Taimyr and Yamal Location Maps

The following two Google maps show Taimyr and Yamal on consistent scales, together with Schweingruber sites in the area.

The Taimyr chronology in Briffa 2000, as you may recall, not only didn’t have HS, but had a notable divergence problem.

I’ve tried to accurately transcribe onto this location map the Naurzbaev 2002 sites (subfossil – white circles; living – three yellow icons), the Schweingruber sites (green). Briffa 2008 reported the addition of the Avam site (yellow labeled), about 400 km from the center of the Taimyr samples. They did not report the addition of the Schweingruber Balschaya Kamenka site relatively near Avam.

[This is what I've figure out so far. The precise network used in Briffa 2000 remains unreported. I can sort of guess by crosschecking the network in Esper 2002, obtained through quasi-litigation at Sciencemag, but there are some puzzles. Briffa 2008 contains no metadata as to which site any given core belongs to.]

There are some Schweingruber sites that seem far more obvious additions to Taimyr than Balschaya Kamenka: for example, the Schweingruber Kotuy River and Kotuykan River sites are slightly uphill from the Naurzbaev Kotuy River samples. The Schweingruber Novoja Rieka site seems to be almost co-located with a Naurzbaev location.

Why did Briffa go all the way to Balschaya Kamenka to add a Schweingruber site, while passing over the nearby sites? In the case of Yamal, where he also omitted a nearby site, Briffa said that they didn’t “simply” didn’t consider the nearby Khadyta River, Yamal site. Perhaps the same thing happened here.

Next here is a corresponding map for Yamal on precisely the same scale. Briffa’s online article made a bit of an issue of the fact that the Schweingruber Khadyta River, Yamal site was “slightly to the south” of the Porza and Yadaya sites – mentioning this not once but twice. However, Khadyta River is obviously far closer to the Porza and Yadaya sites than Avam or Balschaya are to Taimyr.

Polar Urals is also closer to Yamal than Avam is to Taimyr (I’ve got two slightly different latitudes for this site in my data collations – the present NCDC location has a latitude of 66 50N (but my collation of the Schweingruber locations once at NCDC but no longer there has a latitude of 67 50N). For present purposes, both locations are closer to Yamal than Avam is to Taimyr. Briffa said that he didn’t include Khadyta River in the Yamal RCS, because he “simply didn’t consider it”. He didn’t report on his deliberations regarding Polar Urals. Was it not included in the RCS because Briffa “simply didn’t consider it” or for some other reason?

These are elementary and obvious questions. Why are some sites included and some excluded? What are the scientific principles involved? Gavin Schmidt accused me of “randomly” picking a site off the internet, but that is not what I did. Given the precedent use of a Schweingruber site at Taimyr, I looked for the closest Schweingruber site to Yamal. In contrast, Briffa provided no guidance as to the basis for including one Schweingruber site rather than another. Did Briffa “randomly” pick Schweingruber sites to add – right now, we have no way of knowing?

Advocates at realclimate and elsewhere urge us to defer to Briffa’s choices. If Briffa’s articles are to be viewed as a branch of prophetic or oracular literature, then followers are, of course, entitled to defer to his choices.

However, if Briffa’s articles are to be considered as scientific articles, then the selection criteria need to be clearly stated and it should be possible to verify the choices. At present, I am not saying that there were no such rational criteria, only that the articles do not say what they were and, thus far, I have been unable to deduce what the criteria were.


32 Comments

  1. Barclay E MacDonald
    Posted Nov 6, 2009 at 12:50 PM | Permalink

    Nice job with the maps and accompanying explanation. It is a very helpful start to providing layman like me with a clearer picture of the Briffa studies and their progeny.

    “Briffa 2008 contains no metadata as to which site any given core belongs to.”….”Why did Briffa go all the way to Balschaya Kamenka to add a Schweingruber site, while passing over the nearby sites? In the case of Yamal, where he also omitted a nearby site, Briffa said that they didn’t “simply” didn’t consider the nearby Khadyta River, Yamal site.”…”These are elementary and obvious questions. Why are some sites included and some excluded? What are the scientific principles involved?”

    snip – please do not make this sort of editorial comment. You know that this is against blog policies.

  2. steven mosher
    Posted Nov 6, 2009 at 12:59 PM | Permalink

    Thx. Every time I come back to this I still have trouble keeping all the various sites and locations
    right so many details like core counts, paper its used in, start and end dates for the cores in the area.
    Steve is all this info at the tree ring database?


    Steve:
    Only part of it. The tree ring data used in the multiproxy studies has for the most part been conspicuously absent from the ITRDB i.e. Yamal, Taimyr, the Tornetrask update, the Graumlich foxtails (until my campaign),… However, the Schweingruber data is at ITRDB. These maps are a collation from a couple of sources.

    The Taimyr locations are from the location map in Naurzbaev et al (Hol 2002) ; like Yamal, Taimyr is not archived at the ITRB. The Yamal locations are from Briffa (CRU, Oct 2009) and were not previously exactly known. I collate “information” tables and use these for extracting details. These require being able to read a lot of original data, but when done once, stay fixed. Likewise, I make tables of proxies used in each study and refer back to them. Sometimes it takes a while to identify a proxy, but again once done, I can retrieve the information.

    • steven mosher
      Posted Nov 6, 2009 at 4:59 PM | Permalink

      Re: steven mosher (#2), It would be cool if in your data files we had a little HTML table:

      1. Name of site
      2. links to papers using it
      3. Links to CA articles about it
      4. Links to actual data file
      5. Species
      6. location lat lon
      7. link to google map
      8. number of cores.
      9. Chronology ( earliest date, latest date)

      just saying. not like you dont have enough to do

      • MrPete
        Posted Nov 6, 2009 at 5:10 PM | Permalink

        Re: steven mosher (#8),
        No promises, but I’ve been thinking about things like that lately… cross-links, drill-downs, auto-links, etc etc etc… and what it would take to add a layer of “metadata” to CA. Seems there Oughta Be Tools that can do some of that already. Let’s go Unthreaded on this if there’s interest…

      • RomanM
        Posted Nov 6, 2009 at 5:11 PM | Permalink

        Re: steven mosher (#8),

        Sounds like a project for a certain s. mosher… ;)

      • Arthur Edelstein
        Posted Nov 6, 2009 at 11:05 PM | Permalink

        Re: steven mosher (#8),
        Maybe this table would be best posted on Climate Audit 101 and updated as needed?

  3. Joe Black
    Posted Nov 6, 2009 at 1:04 PM | Permalink

    Historic Proxy selection literature:

    Springfield, et al.(1964?)

    Coombs, Stewart, Lennox, Chin & Toomey (1979)

    Longmuir, Mitchell, Faulkner, Wood and McKeown (1976)

    Rucker, Bryan, Fleber, Sonefeld (1995)

    • deadwood
      Posted Nov 6, 2009 at 2:41 PM | Permalink

      Re: Joe Black (#3),
      The first and fourth references provide original criteria, while the second and third are reiterations of the first.

      Given the more recent publication of the fourth, I would normally consider it to be the most relevant to current research, but there still remain some validity to the first, particularly the attention to stylistic elements of the subject matter.

  4. Solomon
    Posted Nov 6, 2009 at 1:08 PM | Permalink

    snip – let’s not go there. This is just piling on.

  5. Barclay E MacDonald
    Posted Nov 6, 2009 at 1:47 PM | Permalink

    Snip is fine. You’re right! Sorry to waste your time. I look forward to any corrections, additions or deletions to the above maps that may become necessary or important.

  6. a reader
    Posted Nov 6, 2009 at 4:18 PM | Permalink

    H&S say they avoid south to north flowing rivers to avoid driftwood. Khadyta looks like it flows south to north. Could Briffa have avoided using it for this same driftwood problem?

  7. Steve McIntyre
    Posted Nov 6, 2009 at 10:54 PM | Permalink

    I’ve transferred OT complaining about CA to Unthreaded.

  8. Posted Nov 7, 2009 at 12:35 AM | Permalink

    FWIW I created a google map of most of the relevant sites. If you want I can either add more, export it or add you to the authorized users.

    • thefordprefect
      Posted Nov 7, 2009 at 7:37 AM | Permalink

      Re: FrancisT (#13),

      Useful map. Those in the headers being non-interactive are next to worthless, sorry!

      Perhaps McIntyre can provide Francis with the same location data as he used?

      If you check out the areas in Francis’ map with zoom then some of the areas refered to by McIntyre are just not the same as Yamal.

      Yamal looks wet and boggy, Polar Urals is in the hills no bogs, most of the others do not show up the same terraine as Yamal however Khadyta River could be ok. Until an interactive map of all sampling locations is provided it is not sensible saying which site is similar to which! just basing similarity of location just on latitude is obviously not sensible. Lat/long, elevation, soil type, water availability, etc. must all come into the equation.
      Mike

      • bender
        Posted Nov 7, 2009 at 8:10 AM | Permalink

        Re: thefordprefect (#16),
        Hold down Ctrl and roll your mouse wheel. Now you can see those titles.

      • Ivan
        Posted Nov 7, 2009 at 8:28 AM | Permalink

        Re: thefordprefect (#16),

        Hmm. If I correctly understand, your assumption is that Briffa took certain data on the basis of your above cited considerations, i.e. similarity of different sites, and not on the basis of some other, more banal criteria. You may be right (we’ll never know since Briffa didn’t explain), but would not be that a really amazing pre-established harmony between the “appropriateness” and “Hockey stickiness” of various data sets? I think the probability of that is very low.

        • bender
          Posted Nov 7, 2009 at 9:23 AM | Permalink

          Re: Ivan (#18),

          ford says:

          just basing similarity of location just on latitude is obviously not sensible. Lat/long, elevation, soil type, water availability, etc. must all come into the equation.

          This is what CA has been advocating since day 1: adequate metadata, fully disclosed site selection criteria, etc. How can Briffa assert “A” is representative and “B” anomalous without referring to the sort of metadata that Steve is just now sleuthing out? ford obviously gets it. Why doesn’t Briffa?

    • Posted Nov 7, 2009 at 8:42 AM | Permalink

      Re: FrancisT (#13), Francis, that’s a lovely clear presentation, but it’s not clear which of the blue points is which, so one cannot check this against Steve’s. Can you update it? Is Steve more correct/detailed?

      • Steve McIntyre
        Posted Nov 7, 2009 at 10:57 AM | Permalink

        Re: Lucy Skywalker (#19),

        I used the Schweingruber lat-longs as archived at ITRDB. Look, this is climate science and just because this is what is archived at ITRDB doesn’t mean that it is where they were sampled. Schweingruber’s lat-long’s for Pike’s Peak near Almagre are about 60 miles off, a point made at the time.

        If the archived Schweingruber lat-longs don’t match the geographic locations of the name, there are no easy ways of figuring out where the sample was really taken. All you can do is plot the locations of rival interpretations and leave the uncertainty visible for the reader.

  9. Peter Dunford
    Posted Nov 7, 2009 at 6:35 AM | Permalink

    Re: Joe Black #3
    Have the links been hacked? They all direct to versions of “I only want to be with you” on YouTube.

  10. Peter Dunford
    Posted Nov 7, 2009 at 6:36 AM | Permalink

    Oh wait, slow this morning…

  11. a reader
    Posted Nov 7, 2009 at 10:40 AM | Permalink

    I’m having trouble finding the actual Khadyta river on Google Maps, but logically it must be part of the Ob River system and thus be affected by that enormous south-north flowing system. According to wikipedia, Barnaul is ice-bound from early Nov.-the end of April, and Salekhard from the end of Oct.-beginning of June. The Bay of Ob is shallow and frozen over from Oct.-July.

    Now imagine what spring and early summer must be like in this area when melt waters start comming up from the south. The flooding must be prodgious when ice dams are blocking the outlets. Unless the Russians have done lots of control damming since the 70’s and 80’s, I assume this area is one big swamp in the summer. At one time the USSR was contemplating reversing the flow of some of the rivers in this system and then sending the water south to the Aral and Caspian Seas to facilitate irrigation.

    • EW
      Posted Nov 7, 2009 at 1:40 PM | Permalink

      Re: a reader (#21),
      I’m having trouble finding the actual Khadyta river on Google Maps, but logically it must be part of the Ob River system?

      Look at the map in Fig. 1 of Hantemirov and Shiyatov 2002. There are all three rivers and sampling sites. You may then compare it with Google Maps – they are rather small and therefore not named. All three flow from North to South.

      • ianl8888
        Posted Nov 7, 2009 at 3:56 PM | Permalink

        Re: EW (#27),

        When using Google Earth, it always helps to turn the altitude record on (then you can easily figure out the river flow directions). I also tend to use the UTM gridding to locate and pinpoint areas and spots. Similarly, using the metric system, as the Russians do, reduces “hunt” time for specific areas

      • Posted Nov 8, 2009 at 4:01 AM | Permalink

        Re: EW (#27), In Fig. 11 of Hantemirov and Shiyatov 2002, where they present the temperature reconstruction, the 20th century is colder than, say, the 18th. This is a Yamal reconstruction. If the Yamal hockey-stick of Briffa et al. is based on the same data, how could they have reached a different conclusion? Is it because they used a different method of age detrending compared to H&S 2002? Or what? What do Hantemirov and Shiyatov think of that? Did they admit somewhere that their 2002 reconstruction was in error?

        I understand one should dig the CA and literature for oneself to find out, but at the same time this (and similar) question (even if trivial for those who know the story from A to Z) might be informative to other CA readers as well, because surely many are like me new to the problem, but would like to know more. A brief hint would be greatly appreciated.

        Steve: H and S use corridor standardization which, as a method, removes centennial variability. Briffa’s RCS has centennial variability, but the extent to which it contains artifacts in any case is the issue.

  12. Erasmus de Frigid
    Posted Nov 7, 2009 at 10:58 AM | Permalink

    According to an article in The Encyclopedia of Earth with contributions from Juday, D’Arrigo, Jacoby and others
    entitled Tree Rings and Past Climate in the Artic:

    http://www.eoearth.org/article/Tree_rings_and_past_climate_in_the_Arctic

    “…tree-ring chronologies were highly correlated across distances up to 200 km (up to 500 km in northern regions)..”

  13. Steve McIntyre
    Posted Nov 7, 2009 at 10:59 AM | Permalink

    You can examine my collation of Schweingruber details as follows:

    download.file(“http://data.climateaudit.org/data/tree/schweingruber.details.tab”,”temp.dat”,mode=”wb”); load(“temp.dat”)

    then do something like:

    temp = details$lat>60 & details$long>88 & details$long<112
    details[temp,]

  14. Kenneth Fritsch
    Posted Nov 7, 2009 at 11:12 AM | Permalink

    I have little to add to this discussion, but to note that I think Steve M’s comment below pretty much sums up the Briffa situation and others like it in climate science.
    .

    The selection criteria and how that process should properly be performed is the all important issue and should be dominating the discussion. And that is why I get so turned-off when the discussion veers off into hurt feelings. Cherry picking is cherry picking, whether intentional or unintentional, and it carries a statistical price. Without an a prior selection criteria, the selection has to be considered cherry picking, whether performed by a saint or a not-so-saintly someone.
    .

    Did Briffa “randomly” pick Schweingruber sites to add – right now, we have no way of knowing?

    Advocates at realclimate and elsewhere urge us to defer to Briffa’s choices. If Briffa’s articles are to be viewed as a branch of prophetic or oracular literature, then followers are, of course, entitled to defer to his choices.

    However, if Briffa’s articles are to be considered as scientific articles, then the selection criteria need to be clearly stated and it should be possible to verify the choices

  15. Steve McIntyre
    Posted Nov 7, 2009 at 11:19 AM | Permalink

    Here is a famous comment by Keynes (1940) about Tinbergen’s econometric models:

    No one could be more frank, more painstaking, more free from subjective bias or parti pris than Professor Tinbergen. There is no one therefore so far as human qualities go, whom it would be safer to trust with black magic. That there is anyone I would trust with it at this present stage or that this brand of statistical alchemy is ripe to become a branch of science, I am not yet persuaded.

    Does Briffa’s RCS rise above “statistical alchemy”? That’s the question.

    • Layman Lurker
      Posted Nov 8, 2009 at 11:18 AM | Permalink

      Re: Steve McIntyre (#26),

      Sorry for OT Steve.

      The parallel with econometrics is something I’ve often thought about when reading up on the latest “statistical” methods used in a climate science paper being dissected. Econometricians are the stewards of the analytical methods used in economics. As a consequence these methods are less likely to be hijacked by economists with political agendas.

  16. Don Keiller
    Posted Nov 8, 2009 at 10:49 AM | Permalink

    Steve, have you seen this paper? It is based on tree-lines and paints a radically different picture of past climate at Yamal.
    The Holocene 12,6 (2002) pp. 717–726
    A continuous multimillennial ring-width
    chronology in Yamal, northwestern
    Siberia
    Rashit M. Hantemirov* and Stepan G. Shiyatov

    Steve: ? This article has been referred to in almost every post. The chronology was done using corridor standardization and this is what accounts for the chrnoology difference – this point discussed in one of the early posts.

    • bender
      Posted Nov 8, 2009 at 1:23 PM | Permalink

      Re: Don Keiller (#30),
      People will be wondering what “corridor standardization” is and why it removes low-frequency variation. It’s called “corridor” because if the plot the raw average chronology alonng with the overall maximum values above and overally minimum values below, the mean chronology appears to travel through a winding “corridor” of max and min values. If you subtract (min+max/)2 from the mean you remove a ton of low-frequency variation attributable to the winding corridor. The result looks like white noise.

Follow

Get every new post delivered to your Inbox.

Join 3,322 other followers

%d bloggers like this: