’s comment is pretty much the same as my own opinion: the current climate (no pun intended, but obviously there) that exists for the review process is marked by somewhat childish behavior of team members which would tend to limit willingness of authors to review others’ work. They would likely be afraid of the consequences. Look how they treated Judith Curry!
Mark T, I’m interested in any articles or documents evidencing the existence of a duty of confidentiality to the reviewers. Reasonable people can disagree on whether it is or isn’t a sensible policy; I’m looking for something else – why do people believe that such a duty exists. The AGU policy doesn’t prove it.
Mark T #96,
I understand the problem, but accepting the current situation of academic bullying (and I think that is a fair description) is I think not an acceptable option. Unfair (and incorrect) reviewer comments which are only intended to inhibit publications will never go way so long as those unfair comments remain hidden from public view. The UEA emails are filled with messages that show how a small group worked behind the scenes to block publication of articles they did not agree with, by most any means possible, including undisclosed direct pressure on editors and publishers. Why does it not surprise me that the authors of RO(10) all believe the same group worked together (again) to block the publication?
It is a rotten process, and will remain that way unless unfair reviews can be exposed for what they are.
97.Steve McIntyre said
December 2, 2010 at 4:41 pm
On this I agree, i.e., I don’t think such a duty exists, I think it is simply courtesy. More on this below…
98.Steve Fitzpatrick said
December 2, 2010 at 4:45 pm
This is at least partly true, though in the mean time real careers could otherwise be at stake. I say partly because we are not talking about normal processes as it is; the current state of affairs is rigged. It will take a lot more than exposing unfair reviews, IMO, to correct this situation.
Unfortunately, without such a courtesy, getting a legitimate review will be all but impossible. Authors will simply refrain from reviewing for fear of being revealed as a traitor to their cause or a childish idiot for being so petty with a nonsensical review (my take on the 88 pages of back and forth is that a fair amount of idiocy was in play.) Take your pick, damned if you do, damned if you don’t. Granted, what currently happens is not much better, but at least there is a chance, albeit small, for legitimacy to win the day as this submission clearly shows.
Perhaps I’m wrong and the courtesy should not be upheld. Perhaps, Steve F., simply posting this stuff will be what turns the worm? Dunno. Steve M. seems almost as if he’s looking for an excuse to do so, or at least, trying hard to find a reason he shouldn’t, hehe.
Having had a career in a non-controversial discipline, I was extrapolating a bit in my previous comment. In a perfect world, I would have no objection to open reviews by identifiable referees. But even in my own mathematical area, I observed some serious resentment in cases where a particularly harsh review had been received by colleagues. Had they known who the referees were, it would likely have produced some unpleasantness.
As MarkT also points out, the climate in climate science is very politicized to the point where people have been categorized by the current powers-that-be into two groups and a paper published (in what once a legitimate journal) smearing those they have deemed substandard. I would hate to see someone reclassified on the basis of positively refereeing a paper.
Remembering my own reluctance to make waves as I was going through the academic ranks, I did not want to see their careers threatened (and I honestly believe that such threats could be the result) by the team and their supporters. Maybe I am just being an alarmist … 😉
That said, I see no objection to mentioning the objections that were raised and any positive supportive comments that may have been received.
Back when Ryan O had written comments at RC, I said something like “I encourage you to submit this work for publication.” I’d glad to see that this work has gone through the peer review process, and I look forward to reading it.
I appreciate also Ryan’s comment that “I would hope that our paper is not seen as a repudiation of Steig’s results, but rather as an improvement” and his emphasizing that their results (evidently) back up our most important point – -the significant warming West Antarctica.
This is indeed the way things ought to work — and evidently do. Too bad Steve McI seems bent on spinning it otherwise. His claim that this new work ‘refutes’ mine is a prime example of why I cannot take him seriously.
Ryan, if you don’t mind sending me a preprint, and a link to your reconstructed data, I’d appreciate it.
I will presumably have more to say after I get a chance to read the paper, but it’ll be a month or more as I’m simply too busy with current projects.
Congratulations! I look forward to hearing more.
Regarding confidentiality of reviewer comments. There is an organisation the UK – COPE (Committee on Publication Ethics). I can find no mention of an obligation to keep reviewer comments confidential other than the possible situation Lucia describes (#80). However some of their anonymised case studies make interesting reading.
http://publicationethics.org/case e.g. reviewer misconduct
It looks to me like you did not read the post above when you say the results “back up our most important point – -the significant warming West Antarctica”. The diagnosed warming is half what your Nature paper said, and is mainly localized adjacent to the peninsular region in the new reconstruction, not spread over the ice shelf areas.
OK, here’s the metric. One boot per year. So the truth took two years putting on its boots.
Congrats to all the authors, which I noticed that Dr. Steig didn’t since Steve McI is one of the authors…
As Steve McIntyre is an author on the paper in question, it seems significant to me that he considers it refutes your earlier work just as much as Ryan considers it an improvement.
Movie-billboards cherry pick reviewer’s comments; scientist don’t.
Are snow shoes part of the metric?
Sure. One can clearly see this in the anomaly maps compared.
I don’t think it’s uncommon that personnal animosities are played out in the peer-review process in all fields, we’re all human after all. I’m sure McIntyre has built up a serious number of climate science enemies. The more interesting reviews would be those that allowed the flawed Steig paper through the review process of what is meant to be one of the most prostigious science journals on the planet. If this is an embarassment to Nature to have a cover story rebutted in such a way then maybe they may start to consider their own process has got a little too incestuous.
Eric Stieg says:
I remember this a little bit differently. It seems to me there was a RealClimate post about how the S09 results were “consistent” with the model results and a technical conversation about the PCA followed. When it got to hot in the kitchen technically for Eric to handle, in usual RealClimate fashion comments were closed and the quote Eric made was (quoting from memory here) : “If you have something to say to me, say it in the literature”
Looks like they did.
I don’t understand why is dr. Steig seemingly incapable of reading the abstract and the overall arithmetical conclusion of the paper. Why isn’t McIntyre’s right to disagree with your assessment that this is, in fact, a rebuttal? We should not be welcoming sophistry argumentations about this: in fact, technically, this paper is a rebuttal of the original S09 one, since it changes core methods and many mathematical results of it, thus substituting it in the peer reviewed literature.
Surely, we can say that many conclusions remain the same, while others don’t. Thus it is clear to me that if Steve wants to conclude that this “refutes” S09, he is entitled to do so, and it is a factual claim, not spin.
Words have meanings. Generally it is a good idea understand the words you have read before offering criticism.
“I appreciate also Ryan’s comment that “I would hope that our paper is not seen as a repudiation of Steig’s results, but rather as an improvement” and his emphasizing that their results (evidently) back up our most important point – -the significant warming West Antarctica.
This is indeed the way things ought to work — and evidently do. Too bad Steve McI seems bent on spinning it otherwise. His claim that this new work ‘refutes’ mine is a prime example of why I cannot take him seriously.”
From a dictionary:
“USAGE Refute and repudiate are sometimes confused. Refute means ‘prove (something or someone) to be false or erroneous’:: attempts to refute Einstein’s theory. Repudiate means ‘reject as baseless, or refuse to acknowledge’”
Both Ryan O’Donnell and Steve McIntyre are correct in their statements. S09 has been refuted, but not repudiated, by OD10.
I do not have a website where I can post the reconstructions yet – still working on that (we are entirely self-funded 😉 ) – and they currently exist only as R objects. Give me a couple of days to save them in a different format and upload them somewhere (like FileDropper or something) so you can get at them. I plan to have them permanently hosted somewhere in the near future, but I have no problem letting you get at them from a temporary location.
As an aside (not that Steve needs me to defend him), I think that dismissing Steve for the reason you state might be doing yourself a disservice. Steve certainly uses different language than I do, but the substance of his message is the same. Whether something is an improvement, refutation, or rebuttal is often a statement of opinion. Indeed, one of the favorable reviewers felt the differences were significant enough to refer to our paper as a rebuttal of yours in the review itself, and another reviewer spent time discussing how our paper addressed the concerns the community had with your reconstruction. While he did not specifically call it a rebuttal, it was clear that he felt the differences were of material importance regardless of whether the regional West Antarctic trend was statistically significant. So while Steve chose to express this using different language than I did, I would not agree that this is a reason for dismissal. My opinion, anyway.
Eric Steig’s comment here (and others elsewhere) are why it is impossible to take him seriously.
I won’t get into arguing about Steve’s intentions here…
At any rate, for archiving data, why not use one of the major data centers? They are not limited, as far as I’m aware to ‘funded’ researchers. Try http://www.usap-data.org or http://nsidc.org
I think we’re going to stick it on a big server site, data, code the whole bit. It’s totally turnkey, in case you were wondering.
Did the very adversarial review process cause you guys to refactor your code to respond more easily to the demands?
You know it really does not matter whether the assessment of Steig et al is handled with kid gloves or straight on. The objection of Steig himself does not matter. Read the table above in the introduction to this thread and pay strict attention to what comes after the plus/minus notation and finally read (again) what Steig et al stated in the paper (Steve M did not just make that up):
“Assessments of Antarctic temperature change have emphasized the contrast between strong warming of the Antarctic Peninsula and slight cooling of the Antarctic continental interior in recent decades. This pattern of temperature change has been attributed to the increased strength of the circumpolar westerlies, largely in response to changes in stratospheric ozone. This picture, however, is substantially incomplete owing to the sparseness and short duration of the observations. Here we show that significant warming extends well beyond the Antarctic Peninsula to cover most of West Antarctica, an area of warming much larger than previously reported. West Antarctic warming exceeds 0.1 deg C per decade over the past 50 years, and is strongest in winter and spring. Although this is partly offset by autumn cooling in East Antarctica, the continent-wide average near-surface temperature trend is positive.”
And if you do not think that that comment does not weigh heavily on the concerns of the precariousness of the West Antarctica Ice shelf and potential sea level consequences, I would advise you to go back to sleep for awhile longer.
That was of course the point of my including the quote from the AGU policy (if somebody can find a similar restriction on authors, I’d love to see it, but I don’t think it exists).
There is an implied copyright/ownership on the part of the original reviewers, so I would suppose that verbatim replication of a review would reply their permission (which could of course be gained through the editor in charge of the paper).
I think transparency benefits the science, anonymity favors the “cabal”. Transparency is their enemy and that of people who play politics.
Does anybody really think the obnoxious over the top censorship of Judith Curry by members of the climate community has in any measure increased their (not her) standing within the science community? Seriously, this is an argument in favor of transparency.
Let those who would behave this way expose themselves for who they are.
*** would require their permission
It’s not sleep that is required it would be study. Most of us in blogland started with zero knowledge of climatology. We studied and learned and filled our heads with data until it made sense. The S09 paper was exciting/depressing in that everyone with common science (sense) knows the world won’t flood if the Antarctic won’t melt. The S09 authors never said the Antarctic will melt but the implication to the layperson was ‘big red plot’. Now the plot has an appropriate amount of blue but if I recall there were many dozens of mainstream papers which carried the ‘red’. Can anyone guess which mainstream papers will carry the ‘blue’.
Unless you have a side profession as a psychic (and are any good at it), I think that’s wise in any case. 😉 Objecting to Steve’s over-the-top, adversarial choice language on the other hand, is understandable.
One can recognize Ryan et als work as an extension of what was really a very good idea on Eric’s part. There seems to be no need to get into cat fights over who has “scored more points” here. Would Ryan’s work exist without Eric’s contribution? The answer is certainly “no,” and Eric deserves praise for his original, creative ideas that led to this extension of his work.
Everyone benefits from the clearer understanding that comes out of both Eric’s and Ryan et al.’s work. That statement’s called “keeping your eye on the ball.”
And Kenneth, I’m not objecting to the non-use of kid gloves here. As you well know, I’m prone to taking them off myself at times.
Of course Eric’s abstract is going to contain errors, if it accurately reflects the content of his paper, and if there are advances in the techniques by Ryan et al that have improved upon his original work. As analogy, if I present in a publication a photo of a planet using a lower resolution telescope, and somebody comes along with a higher resolution scope, is my photo “wrong”?
Certainly, the summary of the results of the new photo would differ from the summary of my photo. Does that “refute” what is an accurate summary of my photo, presented in my original paper?
That’s a relatively close analogy to what we have here. It’s more that the newer tools seem to be better able to resolve finer scale features of warming in Antarctica than the older tools. Should we call the newer tools a “refutation” of the older ones?
Jeff, I think the treatment of Steig’s work by Nature and the MSM, and its perception by the public, is a separate issue from his own writeup. (Given the recent public response to climate change, I would say the objection might well be that people are overly skeptical, rather than not skeptical enough.)
I’ve had work of mine (and my colleagues) that the press got carried with in their reporting of it…. Exaggerating is what they do best.
Re: Ryan O (Dec 2 19:32),
Ryan, I recommend Google Sites for this. You can set up a website – they give useful templates with all sorts of gadgets. And you can upload up to 100 Mb, all free.
Since Eric won’t be getting around to it for “a month or more” I wonder why he is so persistant in acquiring a preprint. The paper will be online in a week.
Maybe he will find some time for it after all.:)
Jeff: (and Kenneth Fritsch): Given that you guys have apparently demonstrated much greater warming that we found (!), in the most critical area glaciologically (the Amundsen Sea embayment), I’m afraid you may well find that the media you complain about will indeed emphasize the red even more than last time. The areas that are blue in your results are too cold to matter even if they did warm up a bit. No one has ever though the Ross Ice Shelf area (blue in your results) was sensitive to increasing temperatures. This is something I pointed out to you in an email or two some year or so ago. Nature has a funny way of conspiring against those that are trying to tell you that nothing interesting is happening!
If you are right about the peninsula, I hope you will help the media take notice. I’m not enough of an expert to say what this means for the future but the data is the data and there ain’t much we can do about it.
Carrick when you say the following:
“And Kenneth, I’m not objecting to the non-use of kid gloves here. As you well know, I’m prone to taking them off myself at times.
Of course Eric’s abstract is going to contain errors, if it accurately reflects the content of his paper, and if there are advances in the techniques by Ryan et al that have improved upon his original work. As analogy, if I present in a publication a photo of a planet using a lower resolution telescope, and somebody comes along with a higher resolution scope, is my photo “wrong”?
Certainly, the summary of the results of the new photo would differ from the summary of my photo. Does that “refute” what is an accurate summary of my photo, presented in my original paper?”
I must say that my point was that it does matter whether you use kid gloves or not, the conclusions of these two papers under discussion here are different and that is clearly seen from the table and colorful graphic above. It also has nothing what so ever to do with what an author may do in the first effort or how it would be interpreted by outsiders or the advantages of those who might follow.
The novel idea from Steig was to apply a correlation of surface stations to satellite ATTR data and then use that relationship to get more complete coverage from the surface stations back in time when the satellite data did not exist and the surface stations were even more sparse than in the calibration period. I am eagerly anticipating what the Ryan O paper has to say about the application of that novel idea and how well it was carried out.
A hot Peninsula and neutral mainland versus a moderately warm peninsula and warming mainland would change how climate scientists think about the Antarctica considerably.
Of interest to me was that you do not obtain a statistically significant warming Antarctica unless you do go back past the start date of the satellite measurements, using the Steig data and methods. In fact you have to go back, as I recall, to a time within 10 years of the 1957 start date for the surfaces stations to obtain that kind of warming trend.
I am primarily interested in the science and less so the scientists and the nuances of publications.
Oh, and by the way, I now have to go back to the Steig paper and find the comment indicating that the less warming peninsula (but not the warming Antarctica) was partly an artifact of the methods applied.
Eric Steig when you say the following:
“Given that you guys have apparently demonstrated much greater warming that we found (!), in the most critical area glaciologically (the Amundsen Sea embayment), I’m afraid you may well find that the media you complain about will indeed emphasize the red even more than last time. The areas that are blue in your results are too cold to matter even if they did warm up a bit.”
I think perhaps you can clarify my undersatnding of sea level consequences of a hot peninsula dropping (melting?) into the sea and the West Antarctica not melting but sliding into the sea. I read an article recently that indicated the peninsula what have a small effect compared to parts of the West Antarctica sliding into the sea.
We must be looking at different graphics. It looks to me (http://i53.tinypic.com/14vh935.jpg) like the embayment area is at least on average, considerably cooler int he RO(10) reconstruction. What do you see that I do not?
The link below is not the article I was thinking about, but it does go on about the dangers of the West Antartica sliding into the sea and raising sea levels by 3 to 5 feet and then way late in the article notes that all the ice in the Peninsula would raise sea levels by 9 inches. I also suspect that when the article writer refers to Antarctica warming that he is using the warming of the Peninsula.
Here is the comment from the Steig paper discussing a trade-off from not using higher order PCs. hat cut off did not capture the variance of the Peninsula, but that was OK because we have lots of temperature data from the Peninsula.
“Principal component analysis of the weather station data produces results
similar to those of the satellite data analysis, yielding three separable principal
components. We therefore used the RegEM algorithm with a cut-off parameter
k=3. A disadvantage of excluding higher-order terms (k>3) is that this fails to
fully capture the variance in the Antarctic Peninsula region. We accept this tradeoff
because the Peninsula is already the best-observed region of the Antarctic.”
As an aside I murdered the satellite acronym, as it was Tir, and the second correlation was with the AWS stations.
Nick and I used to host stuff at DropIo, but they are being aquired.
Nick has looked at google sites. That’s a fine choice. You could also look at driveHQ which gives you a free FTP with 1GB free storage. You could also get more space on google by getting your site through google apps.
Steve Mc makes a good point in his comment (#94) about Nature and their effort to stop Ross and him from posting what reviewers wrote. First, there is no reason to assume that confidentiality works both ways. Second, the entire process requires good faith or it falls apart. If someone is abusing the process, their lack of good faith should be exposed or the whole system suffers. Pretty clearly, that is what has happened with the team as we can read in the e-mails. If people fail to conduct themselves honorably, they should be estopped from claiming that they are still entitled to such treatment.
We don’t even know what the spatial verification looks like yet. Looking back on some of Ryan’s blog posts at tAV, this very area put up some of the weakest numbers (within the context of much stronger overall verification stats than S09).
FTP directory /SAN at n4ftl01u.ecs.nasa.gov
To view this FTP site in Windows Explorer, click Page, and then click Open FTP Site in Windows Explorer.
U.S. GOVERNMENT COMPUTER
If not authorized to access this system, disconnect now.
YOU SHOULD HAVE NO EXPECTATION OF PRIVACY
By continuing, you consent in your keystrokes and data content being monitored.
that’s not very nice.
Steve F. and Eric:
As Eric probably knows ;), I am no glaciologist. I mentioned in one of the review responses (where the objection was that we did not delineate our regional areas based on things important to glaciology) that our paper was not concerned with the physical processes or theories about sea ice, glaciology, the weather/climatological effects of the ozone, or any other mechanism postulated to be driving change in Antarctica. Instead, we were focused on what was done either incorrectly or suboptimally in S09 from a mathematical point of view and what – if anything – changed in the results if these issues were addressed.
From this perspective, yes, I do think we find substantial differences, and the paper could be interpreted as a rebuttal. There are two definite mathematical errors in S09, which relate to how to properly calibrate a response variable to an explanatory variable. Both are errors; only one of them matters in this specific case. Regardless of whether the answers are the same in this specific case, the methodological deficiency needs to be addressed lest it be repeated in an instance where it does matter. In addition to the errors, there are some other choices made which we demonstrate to be suboptimal.
From the perspective of an overall picture of what is changing in Antarctica, I also believe we find substantial differences. The 1957-2006 linear trends depicted at the top of this post have many differences that (I imagine) would be important to those trying to discern the physical mechanism driving the change. However, since two of the differences – i.e., the exceedingly low Peninsula trends and rather high East Antarctic trends in the S09 reconstruction – are not supported by the ground data (both), were already acknowledged by S09 (Peninsula), and did not gain traction in the community (East Antarctica), the tendency may be to say, “Well, we already knew that,” and subsequently discount that difference as being “important”. Nevertheless, they are real differences that may matter a great deal when trying to quantitatively reconcile a reconstruction with a physical theory. With all due respect, Eric, I feel that you are somewhat afflicted by this condition. 😉
From the perspective of the key finding that S09 chose to highlight – significant warming in West Antarctica – there are both similarities and differences. The overall magnitude in ours is half that of S09, and the pattern is substantially different. In terms of Eric’s sea ice hypothesis (which was a pervasive sub-theme for one of the reviews 😉 ), this difference could prove to be important, as the increasing amount of sea ice in the Ross Sea is somewhat at odds with the S09 pattern of warming, which shows a maximum on Ross. In terms of glaciology, the difference is much less because the glaciers Eric is concerned with are not located on Ross.
The key area Eric is talking about (also pointed out by one of the reviewers 😉 ) affects the Pine Island and Thwaites glaciers. Here is the Pine Island glacier (Thwaites is adjacent):
Both flow into the Amundsen Sea via the Pine Island Bay:
If you examine the graphics at the top of this post, you will note that the land area around Pine Island Bay is warmer in our reconstruction than in S09. Since this is an area where increasing temperatures could actually matter (unlike on Ross), Eric’s point is that our reconstruction confirms (with greater emphasis) an implied finding from S09 that these two massive glaciers could be detrimentally affected.
So we have the following cases (and I’ll throw in one of the obvious seasonal differences, too) :
1. If one is a glaciologist, one might consider that the “key” finding from S09 and conclude that ours does little to change that picture.
2. If one is studying the effects of ozone depletion on temperature/circulation patterns (especially seasonal ones), one might consider that the “key” finding from S09 and conclude that ours fully contradicts it by displaying opposite trends and seasonal patterns.
3. If one is concerned about the statistical model used by S09 to achieve their result, one might conclude that several “key” methodological choices by S09 are fully contradicted by ours, while some of the other choices are confirmed
. . . etc.
Different people with different perspectives will come to different conclusions about whether this qualifies as an “improvement”, “rebuttal”, or “refutation”. From each of their perspectives, they are giving an accurate description. As already obvious, I lean toward the first descriptor. Others have different opinions.
The only opinion I would strongly disagree with is that ours is a “repudiation” of S09. I see nothing in our work or in S09’s that could justify that strong of a statement.
Wow!!! Outstanding! Congrats, you and Steve are an inspiration to all of us.
I think your paper will be of interest in the discussion of Arctic warming anomalies – I recall discussion of how circum-Arctic temperature stations are being used to smear the values over the whole polar region. Your paper, presumably is an example of why this needs to be carefully considered.”
While the results of this paper are specific to the antarctic, this is still a very important point because it shows that extrapolation over enormous distances (in the arctic, for example)is not acceptable. Just removing these extrapolated temperatures in the arctic would cool down Hansen’s GISS temperature quite a bit and would make much more sense WRT to both numerical analysis and climate science.
It would be interesting to recalculate the GISS global temp record without the bogus arctic extrapolated temps. 2010 would be quite a bit cooler, no doubt.
Given the unscientific and immoral behavior during the review process by guys like Jones and Mann, it is time to update the review process.
Total transparency is the way to go. This should apply retroactively. Any reviewer abusing the review process should no longer be a reviewer.
These guys are like roaches, turn on the light and they scuttle away. They thrive on these underhanded secret machinations. Turn the lights on by forcing them to do their science business transparently.
Transparency is a big part of how civilization can protect itself from unscientific scientists/academics.
That was very well done, congratulations to all concerned. I think that first graphic at the head of this paper speaks volumes.
AGW proponents are all to ready to use over warm colours in their graphics.
I saw this coming, a bigcitylib over at climateaudit or an Eric Steig here at tAV, commenting outright that the maths may be wrong but that doesn’t matter, as the main result – the warming trend – is confirmed.
My question at WattsUpWithThat remained unanswered and I still would like to know, why on earth did you obviously not include the very different trends with different starting dates in this paper, telling a completely different story.?
If I remember right, most of the warming appeared to have happend between 1957 and 1969. Later starting dates show cooling and after 1980 a strong cooling of around -0.2 deg/decade,
I can only guess that the authors were almost entirely interested in correcting statistical and mathematical errors and not much more, particularly not in the scientific importance of the result.
However, there is no way to reconcile a warming period between 1957-1969 and cooling between 1980-2007 with man made global warming. The discussion here about melting glaciers is bizarre with the background of a strong cooling trend since at least 1980.
I would also like to ask Eric Steig the same question, why he did not display or discuss the cooling trends already in his Nature publication. (the link above shows a cooling of -0.06 deg/decade after 1980 even with Steig’s maths) ?
The errors exposed by the new paper may be due to neglect, insufficient skills or failure to interact with the statistical community, but how can such an omission be explained ?
I am quite proud of you, and this….
It is interesting: a labor of love, understanding, devotion to science, as opposed to a cushy retirement and not having to teach much, anymore….
I confess: I do not understand the science. I do understand the nuance of the picture difference. I understand
the nuance of good statistical modelling vs not.
Did I write: I am proud of you. ….Lady in Red
It’s very clear that McSteve is the Antichrist for the Team. There will never, ever be a group hug between the two antipodes, no matter who’s right or wrong. Having this paper being published is seen as a major defeat in the Team’s locker room.
Hence Eric’s whine above. He needs some cheddar, pairs well.
Only people with a serious case of color blindness cannot see the differences between the “Nature” cover picture (widely used in the MSM) and RyanO’s graphs. Is Cherry picking a “basic” part of Climate Science education?
Eric Steig wrote:
“This is indeed the way things ought to work — and evidently do. Too bad Steve McI seems bent on spinning it otherwise. His claim that this new work ‘refutes’ mine is a prime example of why I cannot take him seriously.”
Wow. Have you been upgraded to the Pope or some infallible divine entity?
Steve McIntyre joined the project exactly because he noticed that your paper was fundamentally wrong. Not only that: he has explained many of us why it is wrong and many of us know it, too.
I was personally totally shocked by a fact – I fully reproduced by my calculations – that the 50 MB file attached to your paper was artificially generated bogus and could be compressed by a factor of thousands because it only combined three eigenvectors, see
This very fact means that the data evaluated in the paper are equivalently strong to 3 weather stations which is just not enough to reconstruct the local temperature history.
I can’t possibly understand how you may deny that the paper McIntyre co-authored shows that your paper has been refuted if this was the main conclusion of pretty much everything that McIntyre did with the issues raised by your paper.
In reading what Steve McIntyre has to say, I think that there is a point that he is making that is not being made here. Many of the results in both the Steig an the RO paper are likely to be artifacts of the mathematical technique. It is a novel technique that has not been subjected to rigorous mathematical analysis. In particular the strong warming in the peninsula is possibly just an artifact of mathematical choices made and these choices were made without the benefit of a complete mathematical understanding.
In regard to teh confidentiality of reviews, in one of the areas in which I worked there was one major reproacher who led a faction that tired to dominate the entire filed. He had made important early contributions but these were being surpassed by later work. He replied by using political techniques to suppress opposing views. I saw him in action. He had professors working in the filed either under his thumb or terrified of his influence on funding agencies.
I was very surprised at the interest he took in my work. He even contacted my funders to indicate how completely wrong and misguided my work was. Given that it worked, my funders were quite puzzled with this and showed me his letters. They thanked him for his comments. However given how explicit he was, I can understand how he terrified people who did not have established results and careers.
So open up reviews could leave junior workers in vulnerable positions and reinforce the control that people like him have over certain fields.
Please guys, you have found surface warmth in the glaciologically important embankment, whereas we had taken it and sort of spread it around continent-wide – so your results only do more of what we set to do anyway.
So it is neither an improvement, nor a rebuttal, nor a repudiation. It is a distraction – from the larger message of Antartic warming, which I am sure you agree with. Science is all about putting together the bits and pieces to paint the larger picture. If anything, the paper only slightly refines Steig et al results. In fact the problem with such papers is precisely that it muddles the picture.
And moreoever, you have shown cooling in areas that do not even matter – only warming is what matters; but the only outcome of this would be that it could be trumpeted on Climatedepot.com which could lead to the whole world being instantaneously brainwashed about the Antartic. Or it could be entered into the Congressional record, which as you know is the worst thing that could happen in the whole world.
Ryan O (139),
Thanks for that. It sounds like your paper is much like the work Steve Mc has done with the hockey stick and what McShane and Wyner did — it is an effort addressed primarily to correcting the math/statistics. If I understand what you wrote, there may well be all kinds of scientific errors in the approach and mistakes in the data about which you have nothing to say. Your attempt was simply to say — if scientists want to do what Steig tried to do and they have this data to work with, this is a mathematically better way to do it. As to whether there are other flaws and issues, those questions remain for other people to address.
I don’t think the hockey team has figured this out even today (witness all the arguments that Steve needs to do his own reconstructions).
There are all kinds of defects in the temperature data. There are problems with the scarcity of monitoring stations. These problems add a large amount of uncertainty which lies outside the very limited parameters of what you were addressing.
Jeff et al: Congratulations on getting your work published. Well done.
I’m struck by the large differences in interpretation. Ryan wrote in this post:
“I would hope that our paper is not seen as a repudiation of Steig’s results, but rather as an improvement.”
That’s in sharp contrast to Watts’ and McIntyre’s post titles. I guess it’s a matter of whether one choses to emphasize the similarity or the difference, but W and McI’s word choice is rather strong in the scientific realm.
I can’t help but thinking that the way this is framed somehow reflects the level of animosity and/or bias of the writer.
Seriously cool. Jeff C, your comment made my day.
Thanks Bart, Ryan and Nic did an amazing job and deserve the credit. In McIntyre’s defense, he has a long history of abusive reviews (this review wasn’t what I would call abusive) including some recent stuff on his Santer rebuttal submission. This review process in this case was over-extended and forced us to relegate what was the main calculation to the SI. Ryan was far more patient than the rest of us at that point because S09 would never have passed review if it had gone through such a process.
Basically we showed 3 methods which all got the same result and we ended up defending it until Ryan and Nic used a fourth method which still received a negative review from the single reviewer based on ever-wilder criticisms. We replied in short – take it or leave it – no more changes. They took it.
I try to write accurate commentary and if there’s anything incorrect in my commentary, I’d be happy to modify the wording. Prior to Steig, quite a few articles had been written on Antarctic temperatures. As I observed when the article came out, it didn’t seem unreasonable that Antarctic temperatures would be increasing along with temperatures elsewhere.
I have a longstanding objection to climate scientists using “homemade” methods, the statistical properties of which are poorly understood, to derive important empirical results. Steig was merely one more incident. raising this issue is entirely consistent with a longstanding policy.
I definitely had an issue with their policies on PC retention, which were contradictory to the policies set out in wahl and Ammann (ex post) to rationalize MBH results. It’s hard for a multivariate method per se to be “wrong”, but their physical interpretation of Chladni patterns and their use of these incorrect interpretations to rationalize PC retention decisions did, in my opinion, rise to being “wrong” – whether or not such language is polite.
Steig’s principal conclusion was to overturn the “conventional” view of strongly warming Peninsula, asserting that temperature trends in West Antarctic as far as the Ross Ice Shelf were anomalously high – higher than the Peninsula (see Ryan’s table). The abstract stated:
These results turn out to be an artifact of retaining too few Chladni patterns (3) as an eigenvector basis.
Re-stated results restore the very strong increase in the Peninsula with West Antarctic trends now similar to those elsewhere in “continental” Antarctica.
It seems to me that the claims about West Antarctica made in the abstract of Steig et al about the relationship of Peninsula and West Antarctic trends – the “contribution” of Steig et al – is refuted in an objective sense.
As noted above, if someone can show why this is not the case, I’d be happy to modify the terms of my CA post to reflect this.
Many congratulations to all who got this paper written, reviewed and published. Really looking forward to getting to grips with it. Time to dust off the text books…. Might even enroll in a MatLab course.. 🙂
Steve M, I guess it’s a matter of style. I wouldn’t use the word “refute” unless the core method was wrong. But what they did was generally correct, there are some details that this paper offers improvement on . That’s not a “refutation” to me. That seems like an overly emotionally laden choice of words for science.
I can think of articles where I’ve refuted other people’s claims, e.g., they say “such and such doesn’t exist” and I prove that it does and why… that’s a refutation. This offers an improvement on their method, rather than a substantive refutation of it (of course the conclusions change as the method improves, again not the same as a refutation. Their conclusions were not erroneously drawn from their results, their results suffered from defects but the general methodology was not erroneous. I agree you are correct their method contained errors as it applied and that there are weaknesses the theoretical underpinnings of the method that they failed to address here. Demonstrating this wouldn’t be a “refutation”, at least to me, unless the method itself needs to be tossed onto the trash heap.
Your language choice continues to reinforce a general perception of you as somebody with an axe to grind, rather than as an impartial reviewer of others works. Since you choose to continue poking people in the eyes, please stop complaining about them treating you equally poorly when given the chance. That’s just human nature at work. What’s good for the goose is good for the gander.
From my experience it’s had the opposite effect. If what the senior person was saying was obviously wrong, his view gets repudiated by the community. (Opening up the review process leaves him open to criticism now, formally not possible.)
The “code of silence” benefits him, not you.
Ryan O #139,
Thanks for that explanation. I could not understand what Eric was saying about warming of the “Embayment”, when what he really meant was warming of the end of the Embayment adjacent to the peninsula. Still, the difference between the reconstructions, even in this small region, looks quite modest.
Steve Mc #157,
It is obvious to any reasonable person who is aware of the history that the self described ‘Team’ has a long treated you badly and (I think) unfairly. You have every justification to be harsh in your assessment of the rather obvious problems with S09. Few would fault you for that. But sometimes grace in victory is very effective.
I am a lawyer who has had 150 jury trials and am in the business of persuasion. When one’s arguments go beyond their most objective characterization, the speaker loses credibility. This is a huge problem for those who seek stringent CO2 controls. (See Hansen.) I believe Mr. McIntyre has done a wonderful job of improving climate science and opposing group think. Since one of the paper’s authors believes that the new paper is an improvement on the old one but that there were substantial problems with Steig’s paper, I would describe the second paper as having refuted a substantial portion of the original work.
As a compromise, I might suggest Anthony’s wording of “rebuts”. There are definite errors in S09. There are also things in S09 that they did correctly, and without them having done it, Carrick is 100% correct that our paper would not exist. However, I do not think that describing this as simply an extension of the S09 method is wholly accurate, either. “Rebuttal” may be a less charged word that adequately describes the relationship of the two papers.
Establishing the custom of publishing review comments seems to be the way to go. But there is an ambivalence to solve:
A published comment can no longer be considered anonymous if we’re to apply something like stichometry to contemporary scientific articles or something more mundane, like our own hermeneutical instinct.
The anonymity of the review process would not be needed if the much obliged “courtesy” was not an excuse to lack courtesy or time when reviewing papers.
A solution would be to create a more objective correction grid for the reviewers.
Another solution would be to create real reviewing blogs.
Refutation or Rebuttal?
Ryan O, I don’t think the problem is with the terminology so much as with the implication that the all of S09 was discredited. I personally don’t see much difference between rebutting or refuting another person’s work. I think any accurate description would limit the discreditation aspect (whether phrased in terms of refutation or rebuttal)to a portion of Steig’s paper.
I consider Steve McIntyre to be entirely correct in his use of harsh words. S09 was clearly post-modern science, i.e. the methods were chosen to get the required outcome. There was no attempt to let the cards fall where they would. It was basically no different form MBH98 etc in not being normal science, as normal scientists understand it. Given Steve’s treatment, as well laid out in the Hockey Stick Illusion, he has every justification to be harsh. To me, S09 should be withdrawn.
First, congratulations to all the authors. I look forward to reading the article.
With regard to the issue of language, note that the title of the paper is “Improved methods for PCA-based reconstruction:…” S09 is the case study. So “improvement” is the main purpose of the paper.
But the improvement does have implications for specific conclusion from S09, which Steve Mc rightly points our were not justified. In fact, his climate auditing experience made him very quick to see that they were not justified, because they failed to take into account the uncertainty introduced by the use of poorly understood methods. And surely if the improved reconstruction does not support those conculsions, they are now also “refuted”.
That said, what conclusions can be drawn based on the current work? I’m very interested to see how confidently RO et al. assert the reality of their reconstructed trends, and especially to get a better understanding of how confident they are in their error estimates. Those “+/- 0.0X” things are always trickier than they let on.
Hopefully you realize that I do not agree with a statement that all of S09 was discredited. As we based our reconstructions on the spirit of the method that S09 used (correcting the deficiencies we noted along the way), to claim that the entire work of S09 would be a contradiction. I prefer the term “improvement” (as mentioned in the top post), but agree that “rebuttal” would be an appropriate descriptor as well. As an aside, one of the reviewers used that exact term in describing the relationship between the two papers.
In the end, my hope is that our paper made a contribution to advancing knowledge of what is happening in Antarctica in spite of the differing ways in which individuals choose to describe it.
I believe the warmists favorite expression is “Debunk”.
Carrick-“if I present in a publication a photo of a planet using a lower resolution telescope, and somebody comes along with a higher resolution scope, is my photo “wrong”?”
You are comparing quanta to qualia. Pictures are qualia, they are not things which can be judged as “this statistically significantly different” because a picture is not a number, it isn’t a measure, it’s just an image. But if I say, such an such thing has a value of, say 2+-.01 and someone says, I have a better method, the value is 2.4+-.01, you can’t say my previous answer was just “imporved” upon. My answer was WRONG. In math, an answer which is not exactly right is wrong no matter how close it is, and in stats if it isn’t at least close enough to be within some confidence interval, it’s very likely to be wrong. So you may say, “this answer is better” well, sure, but if it is better in some measurably different way, that is statistically significant, then one must conclude that the older result was indeed WRONG, even if slightly so. There is only one truth an in an absolute sense that means that even something which is mostly true is still false even if only a small part of it is untrue.
That’s my epistemology anyway.
The word refute or rebut in reference to a scientific paper to me means that the central conclusions are shown to be wrong.
Eric Steig wrote above: “I appreciate also Ryan’s comment that “I would hope that our paper is not seen as a repudiation of Steig’s results, but rather as an improvement” and his emphasizing that their results (evidently) back up our most important point – -the significant warming West Antarctica.”
Yes, there are differences in the spatial distribution of the warming, but in light of what Steig wrote, and in light of what Ryan (the first author of this article) wrote in the post (“I would hope that our paper is not seen as a repudiation of Steig’s results, but rather as an improvement.”) the W & McI descriptions seem far out.
Eric Steig wrote
“… you guys have apparently demonstrated much greater warming that we found (!), in the most critical area glaciologically (the Amundsen Sea embayment)…”
While our paper was primarily about improved methods for reconstructing surface temperatures and I, like Ryan, am no glaciologist, I think it is worth briefly setting out some key aspects of recent evidence and scientific understanding about the Antarctic ice sheets.
A principal concern, often linked with Anthropogenic Global Warming (AGW), is that global sea level will be pushed up by faster loss of ice from, in particular, the West Antarctic Ice Sheet (WAIS). There are indications of faster ice loss in the Amundsen Sea sector, with Pine Island Glacier (PIG) in particular having shown nearly continuous acceleration.
It appears that the floating ice shelves that form where the glaciers reach the coast have experienced greater melting from below, reducing the friction holding back the grounded ice sheet and causing faster flow. The faster melting is due to changes in the circulation of the relatively warm Circumpolar Deep Water. This water has not been warmed by human activities, and it is far from clear that there is any causative link between AGW and the melting from below of the floating ice shelves.
Indeed, research shows that the real reason for the increased ice loss form PIG is that in the early 1970s it ceased to be grounded on the highest point of an undersea ridge, allowing the glacier to flow more freely and warm sea water to flow above the ridge and melt ice upstream of it. The June 2010 Nature Geoscience paper reporting this discovery suggests that the ice loss encompassing PIG and two neighbouring glacier basins arises from a common cause, and that increased ice loss there and at other glacier ocean outlets in Antarctica is due to an unknown external forcing acting over a multi-decadal, perhaps centennial, timescale.
So, although I am no glaciologist, it seems that surface warming in the Amundsen Sea embayment is not the key issue, and AGW is not to blame for the faster loss of ice.
Finally, it is important to recognize that while the WAIS is large, the East Antarctic Ice Sheet (EAIS) is ten times larger (by volume of grounded ice) and is almost invulnerable to melting by ocean water. Further, the surface of the EAIS is very cold and, as our study shows, does not show any significant warming since 1957.
The warming in the peninsula was well known before Steig et al. 2009 (see Leroux various books, springer) and can easily be explained through atmospheric circulation dynamics as warm air moist advections are channelized by relief as a result of stronger catabatic anticyclones coming out of Antarctica. In fact the same dynamic can be noticed along the Chilean coast where temperatures have been cold at sea level -dense cold air- while temperatures have risen in the mountains where warm air is advected against relief as the strong mobile anticyclone progresses north. It is also these warm air advections on the WA peninsula that bring record snowfalls there. Dr. Steig should at least know that depressions affecting the area are deeper and their frequency has been increasing, hardly a sign of global warming!
Steve McIntyre brings up an important point about chladni patterns that IMO cannot be over emphasized. The spatial eigenvectors are *not* reflective of physical processes. That doesn’t mean you can’t assemble a weighted combination of these spurious modes to fit temperature in a calibration period. But all bets are off in the reconstruction period because these patterns are artifacts of autocorrelation on a disk and this is what the infilling of the reconstruction depends on. The peninsula and perhaps the coastal areas have enough predictor data points for some level of confidence in these areas, but what of the interior (including west Antarctica) with the vast unsampled areas?
You may clutch at straws in your orthodox quailings, and Ryan O may be in the grip of his “contributing to advancing knowledge” modesty trip , but the implications of what was wrong with the Nature paper and how O’Donnell et al, addresses that is pretty clear. W and McI are authors of the very paper you want desparately described, cordoned off and pigeon-holed as just a “refinement”.
How many times has the climate establishment allowed any other view other than the proscribed or the authors own views about a paper to stand as the papal decree on what to think about a given paper? Not once, if I can recall. The usual game is to roll up quote dirtballs from the very papers discussion and fling them at anyone who tries to perform their own interpretation. Surprise, Surprise – the doors have suddenly opened up to an independent interpretation and evaluation to what a paper means. (Without even having looked at the paper?!).
You should have been extremely hesistant to be trying to divide up camps between authors based on your perceptions of how painful the results of O’Donnell are. When are you going to stop playing your realclimate/team games? It only hints of a certain kind of desparation.
Keep finding your permutations and combinations of comfort phrases (“What Steig wrote+what Ryan wrote=I dont have to think of what McI wrote”) – it is good fun to watch.
Congratulations, Jeff and co-authors! This is a huge undertaking and a great step for rational inquiry. Regardless of the results, the key thing is that we have some sincere effort to employ real statistical methodologies. Hopefully this will contribute to an improvement in the field.
I’ve been in rougher disputes than Steve M and my lesson (learned from my mentor) is if you stick to arguing the science dispassionately and not engage in food fights, people will put greater weight on your words than your critics (even if as in my case they were Nobel Prize winners).
We can agree errors were made which this paper corrects. Is that a “rebuttal”?
Aren’t rebuttals arguments that are made to counter earlier arguments with which the author disagrees?
I agree that this paper is more than just an extension of their methods, since it offers entirely new and (innovative as used here) methods. I didn’t mean to understate the degree of innovation exhibited by your work.
I cannot speak for Ryan O, but I say we let him bask in the satisfaction of getting his prodigious efforts published and allow him to attempt to keep personalities out of it – at least for the moment. I think he would rather discuss the papers and point to the differences without having someone in the background pounding on the author of the other paper.
I think you can tell someone that they are wrong based on your work and observations without that person becoming your enemy. My wife and I do it all the time.
Steve M, thank you for having refuted the central thesis of the Steig paper, that the ‘Antarctic is warming’.
Thanks you for restoring some of my faith in science and humanity at the same time. This whole history of alarmism spin of climate data has shocked me to the core.
Carrick, the science and the evidence remains unchanged whether it is called a rebuttal or an improvement. I suspect if one wants to keep the lines of communication open with the rebutted or improved for purposes of attempting to have a rational discussion improved will be favored over rebutted.
Was not the term rebutted given a bad name recently by a blogger refering to those he disagreed with as the most rebutted climate scientist?
Also a correction to a reply above to you where I said use of kid gloves mattered when I meant did not matter.
I agree with your comment “Steve McIntyre brings up an important point about chladni patterns that IMO cannot be over emphasized. The spatial eigenvectors are *not* reflective of physical processes.”
You go on to say: “all bets are off in the reconstruction period because these patterns are artifacts of autocorrelation on a disk and this is what the infilling of the reconstruction depends on.”
I don’t think it follows that a valid reconstruction is impossible, provided that enough eigenvectors are retained to represent adequately the correlations between locations where surface temperatures are known and elsewhere, and those correlations are sufficiently high.
Although there are few points in the interior with long surface temperature records (with a particular paucity in West Antarctica), the satellite data shows a reasonably high mean correlation between East Antarctica grid cells, and likewise between West Antarctica grid cells. While Steig et al didn’t retain sufficient eigenvectors to adequately represent Antarctica’s spatial correlation structure, we did.
I’m not particularly tied to any word that describes the relationship of the two papers. The differences in the words used is a question of semantics, and I am happy to let each have his own opinion. In terms of your definition of rebuttal, once the paper is available, I think you might agree that it fits the bill – at least in part. Or you might not. Either way, I do not object to anything you have said about the two papers, as the language I prefer to use seems to be roughly equivalent to the language you would choose.
To Steve & Ryan O, Refute, Rebut or Improve
You have done very important, high quality work, and this will be last post dealing with terminology. Nothing will moderate the behavior of AGW extremists. However, there are a vast number of people who are in-betweeners. If your descriptions of your results are scrupulously accurate and not infused with a hint of rancor, I believe it will add to your credibility with the in-betweeners. If you are gracious, you will be surprised on occasion by the return of the goodwill from unexpected sources. In my legal career, there are many instances where a small kindness on my part resulted in unexpected goodwill many years into the future.
To Ryan: I believe there is a huge difference between an improvement and a rebuttal. An improvement is often thought to be a modification that makes a good product better. A rebuttal is most often thought to be some form of discrediting. I think by using both the terms “rebuttal” and “improvement,” you are causing confusion and that it would be helpful to settle on one accurate description. So far, I have seen no reason why my suggestion, of calling your work a substantial rebuttal or refutation of the previous work would not be useful.
As a followup to Nic’s comment, standing wave patterns in spatial eigenvectors are a common phenomenon (see North 1984) and do not imply that reconstructions are impossible any more so than the fact that one can use wavelet decomposition on a time series (where the individual wavelets have little or no physical meaning) implies that regression analysis using a subset of wavelets is impossible.
The point of discussion with the standing wave – or Chladni – patterns is that the individual patterns are not representative of physical processes. Their sum, however, must be, as the temperature observations that generated the patterns are necessarily physical. Simply because one can find these patterns in singular value decompositions does not disqualify the data from analysis. Attempting to interpret an individual pattern as a physical process is fraught with difficulty.
It is a subtle – yet rather important – distinction.
#174 – You’ve put your finger on it, Layman. There typically is no physically valid way in climate science to derive physical meaning from PCs because physical theory is not advanced enough to derive physically orthogonal results from weighted averages of merely numerically orthogonal principal components.
It’s therefore become a persistent fault in climate science to use arguments from analogy to artificially impose physical meaning onto strictly numerical principal components.
The tendentious abuse of principal components is the only way they can say something that appears definitive. This is a very serious abuse of science. It’s physics by fiat.
#181, Nic, combining all of the principal components merely reconstructs the original data. As soon as some subset of PCs is broken away, the physical meaning is lost. Demonstrating a statistical correlation between the chosen PC subset and some or all of the original data is just a complicated way of asserting that correlation = causation.
This is not necessarily true. With respect to a system with smoothly decaying eigenvalues, truncation is most likely going to yield suboptimal results. In a system with a few standout modes, truncation may be vastly superior to OLS or other shrinkage estimators. There is a vast body of literature on PCA and PCA variants that you should probably familiarize yourself with prior to attempting to boil down the use of PCA to as simple a statement as you propose.
Besides, your statement implies that any regression result is inadmissible, as even when all of the data is used, the predicted estimates are simply linear combinations of the predictors (and contain zero information from the predictands) based solely on the correlation between the predictors and predictands.
Truncated PCA is a useful tool when performed in the proper situations. So is OLS. So is GLS. So is ridge regression. So are any number of other prediction tools. In every case, if they are attempted to be used on a system of data for which they are not appropriate, the results will suffer. If the practitioner chooses adjustable parameters that are not suitable for the system, the results will suffer. Such is the case with S09.
One cannot extend the suboptimal results from the parameter and regression choices in S09 to conclude that the method is generally false.
# 186, Pat, I appreciate what combining the PCs (weighted by the spatial eigenvectors) does, and I think we share the view that there is no physical meaning in the individual PCs.
But if a large enough subset of PCs and spatial eigenvectors is used to capture the vast bulk of genuine correlations between temperatures at different locations, then I see no objection in principle to working with that subset rather than the original data, and there can be advantages in doing so.
The spatial accuracy of the resulting temperature reconstruction is likely to be affected far more by the limited amount of ground station temperature data available and by restrictions in the strength of correlations between temperatures at ground stations and elsewhere, than by working with a suitable subset of PCs. Sure, using a subset will result in some of the fine detail of the reconstruction being wrong, but over most of Antartica there isn’t enough data to get fine detail right anyway.
Ryan, Pat, Nick
In a situation like this, with spurious spatial modes of variability, more eigenvectors are better than few because it is not as constraining in fitting to the sample points in the reconstruction. Couldn’t one make the argument, in the case where spatial covariance modes are known to be spurious, that the normal rules for PC truncation don’t apply? That one should just continue to add PC’s if the fit improves? After all the concern for overfitting is that noise starts to swamp the “legitimate” covaraince.
Sorry Ryan I had not read #187.
No problem. The issue is not whether the spatial covariance modes are known to be “spurious”, because without the very data you are trying to reconstruct, you cannot determine if the empirical covariances are in error. The key question is whether the regularization method suits the data. As mentioned above, for data with clear standout modes, truncated PCA can provide a more accurate estimate of the population statistics than ordinary least squares. For data with smoothly decaying eigenvalues, truncation is not likely to provide much of an improvement, and other regularization methods may work much better.
The question about whether the eigenvectors have physical meaning is a separate one.
Please correct me if I am wrong here, but isn’t the truncation of high order PC’s always intended to allow recovery of meaningful information and removal of non-meaningful noise? I always thought the less noisy the data, the more PC’s can be included without adding spurious noise to the result, and the ‘optimum’ number of PC’s maximizes S/N ratio in the result. Do I have this wrong?
#189, Layman Lurker,
I agree in relation to the satellite data PCs, if the goodness of fit is measured at withheld verification points. And applying Tikhonov regularization – downweighting the higher retained PCs to reflect their higher noise content – makes it feasible and desirable to work with a larger subset of satellite PCs than would otherwise be optimal.
That is, indeed, the intent. However, “noise” on a system is incorporated into all modes, not just the higher-order ones. Take a data set and perform SVD. Record the eigenvalues. Then add noise to the data set. Take the SVD again and record the eigenvalues. You will note that all of the eigenvalues are larger than the first set. This means that each eigenvalue (and, hence, each spatial eigenvector) contains noise. A generic property of this noise is that it is somewhat equally distributed among each mode. So if you have a few very large modes and the rest are small, then truncating the small ones will remove a good portion of the noise without degrading the signal.
If, on the other hand, you have smoothly decaying eigenvalues, choosing a small truncation parameter will remove not only most of the noise, but also a good portion of the signal. Choosing a large truncation parameter will retain most of the signal, but also most of the noise. In this case, smooth regularization (such as ridge regression or a regularized least squares approach) is likely to be much more effective.
None of this has anything to do with whether the eigenvector patterns have individual physical meanings.
Steve M’s critique at CA concerning Chladni patterns was concerned with using a physical interpretation of the patterns to determine the truncation point. Since the patterns are likely to be mathematical (not physical) objects, this practice can lead you astray. It does not indict the general practice of truncation. Rather, it indicts a specific implementation of truncation wherein the truncation point is determined by interpreting mathematical objects as physical ones.
You are right, but with noisy data all the PCs will contain some noise, the higher PCs having a worse S/N ratio. As you say, with less noise in the data, one should retain more PCs. The problem with retaining too few PCs could be viewed as too much of the signal being omitted rather than the S/N ratio being worse. And simple truncation is not necessarily the best approach.
We should submit our comments to a redacting committee, who will then combine them and post them. 😀
The importance of this rebuttal or whatever you want to call it becomes even more obvious when you re-visit the Guardian’s take on the Steig et al. article when it was published – “Research ‘kills off’ climate sceptic argument by showing average temperature across the continent has risen over the last 50 years.”
Presumably various other media outlets had similar reports.
Thanks for the reminder. — wow!
Jeez, I’m late for the party.
From BCL comment # 51:
Congrats on confirming Steig’s basic conclusion.
Let’s put the two studies into medical terms, where AGW = a bomb that has exploded, sending shrapnel into a body, which would be the Earth’s climate. Dr. Steig and his team of doctors makes the diagnosis that, on the left hand, there is wide spread damage and the hand will have to be amputated (this is the part where everyone should be staring at the palm of their left hand). Dr O’Donnell et al takes a closer look and finds that there is some damage to the hand, but most is confined to the area on the thumb, just above the metacarpophalangeal joint (the joint that connects the thumb to the palm). The thumb might have to come off, but the hand is spared. Now, if I’m the patient trying to decide which doctors to trust, which, in this case, do you think is the one I should be listening to????
Perhaps BCL can explain why it took 88 pages of review if it’s confirming Steig’s basic conclusion?
I’m curious if any of your papers submitted in your entire C.V. went through any process wherein the length of comment/reply interchange extended to 10 times the length of the paper itself. You seem to indicate that this is how things should work, do you have any concrete examples of any of your own papers being subjected to such obstructionism, err, I mean thoroughness?
Jeremy: “I’m curious if any of your papers submitted in your entire C.V. went through any process wherein the length of comment/reply interchange extended to 10 times the length of the paper itself.”
Though the question wasn’t addressed at me, I think more than a few papers submitted to Science and Nature would come close to such a ratio (though a factor of 10 surely is a lot). Note that most manuscripts that are submitted to those journals don’t actually get published there, but rather in a more specific journal after having tried at the high impact ones. The total tally of reviews can easily mount that way.
Shub: “Ryan O may be in the grip of his “contributing to advancing knowledge” modesty trip”
Perhaps that’s the distinguishing factor indeed.
Ryan O, Nic L,
Thanks. And sorry for doubling your efforts!
Might I suggest a solution to the rebuttal/refute/repudiate debate between S09 and RO10?
Typically when a paper is cited to prove a point the text will read something like “…bark, bark,bark warming trends in the Antarctic (S09, and other refs).”
If we see in the future papers that reference S09 and do not reference RO10, then clearly the supporting evidence is contained within S09 and needs no further supporting references. If however we see in the future references to S09 and they always contain an associated reference to RO10, or the references only cite RO10, then we can say that S09 has been supplanted/improved/refuted by the later paper.
In five years there should be enough evidence to settle this question, if anyone still cares.
I do not think so. Journal of Climate is a respected Journal. Anybody who points to S09 without any reference to RO10 will be made to look a fool by anyone who brings it up. The two papers are forever locked in an embrace of (if not love) close connection.
When eric comes back..
eric do you think it should have been published?
We all know which merry team of modest geniuses would immediately qualify themselves to be taken with the utmost seriousness by the mere virtue of their humble, patient and openminded attitudes.
As much as anyone would fantasize otherwise, modesty and humility are never an infalliable sign of agreeability, nor is arrogance an undoubtable sign of genius.
“If you can meet with triumph and disaster
And treat those two imposters just the same;
If you can bear to hear the truth you’ve spoken
Twisted by knaves to make a trap for fools,”
Poem “If” by R. Kipling
Cantering down a dark defile of knowledge.
I assume you mean due to the gatekeeping that seems rampant. Most papers, in my experience, if they truly need this much back-and-forth from reviewers are basically rejected. Now it being presumed to be either a miracle of patience or miracle of human reason that this paper made it through such a gauntlet; if there were truly enough wrong with it to justify so much discussion why wasn’t the paper simply rejected by the reviewer in question and the reasons given to the publisher? This would seem to have been a much simpler thing to do and would have avoided many (presumed) months of painful correspondence.
The only conclusion, from my perspective, is that there wasn’t anything worth rejecting the paper over, and the reviewer decided to nit-pick on comments made rather than the substance of the paper itself to obstruct publication. That is all blatant conjecture of course, but this area of science has not demonstrated any form of fair play in over a decade now. They deserve the reputation they have earned.
Steig’s paper flew in the face of decades of research that said the Antactic was remaining stable or cooling slightly and experiencing sea-ice growth rates of instrumental record highs. Now someone comes out and shows mathematical abstractions were misinterpreted to have physical meaning, and it’s normal to expect this painful deluge of reviewer discussion? I have to say, if that’s how Science and Nature work, it’s no wonder they appear horrifically biased.
congratulations to all concerned !
I don’t see any impediment to publishing reviewer comments. It is rare that they are published; but I know no reason why they shouldn’t be, other than tradition and that other people generally are not interested.
“asically we showed 3 methods which all got the same result and we ended up defending it until Ryan and Nic used a fourth method which still received a negative review from the single reviewer based on ever-wilder criticisms. We replied in short – take it or leave it – no more changes. They took it.”
There is a good side to this. If an editor understands the science, they are capable of understanding what the criticisms are, and when a reviewer is being reasonable- or unreasonable. You don’t have to see someone being unreasonable on too many occasions before you decide that a particular reviewer is downright quirky.
After going back and re-reading the post by Ryan and some comments by both Ryan and Jeff something stood out to me: The Reviewer that caused the 88 page wackiness just about ensured that O’Donnell et al got published. He/she put the editor, Dr. Broccoli, in a bad position such that if he denied publication on that reviewers recommendation he would have basically proven one of the major allegations from Climategate of how sceptic (non team) papers get reviewed and rejected. From that imagine a much different headline post here, CA, WUWT and such places as ClimateDepot and even Dr. Curry’s blog with nuggets from the reviewer attached.
To help make my point I went and grabbed the pertinent quotes from both the post and comments:
Ryan O’s post:
To me the editor was forced to get a fourth reviewer because he couldn’t justify rejecting on that one reviewers grounds and he didn’t want to approve with just 2 positive reviews. The reason why he had to do that was something Jeff let slip IMHO:
Notice that they showed that 4 different methods all came up with the same result and that reviewer still wasn’t satisfied. Now Imagine Jeff ranting about that in a post, how after showing the paper with different methods in it coming to the same result, being negatively reviewed showing a 4th method getting the same result and more negative review, totaling 88 pages and still rejected. Can you say: going viral across the blogosphere? Especially since Ryan has already stated that some nuggets from that he is prepared to release some nuggets even now imagine what that would look like in conjunction with a rejection:
So at least that reviewer (Team of reviewers) didn’t learn that lesson of Climategate and basically ensured that this paper got published and at the same time probably pissed off Dr. Broccoli and the Journal of Climate. I wonder how they will be received the next time they submit to JoC?
Nic L #172
For a non-glaciologist you’re doing a pretty good job.
As you say the present theory is that ice mass loss on the WAIS is related to ocean temp, nothing to do with atmosperic temp.
Eric if you’re looking to refresh yourself on this you could try Scambos et al (2004), Rignot et al (2004) or Shepherd, Wingham and Rignot, (2002).
128.Eric Steig said
“No one has ever though[t] the Ross Ice Shelf area (blue in your results) was sensitive to increasing temperatures.”
This quote from Ted Scambos seems to be all over the internet.
“More importantly, regions of the giant “Ross Ice Shelf are just a few degrees Celsius away from being overtaken by the same processes that have destroyed the Larsen.”
try this one
The team’s co-chief scientist, Tim Naish, told The Press newspaper the sediment record was important because it provided crucial evidence about how the Ross Ice Shelf would react to climate change, with potential to dramatically increase sea levels.“If the past is any indication of the future, then the ice shelf will collapse,” he said.
Sedimentologist Tim Naish, Institute of Geological and Nuclear Sciences
That’s right no self-respecting climate scientist has ever speculated about the collaspe of the Ross Ice Shelf.
I think Lucia first touched on this in comment #80. It may be thought to be common practice or a legal requirement to not show reviewer comments. I do not know. But in a practical view why would review comments be improper if the anonymity of the reviewer is preserved? It seems inconsistent to require both anonymity of a reviewer and withholding of review comments.
Imagine a newspaper reporter saying he has an anonymous source but can’t tell you what the source said. Absurd. It is the whole point of being anonymous to be able to show information without impugning the source.
I understand that the actual writing style and use of certain idioms in review comments may give pointers to the identity of the reviewer. Yes, probably can have high confidence of getting some good pointers toward identity. So if anonymity is maintained but reviewer comments published then anonymity might indeed be compromised.
I must say that this anonymity thing seems contradictory to the concept of an open science process to me. Is it just me that thinks this? Because it has frequent precedence in many journals over some longish periods of time does not add weight to an argument that it is the best process.
The question I am most interested in at the moment is the substance of the review objections. This work got pretty thorough discussion on several sites. Did the expert review process provide stunning new insights not provided by blog discussion, or did the technical blog discussions cover most of the issues raised by the reviewers. With your “problem reviewer”, was there anything new or just a rehash of the weak arguments presented at RC and elsewhere ? I am interested because this looks to be a pretty specialized case where the material was given a pretty thorough public review before the formal review. So how well did the public review work ?
I think it would be more appropriate if Ryan answered. He really did most of the work with Nic right behind.
Is your code/method easily update-able, or does it require significant manual data/code work? I’m asking because, if your model remains the “best” way to describe antarctic temperatures in a few years, an updated look with more years of temp/satellite data would be a great thing to do.
Re: John Whitman (Dec 4 10:00)
“Is it just me that thinks this?”
No. But I would have thought anonymity would be attractive here. There has been much horror at the climategate discovery that people get het up about papers and reviewing. But they do, and journal editors have to thread their way through that to get an honest assessment of the real merits of papers.
With anonymity, referees can more easily say what they really think, without having to worry about what might happen when they apply for a job withy Prof X’s lab. etc. It’s a measure against dominance by powerful groups.
There’s the practical consideration that the big problem for editors is finding reviewers prepared to spend the necessary unpaid time for a thankless job. If reviewers have to enter into a bruising personal argument as well, I suspect there’ll be no volunteers.
In business, you have to disclose conflicts of interest. A problem with anonymous peer review occurs if the reviewer has a conflict of interest – such conflicts are not limited to financial conflicts, but include personal bias for or against.
It seems to me that the requirement to disclose conflict of interest should supercede anonymity.
To some extent, journals are getting the quality of work that they are paying for. Reviewers with an axe to grind seem to be far too involved with the review process.
Somehow it seems scientists have lots of trouble understanding the value and requirement of anonymous and pseudonymous blog commenters, but can perfectly understand its value in the anonymous peer review system. The operational principle is the same: to have the freedom of offering one’s opinion unaffected by one’s identity.
It also seems scientists clearly identify conflicts of interest in their political and scientific opponents – the skeptics – “you try to fault the science because of your ideology”. But such a process never affects their own anonymous peer reviewing?
Re: Steve McIntyre (Dec 4 18:54),
“A problem with anonymous peer review occurs if the reviewer has a conflict of interest – such conflicts are not limited to financial conflicts, but include personal bias for or against. “
Reviewers are responding to the editors, who are the ones who must evaluate what they say. And the editors know who they are.
The task of finding reviewers with no biases is often hopeless. More often, editors try to balance biases on the panel, and assess the reviews accordingly.
Re: Shub Niggurath (Dec 4 19:32),
A bad analogy. Journal reviewers are chosen by, and known to, the editors. Journals have reputations. Blog commenters aren’t vouched for by anyone.
Peer review really is a sticky wicket, and maybe there is no perfect solution. But reviewers that are obviously biased for or against a paper would seem to bring little to the party, and may only impede the progress of science. I am astounded that editors appear to go out of their way to sign up reviewers who have obvious conflicts of interest. Why would they not at least try to line up (as much as possible) more dispassionate reviewers? RO10 is a perfect example of the problem… 88 pages? With one reviewer?
“There has been much horror at the climategate discovery that people get het up about papers and reviewing.”
I think it was more than just that; people in every field often think published papers are rubbish (I sure do!); often they are rubbish, and are generally ignored. Peer review is never going to stop this. What offended in the EUA emails was the behind-the-scenes efforts to influence the review policy by pressuring reviewers, publishers, and editors to simply not allow certain papers to be published under any circumstances. It was the attempt by a small group of climate scientists, all with conflicts of interest, to control what gets published. It was wholly inappropriate, and those involved knew it was inappropriate… “don’t give this information to anyone else…” “I’ve learned that one of the reviewers works in your lab..” Seriously Nick, do you think this sort of thing is ever appropriate?
What grade do you guys think you’d get in Steig’s matlab course?
#187, 188, Ryan and Nic, thanks for your thoughtful replies.
Nic, I have no problem with anything you wrote, and agree that we agree about the lack of physical meaning in individual PCs. 🙂
Ryan, I’m not dismissing numerical fits as a class, or PCA as a method of analysis. My own work would be lost without numerical fitting, and I’m very grateful to all the extremely clever people who’ve made those methods possible.
In the case of the work you all did, you’re using temperature data that has discrete physical meaning in and of itself (in distinction to tree rings w.r.t. degrees C, for example).
I’ve not yet read your paper, but from comments here I’m guessing you used the Antarctic AVHRR temperature field to calculate recombination weights for the PCs you extracted from the temperature/time series of the Antarctic surface stations. I’m guessing also that you worked hard to maximize the correlation of the derived continental temperature field with the local field at each surface station. High correlation between the local field and the derived field, around each station locale, lends confidence that the derived field far from any station has some physical meaning.
Here’s my problem: There are a total of 29 Antarctic surface stations, shown here. Of those 29, twenty-five are on the coast and only four are inland (Siple, Byrd, Vostok, and Amundsen-Scott).
As Nic pointed out, data for the interior are very sparse. The area of Antarctica is 14 million sq.miles. That means the uncertainty of your projected (or predicted) temperature field should increase with distance from the anchor data points at the 29 surface stations. The projected temperature field in the interior of Antarctica will be of relatively high uncertainty. Now, we look at your derived thermal map at the head of the post and at the temperature bar. No uncertainty indicators.
In my view, you need to show a 2-dimensional display of the results. D1 is the temperature field. D2 is an uncertainty topology, which would tend to maximize between the stations. That would give us all a good feel for how much confidence to place in the apparent areal temperature distinctions of the false color map. Maybe you’ve done something like that. I’ve not seen your paper yet, and if an uncertainty map is there, I hereby and publicly take back all of those critical comments. 🙂
But there’s one more thing. When I and my colleagues do numerical fits, it’s kept in mind that the physical uncertainty in the results is distinct from the numerical standard deviations of the fits. Typically, the physical uncertainty is larger than the numerical uncertainty from the fit itself.
Looking at your table of trends below the areal maps, the trends and SD’s are quoted to 0.01 (hundredths) of a degree. I’m guessing that these are fit standard deviations.
It’s certainly true that the Antarctic surface station temperatures cannot be accurate to (+/-)0.01 degree. It’s never been shown, to my knowledge anyway, that the Antarctic temperature measurement errors are invariably random and decrement away as 1/[sqrt(N)]. I’d expect plenty of systematic errors, due to wind speed if nothing else, which can’t be assumed random.
Honestly, I don’t think it’s possible to detect a physical trend accurate to (+/-)0.1 C/decade in Antarctic temperatures over the last 50 years, much less to (+/-)0.01 C. A better guess would be no better than (+/-)0.5 C/decade, and that under ideal measurement conditions.
Maybe the satellite data are now that accurate, though I tend to doubt that, too. Satellite instruments are well-calibrated on Earth before launch, but how does the heating/cooling cycle affect response in space? Is that variance stationary?
I don’t believe anyone has ever bothered to monitor the surface station sensor variance as it appears in individual Antarctic temperature measurements. Heck, no one has bothered to measure the systematic variances imposed on temperature measurements from the sensors in USHCN surface stations, and everyone now knows from the admirable work of Anthony Watts and his brigade of volunteers that the USHCN produces the highest precision temperatures in the world. 🙂
Anyway, that’s my worry. None of that reduces my high admiration for what you’ve done. But my mind won’t leave me alone, and I’m very sure you guys know what torture that is. 🙂
“Honestly, I don’t think it’s possible to detect a physical trend accurate to (+/-)0.1 C/decade in Antarctic temperatures over the last 50 years”
Actually, that is pretty close to my limit and I’ve spent the time with the data. If someone put a heat gun to my head and demanded an answer for ‘certainty’, I think you just about nailed it.
Statistically based on variance however, it is about 60% of that.
Re: Steve Fitzpatrick (Dec 4 21:31)
I don’t particularly want to trawl through these guy’s emails again. But I didn’t see anything major there. Yes, they shouldn’t discuss reviews they have done, and shouldn’t gossip about who reviewers are or where they might work. But it’s unfortunately quite common (eg your speculation in #2, I saw MM and IJ mentioned).
And yes, while I think their annoyance over what they see as bad papers getting into journals of societies of which they are members is formally reasonable, it is probably augmented somewhat by tribalism.
89. NVW said “I think your paper will be of interest in the discussion of Arctic warming anomalies – I recall discussion of how circum-Arctic temperature stations are being used to smear the values over the whole polar region(This was a post dated March 25 2010 by Willis Eschenbach on Watts Up With That). Your paper, presumably is an example of why this needs to be carefully considered.”
Then 141. Steve Kock added “While the results of this paper are specific to the antarctic, this is still a very important point because it shows that extrapolation over enormous distances (in the arctic, for example)is not acceptable. Just removing these extrapolated temperatures in the arctic would cool down Hansen’s GISS temperature quite a bit and would make much more sense WRT to both numerical analysis and climate science.
It would be interesting to recalculate the GISS global temp record without the bogus arctic extrapolated temps. 2010 would be quite a bit cooler, no doubt.”
Would any of Ryan O’D et al. care to comment on this point, which I’m also sure is very important? Is the Arctic really warming and, if so, is it really warming at the rate claimed? What are the distances between the temperature sensors whose extrapolations are telling us 2010 has been so warm? I’m not a scientist so apologies if I’m missing something obvious.
For info. re: review – the AMS Council adopted new Guidlines for authors, editors and reviewers on 22 September 2010.
Click to access authorseditorsreviewersguide.pdf
Apologies if these have been referenced up thread and I missed it.
Many editors don’t know in advance who is likely to have an axe to grind, Some authors or papers are interested in papers that show improved methods with new results and are happy to give comments and accelerate the advancement of science. Others are more interested in preserving their own result. An editor is unlikely to know which is which until after witnessing the impeding author in action. So, Broccolli probably had no reason to expect reviewer X was an impeding one.
The AMS policy states:
In philosophy journals, the reviewer so conflicted is required NOT to continue with the review. The above policy appears to permit the editor to retain the conflicted reviewer without any notification to the author of the conflict. The conflicted reviewer can then proceed as though he were an unconflicted reviewer and the responding author lacks this information.
In our case, Reviewer A referred to Steig et al 2009 no fewer than 106 times in his first review, challenging our characterization of their results in minute detail, while Reviewer C referred to Steig et al 2009 twice.
Reviewer A required that the discussion of Chladni patterns be removed. Reviewer B wondered what Chladni patterns were – somewhat belying Reviewer A’s insistence that this sort of issue was already well understood in the literature. (The issue has been understood for many years e.g. Buell in the 1970s, but that doesn’t stop the problem from recurring, as in Steig et al.) If a responding author knows that the reviewer has an adverse interest, then he might contest this sort of reviewer requirement in stronger terms; if the author merely suspects it, then his options are far more citcumscribed.
If journal policies end up permitting reviewers with adverse interest of this type to act as reviewers, then this adverse interest should be clearly disclosed to the responding author.
Saiers, as editor of GRL, allowed authors to “reply” and considered the “reply” in evaluating the “comment”. But the reply was not confused with a review. This seems fairer to all parties than allowing a conflicted party to act as reviewer without disclosure of the conflict.
My views here are influenced by the handling of conflicts of interest in business cases. Acting in the presence of a conflict is strongly discouraged, but, if one arises and cannot be otherwise avoided, the conflict must be disclosed to all parties and all parties must consent.
The discussion on peer review and anonymity here is a really interesting one.
The British Medical Journal (BMJ) some time ago moved to a system where not only are the author(s) made known to the reviewers but the reviewers are made known to the author(s). An interesting commentary on this move can be found here.
Keyboard paged impediment.
Could not be foreseen.
There may be three opportunities for contention over a series of conflicting papers. The first might be in the letters following publication of the first paper. Did Nature prevent a discussion of Steig et al? Were the objections sufficiently focused timely to submittal of letters to Nature or did it take longer?
From the outside, it doesn’t seem unreasonable to have a pre-publication review by a conflicted party especially if it might be assumed that that party is really sharp on the issues raised by the new paper. Who better than Steig, if this idea carries any water?
The third opportunity would be following publication when a hopefully open letters section would allow Steig and others to have another day on the court.
Would you really not want your paper looked at by it’s potentially most knowledgeable critic? Assume, of course, that the critic is able to rise above the level of childish harassment.
Isn’t prevailing in the scientific world overcoming the objections of your strongest, not weakest critic?
Maybe Steig isn’t the strongest critic.
AAAK, the dreaded “it’s” again. forgive forgive.
I should have begun #228 by noting that after reading the comments of Ryan O and Nic L in #187 & #188, and thinking a bit, I realized using the Antarctic AVHRR data to weight the reconstruction of PCs derived from the surface temperature series is physically legitimate. So, my objection in #185 to the abuse of PCA in climate science would not apply to the new work of Jeff, Ryan, Nic and Steve. Thanks for your reasonable responses, guys.
Nic and HR, thanks for the tutorial by Nic and concurrence by HR of the most recently accepted theories attempting to explain the currently held disintegration of the West Antarctica Ice Shelf (WAIS), i.e. upwelling warm sea water that is too old to have been warmed by AGW.
Unfortunately in the link I gave above, the more alarmist view confuses Peninsula estimated warming for that for the West Antarctica and while not linking directly the surface temperature increase to the WAIS disintegration talks about them in the same breath. I have read other alarmist reports that talk about surface WA temperatures and the WAIS disintegration without bothering to get into the latest conjectures on the cause and effect relationships.
I believe that the consensus view of the conjecture on relating AGW to the disintegration of the WAIS is that the upwelling of old water is caused by the increase in precipitation due to AGW which dilutes the salinity in the oceans and in turn speeds up ocean currents and upwelling of the old warmer water in the sea near the WAIS.
Nature and Science create a problem because of their reluctance to take responsibility for comments that rises above a negligible word count – on the basis that this should be done in specialist literature.
Thank you Pat Frank! (#228)
There is a tendency sometimes to crunch the available data to within an inch of its life. No matter how sound the math behind the crunching, it is only as good as the data. And sometimes the data is so sparse and so questionable that the only legitimate answer is “we don’t know”.
That’s why I thought this paper was really limited to one purpose — showing that the stats methodology Steig used was wrong. It doesn’t claim to know or show anything about the science, only that Steig’s crunching was wrong.
#241 Steve McIntyre,
That Steig et al 09 was in Nature suggests that its import (to Steig and likely Nature as well) was not in its specialist workings but in its “unprecedented” conclusions, which to my mind fully supports your characterization of O,Donnell et al 10 as a rebuttal. O(10) seems to evolve and improve Steig’s technique but in so doing disrupts the basis of Steig’s conclusion. Can I assume that Nature refused to publish O,Donnell et al?
It does seem sensible to try to publish in the highest circulation vehicle available regardless of whether the real value of your work is of a more arcane nature. It also seems obvious that Steig’s purpose was to trumpet the conclusions not share simply the new technique.
Have I got the nuances of this substantially correct?
oops. substitute refutation for rebuttal. onset of dementia, don’t you know. Here, they’re coming to take the notebook away from me.
“I believe that the consensus view of the conjecture on relating AGW to the disintegration of the WAIS is that the upwelling of old water is caused by the increase in precipitation due to AGW which dilutes the salinity in the oceans and in turn speeds up ocean currents and upwelling of the old warmer water in the sea near the WAIS.”
The recent finding that the acceleration of the Pine Island Glacier is linked to it becoming disconnected from a submarine ridge, with melting by warm ocean waters evidently having gone on (probably affecting other nearby glaciers as well) for many years before AGW could have had any material effect, appears to rule out AGW as the primary cause. That is the case whether or not AGW affects ocean upwelling, or changes wind patterns to bring deep warm water into contact with the WAIS (another theory I have read), or anything else. But I wouldn’t rely on the consensus view changing any time soon to reflect this unwelcome (to AGW proponents) new research!
#243 J Ferguson
In truth, we did not even try to submit the article to Nature. The length required, and the inability to phrase most of it in non-specialist terms, eliminated Nature as a potential option. The article simply doesn’t fit Nature’s criteria.
Thanks, Nic L for the further clarification. Your original comments caused me to do more background on the WAIS. It takes careful readings (or a careful reader to point it out to me) of the material on subjects such as this one in order not to miss some of the more nuanced evidence and were it fits into the bigger picture.
I should dig out the article that I read that stated that a group of scientists working in the area of science concerned with the disintegration of the WAIS put all the evidence together and then “locked” theselves in room in order to come to a consensus on if(?) and how AGW could be causing loss of ice from areas of the WAIS. I kid you not. The consensus was what I reported above and its published along with a description of the gathered scientists being experts in separate fields that deal with the WAIS concerns and some even experts in two different fields.
Despite the evidence you presented above, how do we go against that impressive array of experts and their consensus?
In addition to the possibility of reviewer bias,I get another impression from these two papers.
The Steig et al 2009 paper suggests that the peer reviewers and the writers did not know enough about the statistics used while the writers of O’Donnell et al 2010 knew more that most of the reviewers. This is where peer review falterss and blog review stands out.
I’m not aware of the article you quote, but I agree that overturning an erroneous understanding based on a consensus of ‘experts’ is a long and difficult process.
An example of an erroneus understanding persisting is the effect of the Gulf Stream on the UK. Most people here believe that the primary reason the UK has mild winters is the warmth brought by the Gulf Stream. No doubt that is why AGW alarmists have emphasized concerns that global warming might cause the thermohaline circulation driving the Gulf Stream to slow down considerably. But according to a well argued and referenced paper (available at http://www.americanscientist.org/issues/id.999,y.0,no.,content.true,page.1,css.print/issue.aspx ) scientific research shows that the Gulf Stream accounts for only a relatively small proportion of the mildness of the UK’s climate. Moreover, this has been known to be the case for many years. But despite that, the myth that we would all freeze in the UK if it weren’t for the Gulf Stream persists – almost everyone believes it.
Nic L here is the link and comment from the article to which I referred above:
And here is the part of the article on how a consensus was formed:
Links Between Atmosphere, Ocean & Ice
Blankenship said when the workshop began, fewer than five attendees suspected this link between atmosphere, ocean, and ice; by the end, all 25 agreed it was the most plausible explanation. He said each person was an expert in one, maybe two areas.
“But to say that atmospheric changes are causing the ocean changes that are causing ice sheet changes, that requires more self confidence than most of the people had,” he said. “That could only happen by bringing together so many people with overlapping skill sets. The result was a surprise and a significant moment. We all agreed that was the most likely answer.”
If you read the paper and all the posts about it on R/C, as well as the previous posts on Antarctic warming and parse them critically a clear story emerges:
The peninsula has been warming since measurements began in the 1950’s but the interior has been cooling for half that time. ‘Yeah, R/C experts knew that.’ No wait, yes of course the peninsula has been warming rapidly but West Antarctica and the interior are also warming! No wait again, West Antarctica is warming “rapidly!”
‘The Antarctic Peninsula, site of the now-defunct Larsen-B ice shelf, has warmed substantially. On the other hand, the few stations on the continent and in the interior appear to have cooled slightly (Doran et al, 2002; GISTEMP)’
‘Their conclusions that West Antarctica is warming quite strongly and that even Antarctica as a whole is warming since 1957’ –
‘So what do our results show? Essentially, that the big picture of Antarctic climate change in the latter part of the 20th century has been largely overlooked. It is well known that it has been warming on the Antarctic Peninsula, probably for the last 100 years (measurements begin at the sub-Antarctic Island of Orcadas in 1901 and show a nearly monotonic warming trend). And yes, East Antarctica cooled over the 1980s and 1990s (though not, in our results, at a statistically significant rate). But West Antarctica, which no one really has paid much attention to (as far as temperature changes are concerned), has been warming rapidly for at least the last 50 years.’
The point of S09 wasn’t that the Peninsula was warming – that was well known. The takeaway was that Western Antarctica and the interior were warming “rapidly”. That’s the way R/C presented it, that’s the way Nature presented it, that’s the way Dr. Steig wrote it there is no other way to read the message that was presented at the time. To read Dr. Steig write that this paper ‘improves S09’ is mind boggling.
Sorry links to relevant R/C posts were mangled.
251, 252 D.Robinson – it’s at times like these I wonder if anyone is taking a mirror of the RC site for posterity?!?
Thispagefrom eastangliaemails has a discussion involving Peter Thorne, Phil Jones and David Parker. The following issues were discussed:
1. Homogeneity of antarctic manned stations. It seems Jones has looked at this issue in the past.
2. AWS data quality
3. Jones mentions that S09’s validity might depend on the assumption that the AVHRR covariance structure must hold in the reconstruction.
IIRC here is an interesting excerpt from a Peter Thorne email expressing some trepidation about whether a proper homogeneity analysis of manned stations in Antarctica would reconcile with S09’s findings. …”a can of worms too far …”
Re: Curious 253
It’s infuriating. At first R/C admonished skeptics saying, yeah it’s cold and cooling in the interior of Antarctica “we knew that” and the models showed it.
A few years later they said wait, it’s not cooling, it’s warming slightly, as the models predicted.
After S09 they said the interior and West of Antarctica has been warming “rapidly” for 50 or 100 years.
Now, Dr. Steig thanks the authors for finding “much greater warming than we found” referring to the increased trend in the peninsula rather than the decreased trend in the remainder of the continent.
Reminds me of something
My “amen” to that. If all of the available Antarctic station records extended back to 1957 AND were intact AND were uniformly accurate in representing the temperature variations of the region instead of camp sites, there still would be the problem of lack of data coverage in the continental interior. The dimensions of Antarctica are such (5.1 million square miles)that the virtual extinction of strong temperature coherence at multidecadal periods within several hundred kilometers means that there is NO physically meaningful information available for the bulk of it bearing on the question of trend. The actual station data situation, of course, is nowhere near that sanguine.
Mathematical cleverness notwithstanding, Steig’s reconstruction had little to with physical reality and should never have been published as such. And fixing his methodological flaws cannot substitute for unavailable data. Kudos to Ryan and his co-authors nevertheless.
We found the continental result to be stable to both covariance and spatial position which are mathematically very different methods. This is a strong indicator that the sparseness of data had little impact on the ability to know trend continent wide. If covariance could distribute the data, noise wasn’t the biggest factor. It would be easy for bad data to cause problems for sure, but most of the interior stations were fairly consistent with one another.
I’m not claiming perfection, more data would be better, but I am confident that the trends are reasonably well known.
The issue I raise involves not just the zero-lag covariance, but the entire covariance function out to lags commensurate with multidecadal oscillations. Satellite data is too short to provide adequate determination and the few Antarctic station-record pairs long-enough for cross-spectrum analysis showed at best marginal coherence at the lowest frequencies. Reconstructing high frequency-components is not enough when the metric of interest is trend. Mere mathematical stability does not address the issue of synthesizing time-series without a basis in coherent measurements.
I’ll check back for your response tomorrow pm.
Does anybody know where I can find a journal link to the paper? It is still not showing up in the early release section of JOC,
It’s not published yet for some kind of technical reason. We believe it will be shortly.
Steve Fitzpatrick predicted RC would either not cover the paper, or badmouth it. That’s doing about as well as the ‘skeptical’ predictions of cooling in 2010.
Ryan O’Donnell: Our paper in the Journal of Climate shows a somewhat better way to look at the same data. Antarctica is warming a bit more in summer, and a bit less in winter in the Ross Sea region. In fall it is cooling a bit more too, and so the overall trends are smaller. Still, West Antarctica is definitely warming significantly, as Steig et al. found. That’s interesting.
Eric Steig: Nice paper Ryan. Thanks for sending along a pre-print.
Steve McIntyre: Hey, we got published in the Journal of Climate! Another paper showing that the “team” made up the data again! (Sotto voce): Ryan says it it is warming a bit more in summer, and a bit less in winter in the Ross Sea region. In fall it is cooling a bit more. Otherwise we get the same results, though the magnitude of the trends is smaller. But West Antarctica is still warming significantly. But I really don’t care. The peer review process is broken, which is why.. umm…our paper was published in the leading climate journal.
Liberal Media: That paper wasn’t published in Nature, so we’re not very interested.
Conservative Media: Antarctica is cooling. Global warming is a fraud.
P.S. For those actually interested, yes, I’ll have more to say about O’Donnell et al., but overall, I like it.–eric”
Are you being sarcastic or an idiot?? The distinction is difficult to discern.
Are you dyslexic, or just illiterate?
Here, let me help you: the link is to Eric Steig’s (first) post on Real Climate about O’Donnell et al.; the quote is the money shot from his post with respect to Steve Fitzpatrick’s prediction.
Steig’s post is peppered with sarcasm (I’ll leave it to you to figure out who are its targets, as homework) but overall, Steig likes the paper. Fitzpatrick predicted either no coverage or ‘badmouthing’ of the paper. Clear now?
Who predicted cooling?
I am utterly confused.
Neither, but I do question your moronic paraphrase. Furthermore, you have not been paying attention if you really think Steig likes their paper. They spanked him publicly and now he is trying to spin it as if their results confirm his. PR, nothing more.
Has anyone looked at what portion of the continent has statistically meaningful warming, versus statistical meaningful cooling? If I recall, the decadal trends are accurate to roughly +/-0.1 deg C.
It appears from the maps, that O’Donnell found statistically meaningful warming in about 30% of East Antarctica, with less than 5% showing meaningful cooling. The previous paper by Steig had far less resolution, and showed less than 5% of East Antarctica showing meaningful warming (with the rest of the area not showing any meaningful warming or cooling trends).
In West Antarctica, both the O’Donnell paper and previous work by Steig show statistical warming over 70% of the area, but the O’Donnell paper shows much warmer anomalies across much of West Antarctica, particularly the Antarctic Peninsula. O’Donnell shows most of the Peninsula warming over 0.4 deg C per decade.
So the lead paragraphs in an accurate news article on the O’Donnell paper would read something like this:
NEW STUDY SHOWS SIGNIFICANT WARMING OVER MUCH OF ANTARCTICA
A new paper published in the highly regarded Journal of Climate shows statistically significant warming in over 70% of West Antarctica, consistent with a previous study. However, the new study shows much higher warming throughout the Antarctic Peninsula than the previous work showed. West Antarctica has some of the most threatened ice sheets, glaciers, and ice shelves on the continent.
The new study also shows significant warming over 30% of the much larger East Antarctica ice sheet, whereas previous studies showed no statistically significant warming. In contrast, less than 5% of East Antarctica showed significant cooling, inconsistent with some forecasts expecting significant cooling due to effects from the ozone hole over the South Pole.
Is this a reasonable summary of the new results?
268 Zinfan – please can you rework it to explain what “statistically meaningful/significant” means? Also I think you should not use qualitative terms such as “much” when the numbers are more illuminating and I think you should explain the nature of the “threats” facing the ice sheets etc. If possible an historical reference to other periods in history would also be useful to give context. And don’t forget when you quote the reported trends report the confidence intervals too. And all this before the paper is published!!
Curious, I remember a key criticism of the Steig paper was that the cover display on Nature showed most of East Antarctica as slightly orange, inferring that it had warmed. The critics (including some of the current paper authors) quite correctly pointed out, that the photo was misleading since Steig’s calculated warming for those areas wasn’t statistically significant. I recall some critics suggesting that the entire range from -0.1 to +0.1 should be displayed by white pixels, so that the display would be more accurate. I found this argument compelling, so I am using the same logic now, to interpret the data.
Regarding threatened ice shelves, there have been a number of general articles written about the loss of several large ice shelves on the West Antarctic peninsula, and of course the loss of the Larsen B shelf has been widely publicized. The possible deterioration of the Pine Island Glacier complex elsewhere in West Antarctica has been studied and reported as well. Of course, the loss of these are tied primarily to ice melt at the grounding line, and the rising temperatures in the Southern Ocean waters, and not to surface temperatures, but in a general article, these features are useful to identify to the public where the warming trends confirmed in this paper by O’Donnell et. al. are located.
270 Zinfan – thanks, but where are you expecting your news article to appear? Depending on the outlet, not everybody will have the background to hand. Nor will they have views on the meaning of “statistically meaningful/significant”.
As far as Nature’s cover goes, I remain surprised that it did not have an embedded scale and reference, as it was relayed to other news media who unthinkingly reproduced it.
Hopefully your intention is to “inform” rather than to “persuade” and you will be able to write your news release accordingly. And obviously that will have to wait until we’ve actually had the chance to read OLMC10 for ourselves.
Re-created image for the RLS reconstruction, 1957 – 2006, with areas of statistically insignificant trends overlaid in gray:
Nick Stokes, what if you have a biased Editor and at least one biased reviewer?
General question: Do reviewers of a paper see each other’s comments?
This changes from field to field and sometimes journal to journal, but in this case, yes, I believe the reviewers were allowed to see each other’s reviews as well as our responses.
I wonder if there have been reviewer wars.
Re: Jeff Alberts (Dec 12 01:06),
Editors are in charge. That’s the way it is.
There’s a free market in journals. They rise and fall by their reputations. They need good papers, and do not succeed if they reject them. J Clim is highly reputable. They accepted OLMC10.
Eric’s moment of infamy was @ 5:36 PM on 12/2/10, comment #102.
He’s in the books, now.
hello there and thank you for your info – I have definitely
picked up something new from right here. I did however expertise some technical points using this website,
since I experienced to reload the website lots of times previous to I could get it to load properly.
I had been wondering if your web host is OK?
Not that I’m complaining, but slow loading instances times will often affect your placement in google and can damage your high quality score if advertising and marketing with Adwords. Anyway I’m adding this
RSS to my email and can look out for a lot more of
your respective exciting content. Ensure that you update this again very soon.
Wow, marvelous blog layout! How long have you been blogging for? you made blogging look easy. The overall look of your website is wonderful, let alone the content!. Thanks For Your article about Doing it Ourselves the Air Vent .
Wow, wonderful blog layout! How long have you been blogging for? you made blogging look easy. The overall look of your site is wonderful, let alone the content!. Thanks For Your article about Doing it Ourselves the Air Vent .
Hey there! I know this is somewhat off topic but I was wondering which blog platform are you using for
this site? I’m getting fed up of WordPress because I’ve had problems with hackers and I’m looking at alternatives for another
platform. I would be great if you could point me in the direction of
a good platform.