Briffa’s new paper on Yamal is online today here, together with Supplementary Information here.
Yamal has been a longstanding issue at Climate Audit. The new article appears to be their long awaited response to criticism from Climate Audit (though this criticism is not referred to anywhere in the aticle.)
In resisting FOI requests for their withheld 2006 Yamal-Urals regional chronology, CRU said that it was incomplete, as they were continuing to work on its development. However, they did undertake to disclose the 2006 regional chronology as part of the present publication. On my first reading, instead of living up to their undertaking to develop a regional chronology, CRU has instead provided reasons against using a regional chronology and do not present one in the paper – instead focussing on a variation of the original Yamal chronology.
In resisting the FOI, CRU said that production of the 2006 regional chronology would damage the reputation of CRU scientists. The 2006 version appears to be the “Urals raw” chronology illustrated in SM9 as Greater Urals (shown below), though it is not identified as such in my first reading. Readers can judge for themselves whether their foreboding was justified.
Figure 1. Compare to GU2 top panel.
Readers who are convinced by Briffa’s arguments against a regional chronology may well wonder whether, for example, the Avam-Taimyr regional chronology of Briffa et al 2008 would pass corresponding tests, since no similar analysis has ever appeared in Briffa articles in which he presented earlier regional chronologies. Or whether these tests only became of interest when the regional chronology went the wrong way.
CA readers will recall the original controversy in September 2009 over the Schweingruber Khadyta River series in Yamal, a controversy on which I’ll review in the present context on a subsequent occasion. Leaving nothing on the table, Briffa excluded the Khadyta River from the present reconstruction, pointing out that recent trees in this area had been growing poorly (thereby lowering the late 20th century uptick.) They state:
the site report (and statistical evidence) demonstrating the anomalous “signal” in the Khadytla data lead us to omit them from the new Yamal chronology constructed here (see SM5 SM2 part YT3 for details)
Climate Audit 
109 Comments
“the site report (and statistical evidence) demonstrating the anomalous “signal” in the Khadytla data lead us to omit them from the new Yamal chronology constructed here (see SM5 for details)”
Otherwise known as excluding data that contradict the favored hypothesis. That also, in this case, contradict the prevailing assumption of uniformity that underlies the entire field of tree-ring paleothermometry.
And this is unreservedly submitted, gets through peer-review, and past journal editors. It’s beyond shameful. It’s beyond incompetence.
Re: “Briffa excluded the Khadyta River from the present reconstruction,”
It appears that Briffa et al. have a different view of how “climate science” should operate, than the view of how physical science should operate, as physics Nobel laureate Richard Feynman . Feynman observed the lack of a
Slightly OT but is it possible to derive from these data a figure for ring width sensitivity – i.e. rings y mm wider now than they used to bewhen the temperature was x degrees cooler? I note that one of the diagrams in the published paper quote ring width as ordinate (though without units so could be some sort of index). Such a number(which I’ve not found elsewhere) would enable a back-of-envelope guesstimation of how much extra carbon is sequestered by forest from the atmosphere due to temperature rise and whether this is a meaningful negative feedback. Of course it rather assumes dendrochronolgy is “true”.
“pointing out that recent trees in this area had been growing poorly… ”
Oh, my! Data happens. As a man once said, the data is the data.
The trees were sick….sick with worrying about being abused for dubious research.
They are probably feeling better now and growing better by the day.
I spent years instructing students, undergrad and postgrad, that you mustn’t omit data without independent good reason, and that even then you have to report them in an appendix: omitting data just because they are anomalous is fail-your-project behaviour. God, the intellectual standards in this field stretch from utterly shabby to blatantly crooked.
I teach secondary school kids that you don’t get to avoid data you don’t like the look of. It really is the first rule of statistics.
Ah, but you were teaching Science. This is something else entirely.
Don’t we know that Briffa has excluded this data because he has reported the fact? He’s allowed critics to discover this point and comment on it. He has essentially done what you’ve asked your students to do why are you complaining on this point?
Because he hasn’t done what I ask my students to do. At a minimum I ask them to give the results with and without the excluded values. And they are not permitted to exclude data merely on the basis that doing so gives a better result in any case. Statistics becomes a mockery when you are allowed to exclude data on that basis.
Moreover, documenting that you have committed a data crime does not make it acceptable. Any more than admitting to a murder does not prevent you from being prosecuted.
Your Yamal ‘controversy’ is of great sentimental value to me, as my original introduction to alternative thinking on AGW, and Climate Audit!
As a newcomer to the internet (2006), I was not aware of Blogs in general, and used to follow the Guardian(UK) newspaper as being the only ‘broadsheet’ interested in the environment – and I guess I believed most of what was written.
Then, one day a rant (I think from Monbiot) about a dastardly Blog disputing Briffa’s Yamal paper, and naming the Blogger culprit – Steve McIntyre of Climate Audit: I rapidly found said Blog, and thence WUWT, etc. etc.
Thus, when ‘Climategate’ broke not long after I was able to watch and better understand the issues of doubt, and dubious behaviour amongst the scientific ‘alarmist’ community.
Many thanks to you, SM, and long may you continue to try to keep scientists honest.
Briffa’s exclusion of Khadyta data reminds me of Esper’s wonderful dictum:
Yup.
Oh my goodness. Did he really say that, on the record? Ouch!
He is wrong, though, to claim that it is unique to dendroclimatology. Just look at Lewandowsky’s work.
There seems to be a whole pseudoscientific discipline springing up about why it is OK to remove data which detracts from the ‘message’ of a study. Of course, the honest thing to do is to remove whatever you think should be removed, but to spell out what it is and why you removed it. And to leave an audit trail so that what you did can be checked subsequently. That’s how I ran research projects under my care.
Post-modern science has a lot to answer for.
That is damaging. Probably more so than any undeleted emails at UEA.
In, say, mass-spectrometry of peptides, the biologist or synthetic-chemist is afforded a plethora of plausible ways in which to arrive at the expected mass/charge-ratio of the analyte.
Knowing precisely the number that is “wanted” is properly seen as a ‘curse’ by the diligent researcher, not an opportunity.
johanna, I would respectfully disagree that this (rightfully named) discipline is “springing up”.
Indeed I would go so far as to suggest that it may well be “mature” enough to warrant a [very recently redefined] “Progress Article” in Nature Geoscience.
And if a member of the prestigious Nature group has so declared, can their peers be far behind? 😉
It seems we have a new definition on which we can depend. Polyglotology: The flexibility of the meaning of commonly understood denominators, in the post-normal “scientific” practice of pick ‘n choose.
Hilary
Yes, well, such practices used to be called by more unflattering names, which I won’t mention because they will activate Steve’s site filters and put me into moderation for who knows how long.
The “bar” is more like a “blur” these days.
Johanna: Not only did he say it on the record, he was proud of it.
Re: Jimmy Haigh (May 26 12:05), he certainly said it on the record: see the paper at Tree Ring Research by Esper, Cook, Krusic, Peters and Schweingruber, and look at page 92.
kerassi-chronology perhaps?
I was thinking dendrophrenology 🙂
I’ve been reading a book about deconstruction (Derrida, Foucault), the barbarism of it. “Once at home mostly in philosophy and literature departments, their nihilistic tenets are cropping up further and further afield: in departments of history, sociology, political science and architecture; in law schools and–God help us–business schools.” With deconstruction “everything is possible and nothing is true.” And so to Esper quoted here and to Briffa in Briffa 2013. There is no external reality with these folks and so “everything is possible” (even the ridiculous). The insidious tentacles of deconstruction have reached the shores of climate science–so is my take on it.
Slightly OT. You may well be entertained by the history of the “Dada Engine,” a rather cunning bit of software that, via the “postmodernism generator,” can produce “postmodernist” gibberish that is often indistinguishable from “real” postmodernist efforts. See here: http://herbert.the-little-red-haired-girl.org/en/dada/
Legend has it that the original programmer submitted one of its products to a small postmodernist journal, where it was published, the editors hailing the author as an “important new voice.” They were not aware that the “paper” was computer generated and content free.
Any “post-” discipline such as post-normal science, post-processual archaeology, etc., has fallen afoul of the meme.
Resisting the FOI sounds a lot like taking the Nickel.
Johanna
You call the practice of suppressing data that detracts from the message of a study a “Pseudoscientific” discipline? “Post-scientific” discipline is perhaps a better description of it.
Stephen Prower
Stevenage
Saturday 25 May 2013
“Allowing for chronology and reconstruction uncertainty, the mean of the last 100 years of the reconstruction is likely warmer than any century in the last 2000 years in this region.”
Translated- since we don’t really know what this reconstruction means in terms of past temperatures, we can use this uncertainly to justify whatever conclusion we want.
And we do.
The entire justification for the omission of Khadytla in the SI seems to be that “the Khadytla trees (red) show up as having an anomalous slope.” Eyeballing the various curves, it would be reasonable to at least hypothesize that the onset of post-LIA warming may have been staggered by weather, with Khadytla warming first. Instead, the authors apparently decided that Khadytla just can’t grow good trees anymore for reasons which stimulate zero curiosity in these dendrochronologists.
The Greater Urals picture looks very coherent with historical data for that area:
1000 CE Vikings move out of Scandanavia into Russia, Britain and Greenland. There is
a pronounced MWP and then a strong cold dip in the 1700 and early 1800s with a large
warming in the 20th century. I see nothing that would be the cause of excluding this data. If the 20th century data did not come up to a level they were happy with, I would chalk it up to the region warming too much, causing the expansion of the forest either
up in altitude or northward, with the trees in the original area operating out of their linear response zone. Once again, this could also have happened in 1500 and 1000 CE.
No cause for alarm, please show all of the data.
SteveM , the Briffa paper and SI that you have linked here have an interesting (to me) side story. The withdrawn paper coauthored by J Gergis on reconstructed Australasian temperature claimed to have used the Melvin/Briffa signal free Regional Curve Standardisation (SF-RCS). When I emailed Melvin concerning this claim he told me that the Gergis paper could not have used the signal free version since he had not made the code for doing it available to anyone and in fact told me he would provide it to me when it was in proper shape to make it public.
In the linked Briffa paper that the authors comment that:
“This “signal-free” standardisation can be applied to both curve-fitting and RCS standardisation (Briffa and Melvin, 2011; Melvin and Briffa, 2013a), and throughout this study we have used signal-free RCS standardisation (SF RCS)”.
Later on in the paper they comment that:
“The RCS application we have used, i.e. two-curve, signal-free RCS has reduced the potential influence of “modern sample bias” that might otherwise exaggerate 20th and 21st century levels in the TRW data. Whether this has overly suppressed the extent of the “real” increase in modern tree growth (and to an extent reduced the scale of the lower growth of the 17th century) is unresolved. Similarly, the transformation of tree indices to have a normal distribution prior to their being averaged to form the chronology is shown to further reduce the apparent high level of tree growth in recent decades, merely because these recent values are among the most positive in the chronology series.”
The paper indicates that the extraction of temperature effects from TRs is very much a work in progress. Unfortunately we cannot yet duplicate the authors efforts here as we informed by the linked SI ( I assume this is the code I had requested of Melvin):
“The tree-ring chronologies were constructed using the CRUST (CRU Standardisation of Tree-ring data) software. A separate article has been submitted for publication that describes the processing methods that CRUST implements, and the CRUST software (including source code) will be made available when that article is published.
Melvin T. M. and Briffa K. R. (submitted 16 February 2013) CRUST: Software for the implementation of Regional Chronology Standardisation: Part 1, Signal-Free RCS. Dendrochronologia (under review).”
As noted by others here this paper eliminates proxy data after the fact with a hand wave at it being unacceptable. Published temperature reconstructions tend in entirety to have this problem of not making the selection a prior and then using all the data selected a priori. It is good to hear that TR chronologies are not cast in stone even though the basic problem has not yet been addressed.
bit O/T aslo
but i read a article/abstract ? talking about soil/tree growth problems due to man made slash & burn in Swedish tundra(i think).
can’t find it now to link to, sorry.
ps. thanks Steve & contributers for getting me interseted in treelines etc.
To paraphrase an non-remembered CA wag (to me anyway):
If you’ve seen an amazing Briffa proxy reconstruction, you’ve seen Yamal.
Gawd was I envious of that one. It came out of the blue at the Bish’s, I believe.
======
If you’ve seen ONE, you’ve seen Yamal.
In resisting the FOI, CRU said that production of the 2006 regional chronology would damage the reputation of CRU scientists.
And of course the reputation of CRU scientists is far more important than the integrity of science.
“12. Discussion and conclusions
No tree-ring chronology, multiple chronology dataset, or climate reconstruction based on them should be considered final in the sense that they are “fixed in stone”……..”
You just have to read the rest of the above section of the paper to realise what a waste of money dendrochronology is and that it is pseudo-science when applied to climate reconstructions.
They would if they were petrified!
Demdrochronology is not a waste of money, used for what it is good for: dating wooden objects, buildings with timber elements, and calibrating C-14 dates. It is a key tool in the American Southwest and has useful in European archaeology as well. That is denrochronology. This use of tree ring variation overtime as a climate proxy is just that and it is not dendrochronology. I like the term TerryM suggested: “dendrophrenology.”
I’m afraid the word that came to my mind was “dendropornography”.
dendronoknowledgey
I used to think that “chronology” had something to do with dating events; therefore I do not understand why the ordinate is labelled “chronology” and what the units of measurement are.
Re: Alexej Buergin (May 26 03:05),
Read this site (or others) to learn more.
Traditionally, the tree ring measurements were for chronology analysis. The interest was in accurate year-by-year analysis of events that could be detected in the life of the tree.
Now, this has largely switched to dendroclimatology — the study of climate via trees. Or treemometers as some like to say 🙂
That is what I thought, too. But should the ordinate then not read “temperature” instead of chronology, with units like °C or °F or K?
All you kids out there: if ya wanna make cherry pie, ya gotta pick cherries and everything like dat!
See https://climateaudit.org/2006/03/07/darrigo-making-cherry-pie/
Forgive me for mixing my metaphors, but I get the feeling Briffa is between a rock and a hard place; he wants to come in from the cold, but can’t very well bite the hand that feeds, and therefore is standing with one foot on the rowboat and one on the shore.
At least he seems more honest than others about what he is leaving out, in order to “draw out the signal” he is seeking to reveal (and wants to see.) The problem is that even an average Joe like me, who has worked cutting down trees, and has looked at tree rings, since age ten (1963) can see the problems Briffa is going to slip into.
When I was just commenting on these problems on WUWT in 2009 Anthony made a post of my comment, and it drew a surprising 199 other responses. http://wattsupwiththat.com/2009/10/02/a-hands-on-view-of-tree-growth-and-tree-rings-one-explanation-for-briffas-yad061-lone-tree-core/
I’m afraid that if Briffa really wants to come in out of the cold, he is going to have to make a clean break with his old pals.
It is hard, but can be done. In my younger day I hung out with a bad crowd who were a lot of fun, but up to no good, and got me into such trouble I had to walk away from their society. By then I had such a bad reputation that it took decades to clear my name, (and some still don’t trust me.) However I felt much better for doing so.
The concept of removing samples so as to “enhance a desired signal” is so contrary to proper research, I immediately wonder about the frame of mind behind such an obviously ridiculous statement, especially as there was no chance to get away with it unscathed.
snip – I discourage this sort of editorial comment- generalizing from one incident to blanket condemnatinons – , though enforcement is inconsistent.
In 2010, the Guardian published an article that refuted the claim that there was anything wrong with the Yamal data or their use. How was the columnist convinced?
“Briffa denies any wrongdoing. He said “we would never select or manipulate data in order to arrive at some preconceived or unrepresentative result”.
There you have it.
Although S. McIntyre is mentioned more than once in the article, he was not asked to comment on the subject.
http://www.guardian.co.uk/environment/2010/feb/09/yamal-climate-tree-ring-data-withheld
“we would never select or manipulate data in order to arrive at some preconceived or unrepresentative result”. and yet that is exactly what happens when you drop “outliers” merely because they are not growing well recently (or some other excuse). If the outliers were dropped because they just had been attacked by insects, or had recently had a dirty factory built nearby, maybe. But dropping sites to “enhance a signal” is exactly what Briffa claimed they would not do. How do they know there is even a signal to be found? Lots of these reconstructions have verification statistics verging on 0. Not inspiring of confidence.
Briffa showed himself to be in error, with the words Steve quoted:
“the site report (and statistical evidence) demonstrating the anomalous “signal” in the Khadytla data lead us to omit them from the new Yamal chronology constructed here”
Briffa’s problem is that, recognising an abnormality when the trees could be observed, he is unable to discern if a similar abnormality ever happened before the trees were observed. He cannot distinguish a climate signal from a growth-related signal from another cause, and he knows other causes to be present.
The Guardian was a bit dumb to ignore this obvious deficiency. It knocks out the whole Yamal exercise, not just the Yamal super tree.
I’m not following this too closely, but how could cores be taken “before the trees were observed”?
It seems the original researchers who took the cores would be the ones who observed and decided the trees were not suffering from “abnormality”.
So would this exclusion be data disrespect of the original researcher(s), in some ways like siltdown Mann’s treatment of Tiljander?
I believe what Geoff means is before the period of instrumental records.
Yes, it is disrespect to the original researchers.
> siltdown Mann
Brilliant
It was not original. I believe Moshpit gets the credit again, but I’m not sure.
yes that’s a moshpitism
hmm
I may have used it before then
AS mentioned elsewhere
Hot dry drought like summer= No tree growth (or any growth of plants)
Lukewarm wet summer = some tree growth
Hot wet summer = Lots of growth
This is of course considering LOCAL weather conditions as well.
Ergo Dendrology climate Science = BS
Steve writes “Briffa excluded the Khadyta River from the present reconstruction, pointing out that recent trees in this area had been growing poorly ”
Perhaps they’re looking more like swiss cheese than trees these days.
Just one more core damnit. The signal has to be here somewhere!
Lots of smoke and behind the smoke : a gold mine.
An example with the file polar.mxd recovered from appendices: http://imageshack.us/a/img23/4564/polarh.png
(45 cores for the twentieth century)
The divergence here seems to appear around 1995 and it is clearly a problem with stations and not with MXD.
Steve: Yes. There’s lots of information here. For the first time, there is something approaching a technical publication of this data. I need to give some recognition to this point in my commentary. I will do so – I’m pressed on other matters right now.
That image poses some interesting dilemmas. It certainly suggests that CRU has some work to do.
Steve: actually based on this graphic, this MXD is an amazingly good performance as proxies go. Note that the main Briffa reconstructions are RW rather than MXD
PHI above re graph and yet it shows no deviation from 0C anomaly. This to me is the important point.
The conclusion to Briffa’s paper includes ” there is no evidence of “divergence”, i.e. any late 20th century underperformance in tree productivity compared to that expected on the basis of increasing summer temperature”.
Does that mean that there was never a need to hide the decline? Is the main purpose of this paper therefore to bring those Yamal trees back into play for future proxy-based papers?
The divergence is a problem with instrumental data (of stations). Where (and when) perturbations are important, there is a divergence. The start date varies:
– Roughly 1890 for the Alpine region (http://img38.imageshack.us/img38/1905/atsas.png)
– Roughly 1920 for the continents of the northern hemisphere as a whole (http://img708.imageshack.us/img708/1363/anomthn.png),
And much later for Urals.
Phi, you have cracked the “divergence problem”.
It is not the treemometers that are at fault, rather that the instrumental record is.
Far too many post-observation “adjustments”.
Some clarification. It seems that all the interesting MXD material for the twentieth century is in the polar.mxd file. The series have been normalized in the period 1951-1980 before being averaged. The temperature anomalies are those of JJA. Some years are missing for HadCRUT, I could recover some by change in the reading program (http://imageshack.us/a/img21/1076/polar2.png). Still missing are 1905, 1907, 1961, 1962 and 1994. With these completed data, we see that the divergence appears earlier, around 1990.
Another observation (expected), the warming trend from 1900 to 2006 in Briffa 2013 is entirely the result of the RCS process. Raw TRW show no significant trend over this period (ref. polar.raw).
The difference between the criteria used to reject root collar samples and the criterion used to reject the entire Khadytla dataset provide a good answer to the previously discussed CA topic of “when should you reject data a posteriori?”
Steve: I prefer to develop criteria on a sort-of common law basis: i.e. develop principles from cases. I agree that there is a useful distinction between the two examples. However, note the impact of Briffa’s rejection criterion on strip bark bristlecones, which are far more radially deformed than root collar. As too often, the Team is very alert to problems with data that gives them the “wrong” answer, while blind to worse problems where the problem goes in their favour. I’m working on a detailed exposition.
Can someone explain to me how you go from this: http://img827.imageshack.us/img827/640/ymlall.png (TRW normalized over 1951-1980) to graphics of Briffa 2013 ?????
Oh, I found and it is interesting. This is due to a massive addition of young trees during the first half of the twentieth century. In fact, it is a lack of correction by the RCS method which is the cause.
The corresponding graph:
We see that this is not the only problem.
The interpretation of my previous graph is not trivial. What can be drawn is that the new data used to complete the study until 2005 are based on trees whose growth is higher than the average. The following graphic shows it most simply: http://img69.imageshack.us/img69/306/trwyml.png
The singular nature of the TRW evolution in the twentieth century for Yamal comes therefore from a selection bias of trees.
That’s it. In this whole, it is obviously the comparison MXD-HadCRUT-UAH which is most fertile.
This will probably be my last post, I thank the few people who have bothered to try to understand my approach.
Goodbye
Steve: be patient. it’s an interesting topic. I’m a little preoccupied with other things right now tho.
Article now at RC. They mention CA but dare not link to it. One commenter has already questioned the post hoc exclusion of data.
http://www.realclimate.org/index.php/archives/2013/06/yamal-and-polar-urals-a-research-update/
Thanks for the link. This should get interesting given the rather gratuitous comments about criticisms of their earlier work.
The Real Climate post is being flogged by Scott Mandia, Michael Mann, and Dana Nuccitelli as some final refutation of everything related to Briffa and Siberia from Climate Audit:
https://mobile.twitter.com/AGW_Prof/status/341589957175017472
I very-much would like to hear Steve’s response to this post myself. Briffa et al has indicated that they had published their caveats and misgivings about the Khadtya data long ago, and had obviously had been working on it in the meantime (the reason for their FOI victory?)
Still, I believe the publishing of data about which even one might have misgivings has some positive effects. Perhaps others may more speedily arrive at similar conclusions, or discover other things in it. Data itself is a worthy thing to be published, yes? However, it’s probably also a good idea to leave the purview of the folks who did the work of obtaining it to have the best say when this happens, though best practices would indicate a favoring of the sooner rather than the later.
I’m also a little confused about the multiple uses of the word ‘replication’. I might also ask about this over at RC, but I keep getting the idea that they’re looking for a particular signal rather than fully examining one pre-defined sample set and taking whatever comes. But, Tom insists this is not occurring, so there have to be ways to reconcile all of that.
The Q and A you initiated on Real Climate was useful. You asked good questions and got what seemed like reasonable answers.
I’m still a little iffy on his response to one of my questions (#2) …
“Our approach was to begin with all samples unless there was a reason not to (e.g. root collar samples), but then to consider common signal between independent subsets to identify any issues. None of our selection was based on correlation or non-correlation with temperature observations. The comparison with temperature came only after we had finalised the chronology.”
From what he’s saying is that they ‘begin with all samples’ and next identify any issues before looking at the data they have within them… However, it’s after they finalise the chronology that they see whether the data doesn’t ‘replicate’, leading to the assumption that their previously identified ‘issues’ are actually issues worth discounting the sample over. At least that’s how I read it.
If that’s the case, that’s still dribbling over into the post-hoc category the way I’ve got in my mind– because you could theoretically take a potential issue, like with Bristlecone strip bark, and see that the results have a more robust ‘replication’ (to use his word) when included vs. excluded… and therefore merit inclusion (which is what Michael Mann has advocated before). Should not the sample integrity be preserved over the signal integrity?
This is why I made the comparison to med/pharm research (that has so far gone un-touched)… Imagine if a drug company were to survey 10,000 people from a group that they feel would best highlight the potential utility of their product. They then identify as many people who exhibit a ‘replication’ that is desired and file that as a signal that their product is suitable for mass acceptance. I know they (dendroclimatologists) feel that their signal replication is strong within a sample size restricted to suitable tree-mometers, but that’s obvious when they reduce the sample size to those that most likely exhibit the signal (seems round-robin-y). Anyway, would not at some point the sheer amount of trees you have to go through to get this signal to show itself– even within these specified locations/conditions, start to matter? It’s like a handful of trees that are dominating the say-so for what could be thousands(millions?) of trees. Jesper says this is ‘unique’ to the field of dendroclimatology, but should there be a larger scientific body that independently concurs? It seems like a pretty liberating hallowed status that many science fields would love to enjoy.
Salamano, your question about strip bark was exactly on point. I notice that they have been completely unresponsive to this question. It’s worth re-iterating.
The problem of radial deformation in strip bark is a long standing issue at Climate Audit. Bristlecone deformation is far more severe than the radial deformation that Briffa took exception to.
On the standards now advocated by Briffa, the Mann reconstructions are out the window.
Of course, I anticipate that Briffa et al would be completely opportunistic on this issue and only object to radial deformation when it gives an edge to medieval trees and not object when it gives an edge to modern trees,.
Salamano, you have to parse anything that Briffa and Osborn say as minutely as if it were by Gavin Schmidt. They appear to claim that the root collar analysis wasn’t post hoc with their language readily suggesting to an unwary reader that they identified the root collar issue in 1999 and took the decision to proceed no further at the time. I’d be very surprised if this is what actually happened. They told Muir Russell that they’d never considered updating the Polar Urals dataset – a claim that is inconsistent with having turned their minds to root collars in 1999 or so. My own interpretation is that their interest in root collars is very recent and arose only after they were aware of the high apparent MWP with the inclusion of the 1999 data.
Re-read their statements and see if you agree.
To one of the points, Tom Osborne said:
“The fact that some Polar Urals cores were taken from root collars was noted at the time the samples were taken — please see figures PU05 and PU06 of SM3 at our supplementary materials webpage:
http://www.cru.uea.ac.uk/cru/papers/briffa2013qsr/
Once aware of this information, the decision to exclude the root collar samples was taken on the the basis of the likely incompatibility of the ring widths from root versus trunk samples. We did attempt an adjustment to compensate for different mean ring widths between root versus trunk samples, so that the root data could be retained and used, but without success due to the added complication of difference growth trends as well as mean growth rates. These attempts are documented in SM4. For tree density (MXD), such an approach did work and we could therefore use the MXD data even from the root samples.
For the second case, the potential problems with the Khadyta River site were also noted at the time the samples were taken — please see SM2 (pages 1, and 5 to 8).”
So he is stating that there’s some data caveat that was published a while back, or perhaps noted by the original authors of the data, or new co-authors to this publication noted back then but were not on the other Briffa work to be on record back then. So the “we” could represent a changing group of people over the years (unless it’s always understood to be some larger Body Scientifik or something).
I’m seeing the introduction of the idea that there needs to be a “compatibility” between ring widths between root and trunk samples. This may be well known and standard within the dendro community, but I haven’t heard this explanation before (particularly considering its potential applicability in strip bark scenarios). If this were a well known reasoning, then I would assume that Mann would have gone to it in the case of Strip Bark, rather than simply stating that including those samples shows better statistical conformity than excluding them.
There also is this idea that the potential issues with the ‘bad’ tree data had been known by the “Russian Colleagues” with direct knowledge of the individual trees themselves. I think this is another one of those deferring to the original purveyors of data bit that permits scientists to use/not use various proxies, cores, etc. for whatever the original author allows. In order to get at that argumentation, perhaps it necessitates conducting research at the site as well.
I was particularly curious about the introduced knowledge of the ‘dying trees’ being unsuitable for use in the sample. In my opinion, what dendro-guy would bother coring a dying tree? More to the point, who’s to say these other trees are presently any more or less healthy? Do they geotag all these trees. Perhaps Yamal needs another visit…
Steve:
no such caveat was “published”. Watch the pea. There were observations in the field notes that, looking back, are now interpreted as caveats, but these problems were not disclosed at the time. As so often, Osborn is trying to spin.
elsewhere Melvin and Briffa have observed the existence of “Modern Sample Bias” which arises from a tendency of dendros to sample the most imposing modern trees, whereas subfossil trees are a more universal sample. Presumably subfossil collections include everything including trees that weren’t flourishing. Excluding “unhealthy” trees seems like the sort of ridiculous ad hockery that characterizes the field.
It was not noted in SUpplementary information at the NOAA archive or anywhere else. The supporting information is merely a private email from the Russians to Schweingruber. Were Briffa and Osborn copied on this? Did Briffa and/or Osborn take notice of this at the original time? I am very very dubious of this.
I’m still curious about the ‘replication’ bit…
For example, Osborn says:
“Ideally dendroclimatologists would like to avoid using such samples (if their interest is in the longer timescales). But if the sample count is low, then comparison with other trees may be a reasonable approach to deciding whether non-ideal samples could be used to increase the replication. We cover this quite a bit in our paper — looking for common behaviour between independent sets of data as one measure of confidence (Figure 5f shows the high standard deviation 900-1100 and 1400-1600 when roots are included, and thus lower confidence in the chronology mean value).”
I’m sure some folks here find the term “increase the replication” suspicious, but is this actually a common thing within other areas of science, let alone dendro? The idea of working backwords through lower-quality materials in hopes to get more of the target signal (otherwise known as ‘common behavior between independent sets of data’)… Is there something I’m missing here? Each new sample that is looked through contains seemingly orders of magnitude more of rejected trees amidst a larger sample. 4-264 years of data with “samples from less than 10 trees” (even including ‘non-ideal’ specimens)?
The incidence of twins is about 33 out of every 1000 births, Siamese twins are something like 1 in 50,000 … It’s like going out there and finding the Siamese twin ‘replication’ and calling it a robust signal of some greater pattern significance to characterize world population. Or is it? Obviously we do have this temperature appearing in a myriad of other sources, which breeds more confidence when it shows up here. But what is the line where it becomes significant and not a random outlier or corrupted sample? Perhaps some dendro folks feel that if there’s nothing wrong with the sample, even ONE tree will suffice and will not be a product of random chance. There seems to be extra ‘skeptical’ work among dendro types to indentify/explore issues with non-replicating trees than with ones who do, but at least Osborn is saying there were some caveats indicated prior to seeing the results– an escape hatch perhaps? That’s probably too brash of a statement.
Their latest response to your latest comment at RC has left me scratching my head. Seems more like hand-waving than a response. At the same time, assuming strip bark is associated with a sick or damaged tree, by their logic, shouldn’t all these be removed from existing data sets?
Some here seem to think that Briffa,at al are under an obligation to include all tree rings everywhere. That seems…odd. I would think that part of the researcher’s job is to look for those data sources that are reliable and useful and include them while excluding sources that are spurious. Is that a bad way of looking at this?
And do I understand that the various FOI requests on this topic have been fully and finally determined to have been inappropriate?
Steve: The FOI story is not accurately described by Briffa. I succeeded on part of the request but appealed on the point that I lost on. The UEA gave multiple excuses and nearly all of their excuses were rejected. However, one of their excuses succeeded. I think that the Tribunal was incorrect on their reasoning on this remaining point, but the matter is moot since CRU has now released the data, something that they had undertaken to do in connection with my request. Reasonable people can disagree on wHether they would have released the data without the FOI. I happen to think that they would never have released the previously withheld results. So in this case, I’d say that the the FOI was relatively successful.
As long as the rationale is stated a priori, is consistently applied and is independent of the dependent measure I would think that you could drop the proxy. In fact, in some cases with known false signals, e.g., the case of strip barked BCP, you should drop the trees so affected. In Osborne et al RC piece, the argument for dropping trees where the core was from the root collar should work as long as all cores taken from root collars are (a) identified and (b) dropped. I am assuming that the distortion of the metric of those cores taken from the root collar are indeed independent of climate variables and reflect other biological processes akin to the effect of strip barked trees.
This probably is going over old ground and I may be missing something, but Osborn’s response (#20) at RC (http://www.realclimate.org/index.php/archives/2013/06/yamal-and-polar-urals-a-research-update/comment-page-1/#comment-342096) to the issue raised by MJB (#18) seems a little odd. Can anyone clarify? Wouldn’t their approach highlight outlier trees unless there was a large sample? Are magical trees a potential problem?
Steve: calculating a tree ring chronology allowing for all the random effects is not easy. I’ve posted on this from time to time. Osborn’s point that standardizing on each tree eliminates long-term variability is one that I don’t take issue with. However, once you abandon that approach, allowing for different sites and microsites is not easy. Previously, CRU had never provided anything remotely approaching a description of their behind-the-scenes adjustments and bodges.
They open the door a little in the present article, but much remains unclear. They say that they separately standardize slow and fast-growing trees, but thus far I haven’t seen a definition distinguishing the two. There’s a lot of poorly organized material to digest so it may be in there somewhere.
Look at Supp Mat 4 page 4 for a flow chart of truly baroque adjustments in the present case.
One of the fundamental problems is that they are dealing with a fairly difficult statistical problem without any reference to statistical techniques known to statisticians, instead using all sorts of ad hockery. Some techniques can probably be translated to sane statistics, but it takes time to do so. It’s too bad that they’ve been so resistant to outside statistical advice.
Steve:
Thanks for the response. I guess I do not see why standardizing on each tree would remove the long term signal anymore than using anomalies would, assuming that separate standard adjustments are made for the age of the rings.
From the narrative at RC:
“As reported in BEA13, through collaboration with our Russian colleagues who have extensive knowledge of tree-rings in this region, we have learnt that the Khadyta River site has problems related to the particular site conditions that differ from other sites in this region, and maybe influenced by changing permafrost.”
Had to chase the pea a bit on this one. The statement above apparently refers to the single line Steve quoted: “the site report (and statistical evidence) demonstrating the anomalous “signal” in the Khadytla [sic] data lead us to omit them….” The three words “the site report…” apparently can be made to pass as disclosure that the rationale for omission was poor health. In any event, to exclude sick trees from a larger data set requires that you be able to show that the rest of the trees that you did use were healthy throughout the entire time series.
they also falsely stated that Khadyta River was atypical of data in the region. Another faabrication. In fact, it is very similar to other Schweingruber series.
So you are publicly accusing them of falsehood and fabrication?
When will you (and your friends here) do your own temperature reconstruction? You must have a lot of data by now, a lot of knowledge perhaps, so when will the world see a reconstuction in a peer reviewed journal? Or are you afraid of the result?
I think the point is that reconstructions are critically dependent on more or less arbitrary decisions on what to include and how. A wiggle of any general shape can be produced by a selection that can be justified in an arm-waving kind of way.
It seems to me that it’s not just ‘having data’…
If you’ve read many of CAs posts over the last months (including this one), a lynchpin in the reconstructions boil down to personal communication by the original purveyor of the data specimens or a declaration of their usefulness that tie the hands of potential reconstructors.
You’ve got ice cores from the same region in South America where the originator declares one as fit for temperature and another not; you have in this situation apparently the saga of scientists trapsing through northern Siberia unwittingly coring dying unhealthy trees–
Sometimes the problem is not devising the reconstruction itself, but rather questioning why certain sources are good and others are bad, beyond the simple sayso of the folks that got them. You obviously can’t fault scientists that choose one proxy over another if the original author says they don’t like the specimen they $pent the time to get (for temperature), but sometimes too we have a situation where the original author says “don’t use this” (strip bark, Tiljander, etc.) and yet it’s done anyway, sometimes for little more reasoning than the results are statistically more robust with them than without them.
There are all sorts of places you can go online to grab a bucket-full of proxies and have at it. I don’t think the discussion is primarily in that area.
Surely every site has “problems related to the particular site conditions”? I doubt any site is perfect, nor any data perfect in gathering and/or measurement.
Is there a site report that doesn’t express a caveat or two? Or, rather, would one trust a site report that didn’t bother noting potential problems?
Jim Bouldin has written two posts on the subject: one and two, with apparently more to come. The second one points out a bias in the Polar Urals sampling which may affect the quality of results.
It makes for an interesting exchange at RC. Commenter Ruth asks:
“When you discard 20th-century non-climate signals, then compare with a medieval sample where non-climate-variation isn’t excluded, you will probably get an enhanced climate signal in the 20th century. Won’t you?”
An elegant way to say what I wrote earlier in this thread. Jim Bouldin responds affirmatively:
“Your point is a good one on a couple of levels.”
One of the levels is the engineering subdiscipline known as “Design of Experiments (DOE).” The author team recognizes the DOE issue – without explicitly mentioning it as such – and discusses how you could fix the shortcoming. However, their attempt does not really suffice, partly because the modern warmth at northern treeline should make the trees healthier, not sicker, which goes to the core (sorry) of the argument that these are treemometers. This logical flaw, combined with the DOE issue, means there is a strong burden of proof on them to show that other periods of poor growth would be recognized and discarded. The fact that they found none means they did not shoulder the proverbial burden. But at least they took the question seriously!
The public discussions on this topic are just completely screwed up, at least partly because the science itself is a mess. I feel truly sorry for the interested public.
Dr. Bouldin,
I have now read through your essays on analytical problems in treemometry, very impressive work, FWIW. In fact, I think Steve would agree that it rises to the level of “an engineering quality assessment.” By that I mean that you make a serious effort to understand all the variables and how they interact. I have been guilty of cynically referring to “the silence of the dendros” for not standing up for the intregity of the science. Now that I have found your blog, I won’t be able to say that anymore.
Thanks for the good words Matt, I do hope those articles were helpful.
In my view one of the single most important practices in all of science is thoroughly exploring all possible uncertainties that many exist (and quantifying them to the degree possible) in any given data set and/or analysis. Perhaps that’s one of the roots of true skepticism. You have to work at it, but if you put that rule near the top of the “to do” list, you save everyone a lot of grief in the long run.
And part four of my comment on Briffa et al is now up by the way.
I’ve left this comment at RC:
#50 “Tim Osborn, Tom Melvin
Thank you for this additional information.
Is there any other tree you could remove from the chronology which would make more than a 1C difference to the result as the removal of YAD061 does according to your linked plot?
You state: “Of these 18 trees with the highest peak index values, 8 peak values occur in the 20th century and no more than 2 occur in any of the preceding 20 centuries.”
Please could you provide a table or plot showing the positions and durations of the samples from those 18 trees in the chronology.
Regarding the prevalence of peak values in the C20th, to what extent may they be due to increased CO2 fertilisation rather than increased temperature?”
Could Steve shed any light on the ’18 trees’ and their positions in the chronology?
Tallbloke,
Your questions bored a hole through their arguments, so they put your post in the proper place. It would have been interesting to see the response but thats how it goes over there. Like most of the threads at RC, this one quickly devolves into painfully silly jabs about deniers from the usual handful of commenters.
Thanks Matt. Yup, it’s gone.
amazing (but not surprising) — why can’t RC permit such concise fact-oriented questions to stand for discussion?
One may disagree with any aspect(s) of what was posed, but the comment was polite, inquisitive, fact-oriented (what are the real facts here, please guys?), and on relevant (highly) issues. If public education and “science communication” does not include informative answers to such questions then how is what RC does to be distinguished from “propaganda”??
Gavin and I have some ‘history’. clearly, the water that went under my bridge is a stagnant pool under his.
This is why I no longer comment at RC. You saw what happened to Clive Best on this same thread. The problem with RC is the peanut gallery of name calling and snide personal attacks and just plain political diatribes. I remain surprised that Gavin and the team doesn’t reign it in to increase their credibility. But hey, it shows you who the real target audience is.
Incredible. Or actually, to think about it, not really. And that’s why I’ve claimed for years now that RC (and maybe Tamino) is probably the greatest skeptics generator out there. A strange kind of a situation actually, as I think that (climate communicator) Gavin is an incredibly intelligent person. How on earth does he not understand this?!
Yup, and as a relatively recent visitor to the climate wars, I can say that it is misbehavior by quite a few on the “alarm” side that first aroused my suspicions that all is not good in climate world…. Honest people fully dedicated to seeking facts and sound science do not behave in ways that we have seen too often now. Something else is going on….
the yamal trees are descendants of Treebeard the Ent. they walked the earth and thus their rings are global not regional. only stationary trees can be expected to reflect regional temperatures. all other trees will have rings that reflect global temperatures.
I have a couple of links here that lead to the code used by Briffa and Melvin for their new and improved RCS:
http://www.cru.uea.ac.uk/cru/papers/briffa2013qsr/
http://www.cru.uea.ac.uk/cru/papers/briffa2013qsr/YamCRUST.zip
This is from the RC Yamal thread:
You can see that the mean yearly temperature anomalie from 1920 – 1950 is about 0,5 C higher than the mean summer temeprature anomalie in that period. I wonder how they would figure out past yearly temperatures only having tree ring derived summer temepratures.
5 Trackbacks
[…] Read his entire essay here: Briffa 2013 […]
[…] to "draw out the signal" he wants to see by including some tree-rings while excluding others. https://climateaudit.org/2013/05/24/briffa-2013/ I left this comment on the Climate Audit site: Forgive me for mixing my metaphors, but I get the […]
[…] https://climateaudit.org/2013/05/24/briffa-2013/ […]
[…] for the last ~ one to two thousand years. There have been two blog posts on the paper already (here and here), both dealing mainly with issues raised by Steve McIntyre in years […]
[…] in SM9 as Greater Urals (shown below), though it is not identified as such in my first reading. Readers can judge for themselves whether their foreboding was […]