Siberian temperatures are an interesting case study in CRU gatekeeping. As reported a few days ago here, in an email of Mar 31, 2004, Jones advised Climategate correspondent Michael Mann that he had “gone to town” in his rejection reviews of submissions criticizing CRU’s handling of Siberian temperatures.
Today, in a Climate Audit exclusive, we provide you with the rejected paper (by Lars Kamél), one which seems like it would have been a useful contribution to the peerreviewedlitchurchur.
Jones’ Climategate statement was:
Recently rejected two papers (one for JGR and for GRL) from people saying CRU has it wrong over Siberia. Went to town in both reviews, hopefully successfully. If either appears I will be very surprised, but you never know with GRL.
In the subsequent comments, Lars Kamél reported that his 2004 submission to GRL on Siberian temperatures was almost certainly one of the two articles on Siberia where Jones’ adverse reviews had prevented publication.
Kamél sent me a copy of the 2004 submission which I’ve placed online here.
Kamél applied the homogenization technique of Vincent (1998) – used for Canadian station data – to a network of stations in southern Siberia around Lake Baikal (90−130 E; 40−75 N). Kamél reported that in the innermost portion of this region (100−120 E, 50−65 N):
the number of stations increased from 8 in 1901 to 23 in 1951 and then decreased to 12 from 1989 to present. Only four stations, those at Irkutsk, Bratsk, Chita and Kirensk, cover the entire 20th century.
Relative to CRU, he found that the trend in his results for the period 1901−2002 was 0.33−0.62 K/century less for calendar year data relative to CRU.
He speculated that the reason for the difference was that CRU contained “too little correction for urban warming”:
The reason for the differences, compared to the CRU calculation, is not known, but probably it is because the CRU compilation contains too little correction for urban warming. It is unlikely that the small modifications made to Vincent’s method could have created any non−climate cooling trend. There is at least one further reason to believe that the mean region had a very small warming in this period. There is one “rural” location (< 10,000 inhabitants), Kirensk, that have a record which covers the entire period. This record shows no significant temperature change at all.
From this spot check, Kamél recommended that the surface record be checked “in more regions and even globally”:
The result presented here does, however, suggest that the surface record should be checked in more regions and even globally.
Kamél said in his email that he no longer had the reviews, but still had his response to review comments here).
On Dec 15 2009, after noting the availability of new data from the UK Hadley Center in the wake of Climategate, IEA in Russia reported that CRU’s selection of Siberian stations “exaggerated” warming and recommended recalculation of CRU results:
IEA analysts say climatologists use the data of stations located in large populated centers that are influenced by the urban-warming effect more frequently than the correct data of remote stations… The scale of global warming was exaggerated due to temperature distortions for Russia accounting for 12.5% of the world’s land mass. The IEA said it was necessary to recalculate all global-temperature data in order to assess the scale of such exaggeration.
CRU’s handling of Siberian temperatures had been questioned a number of years ago by Warwick Hughes e.g. here. Hughes’ request for station data was infamously rebuffed by Jones as follows :
We have 25 or so years invested in the work. Why should I make the data available to you, when your aim is to try and find something wrong with it.
As I’ve noted on other occasions, it seems evident to me that temperatures have warmed since the 19th century. Personally I’m more interested in the comparison to the 11th century. However, as a matter of craftsmanship, it seems to me that one can reasonably inquire into the allocation of 20th century warming between the period leading up to the 1930s and the modern period.
CRU’s policies of obstructing critical articles in the peer-reviewed literature and withholding data from critics have unfortunately placed issues into play that might otherwise have been settled long ago.
In this case, as with interference torts, it’s hard to assess the precise damage of the interference. In the case of another paper (Aufhammer et al ), obstruction has delayed publication of the paper by six years but the authors are still endeavouring to get the paper into print. This was not the case with the Kamél paper; Kamél himself had abandoned the field.
Perhaps publication of Kamél’s paper would have inspired others to critically examine the CRU temperature data. And perhaps no problems would have been encountered.
However, as long as obstruction and withholding incidents mar the research record, it’s premature to claim, as Allen and von Storch did recently in a prominent trade journal that there are “no grounds” to question the validity of the CRU temperature history.
Climate Audit 
144 Comments
"Correct within an order of magnitude." Wrong.See my note on Menne et al 2009 “homogenization” in the Pielke on USHCN thread.
“Relative to CRU, he found that the trend in his results for the period 1901−2002 was 0.33−0.62 K/century for calendar year data relative to CRU.”
Maybe I’m just being thick but this sentence is hard to follow. Does it mean that the trend shown in his results was .33 to .62 K/Century less than the trend shown in CRU?
Steve: fixed
The temperature deviation trend line is described as 0.33 degress minus .62 degrees per century after the start date. At time zero, the trend line is at .33 degrees above norm. At one century, the value is at negative 0.29 (0.33 minus .62 times 1 century) degrees.
I found the sentence awkward as well.
Siberia, Fennoscandia, Australia + NZ, USA, …
Where does it stop?
Time to examine this “homogenization” method very, very closely.
It looks like they are playing with a loaded deck. They correct for stations moving out of the UHI in the 20th c. and fail to remove the trend caused by land-surface change. Hence the residual spatial patterns of waste heat noted by McKitrick & Michaels. (The ones whose pattern is NOT removed by adjusting for spatial autocorrelation, Dearest Gavin.)
“loaded deck”?
or possibly ‘marked dice’, perhaps … 😉
RE: **In this case, as with interference torts, it’s hard to assess the precise damage of the interference. In the case of another paper (Aufhammer et al ), obstruction has delayed publication of the paper by six years but the authors are still endeavouring to get the paper into print. This was not the case with the Kamel paper; Kamel himself had abandoned the field.**
Correct. It is hard to assess the damage. But it is likely significant. We have had many “independent” papers accepted which could all be summarized in a few papers. With the Team and associates in control of “peer-reviewed” papers, the damage will continue. The blogs have brought this to light. Now is anyone up there listening? Or will this continue?
The economic damage has been significant. snip
Steve: C’mon. Keep in mind that we’re talking about one study by Lars Kamel on CRU in Siberia. Stay focused.
“Trade journal!”
More great wit!
Kamel’s paper would have been stronger if he had presented some kind of data for urbanization at each of the Siberian stations to support his speculation. Are the reviewers comments still available and was this one of them?
Steve: he doesn’t have the reviews anymore. I’ve added a link to his response to reviews, which he still had and which he had sent to me.
It might be possible for him to request copies of the reviews from the editorial office of the journal.
It seems like “unnecessarily negative” reviewer-2 was quite demanding of detailed explanations and supporting data. If it indeed was Jones who was this reviewer, the evidence points to pretty transparent hypocrisy. Kamel is amenable to revising his paper and to providing his adjustment data online. Nothing impolite in his reply, either.
Unless funding really starts to flow into investigating the shortcomings of CRU temperature records, suppression of papers, etc., I don’t see any improvement in how climate science conducts its business coming anytime soon. Governments just don’t believe there’s any benefit from getting to the bottom of this rotten barrel. Indeed they seem more intent on whitewashing the whole mess.
I guess all we can do for now is too keep gathering evidence and hope it gets too big to ignore.
I find it strange that people like Kamel don’t make more noise about this. They just seem to give up and quietly go away. This only serves to embolden the bullying mentality that pervades in climate science.
I agree. I could say a lot but I’ll just summarise it this way: “All that is necessary for the forces of evil to win in the world is for enough good men to do nothing”.
It seems to me that there is a huge bestseller of a book in the story of the Climategate e-mails and the damage done to science by EAU/CRU & Mann et al. That would add value (and participation) to the discussion among both scientists and the educated public.
I wish that it would be too but I doubt it. You are over-estimating the detailed interest in this topic. A good indicator of interest might be the number of hits on You-tube for Steve’s performance last night. If it goes viral, then you are right. Somehow I doubt it.
It also seems important to get the regional/continental temperature records correct, as the calibration and validation of models depends on them…the argument that everything washes out in some global temperature anomaly and therefore that it’s not important to get any particular record right is ridiculous and certainly doesn’t breed confidence in the process….
Phil Jones wrote:
“We have 25 or so years invested in the work. Why should I make the data available to you, when your aim is to try and find something wrong with it.”
Do we know Hughes’ reply to this? E.g., “Because, Phil, I thought you were a scientist? (Because not only doing their research but then indeed sharing their data with others who devote themselves to finding something wrong with same is precisely what scientists do.)” ?
As someone who has only recently gotten interested in this stuff and is trying to stay as agnostic as possible about its substantive merits until I understand more I do have to say I found this email from Jones to be far and away the most troubling if not damning of him in all that recently disclosed material.
Of course the “trick” email is bad (even if I think that its real badness lies not in the “trick” part but the “hide” part instead.) Nevertheless, it still would be hard to totally indict someone and see them removed from their position on the basis of what can at least arguably be said to be just a loose, casual use of potentially ambiguous words in what was, after all, just a loose, casual email to a friend.
What Jones’ other “why should I share” email does for me at least is go very very much further. In fact for me it just way bolstered the sense I was already getting that a number of these “pro-AGW” climatologists felt for whatever reason that they could … “move beyond” (if not rejected outright) the old classical standards of science. (And that they could do so in a fairly wholesale fashion too.)
In any event if I were on the East Anglia committee reviewing this stuff it is this “why should I share” email that would lie at the absolute center of my concerns. One way or another it seems to me there are only two possible answers to it: Somehow either Dr. Jones just doesn’t understand the most fundamental nature of modern science or else he feels that for some reason he isn’t bound by same. Which is it, why, and what has the been consequence of either would be the focus of my most pressing questions at least.
East Anglia would seem to have a very grave decision to make as to what kind of institution it wants to be.
TomB, the “why should I share” statement made to Warwick Hughes has been known for years, and was not part of the Climategate emails.
Well thank you Al, I didn’t realize that even though of course that just makes sense. In a way even more appalling though that Jones would not be saying this in a private context but openly, where he could expect this view of science to be made public. Amazing.
But why was the Warwick Hughes email not included (what is being hidden?)
But, but where are all the FOI requests or communications of FOI requests by the university to Jones?
Have these been fitered out by the hackers, and if so, what are they hiding???
Many people think that this was a package of material selected by someone at CRU in response to an actual or possible FOI request. That seems reasonable.
For what it’s worth, given how damaging some of the material is and how keen Jones etc were to delete inconvenient files, I wonder if it was a “first draft” selection by a dogsbody, intended to be further whittled down by someone higher up.
If this was a folder selected by such a person on the inside then leaking, coming across accidentally or, less likely, hacking the folder was much easier – no effort in filtering required.
Of course, a possible answer to the puzzle of the email and code folder is that the CRU – or perhaps the UEA – was already carrying out its own internal investigation which, for obvious reasons, they didn’t want to become public. This could have been undertaken with or without Jones’s knowledge but instigated in response to the growing external unease about the controversial role the Unit was playing in the science (and especially concerning the mounting pile of refused FOI requests).
As often happens within the closed walls of institutions like universities, rumours and whispers could seep out about the existence of such an investigation and the existence of a mythical folder containing some pretty damning material. It would only take one enterprising ‘nobody’ on campus to tune into this gossip and set himself the challenge of hunting down evidence. Once found, he would realise that immortality (and eventual riches) was only a few simple mouse-clicks away via an anonymous Russian server.
There is another explanation. Someone (hacker or mole) identified emails that Jones had deleted in his mail box as too sensitive to reveal but which were available on a back-up server.
This is not a small area, 40 deg to 75 deg lat and 90 deg to 130 long is an enormous
area about the size of the western US and western Canada as far east as the great lakes.
I hope this is not the final version that was sent off….too many grammar and spelling
mistakes and the graphs look like they were done with an ’87 Commodore PC. This report needs
some work.
Lars,
Than you for the article. Your paper is straightforward enough, I’m very curious what the review comments were that prevented publication.
Hmm, looking at the graphs I would say that the reviewers have a point. If the graphs show what they say they show then the second one (showing the difference between Kamels’ approach and the CRU approach) suggests that more recent temperature data from CRU has less “error” in it than older data. However, Kamel suggests that inclusion of UHI affected sites has caused the error – if that was the case then the error would be bigger in the more recent data, as the sites get more impacted by UHI over time. Therefore the graphs may be correct but the conclusion that Kamel comes to appears illogical – sufficient reason to throw out the paper I suspect, even if the basic analysis is correct and reflects what we have seen elsewhere. Of course, if the conclusion is illogical then perhaps the data analysis wasn’t top notch either…
A more valid conclusion to come to on the basis of these graphs is that the more recent data can be relied upon but the older data is unreliable, perhaps leading to 0.75Kelvin error in the worst case. The CRU data tends to show the start of the last century as being cooler than it really was.
You might also notice that the CRU surface data is at its most reliable during the period when it happens it can be corroborated by satellite data (about 1977) but is progressively less reliable as we go back further in time. With data before 1960 the CRU temperature data is almost always showing the temperature to be cooler than Kamel’s analysis. Naturally this contributes to a significant uplift to the perceived warming trend from CRU. With Kamel’s analysis we see no warming trend until suddenly in the 1990s there is an uplift of about 1K. However, this assumes Kamel’s analysis was correct.
Ryan, I thought the satellite data is actually calibrated against the surface data to get their readings, which necessitates that they agree. Satellites don’t provide temperature readings like a thermometer, unfortunately, just microwave emissions which are used to infer the temperature after calibration. Someone else can correct me as my understanding is probably wrong.
“then the second one (showing the difference between Kamels’ approach and the CRU approach) suggests that more recent temperature data from CRU has less “error” in it than older data.”
This is a misreading. That there is a difference between the two is to be expected. The magnitude of the difference doesn’t indicate whether a time period is good or bad – the difference is an artifact of the analyses. That the difference drifts over time indicates there is a serious problem. Since it’s all from the same underlying data set, it’s an analysis issue. It’s tough to argue with blocking spatial and environmental effects in an analysis, so the CRU analysis is presumed to be flawed in some way. It isn’t up to this author to show how it is flawed – that’s CRU’s problem to find and fix.
Don’t get me wrong, the analysis was done in somewhat different ways and the results correlate in later years (suggesting that the analysis is not completely off the wall) but not in the early years. As you say, this suggests that at the very least different methods of analysing the data give very different results – which calls into question any specific approach you care to use.
However, the conlusion that this was due to urban heat island effects makes no sense – urban heat island effect would normally give more error in the later years as areas become more urbanised, not less. Therefore the reviewers would have had every right to challenge this paper on that basis. It needed to be re-written and re-submitted, but it seems the author never did so.
We’re probably using error in different senses? I think the trending by its self is enough to suggest further work is necessary, but not lead to the conclusion that the trending is due to heat islands. I don’t think this was a valid conlusion from this effort, more like a hypothesis to be tested in some follow on work.
By error, I’m assuming you mean how far away from 0 the data is. I read this value as the difference between biases in analysis techniques. That the CRU analysis starts half a degree colder than this analysis and trends upwards leads to the observed difference. This is completely consistent with heat island effects coupled with the initial difference in analysis bias.
There is no way for me to tell which technique is the least biased. I checked Vincent and there was some work to show limitations of the technique, but not a detailed comparison to other methods. Since the temperature change of interest is small relative to the variation in the data, everyone is trying to pick out an unknown signal from a relatively high signal to noise situation. The likelihood of having artifacts due to the analysis is huge.
Kamel says
Overall this is true – and it still looks to me as if it could be UHI. Moreover, the point of change in pattern fits the 1989 collapse of the Soviet Union, which we know means discontinuity in the records. Plus, there are stories that some Russians report lower temperatures in order to get district heating allowances. Complicated!
Still, the presence of the anomaly should have triggered investigation, not suppression.
Sorry to disagree! Kamel’s simple approach includes no correction for UHI at all, so if CRU does what it claims to do, namely properly compensating for UHI, one would expect the two to diverge rather than converge in recent years, with CRU being progessively colder than the uncorrected “simple” version. On the contrary, CRU somehow seems to exaggerate recent UHI rather than correct for it (in this dataset at least), or at least corrects less and less approaching the present.
“it’s premature to claim, as Allen and von Storch did recently in a prominent trade journal that there are “no grounds” to question the validity of the CRU temperature history”.
Yet looking at the linked article (Nature News) nobody has commented or raised any objections. Shouldn’t someone with a degree of credibility at least post some kind of contrary opinion?
I was going to post a comment (not that I possess credibility), but then I noticed the caption: “Reader comments are usually moderated after posting.” The latter immediately made me somewhere leery; perhaps they deleted or modified comments which challenged the opinions of the authors. Then I decided not to send a comment after reflecting that this was Nature, where an unsavory unidentified editor recently wrote unprofessional remarks about “deniers”
I tried to comment but couldn’t as you have to be a paid-up subscriber it seems. Unless I’m just technically incompetent.
“Undoubtedly, it is a well tested method, since it has successfully been used to construct the Canadian Historical Temperature Database (Vincent and Gullet 1999). This techniques seems, however, to have been used very little in other regions of the world.”
I doubt that this amount of testing makes it an “undoubtely” well tested method.
To a non-expert yet technical person like me, the article seems to draw weak conclusions at each step of its reasoning.
Here is the orignial article on inhomogeneity detection by Lucie Vincent:
Click to access TechniqueIdentificationInhomoCanTemp.pdf
Perhaps Norbert can audit for us?
Observe how she uses inhomogeneity analysis to cool down the 1940s warming at Collegeville, Nova Scotia, Canada. The same pulse of warming that worried the climategatekeepers.
Since when does an article by the author of the method (Vincent) provide something that could be called an “undoubtedly” well-tested method invented by the same person (Vincent) ? Vincent testing Vincent?
No need to audit, this appears to be a new method, perhaps a good one, but not one which in any sense that would deserve the expression “undoubtedly well-tested”.
I never said it was “undoubtedly well-tested”. Kamel said it, and he is Swedish author for whom English may not be his first language. His point is that he applied a method that at least has some legitimacy/credibility. Did he overstate teh degree of legitimacy? Maybe. But so what? This paper does not live or die on this assertion!
.
Your point has been made already and it is a red herring. Read the paper and read Vicnent. Then tell us if Kamel is not correct.
Just to be clear: The article by Lucie Vincent appears much more professional. Compare it to the one by Kamel.
You ask “so what”? If anything is a red herring, then your explanation that the author doesn’t speak english well enough to know what he is saying.
Norbert, if you read to the end of the Kamel paper, you will see his acknowledgements.
“Acknowledgements
No funding was received from any institute, company or other sources to do this research. Dr. G. Henriksson is thanked for interesting discussions, for assistance in software developement and for creating a good atmosphere for doing research.”
Kamel is the lone author. It is entirely possible this was a graduate level paper/dissertation by him. The acknowledgement of a single professor is something I have seen in many first dissertation presentations.
As bender mentions, Kamel is Swedish, and other papers I have found by him are written in Swedish, not English. As this was a paper for an English journal, I’ll give him some latitude with regard to grammar in writing in what is probably a second language at best. As for professional quality, this was the first submitted draft of the paper, which is expected to be rougher than the final product. Vincent’s paper is the finished product, which means it had an editor examine difficult sections and discuss with the author possible corrections to make the presentation better.
Norbert, I don’t see your criticisms as impacting the validity of the work. Basically, you think the authors aren’t very good writers in a second language. This is nothing compared to plenty of papers written and published, for instance, by Japanese researchers. The reviewers normally help the foreign authors clean up the paper – it’s more important to get the work published than to harp on the authors language usage.
To me the paper seems unnecessarily brief. There are hints that a considerable amount of work went into this, yet there is no evidence to support the checking of individual series, amount of in-filling, and sensitivity checks. It would seem difficult to reproduce this work from the detail given, and I suspect even the author would struggle today – so on that basis rejection would be valid (but supporting scripts and data would be sufficient).
Looks like an interesting discrepancy to be explained though. I think it is acceptable for this paper to fail to come up with a detailed justification, or proof of the UHI link (and I am not sure if the recent agreement disproves that – it could just be a baseline effect)
Isn’t it interesting that your concerns, valid as they may be, apply equally to many other published papers.
Differential pass-through.
Sure, I wasn’t suggesting that this paper shouldn’t have been published eventually (or even that it’s worse than the average paper that does make it through review). The point I was trying to make is that the rejection isn’t particularly good evidence of bias, even if it does show how that bias took place. I’m not sure how many of the readers here would make the distinction.
I agree. I would evaluate the paper on its merits and forget about this hunt for bias. Biased reviewing is a fact of life in science. I know a lot of people can’t seem to accept that. You’re much better off taking what’s good in these papers and figuring out how to make it better.
I agree as well! I really was making a note of interest 🙂
Keep in mind that this was the initial submission. At this stage, there is the option to request changes, including added detail. The only valid reasons for an outright rejection at this point would be either lack of relevance – improbable in this case – or some fundamental flaw in the methodology. It would be very interesting to see whether the actual reviews contain a convincing criticism of methodology.
One should ask – where’s the code, where’s the data? Please
Lars says in response to his reviews:
I can easily supply all the data and logs through a ftp – site, though.
Since this particular paper was hamstrung, I would have no idea where that data and code would be now though.
Kamel should have submitted to a more local journal. GRL was a poor choice for such a mundane* analysis. Can anyone name a journal that would have been more appropriate? (Perhaps a technical report of the Russian IEA?)
.
*Not to be perjorative, but this is pretty mundane stuff. Worthy of reporting on, but not in GRL, where a premium is placed on novelty.
Nature – like Steig et al.
Is there a Russian/Siberian Journal of Climatology? Perhaps Anastassia Marakeva can comment?
The article didn’t pass GRL reviews. Would publishing it (as is) in a Russian IEA report make it any more valid? No.
Frankly, as a publishing academic and reviewer I would be shocked to find my anonymous reviews of any given paper in the public domain; likewise, I wouldn’t be waving around reviews of my own rejected papers. If it wasn’t good enough to publish it should be reworked and submitted elsewhere. That’s how science works.
Joe, there is a presumption in your comment that the analysis in this paper is not valid. After a first reading this is a straightforward analysis that reveals significant differences between the reported trends and those reported by CRU using different homogenization schemes.
I agree with Bender that it probably is not suitable for GRL being a rather straightforward analysis of data.
As for reviewing I always ask that my reviews of papaers are not anonymous to the author.
I think it is fairly clear from the article and the comments on the thread where this was introduced that the author was probably fairly inexperienced in submitting papers. Does that make his work any less valid? No. Does your comment make your own work any less valid? Maybe. I think we are agreed that the paper needed some work, but the impression received by the author seems to be that the reasons for rejection were sufficiently ideological for any re-work to be a waste of time.
The ‘That’s how science works’ meme is a little tired. Science doesn’t work – people do. People also lie.
I don’t think the article needs more work. I think the author needs more experience.
Same thing. That author needed an experienced co-author. Stop splitting hairs.
p. 8 reads:
“The reason for the differences, compared to the CRU calculation, is not known, but probably it is because the CRU compilation contains too little correction for urban
warming.”
.
A very weak line. This should have been taken by Jones as an invitation to sit down together and try to figure out the source of the discrepancy. Instead, Phil “went to town”. In many other fields, reviewer Jones would have gently rejected the ms, but joined forces with the author and become an eventual co-author.
And together they would have disproved Einstein’s relativity.
You’re full of humor. Short on analysis.
bender, I think you’ve got to the heart of the matter here, plus what would/should be good/normal practice, very succinctly.
Thanks, Lucy. I hope my voluminous contributions in the last 48h have at least a little value.
“The article didn’t pass GRL reviews. Would publishing it (as is) in a Russian IEA report make it any more valid? No.”
.
Maybe it was correct, but too mundane. Maybe it just lacked novelty. You haven’t seen the reviews and you haven’t pointed to any errors in the paper. If you can, please do so.
To further explain, I’m not making any assertions as to the correctness of either the Kamel or CRU analyses. What this paper does highlight is the fact that different homogenization schemes (and we don’t know the details of the CRU value added product) produce different results.
For me this poses serious questions about the effect of the different schemes that needs addressing.
Compare your argument to the Teams argument that they needn’t provide details for how they came up with their numbers, and from exactly what datasets, because attempts to replicate their work should be done independently. Your argument shows a bit of how shallow their argument is.
Tim Berners-Lee invented the world wide web. He submitted his first paper on it to the Hypertext conference. It was rejected. The next Hypertest conferecne was concerned with the WWW.
I have seen similar rejections published on the web, Papers in engineering were rejected nd later found to be the basis of major work. Anotehr such rejection was posted prominently on the author’s office door.
I am unconvinced that “novelty” is a particularly relevant criterion in most climate studies. In a lot of cases, we’re talking about collecting and analysing more data – nothing particular novel.
Lots of times, it seems that authors are trying to shoehorn mundane (but useful) practical results into “novelty” criteria – serving neither especially well.
Personally I’d prefer a lot more technical reports and fewer pointless attempts at statistical “novelty”.
Agreed, on all 3 points. But don’t underestimate the novelty criterion and the lip service paid to it.
“Novel” methods enable them to recycle the same data and results many more times, creating an impression of a “mass” of independent analysis. Just windowdressing.
It seems “novelty” can be used to describe either methods or conclusions. Even if the methods are not novel, if the conclusion goes against the mainstream view of global warming, I would think such a novel result would be welcomed by GRL.
Putting a premium on novelty is a way of opening up the scientific debate. It is exactly the opposite of the kind of climategatekeeping CRU was guilty of.
“it seems evident to me that temperatures have warmed since the 19th century”: I’m beginning to wonder whether the effect might happen to be concentrated in the Northern Hemisphere, specifically Western Europe and North America. That might explain the doubtful treatment of data from Australia, NZ, Siberia and so on. Perhaps they’ve set out to exaggerate the effect in Western Europe and North America, and invent it elsewhere? Because if it isn’t really global, their whole case collapses.
“Perhaps they’ve set out to exaggerate the effect”
.
After reading Harry.txt, do you really think that they could “set out to do something” and accomplish it? Just kidding, of course.
.
But isn;t it just as likely that they stumbled onto a pattern and the confirmation bias took over? What is their proof that “homogenization” homogenizes and “UHI-correction” UHI-corrects? In the absence of such proof, it seems possible that “homogenization” urban-warms otherwise cool records and “UHI correction” does little to reverse the artifical warming.
dearieme, you should read or reread Steve’s posts here on “Where’s Waldo?” Steve looked at different continents to see if they had exceptional warming. If I remember correctly, North America, Africa and Antarctica did not show exceptional warming. Warming was exceptional in Russia (Siberia esp. IIRC). We now know the Russians think their data was fudged. China was another controversial area. Douglas Keenan brought an allegation of fabricated claims/data against Wei-Chyung Wang (a co-author with Phil Jones) regarding temp data from China. See http://www.informath.org/
A potential discrepancy in Siberia of the size reported here is a very important issue regardless of whether it’s as “good” an article as Vincent 1998.
The problem with the review process here is that there is evidence that reviews are tainted and that Jones’ primary interest was not to improve the Kamel article, but to keep his Siberia findings out of the literature.
Again, I hope it’s clear I agree.
How on earth did this manuscript fall into Phil Jones’ hands? Is it not clear that he would have had difficulty providing an objective review? He says he “went to town” on it? What a surprise. Which Editor at GRL passed this on to Jones? I can not agree with that decision.
That Phil Jones was selected as a reviewer in the first place is perhaps more telling than anything else here.
IIRC it is quite common for the author of the paper being criticized to be one of the reviewers. But my understanding is that the editor is expected to take the fact that he is being criticized into consideration while reading his comments
To be blunt, IMHO, Jones should have recused himself from the reviewing process. How could he maintain and unbiased position? Even if he was a saint, he should have recognized the compromising circumstances.
Agreed. That was my point.
I like having the author under criticism being one of the reviewers – it avoids some of the back – and – forth after publication, and can improve the quality of the paper substantially. They just shouldn’t be considered when deciding to publish or not publish.
On the industrial side of things, having an author under criticism review submissions may not be practical due to early disclosure of information which may have competitive significance, but that issue arises in any of these reviews.
Do you like good papers being suppressed just because they question a consensus?
To be responsive, I think papers should not be supressed. As an aside, the idea that peer review settles the science doesn’t agree with me, either.
Bender, I believe consensus science doesn’t exists. Science is science, and a consensus is, by definition, a political outcome.
Seems to me that until a legitimate scientific body(one that is truly agnostic, but how do you find these) gets into verifying all the temperature data used to generate the theories and proof of theories, nothing else matters. As a non-scientist, I am at the point of believing nothing I read, even the GW deniers, until I see analysis based on confirmed and proven data. If the data cannot be verified as accurate, why use it? Why take guesses at the inaccuracies of certain data only to use that data in a subsequent study and then another subsequent study. No one can trace the original data and verify anything. I call that BS. So, Rule 1, use no data that is unsubstantiated as valid by multiple sources. Rule 2, use no data that is unsubstantiated as valid by multiple sources. Thanks to those of you dedicated to verifiable truth.
Well, you have a point, sort of, but what exactly do you mean by “verifying”, “proving” or “confirming” data? The closest thing that could be done in hindsight would simply be to make all original station records available, along with any documentation pertaining to the history of each record. Which ones of these then make the cut for inclusion in a regional or global reconstruction will always remain a matter of judgment.
With his medical background, Michael Crichton borrowed from that field to posit that maybe the process should be broken down to multiple blind groups, one to gather the data, one to analyze the data and another to synthesize the findings. Very different that climate science and many other branches, but when so much is on the line, it’s at least an interesting concept. It certainly would break up the centers of power.
I think transparency alone would suffice. Also, the medical analogy strikes me as false – it sounds more like division of powers.
IMHO, transparency is overrated. If papers are being used as a basis for action, transparency only gives some kind of guarantee if the time required for evaluation and criticism / resolving any issues is significantly shorter than the time during which decisions would be made. This is not the case here, partly because climate science is an immature discipline, the decision/action seems to be running ahead of the science.
Having read Lars Kamel’s response to Miyahara (handling editor at GRL?) it is not immediately clear to me that the paper was rejected. It looks possible that GRL might have invited a resubmission having addressed the reviewers comments. Under these terms it is usual for the journal to send the article out for re-review to the same reviewers. Lars Kamel will be able to confirm whether this is true or not. It does seem from his response that the technical issues were not insurmountable.
I think this is a potentially important paper because it apparently highlights the fact that different homogenization schemes result in widely different results. This is of critical importance.
The best response to Lambert, imho… The problem is not the discrepancy post 1960, but the scientific methodology. Very interesting view point in the homogenization/correction of data by CRU.
We have 25 or so years invested in the work. Why should I make the data available to you, when your aim is to try and find something wrong with it.
And when the “skeptics” provide a “negative” study, because they release all data and programm? In this case but also whit Scafetta. Why McIntyre reanalyze the IEA’s study while not reanalizy the CRU’s data, for example?
When dealing with authors whose writing skills are limited in English; i.e., when English is a second or third language; one way to improve the author’s dialogue is for the author to write the article in his/her native language and then to send the article to a professional science translator. When the latter product reaches me, I or one of my colleagues possess enough expertise to edit the translation for appropriate text errors. This method has been a great asset in my realm, especially for papers originating from Asia.
Great post Steve.
I like the way you call out the CRU and attribute their refusal to submit data and methods to “CRU’s policies”.
Standard operating practices, no?
Thanks again for all your work!
A note if I may. I think the issue of the sad state of peer review lies at the climate scientific journals’ feet.
If a journal wishes to be an advocate, one way or the other, then the serious climate scientists have a choice to make.
From my limited perspective, most of the popular journals are beyond repair.
I think a new climate science journal with demanding archive and disclosure policies, unbiased reviewers and editors could be launched.
Wouldn’t the unbiased and serious scientist be attracted to such a publication?
Steve wrote,”Kamél himself had abandoned the field.
This phrase may convey a personal tragedy. Does anyone know whether Kamél abandoned the field in despair over the unfair treatment of his manuscript? Was that his only experience, or were there others? Did he decide that Climatology was corrupt, where honest work and honest workers were done violence?
Steve, did Kamél mention anything of this kind to you — that you can share?
If so, I wonder how many other good scientists were driven from the field by the integrity-violating actions of the partisan few-in-high-places.
Lars has been around a while. He used to post sometimes on the other climate board I used to be on. He was always quite knowledgeable and a AGW skeptic. I expect he’ll come here and answer some questions.
Perhaps this OT but it appears both the US and most of Europe are currently suffering some of the coldest if not the coldest weather in history. Also, the Arctic area appears to be at or near its largest extent. Countries like Australia are currently experiencing mild to normal summers. So I expect by the end of this month the land surface temperatures reported by NASA should indicate a significant drop in the running average. If they don’t I will be convinced their methods of correcting the data are false.
My bad – I meant to say Antarctic, not Arctic.
The amount of leverage that the Team had with the journals is something I’d like to see more deeply explored by the scientific community. The control apparently exerted over some of the pubs in the emails is appalling.
So the question is: Could honest pubs with backbone have functioned as a check on many of the scientific abuses of the Team?
Steve and All,
As someone from the liberal arts it’s a struggle to come to terms with this issue of Climatology and Climate Modeling. I want to say how much I appreciate the work that M & M have done and all those who try to honestly seek the truth. This may be a bit off topic, but the thing that disturbs me the most in all that I have read is the small number of surface stations in large areas of the planet, especially those with records that go back for any time at all, the lack of standardization of the ones that do exist and the massive issue of URH effect. With so few records for a cold and massive place like Siberia, can the missing data really be extrapolated for such a large area and have even a reasonable degree of accuracy? How can anyone really say that the world’s climate is heating, cooling or remaining the same when the surface station network appears to be woefully inadequate? Shouldn’t everyone involved take a humility pledge and say that “based on the very limited surface station data we have, the world appears to have warmed slightly in the past hundred and fifty years, but a precise method isn’t possible.” Or, when discussing proxies, admit that everything doesn’t match up precisely, so say that “tree rings and other proxies also seen to indicate some warming, but the calibration of these proxies to thermometer data isn’t an exact science?” In history we are taught to say, “this is what we know and this is what we have surmised based on incomplete evidence.” I just can’t understand anyone using the data I have seen to deliver pronouncements with the degree of certainty in climate science which may more properly be termed “climate studies” because of the state of the available data in a layman’s opinion.
Hi Jeff:
As someone also from the liberal arts side of things somewhat new to getting into this too your point is exactly something that rang a big bell for me as well, and has had me wanting to ask question concerning same for the techie-types here:
As Jeff says it’s obvious that the number of these surface stations in some huge whole regions at least (not to mention the even smaller number of oceanic stations) is incredibly small in some instances. And indeed I believe I’ve seen some talk about the number of some of the tree-ring samples that just strikes me as incredibly small as well.
So okay, I understand the idea—Jeff’s really—of merely saying “okay, this is what we got even if it’s not all that much but it’s at least worth talking about….” But that’s not what’s been done is it?
So … in statistics isn’t there some *minimal* level of sample required before you can say *anything* at all really? Or, I can’t imagine with the small number of samples that I’ve read about here and there that went into these reports that the IPCC uses that statistics doesn’t provide some simply incredibly huge “margin of error” figure also to go along with same, no?
I.e., for any statistically derived figure isn’t there also some “margin of error” figured for same?
From some of what I’ve read—if I understand it right—I’d have to believe that the samples are so few you couldn’t even begin to calculate the margin of error much less any data of any reliability at all, and if statistics says you can, then my God that margin of error figure would be so huge that this and this alone would cause people to just laugh at the underlying suggested statistical “finding,” no?
Jeffrey,
You will find a lot of analysis on that subject over at William Briggs site. He just finished a 5 part article on why everyone thinks they are too sure about homogenization results. He appears to be a very good statistician and writes analysis of AGW and other topics.
Thank you so much for the Briggs URL. This is just the kind of thing I’ve been looking for – I have a science degree, but not much statistical awareness. Hopefully, this will help bridge the gap.
Jeffrey,
For someone from the liberal arts who is struggling with climate issues, you just made several excellent big picture points about the topic in this post, and in AGW science in general. I haven’t checked out the link provided by James yet but I suggest reading for a thorough description of what is done vs. what should be done with temperature data. The paper discussed in this post takes what I would call a very gentle approach to countering temperature data presented by the CRU vs. what AJ Strata has to say about temperature readings and adjustments in general.
I’ve read quite a bit here and at Anthony Watts’ site even before Climategate broke, and I’ve always had a problem hearing about temperature estimations, extrapolations, interpolations, etc. (I doubt that most people who believe in AGW have any idea what is being done “behind the scenes” to produce the various graphs shown to support the theory.) I’m with Strata that the way to do things is with real measurements. My thought is performing correlation exercises with existing stations. Correlations would help considerably in determining whether past temperature adjustments were valid.
Many of the stations in Siberia discussed by Kamel will probably still be in service, and there should be records for most of the previous station locations. Each of these locations (existing and past) should have additional temperature sensors installed in their vicinity to verify the past readings. As you point out Siberia is a vast area, so additional sites to provide better coverage would also be extremely helpful, and could also serve as an iron clad check of whether various homogenization routines (such as those used by the CRU discussed in the paper here) are valid. This won’t be as good as having unadjusted and unaffected (e.g., by UHI effects) long term histories at thousands of location across the globe, but it will help.
Oops on the hyperlink in my post above. A fix would be appreciated.
I would judge that the paper in question in this thread is rather obviously not a complete work.
Like other references I have seen of late to differences in the Russian region between CRU temperature data and other derived data sets, my first impression is that with these large geographical areas and limited number of stations and historical time coverage is CRU making any claims for the data being able to measure trends with any reasonable degree of certainty?
Below is what I extracted for the CRU land and sea temperature series from the KNMI climate data repository linked below:
http://climexp.knmi.nl/selectfield_obs.cgi?someone@somewhere
The region 40-75N and 90-130E: 175 stations which are in China, Korea and Mongolia with only 54 in Russia. The trend slope from 1909-2008 is 1.74 degrees C per century -/+ 0.47; p = 0.000.
The region 45-75N and 95-125E: 78 stations in China and Mongolia with 42 in Russia. The trend slope 1909-2008 is 1.57 degrees C per century -/+0.55; p = 0.000.
The region 50-65N and 100-120E has 24 stations in Russia. The trend slope for 1909-2008 is 1.73 degrees C per century -/+ 0.60; p = 0.000.
I used the default 30% valid data points for the KNMI extraction. Even with that the first second and third series above had 1 year, 6 years and 1 year, respectively of missing data for the 1909-2008 time period.
The 24 station coverage for the last series, 50-65N and 100-125E, was as follows:
1928-1991, 1933-1991, 1913-1991, 1932-2009, 1928-1900, 1892-1991, 1936-1989, 1901-1991, 1938-2009, 1951-2009, 1935-1991, 1937-2009, 1938-1990, 1938-1990, 1951-19901, 1898-2009, 1910-1991, 1857-1991, 1936-1989, 1886-1990, 1890-1991, 1843-2009, 1895-2009, 1901-2009
“as Allen and von Storch did recently in a prominent trade journal that there are “no grounds” to question the validity of the CRU temperature history”
Jones supporters should welcome a chance for vindication on the more serious implications.
As it is it continues to have the drama and flavor of a slow speed Bronco chase complete with folks hanging from the overpasses urging a run for the border.
All these homogenisation schemes are far more complicated than they need to be. We already know the temperature today, so all we need to know is the trend. If a site is believed to have reliable data for a given period during which no change appeared to have happened then the trend in the temperature during that specific period is reliable. It could, of course, be affected by UHI – but this is something that can be checked by making observations today within the UHI and just outside. Put all this data on the trends together and you will quickly build up a picture of how we got to todays climate from the last 100 years.
Such an approach minimises any tampering of the raw data and only relies on the detection of discontinuities in the data to suggest that something dubious may have happened to the site at that point in time. The treatment of the raw data will then be completely transparent.
On the flip side, it appears Mr. Phil Jones did not feel the need to “go to town” with all of his reviews:
“Also just sent back comments to Mike Mann on the paper by Tom and
you factoring out ENSO and Volcanoes. Felt like writing red ink all over
it, but sent back a short publish suject to minor revision to Mike. This
is the first time I’ve ever reviewed one of Tom’s or your papers !
Copy of what I sent is attached. I forgot to sign it before sending it !”
http://www.eastangliaemails.com/emails.php?eid=202&filename=.txt
No we don’t, because there is no “the temperature”.
snip – OT
Re Gary’s post” “Kamel’s paper would have been stronger if he had presented some kind of data for urbanization at each of the Siberian stations to support his speculation.
Have to agree. Given the many geographic, political and climatic challenges, an audit by the comrades/sympathizers of one Antonin Wattsupwikov would be an epic undertaking.
I find this single comment of most interest in your entire post.
“This was not the case with the Kamél paper; Kamél himself had abandoned the field.”
In the area of finance we struggle constantly with data problems caused by “survivorship bias”.
Intuitively one would have expected such bias to have been occuring in the climate science literatue. This is an interesting confirmation of the fact.
Let’s be honest – as several people have commented above (Gary Edef Ryan Sean Kenneth) this is not a very good paper. Only one region of Siberia is considered, where the big cities are (cherrypicking). The methods are described rather vaguely without full details of the stations used. What exactly was the primary data source? And the writing isn’t very good.
Of course, many of these criticisms also apply to published papers by “The Team”!
I quite agree. This paper was rightly rejected by the review process. It should have been re-written and re-submitted. Thing is that “the team” submitted papers with exactly the same flaws – erroneous data, missing data, reachy or irrational conclusions, misleading or confusing graphics – but then published because they were reviewing each other’s work and leaning on the publishers. Hardly surprising that these flawed papers were then exposed by the likes of Mr McIntyre.
It would take a lot of work to expose some of the other flaws in these papers. Apart from MrMcIntyre, who else has the time? There is confirmational bias here too – those that are in the team are getting paid to produce this tripe and are driven by a supposed need to save the planet – but nobody else has the time or the financing to counter these claims nor is anyone sufficiently driven to, um, not save the planet, cos it doesn’t actually need saving. Difficult to get excited about isn’t it? “Save the Planet!” is just so much more pithy and motivating than “Don’t bother saving the planet because its doing just fine without our meddling”. See the psychology/sociology here? You don’t need a conspiracy theory, just like-minded people being filtered out over time and finding they have a lot in common and then confirming each other’s point of view and believing they have a “consensus”.
Eric, you are preaching to the converted here. Why not post a comment on the Nature page or on von Storch’s Klimazwiebel blog (one of the Anon comments is from me).
Ryan,
You say “However, the conclusion that this was due to urban heat island effects makes no sense – urban heat island effect would normally give more error in the later years as areas become more urbanised, not less.”
There is another explanation that I have proposed elsewhere after studying possible UHI (and confirmed UHI) in some cities in Australia. The explanation is that towns with the temperature sensor located – let’s say centrally and badly – can “max out” on UHI early in their histories, so that one should not expect recent further UHI. This regards UHI as being fairly specific for the location immediately around the weather station, so that UHI in far distant suburbs does not add much temperature to what is already there, though it might cause the effect to last longer each day. I suspect Melbourne to have more or less maxed out in the 1930-40 period. In addition, the comparison stations nearby tend to get urbanised later and so provide a lesser negative correction with time.
The USA and Australia, with vastly different population densities, are not the same kettle of fish. Nor is Russia. We and Russia seem to have in common that various adjusters have discounted many, many stations in a manner that sometimes appears to be capricious rather than systematic.
Yes, I have seen that suggested before and it makes a lot of sense – it also complicates matters radically. Makes any UHI affected site more or less unusable.
I also wonder if Siberia towns show one hell of a lot of UHI. -50Celsius outside, but buildings held at 20Celsius might suggest Stevenson screens that are severely impacted. Also, would you want to put a Siberian Stevenson screen a long way from civilisation? I doubt it!
You don’t need a massive UHI to get a massive waste-heat warming effect. Also, don’t discount the homogenization process as a source of warming. Think about it. A station moves out of a waste heat zone in the 1940s or 1970s, into a more rural setting. The records show a jump step to cooler. The homogenization process turns around and removes this jump step down thus warming up the recent part of the record. Then, when it comes time to removing UHI you should remove exactly what you added in IF you are correctly measuring the land-use/waste heat effects. If not, you are warming the more recent parts of the series. But how can you add back in what you take away when the homogenization process is a *step* function, and the UHI estimation is a *trend* function? See the problem?
.
Numerical toy example. Suppose homogenization warms by a step increment of one degree, but UHI trend estimation takes off only a tenth over that same time frame. The record is now 0.9C warmer than it should be.
.
What % stations exhibit that kind of move, from inside UHI to outside? If you don’t correctly estimate UHI you will double-count it! Once for space, once for time.
“However, the conclusion that this was due to urban heat island effects makes no sense – urban heat island effect would normally give more error in the later years as areas become more urbanised, not less.”
Ryan, I do not see how one can determine when UHI would effect the temperature trend. It is the change in UHI that will effect the trend and most of that change could have occurred many years ago. Quite frankly I see UHI corrections, as carried out by the major temperature set owners, as a rather crude application that does not take into account a direct observation of any micro climate effects.
It is simply too easy to sit in a academic tower (or at my computer for that matter) and play with data and particularly so when we consider the findings of the Watts team in their CRN on site quality control evaluations and posted results.
Having said that I plan to post more on the regions noted in the paper here with regards to station locations and looking for change points in the data.
For what it’s worth, there has been a rather dramatic shift in the numbers of people in Melbourne who are now living in high rises rather than single homes. I lived for a year on the 20th floor of one such building, the footprint of which is smaller than the suburban lot of my current home. The difference in waste heat would be massive.
BTW, the main current problem with temperature reconstruction here seems to be the unwillingness of the holding authorities to come forward with definitive, clarifying statements. Willis’ many-comments posts on Darwin are surrounded by supposition and guesswork when a release by the Bureau and GHCN would place matters on a more firm basis.
The data, after all, belong to the people who paid for their collection. The guardians in an ideal world would be frank and helpful.
“Went to town in both reviews, hopefully successfully.”
Isn’t the glaring question, what the definition of “success” is to these people who are supposed to be scientists?
“” I will keep them out somehow – even if we have to redefine what the peer-review literature is !””
“”Tim, Chris,
I hope you’re not right about the lack of warming lasting till about 2020. I’d rather hoped to see the earlier Met Office press release with Doug’s paper that said something like -half the years to 2014 would exceed the warmest year currently on … See More record, 1998!
Still a way to go before 2014.
I seem to be getting an email a week from skeptics saying where’s the warming gone. I know the warming is on the decadal scale, but it would be nice to wear their smug grins away.””
all quotes from Phil Jones.
Also this email, do these “Team” players go to town” in the opposite way as well? Here is the link (and Jones says he is “just a referee” here) where Jones is helping an author with a paper :
From: Phil Jones
To: Benjamin Santer
Subject: JGR paper
Date: Fri Aug 18 17:19:46 2000
http://www.eastangliaemails.com/emails.php?eid=179&filename=966633586.txt
“Went to town in both reviews, hopefully successfully.”
.
An objective reviewer would “go to town” but “let the chips fall where they may” (in terms of editorial decision). They wouldn’t “hope” for a particular editoiral decision … unless they were engaged in a conflict of interest.
Actually, looking at the first Kamél graph there is an even more obvious “elephant in the room”. As I noticed in passing in previous comments, there is a clear discontinuity that occurs at about 1988 of about 1K. It doesn’t matter if you look at the Kamél data or the CRU data. Now that wasn’t caused by either gradual increases in CO2 or urban heat island effect. My guess would be that it was caused by the Russians swapping out manual mercury thermometers for electronic thermometers in all the sites in the region in one year, but the actual reason isn’t that important (except to prove a point).
Fact is that if you want to homogenise the data then that discontinuity should be removed – it clearly has nothing to do with CO2. Funny kind of “homogenising” that leaves obvious discontinuities in the mean data. Thing is if you take that discontinuity out and shift all the remaining data after 1988 down by 1K to homogenise the means before and after 1988…. where’s the global warming????
Looks like yet another smoking gun….
The way to prove it would be as follows:
take the raw CRU data
use an 8th order Butterworth or similar low-pass filter on a 3 year interval to remove most of the high frequency “noise”.
check for the discontinuity in 1988
find the difference in the means
subtract the difference in the means from all the data after 1988 to give an adjusted series
check for any temperature trend in the adjusted series.
Wish I had the time.
I’ve plotted the GHCN data from the region in this paper at tAV.
This may seem simple but why homogenize the data at all?
AJ Strata says the following on his blog link http://strata-sphere.com/blog/index.php/archives/11824
“They make adjustments or homogenizing stations or fill in grids with pretend stations. A total unscientific joke. The measurement is the measurement. It has a fixed accuracy and uncertainty. Each station has a unique accuracy and uncertainty due to its siting, technology and the accuracy with which its readings are made each day. (geeked out again: if there is a drift in time when you read the sensor, or that sensor’s reference to UTC (the world wide time reference) is unknown or dynamic, then you increase the error bars to the measurement). If the sensor is sited wrong or has problems you extend the error bars. You DON”T adjust the data!”
Is it possible to construct global temperature data sets without “homogenizing” the data?
I can think of several reasons to do so 1) You could provide graphs of all the raw temperature data to test if there has been warming, 2) you could provide graphs of Rural stations only to test if there has been warming, 3) you could provide graphs of City stations to test if there had been warming, 4) you could compare 1,2,3 to graphs of each other and to temps taken just at airports. Perhaps the Global Temp datasets could be offered in these subsets in their raw form and allow people who were interested in using them to apply their own homogenizations that could then be examined by those that analyze and review their studies of their temp data.
It just seems to me to make more sense to provide data available to science in a raw form and leave the “tweaking” to those that will use it and be responsible for explaning and justifying how it was tweaked for the purpose it was used for.
Is there any concensus for doing what I propose?
Thanks
Edward
Bender, the change point homogenization that I am familiar with uses differences in the temperature time series for adjacent stations. The assumption I guess is that a non climatic effect on one station will not correspond coincidently with other stations in the general locale. Also assumed is that the climate is not very local – and of that assumption I have not seen good evidence – or in other words can a non climate and climate temperature change be confused. The global temperature time series has at least 3 rather well defined change points.
The Version 1 to Version 2 adjustments for the USHCN temperature series were made using change point analysis for individual stations for the Version 2. In my analyses I found some very large trend changes between Version 1 and Version 2. Also I found the Version 2 stations with the largest changes from Version 1 could continue to have a number of change points in the series and not those necessarily corresponding with those for the entire US time series.
I have read the foundations for change point analysis and the USHCN applications but I think a rigorous analysis of this system is yet to be done. Interesting though that USHCN used change point because they evidently had major doubts about the meta data that was used previously to make homogeneity adjustments.
Problem is that the station data is only taken to the nearest tenth of a degree. So any correlation that is used to detect similar sites must allow discrepencies of more than 0.1Celsius, otherwise you won’t ever get a correlation that you can use to do the back-filling of the data.
Thing is, that sites are often moving every ten years or so, and in ten years the climate signal you are looking for has probably only changed by 0.1Celsius. So in order to reliably back-fill the data from another site you would really want a correlation that is an order of magnitude greater than the signal your are expecting to look for – so the correlation should be 0.01Celsius. Clearly that is a contradictory requirement.
It follows from this that you cannot really use this method to backfill data where a site has changed. It might look as if it works because if you correlate a site to 0.1Celsius the error will seem small compared to the variation in the raw data (which is typically about 1-1.5Celsius from one year to the next), but what you are actually doing is correlating the large “weather” signal but not to a sufficient degree to correlate the underlying small “climate” signal.
I prefer filtering out the 2 year and 5 year cyclical variations, deleting the data from any sudden discontinuities within that data or significant discrepancies with respect to nearby sites that occur within a 2 year period, and then measuring the gradients between the discontinuities. This is a far more valid approach, and more transparent. But to be frank, if you are measuring only to 0.1 Celsius and the site is changing by +/-0.5Celsius every ten to 20 years due to a relocation or equipment change and and the signal you are looking for is changing by only 0.1Celsius per decade and the whole thing could be impacted by UHI in any period by anything up to 0.3Celsius per decade then it’s really a hopeless cause.
Basically these climate scientists are being paid to chase rainbows. We should have paid them to set up proper climate monitoring stations guaranteed not to change for 100 years and not to be affected by UHI and then sat back and waited. Only they tell us we can’t afford to wait….
Thanks for this.
I have linked the more complete information on the region bounded by 50-65N and 100-120N that I extracted from the KNMI climate data repository linked here:
http://climexp.knmi.nl/selectfield_obs.cgi?someone@somewhere
The station data is in the table linked below and shows the latitude and longitude; the start and end dates; the years with data, and the population rating. The bottom most link shows the coverage of the stations in the region in a graph with the bounds given above.
The graph has little bearing on the comparison between the CRU and the “Vincent” adjusted data but is shown to illustrate the sparse station coverage in that part of the world.
The table shows that the rural station designation is used for 14 of the 24 station locations. Stations with populations greater than 200,000 numbered 4, while the remaining 6 stations had population between 10,000 and 28,000. Notice also that the Time labeled column, which gives the years of coverage for the start and end dates, is often less than the start/end dates would imply and means that intermediate years are missing.
Obviously sparse spatial data that show large trend differences between stations has to increase the uncertainty of any trends assigned to the entire region.
I thought the global temperature data gathered from various countries from 1850 to 19?, used by Hadley was deleted. Only the “adjusted” data exists. Not so?
As a non-scientist I read through the paper and TBH can not believe that such a half-finished, poorly constructed paper, with cherry-picked, uncompared data would be submitted in the first place.
I am completely sceptical of climate change being man made, but this is clearly a case of a poor paper being rightly shot down in flames.
Move along, nothing to see here.
Just because it is raw does not mean it is wrong. Doo NOT move along. Stop right here. Dig.
Lucie Vincent’s (1998) example of Collegeville station was also cherry-picked. That didn’t stop her paper from being published.
But never mind the style of writing. Is the magnitude of apparent bias not extremely curious?
Edward McDermed said:
“Is it possible to construct global temperature data sets without “homogenizing” the data?
I can think of several reasons to do so …”
I can say from experiences in Finland that the raw data must be homogenized. Published (or at least archived) daily temperatures have been differently calculated through times. There is for example a difference in summer mean temperatures of 0,3 centigrade before 1940 or so depending on different calculation methods. To correct for that you must first recalculate daily temperatures for ALL hourly measurements in the same way. After that you MUST check for known movements of termometers. We have in Finland quite a lot of measurement localities but only six homogenized localities. Nearby stations can be compared if they have not moved and the hourly measurements can be recalculated.
(comment: “hourly” measurements means every 3, 4 or 6 hours).
A Warwick Hughes email about Siberia is here http://www.eastangliaemails.com/emails.php?eid=189&filename=969308584.txt. It mentions stations in the same area as Kamel was concerned about e.g. Irkutsk.
Chita, in the east-adjoining province, has the same urbanisation issue as Irkutsk
Guardian’s Fred Pearce just found that peer-review process is flawed while reading about the rejection of Kamel’s paper. But it seems that he also conveniently forgot to acknowledge his sources of knowledge:
n March 2004, Jones wrote to Professor Michael Mann, a leading climate scientist at Pennsylvania State University, saying that he had “recently rejected two papers …..[snip]…. But the Guardian has established that one was probably from Lars Kamel a Swedish astrophysicist formerly of the University of Uppsala. It is the only paper published on the topic in the journal that year.
So Guardian has established – who woulda thunk it…
http://www.guardian.co.uk/environment/2010/feb/02/hacked-climate-emails-flaws-peer-review#post-area
Lars Kamel:
Now that we have seen what Phil Jones is all about, have you decided to follow up in any way with this manuscript of yours?
21 Trackbacks
[…] Climategatekeeping: Siberia […]
[…] Solar Cycle 24 finally starts to ramp up, rejected Siberian paper published […]
[…] Climategatekeeping: Siberia Siberian temperatures are an interesting case study in CRU gatekeeping. As reported a few days ago here, in an email of Mar 31, 2004, Jones advised Climategate correspondent Michael Mann that he had “gone to town” in his rejection reviews of submissions criticizing CRU’s handling of Siberian temperatures. […]
[…] The Siberian station locations from the GHCN network are plotted below. These stations represent a small subgroup of the continent (and the planet). The region is from 50-65 latitude and 100-120 longitude as defined by Lars Kamél in his criticism of CRUtem. The paper was apparently blocked from publication through intense efforts by Phil Climategate Jones and only now the paper is presented at CA. […]
[…] ClimateGateKeeping: Siberia […]
[…] the Climate Audit blog have long accused him of preventing critical research from having an airing. McIntyre wrote on his web site in December: “CRU’s policies of obstructing critical articles in the peer-reviewed literature and […]
[…] Climate Audit blog hit daylong accused him of preventing grave investigate from having an airing. McIntyre wrote on his scheme locate in December: “CRU’s policies of obstructing grave articles in the peer-reviewed literature and […]
[…] the Climate Audit blog have long accused him of preventing critical research from having an airing. McIntyre wrote on his web site in December: “CRU’s policies of obstructing critical articles in the peer-reviewed literature and […]
[…] the Climate Audit blog have long accused him of preventing critical research from having an airing. McIntyre wrote on his web site in December: "CRU's policies of obstructing critical articles in the peer-reviewed literature and withholding […]
[…] […]
[…] […]
[…] […]
[…] […]
[…] […]
[…] […]
[…] […]
[…] […]
[…] […]
[…] […]
[…] […]
[…] […]