Climate, paleoclimate, huevos rancheros, and general asymmetry

Talk about Tiljander

with 9 comments

Stoat moderates a discussion on the proxy records from Tiljander with a higher-than-normal signal-to-noise ratio (due, in part at least, to considerable moderation) than is commonly found on the blogs.   A good case study on the path to how this might be done more regularly …

Written by delayedoscillator

November 3, 2009 at 4:41 am

Posted in Paleoclimatology

9 Responses

Subscribe to comments with RSS.

  1. DO,

    1. It was a pleasure to stumble onto this blog. You take serious methodological issues seriously, which is refreshing. From here, I found the blog of one of your commenters, Jeff Id, who seems to share that trait (even if sharing little in terms of specific interpretations). I hope you and he manage to co-exist, and continue to point out possible holes in one another’s methods and reasoning. I surely will not comment, as I lack the expertise to make a contribution. But it can be very helpful to eavesdrop, in order to get a sense of where the points of disagreement are. And where there is some alignment of views.

    2. It is also refreshing to find a climate-related blog where remarks pertaining to motive are not allowed, across the board. That seems to improve the S/N ratio.

    3. Stoat has had three posts on Tiljander; you link to the second one. The third is Tiljander, again. I’ve been heavily represented in those threads, so thought a meta-comment here might be on topic.

    This is my first WordPress comment submission and there is no Preview, so I’ll press Submit and (hopefully) continue…


    November 6, 2009 at 9:42 pm

  2. 4. Re: “A good case study on the path to how this might be done more regularly …” — it depends on one’s objectives.

    Let me back up. I came to the Tiljander controversy a week or so ago, because I wandered onto some blog that linked Mann et al’s rebuttal to McIntyre & McKitrick’s Comment, which criticized Mann et al (2008, PNAS). One of McIntyre’s points was straightforward: Mann et al’s “non-dendro network uses some data with the axes upside down, e.g., Korttajarvi sediments, which are also compromised by agricultural impact”.

    Mann et al’s response was intriging. “The claim that “upside down” data were used is bizarre. Multivariate regression methods are insensitive to the sign of predictors. Screening, when used, employed one-sided tests only when a definite sign could be a priori reasoned on physical grounds. Potential nonclimatic influences on the Tiljander and other proxies were discussed in the SI, which showed that none of our central conclusions relied on their use.”

    So there’s a straightforward detective story. McIntyre says “proxy misused,” Mann rebuts “not so.” Is it possible for an interested observer with broad scientific knowledge to make heads and tails of this conflict?

    As you will know from reading Stoat, my own conclusion on this narrow point is, “yes.”

    Which brings me to comment on your comment on the Stoat threads. As somebody who was (hopefully) data-driven, and who came to conclude that on the narrow question of the use of the Tiljander proxies that Mann is in the wrong, I found Stoat moderation policy to be somewhat onerous. Those threads did not constitute a level playing field. Technically-themed comments from one side but not from the other were edited, or failed moderation. Post-submission comments added by the blogger–as is his prerogative–may have rendered the less favored side’s arguments less readable.

    Editing: This was clipped to this.

    This failed moderation on the basis of lack of substance.

    Readability: This became that.

    All that said, I recognize that “Full and fair airing of a defined technical issue, with emphasis on logical arguments, appropriate citations, and a restrained, cautious tone” is not the only possible objective for any particular discourse.

    [Oops, spam filter caught this because of too many links]


    November 6, 2009 at 10:19 pm

  3. … and on to Part Three of my comment, then.

    DO, so having read the three Stoat threads on Mann et al (2008)’s use of the Lake Korttajarvi varve proxy series, what do you think?

    * There’s Stoat blogger WMC and some well-informed regular commenters–many of whom are also regulars at–asserting variations on the theme that there really isn’t a problem. Or that if there was a problem, it’s been resolved. And/or that it doesn’t affect the paleoclimate reconstructions in any case.

    * There’s a second set of well-informed commenters–many of whom are also regulars at–asserting that there are problems with Mann et al’s use of the varve proxies, specifically in their calibration (4/4) and orientation (~3/4). And that the problems should be resolved by Corrections to the peer-reviewed literature. And that wrongly-used proxies may or may not have affected the paleoclimate reconstructions; one can’t tell. And/or that irrespective of the effect, such errors in the literature shouldn’t be left to stand.

    WMC’s moderation policies have been generous enough, or variable enough, that my own views have been adequately expressed. Though those threads are not a level playing field, in my opinion.

    By the way, I only started commenting at as a result of Stoat‘s moderation. I would have preferred, but their reputation for snipping dissenting comments deterred me.


    November 9, 2009 at 4:25 am

  4. DO,

    Thanks for allowing the prior dissenting remarks. It sets a good precedent for the paleoclimate community to foster open discussion, even on a “meta-issue” like comment policy. To be clear, my 4:25am, above, was written when the preceding 10:19pm was still in spam purgatory.

    Do you have an opinion on the use of the Lake Korttajarvi varve series in Mann et al (2008)?

    [snip — this is a bit more meta than I really want this discussion to go. Stick to science, as you did much of the time at Stoat]


    November 10, 2009 at 12:02 pm

    • AMac,

      There seem to be three issues (two interesting) with Tiljander’s proxies.

      [1] Were they used ‘upside down’ (that is opposite from the interpretation of the authors)? I’m with Ari Jokimäki on the original Stoat thread in not being convinced this is the case. They are plotted as they appear in the original paper in the Supplemental Material. Without knowing the weights applied in the regression, it is difficult to know for sure how these proxies influenced the reconstruction precisely (can you point me to a trustworthy source who has determined the weights from the PNAS paper’s code?). The new Figure S8 seems to show that if you pull out these and other proxies and all the tree-rings that you get a warmer LIA and a colder MCA, which argues that the Tiljander proxies didn’t influence the variance much at all. There is another potential concern, but lets put that in [2], below:

      [2] Non-climatic disturbance of the proxies. This to me is a larger concern. Kaufman et al. truncated their binned Tiljander proxy record to avoid this, which makes sense to me. If the recent rise in many of the lake sediment proxies from Lake Korttajarvi is the result of non-climatic human influence, this will have an effect on the calibration.

      [3] The proxy interpretation itself. In the original paper, a lot goes into interpreting the Lake Korttajarvi proxies in terms of climate conditions — the x ray density, the light sum and dark sum, and the stratigraphy (their Figure 5, all the way to the left). Tiljander et al interpret the light sum and density and varve thinkness as reflecting colder conditions (so that thinner varves, lower density = warmer), but when I first looked at their Figure 9, I interpreted it to suggest that dark sum and light sum covary — the caption to that figure seems to indicate that DS is calculated with respect to LS, however, so it is not entirely clear to me the relationship between the two components of the varve. In any case, because of the human influence to which the authors ascribe the recent changes in the lake, it isn’t possible to validate their paleolimnological interpretation against instrumental data.

      Finally, sorry for the [snip], but I’m not interested in hosting such discussions nor opining myself about the sociology of science.


      November 10, 2009 at 1:38 pm

  5. > Finally, sorry for the [snip],

    I thought my points were relevant and logical extensions, but it turns out not to be a place you want to go. To me, as a commenter, that use of [snip] is quite acceptable. It’s different from using your power as host to “tilt the playing field” on the specified topic of interest so that some parties get a favored hearing.

    > [1] Were [the varve data] used ‘upside down’?

    Starting with areas of agreement: I agree with you and Ari Jokimäki (original Stoat thread) that the raw values of the four series as used by Mann et al were in the same orientation as was presented in Tiljander et al (2003). E.g. if X-Ray Density for the 1850 varve is “97” in Tiljander, it is “97” and not “-97” in Mann et al (2008).

    Where I differ is at the next step: Mann et al’s assignment of climate information to the varve data series.

    It seems unnecessarily tedious to go into detail as I do in the remainder of this comment. However, this is the point at which the Stoat discussion foundered. Further, I don’t see that Ari Jokimäki addressed this issue in the Stoat thread.

    X-Ray Density. Varve XRDs increased from the 981-1040 period to the 1041-1140 period. (The 20-year average values starting 981 are 61, 56, 55. For 1041 on, they are 65, 66, 70, 72, 76.) Does this trend suggest “warming” or “cooling”?

    Tiljander et al interpreted higher XRD values as reflective of a more severe climate, and that more severe means cooler.

    Mann et al do not explicitly state their interpretation. However, there are lines of evidence that their reconstruction took higher XRD to mean “warmer.” This would be the “Upside-Down with respect to Tiljander” orientation.

    (1) The description of the calibration procedure in Mann et al SI, page 2. Over 1850-1995, global temperature rose. Mann et al searched for proxies that showed rises over this time that might correlate to this global rise. The 20-year averages of XRD 1861-1980 are 100, 92, 99, 110, 85, and 122. The 1981-85 average is 126. I don’t know the post-1985 numbers, but Tiljander Fig. 5 indicates that the 11-year running average rose over 150. Mann’s description of the calibration and validation procedure very strongly suggests the use of the {higher XRD :: warmer} orientation.

    (2) Mann et al’s Figure S9 displays the long-range proxies that passed the screening step. tiljander-2003-xraydenseave is shown oriented with larger values nearer the top of the page, consistent with {higher XRD :: warmer}. To my knowledge, nobody has credibly suggested that any of the proxies in Fog. S9 was interpreted by Mann et al in the fashion of {lower :: warmer}.

    (3) In their Comment, McIntyre and McKitrick alleged that Mann et al used Tiljander proxies in an Upside-Down orientation. In their Response, Mann et all could have clearly indicated that they did not do so. Instead, they made a cryptic statement. (For back-and-forth on the meaning of “bizarre,” see the Stoat threads.)

    My summary of the Lake Korttajarvi proxies follows.

    tiljander-2003-xraydenseave is X-Ray Density, Tiljander Fig. 5, units are grey values. NOAA archive, annual values, units are grey values.
    Tiljander – {lower XRD :: milder climate}
    Mann – {lower XRD :: lower temperature}

    tiljander-2003-lightsum is LS, Tiljander Fig. 9, units are grey values. NOAA archive, Mineral, annual values, units are mm*1E-4.
    Tiljander – {lower LS :: milder climate}
    Mann – {lower LS :: lower temperature}

    tiljander-2003-thicknessmm is Varve Thickness, Tiljander Fig. 5, units are mm.
    Tiljander – did not offer interpretation
    Mann – {lower thickness :: lower temperature}

    tiljander-2003-darksum is DS, Tiljander Fig. 9, units are grey values. NOAA archive, Organic, annual values, units are mm*1E-4.
    Tiljander – {higher DS :: milder climate}
    Mann – {higher DS :: higher temperature}

    Other points addressed briefly in the next comment.


    November 10, 2009 at 5:02 pm

    • > it is difficult to know for sure how these proxies influenced the reconstruction precisely… The new Figure S8 seems to show [features which argue] that the Tiljander proxies didn’t influence the variance much at all.

      Agree. However, as I (tried to) say at Stoat, “The Figures (plural) S8a that show those reconstructions have issues of their own… But here, in these Stoat threads, informed parties can’t agree on answers to simple questions… So I’ll decline to extend this comment to cover the broader “Results” argument.”

      > can you point me to a trustworthy source who has determined the weights from the PNAS paper’s code?

      Sorry, no.

      Re: your point on Kaufman et al — Agree. In addition, Kaufman’s correction puts at least some of his Lake Korttajarvi proxies in Upside-Down orientation with respect to Mann et al. Logic suggests that where they disagree on orientation, one or the other must be wrong.

      Re: your point on DS — agree. It’s unclear to me that DS is a sufficiently independent proxy to be given a weight separate from LS.


      November 10, 2009 at 5:14 pm

  6. {pedantry alert}
    To avoid ambiguity, I should have written in the 5:02pm comment supra —

    (1) The description of the calibration procedure in Mann et al SI, page 2. Over 1850-1995, global temperature rose. Mann et al screened for informative long-term proxies. For a proxy to pass the screen, it would have to either rise or decline over the 1850-1995 period, and show significant correlation to the global rise in temperature.
    {/pedantry alert}


    November 10, 2009 at 6:44 pm

  7. [snip – needless to say, I won’t be publishing stolen emails you attempt to reproduce here]


    November 21, 2009 at 6:33 pm

Leave a Reply

Please log in using one of these methods to post your comment: Logo

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

Google photo

You are commenting using your Google 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: