the Air Vent

Because the world needs another opinion

Circling Yamal – delinquent treering records?

Posted by Jeff Id on September 25, 2009

An interesting post from Lucy Skywalker who is exploring the depths of the Kaufman Antarctic proxy reconstruction.  Partially in response to some of the ridiculous criticisms from Tamino, this post explores a little further into the Yamal series.

The original Yamal study is here and shows no trend.  The data is here and the spreadsheet containing the data with the new Yamal in series 22 for the obviously flawed Arctic temperature reconstruction paper is here. (My sincerest thanks to Bill Illis for providing the links on WUWT)  Yamal is plotted in this post by myself here and the post demonstrating that the unprecedentednosityness vanishes when Yamal is removed is here. or replaced with equally reasonable proxies here.  You can do your own reconstructions by simply averaging the rows of the spreadsheet linked above so you can see yourself what removals or additions of Raman noodles do to the graphs.  I would be remiss not to mention the Erice presentation by Steve McIntyre, who is the layman’s portal into the spider caves of proxy science.  It’s towards the end of the presentation in the link.

Lucy’s original post is linked here.


Treering proxy temperature measurements from Yamal have a pronounced upturn that helps to create a hockey stick shape with the blade rising in the 20th Century. These were used in Kaufman 2009 to show the Arctic apparently warming fast after 2000 years of steady cooling. This warming starts after thermometer records started. The pattern should therefore show in the thermometer records, especially around Yamal. But it does not.


Click to link to Lucy's original post and interactive graph.

This is more work in progress, following Circling the Arctic, in particular I was inspired by Bill Illis (05:42:38 on WUWT, 9/9/09) who provided links to the Yamal material and Kaufman’s data. I am still learning a lot, and am open to being corrected / improved (email preferred). My motivation for doing this was my certainty that Briffa’s Yamal data should never have been used in Kaufman 2009, and that visual comparison to thermometer data can help people see this easily. John Daly collected records, apparently from NASA GISS, that were reliable, longstanding, not-moved rural stations: the Vardo group, the Kanin Nos group, and the Salehard group. Today the NASA records all start at 1880, thereby losing a vital cornerstone of information I find trustworthy. For the early Arctic records were kept by those who had a big reason to get them correct: the life of their community.

I’ve been accused of cherry-picking, but I maintain that Daly’s criteria are actually good science. Understanding is helped by individual stations, names, places, and pictures, not just nameless averages. And “auditing” climate science is about reconnecting professional standards and the best specialists with basic principles and good amateurs and commonsense, where user-friendliness is essential. But to forestall objections, I’ve included records here from four Russian cities that were specified in response to “Circling the Arctic” at WUWT. This lengthens the discussion but does not undo the conclusion, in fact it makes it more – what – “robust”.

There are a number of important details to glean from these records. On first sight, it looks as if the treering proxy record (the “rearing snake” shaded faint against each thermometer record) is way out of kilter, excessive by comparison with all thermometer records, and patterned differently, but it’s good to look at details with an open mind. The treering record shows temperatures relative to a mean, not an absolute level. All my thermometer records with treering record faint overlays are identical scale, with 10-year divisions from 1841 to 2011, and 2ºC divisions from -18ºC to +4ºC.

  • There is one interesting piece of evidence from this Hockey Stick, that the Air Vent shows even in assemblages of less-distorted proxy records. The pattern of temperature change still make a sort of Hockey Stick: slower fall, faster rise. Now observe: this pattern appears again and again in the ice ages temperature sequence. Temperatures in general appear to rise rapidly, compared with rates of temperature decline: a natural pattern.
  • What the Yamal treering evidence shows, however, appears to be a very exaggerated distortion.
  • Moreover, the statistics method used by Mann et al appears to be capable of producing a hockey stick out of any data – see Jeff’s June series of posts starting with Hockey Stick Revisited part 1.

Now to the thermometer records.

  • Savour the environment of Yamal, just north of the Arctic Circle. It is flattish, watery, permafrosted tundra with trees in a few places, fairly maritime, bordering a sheltered area of the Arctic Ocean to the north and the Siberian landmass to the south.
  • Bjørnøya, Vardø and Murmansk are surprisingly warm – Murmansk has a similar latitude to Ostrov Dikson and Hatanga. Perhaps proximity to Gulf Stream waters, and distance from continental centre of mass, raise their mean temperatures. Siberia has an extreme annual temperature range due to landmass; and the greatest departure from “norm” is the winter cold, not the summer warmth.
  • Vardø: the record shows virtually no trend at all, certainly no marked change into uplift from 1900 or so. But its detailed patterns correspond very well with those of nearby Murmansk, and reasonably well with Bjørnøya. Vardø lacks the decadal fluctuations shown by Bjørnøya, but Murmansk, that resembles Vardø very closely otherwise, shows decadal fluctuations.
  • Kanin Nos, Ostrov Dikson and Ostrov Vize temperature patterns all have high correlation with those of Bjørnøya and Murmansk, both short-term and over decadal fluctuations.
  • Salehard, Turuhansk and Hatanga likewise show short-term and decadal fluctuations similar to Bjørnøya and Murmansk.
  • From such similar patterns – with fluctuations but without sudden large upswing around 1900 – one can IMO reasonably infer from Vardø the sort of pattern all these other stations would show.
  • The uniquely steady temperature at Vardø puzzles me. Does it indicate comparative freedom from Arctic sea ice?
  • The Irkutsk group: these are similar to each other, both in short-term fluctuations and in overall pattern; they resemble all the other records less, both in details and in overall pattern. Do they indicate continental, rather than Arctic maritime patterns? Is this difference also a hint about (delayed) effects of PDO and AMO on Arctic maritime temperatures?
  • The Irkutsk group were chosen, with records since 2002, to answer allegations of cherrypicking. They are furthest away and their environment is very different – mid-continental. All these cities lie on major railway lines. Recent years since 2002 are actually irrelevant since the big proxy upsurges started around 1920, 1950, and 1980. But even with their steady rises since 1950 or so, Irkutsk, Omsk, Barnaul and Krasnoyarsk all show considerably less warming than the Yamal treerings, and a different pattern.
  • Note the typical interannual variation of temperatures: the colder the mean, the bigger the fluctuations in general – and the harder to eliminate “noise” signals. Or perhaps those fluctuations are evidence to a different beat, that of cycles? Would they yield significant correspondence to solar cycles? Ocean cycles?
  • There is close correspondence of temperature patterns in each of these four groups, thus confirming the trustworthiness of all.
  • “Fluctuation” is far more striking than “trend”. Any trend measurement must take full account of natural fluctuations in the Arctic.
  • Salehard borders on Yamal, yet its thermometer record is strikingly different from the treering record. And since the pattern at Salehard is backed up by Murmansk, Bjørnøya, Vardø, Kanin Nos, Turuhansk, Ostrov Dikson, Ostrov Vize, and Hatanga, it seems reasonable to conclude that the Yamal treering record is the one that is suspect.

Click to expand

Click to expand

The tree ring proxy temperature record stands out by being utterly different. It would not matter that several years are effectively smoothed into each record; what matters is the range, and the comparisons of its downs and ups with the temperature records. It is wildly higher than even the biggest fluctuations. Salehard should show a very similar temperature pattern. Imagine if the Salehard record were smoothed to the level of smoothing of the Yamal proxies, its range of fluctuations would largely disappear. So the Yamal fluctuations are even further out of order.

Omsk, Barnaul, Krasnoyarsk and Irkutsk show a similar longterm pattern – level until 1950 or so, then a rise of ~2ºC to 2008. Short-term, the patterns sometimes match and sometimes diverge. The longterm rise could be UHI. Or it could be a natural temperature rise that has occurred, that the continental landmass reflects but the maritime locations, influenced by oscillating ocean currents, do not. It is still not the IPCC graph pattern, which rises from 1900, dips from 1940, and rises again from 1970.

Omsk and Irkutsk are completely different to the pattern of all the other stations here from John Daly: Vardø, Murmansk, Kanin Nos, Salehard, Turuhansk, Ostrov Dikson and Hatanga. These stations show very similar cycles that appear to be within range of the cycle that is visible below in the Bodø record, which closely fits the IPCC graph, and also fits a 60-year solar cycle. In the unadjusted Bodø record, the 1930’s were warmer: this suggests UHI contaminating an otherwise reasonable IPCC record.

This has been adjusted 0.5ºC downwards in false compensation These records show a 2ºC divergence that is probably UHI

I’m not able to do trend lines and smoothed graphs. But these graphs speak volumes. Sun, sea, and continental effects are obvious factors to examine further. Temperatures in maritime locations are strongly moderated by the oceans, rather more so, it seems, in the Arctic winters than in the Arctic summers. We are learning about the effect on land temperatures of ocean currents oscillating between warm and cool, and the lag between equatorial seas warming, at least correlating to the Sun (even if mechanisms are lacking so far), and warming at higher latitudes due to ocean currents moving polewards.

The puzzle is, what actually happened to distort the Yamal treering records – assuming it was not just a fluke or mistake? All larches are unsuitable for use as temperature proxies. They are sensitive not only to temperature but also to moisture and CO2. Could the CO2 or moisture level have risen in Yamal in the twentieth century?

26 Responses to “Circling Yamal – delinquent treering records?”

  1. There are many posts about Briffa’s Yamal at Climate Audit over the past few years. Together with the Graybill bristlecone chronologies, it is crack cocaine for the Team. You ask:

    The puzzle is, what actually happened to distort the Yamal treering records – assuming it was not just a fluke or mistake?

    There has been a break in the case in the past few days. For many years, Briffa refused to disclose the measurement data used in his Yamal chronology. Science refused to require him to do so even though it was used in Osborn and Briffa (2006) – on the basis that it was up to the predecessor journal (Briffa 2000) to do this.

    Last year, Briffa used Yamal in an article in Phil Trans B(Roy Soc) and, after I complained to them, they required to Briffa to archive the data. Briffa asked for extra time to do so and as of earlier this year hadn’t done so. A few days ago, I checked and the measurement data was quietly put online some time this year (without anyone notifying me needless to say.)

    I’m presently examining the data and expect to do a new post on Yamal at CA in the next few days, maybe today.

  2. Jeff Id said

    Sounds like a fun post Steve. I’m looking forward to it.

  3. wattsupwiththat said

    Here’s some history and background on Yamal

    The peninsula has quite a lot of drilling:

    In the oil and gas field of the Yamal Peninsula, Gazprom says steam injection is used to make the wells more productive. I also saw something about injecting some heat into the cross country pipelines to keep them operating.

    Drilling in permafrost isn’t easy. Of course we know that oil and gas production with steam injection wells are a uniquely 20th century thing. With the advent of such things in a place that was mostly undeveloped and uninhabited in the 19th century, it would seem like a potential influence on man and associated moisture on trees in the region.

    Depending on where the Yamal trees were sampled/cored, they might very well be close to or downwind of a new development/settlement. Do you have a location?

    Since Gazprom calls it a “mega project” it seem fairly widespread.

    As we know, establishments of settlements means establishment of a water supply. The Russians are famous for leaky steam heating too. With so much industry and mad dash expansion, there are lots of ways to add moisture to the local atmosphere.

    With all that energy used to keep warm and do drilling etc, it would also seem there will be a CO2 plume on Yamal.

  4. wattsupwiththat said

    Forgot to mention, I was responding to Steve M in comment 1

  5. the statistics method used by Mann et al appears to be capable of producing a hockey stick out of any data

    This is not relevant to Briffa’s tree ring chronology at Yamal and is a distracting point in this context.

  6. Jeff Id said

    Isn’t it true that Yamal (Briffa style) was pre-sorted by correlation to temperature?

  7. #6. I still don’t know the anatomy of the Yamal chronology. I’m working on it.

    But don’t confuse the making of a tree ring chronology with the making of a multiproxy composite. They are different animals.

    Obviously the Yamal chronology looks fishy to me, but the problems are going to arise at a different stage than (say) Mannian CPS or something like that.

    In a very general sense, one has to be alert for biased picking, but proving it requires careful parsing. I’ve reported at CA on some gleanings from Yamal ID numbers – afterwards, such parsing can be done quickly, but it takes a while to figure out the ID schemes and whether they have information that’s usable for proving that things were picked.

    I think that I’m going to be able to prove something pretty interesting on this file over the next few days.

  8. Jeff Id said

    I thought there were some good quotes of the authors saying they were rejecting treering data based on correlation to temps. When I went back the other morning, I couldn’t find them on CA.

    Of course you’re right that CPS has it’s own set of issues, I also take issue with sorting and scrapping data by correlation. I imagine you do as well, and for that portion of Mann08 it looks like Yamal is similar. IMO you can see it’s been done visually as well as by the number of proxies remaining as you point out in your latest post.

  9. RomanM said

    I too have been looking at the data. One of the things I did was to take the simple average of the proxies available for each year to create a reconstruction (tried with both scaled and unscaled data)

    From there, I did Kaufman’s decadals and compared them to his version of Yamal. There were sufficient differences to know that the original Kufman sequence wasn’t done that simply. However, the interesting thing was that my version “diverged”drastically after about 1800. results in the 1900’s were in the same range as earlier results lacking the “hockey stick” shape. Italmost looked klike someone had scaled the tail substantially in a linear fashion.

    I’m not quite there on getting a handle on the differences so I haven’t gone to the blog on it yet.

  10. #8,9. Kaufman’s Yamal corresponds exactly to the Briffa 2008 archived version using my implementation of Kaufman’s decadizing and scaling. (This is one of the type series that “proves” that the implementation is correct.)

    Jeff, the surmise that Briffa has not used all available Yamal measurement is still a surmise. We don’t KNOW this for sure (though it looks likely).

    We also don’t KNOW that the Yamal measurements have been screened according to their correlation with temperature as you surmise in #8. Personally, if screening has taken place (and this seems likely), I’m dubious that it would have been done on the basis of correlation to gridcell temperature. (You’re over-assimilating to one species of cherry-picking). It’s possible – and I submit more possible – that Briffa has simply over-selected series where ring widths go strongly up at their close (without specifically considering correlation.)

    I realize that the net result will be similar. However, I think that the operational form of cherry picking will be even simpler than correlation picking (though the net result may be similar if the temperature trend is up). The distinction may seem pedantic, but I think that it’s a valid one.

  11. Whatever Kaufman may or may not have cherrypicked regarding treering data, the evidence I brought up here shows (to me) that he’s cherrypicked even more basically in ignoring those thermometer records which stand united against his proxy.

  12. Jeff Id said

    Thanks Steve,

    I understand your point and went again to find the quote on it I thought you had posted before. In fact it was from the Jacoby tree rings rather than Briffa Yamal.

    We strive to develop and use the best data possible. The criteria are good common low and high-frequency variation, absence of evidence of disturbance (either observed at the site or in the data), and correspondence or correlation with local or regional temperature.

    It’s difficult to keep them all apart. The Yamal series has too much similarity with an aggressive large scale mathematical sorting operation for me to guess that it was eyeballed but either could produce the same effect and you’re right we don’t ‘know’ it.


    They didn’t have to cherry pick temps, they simply used the overly secretive HadCRUT data.

    We used the spatially averaged summer temperature for all land area north of 60° latitude from the CRUTEM3 data series (14).

    That’s what makes this paper so difficult for a skeptic to accept, it has layers of climatology approved data all containing known problems many of which are warming biased. Steve’s point that Yamal may not have been sorted mathematically but rather visually is actually worse.

  13. Jeff Id said


    That sounds interesting. I’ll be watching for your result.

  14. My post on Yamal is now up. The stated reasons for the Yamal sorting are that they selected long (200-400 trees). However, a sensitivity analysis using a Schweingruber sample from the Yamal with an unselected population yields a totally (“remarkably”) different result. Take a look.

    It’s possible that the picking was not done visually but for age – but the motives hardly matter. The results speak for themselves.

  15. #12 yes, Jeff. Right now I’m rolling up my sleeves for a third project that can build further on older temperature records to bridge the unholy splice point between local record-keepers who trusted the evidence of their own instruments, and city record-keepers who trusted “Science”. I’m going to try to bypass CRU and GISS to get as many and as full old UK records as possible. I’ve realized that the records GISS supplies are sometimes (or often – or always) truncated at BOTH ends. And that’s in addition to the universal skewing. But at least, half a record (GISS) is better than no record (CRU).

    It looks like the Divergences precede the discovery and harvesting of Russian gas.

  16. At Climate Audit, Geoff Sherrington says I’d be trying to crack the CRU temperature problem as well as the dendro problem, because in combination they might tell a stronger story than each alone.

    Steve snipped my ref. there to this post, understandably as OT for his post – yet Geoff’s point is important IMO. And I hope this work helps to flush out that “unavailable” CRU temperature data into the public domain.

  17. Kenneth Fritsch said

    My analyses are in essential agreement with Lucia’s comments.

    My most recent post at CA is awaiting review for posting. K09 authors have presented discussions that are limited to temperature anomalies for the whole of the Arctic area and fail to talk about the local and regional temperature trend differences. I think the authors do this since the CPS method assumes that all proxies are responding to the same change climate conditions. They also assume a good correlation of summer temperature anamolies to the annual ones so that they can connect with an importance to AGW policy.

    My break point analysis of the proxy, composite and instrumental temperature series also fail to confirm the long downward sloping handle of the hockey stick and than an abrupt 50 year upward slope for the blade as proposed by the K09 authors.

  18. Jeff Id said

    #17, I’ll look for it.

  19. Steve Mosher at the CA “divergence” thread put a spot on the Google map for the Yamal treering work. The trees here are in a broad shallow valley (2 km wide) in what seems to be an essentially very flat landscape, where the river meanders and frequently changes its meanders. Treeline limits are sharply defined in this whole area, which suggests to me that temperature is NOT the deciding factor, even if it is incidental. Moisture may come into it though since one valley is barren, it doesn’t seem to be the sole factor.

    Just thoughts.

  20. Kenneth Fritsch said

    Jeff ID my post was posted at Post #95 at

    I wanted to post some Arctic temperature data that I had obtained from the KNMI web site that in my mind puts a different spin on the K09 reconstruction then do the authors of that paper.

  21. […] That study was tremendously well done, with over 2000 cores, seemed pretty germane to the issues of paleodendroclimatology we’ve been discussing as of late. Jeff Id touched on it breifly at the Air Vent in Circling Yamal – delinquent treering records? […]

  22. thanks a lot .

  23. thanks a lot ..

  24. […] The ultimate example of the use of outliers in climate science has to be the Yamal hockey stick, where one tree came to represent the entire global temperature (see also: here). […]

  25. […] The ultimate example of the use of outliers in climate science has to be the Yamal hockey stick, where one tree came to represent the entire global temperature (see also: here). […]

  26. […] The ultimate example of the use of outliers in climate science has to be the Yamal hockey stick, where one tree came to represent the entire global temperature (see also: here). […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: