AMac: Upside Down Mann Lives on in Kemp et al 2011

AMac:

Yesterday, Kemp et al. 2011 was published in PNAS, relating sea-level variation to climate over the past 1,600 years (UPenn press release). Among the authors is Prof. Mann. (Kemp11 is downloadable from WUWT.) Figs. 2A and 4A are “Composite EIV global land plus ocean global temperature reconstruction, smoothed with a 30-year LOESS low-pass filter”. This is one of the multiproxy reconstructions in Mann et al. (2008, PNAS). The unsmoothed tracing appears as the black line labelled “Composite (with uncertainties)” in panel F of Fig. S6 of the “Supporting Information” supplement to Mann08 (dowonloadable from pnas.org).

This is one of the Mann08 reconstructions that made use of the four (actually three) uncalibratable Tiljander data series.

As scientist/blogger Gavin Schmidt has indicated, the early years of the EIV Global reconstruction rely heavily on Tiljander to pass its “validation” test: “…it’s worth pointing out that validation for the no-dendro/no-Tilj is quite sensitive to the required significance, for EIV NH Land+Ocean it goes back to 1500 for 95%, but 1300 for 94% and 1100 AD for 90%” (link). Also see RealClimate here (Gavin’s responses to comments 525, 529, and 531).

The dependence of the first two-thirds of the EIV recon on the inclusion of Tiljander’s data series isn’t mentioned in the text of Kemp11. Nor is it discussed in the SI, although it is an obvious and trivial explanation for the pre-1100 divergence noted in the SI’s Figures S3, S4, and S5.

Peer review appears to have been missing in action on this glaring shortcoming in Kemp11′s methodology.

More than anything, I am surprised by this zombie-like re-appearance of the Tiljander data series — nearly three years after the eruption of the controversy over their misuse as temperature proxies!


57 Comments

  1. Posted Jun 21, 2011 at 4:22 PM | Permalink | Reply

    Ya, The first thing I did was read the SI to look for Mann’s handi work.

    It’s quite an interesting read, especially about the divergences

  2. Howard
    Posted Jun 21, 2011 at 4:22 PM | Permalink | Reply

    This explains why so much effort has been put to defending the Hockey Stick at all costs. Who knows how many other future paleo-climate studies will use the stick as a divining rod.

  3. pax
    Posted Jun 21, 2011 at 4:26 PM | Permalink | Reply

    When reading about paleo I always end up having to decide between epic incompetence or deliberate manipulation. Still undecided.

    • stan
      Posted Jun 22, 2011 at 7:40 AM | Permalink | Reply

      Pax,

      Incompetence and manipulation are not mutually exclusive. Problems caused by the former are often “solved” by resort to the latter.

  4. Hoi Polloi
    Posted Jun 21, 2011 at 4:56 PM | Permalink | Reply

    I’m sure that Mann has registered his Hockey Stick at the US Patent and Trademark Office .

  5. Jeremy
    Posted Jun 21, 2011 at 5:03 PM | Permalink | Reply

    They’re not even misused properly, as the divergence of the temperature proxy and the blade end of the sea-level proxy data are out-of-sync such that the start of sea level rise predates most of the man-made CO2 that is supposedly responsible for the temperature rise. Not to mention the second 500 years of the temperature proxy show a decrease, while the sea-level plot shows a small increase. So they are lock-step-in-sync in the 1800s and onward, but they had a different sign for about 500 years at the outset…

    I mean what I say, except when I dont.

  6. Posted Jun 21, 2011 at 5:27 PM | Permalink | Reply

    Hmmm … well, if nothing else, I suppose one must give Mann credit for his dedication to one of the key principles of conservation: Recycle. Recyle. Recyle.

  7. Posted Jun 21, 2011 at 5:36 PM | Permalink | Reply

    Alas, the links didn’t come through when Steve McI turned my comment into a post. An HTML-ified version is up at my blog as The Tiljander Data Series Appear Again, This Time in a Sea-Level Study.

    I had wanted to inquire about this matter at the relevant RealClimate.org post by Stefan Rahmstorf, but my name or email or IP address seems to lead to automatic rejection of submitted comments (or perhaps the site’s software is on the fritz).

    In particular, Kemp et al’s S.I. Figures S3, S4, and S5 show that the correlation of reconstructed sea level with reconstructed global temperature is pretty good from the present to ~1100 AD, then gets progressively worse as one goes back to the starting point of 500 AD. So, let’s assume that the basic idea of the paper is good, and that the t’s are crossed and the i’s are dotted. In that case, the authors have to explain this pattern of divergence. And they address the point in the SI.

    An obvious possible explanation is that, as one moves back from 1500 AD to 500 AD, the Tiljander ‘proxies’ assume more and more weight in the EIV reconstruction that Kemp11 employ. The inclusion of Tiljander could make the early segments of the recon “worse-warmer” or “worse-colder” — the latter, I think. (That’s the one that would account for the divergence, in this scenario.) It was disappointing to see that this wasn’t addressed.

    After almost three years of Tiljander-inspired ruckus, I would have expected one or more of the editors or peer-reviewers to have picked up on this point.

    - – - – - – - – - -

    Anyway. Not to be too picky, but only two of the four (actually three) Tiljander data series are oriented by Mann08′s EIV and CPS reconstructions in an “upside down” manner. One is rightside-up with respect to Tiljander03′s interpretation, and Tiljander03 offered no interpretation for one (thicknessmm).

    The bigger problem is that all four (three) series are uncalibratable to the instrumental record, due to progressively greater contamination by local activities (e.g. farming, eutrophication, road-building), starting in the mid-18th century and continuing through the late 20th century.

    Also, the opinion of Tiljander and her co-authors notwithstanding, I haven’t encountered persuasive evidence that these varve series convey much of a temperature signal in the pre-1720 era. In either orientation.

    • Salamano
      Posted Jun 21, 2011 at 6:43 PM | Permalink | Reply

      “After almost three years of Tiljander-inspired ruckus, I would have expected one or more of the editors or peer-reviewers to have picked up on this point.”

      What if the editors decided only “true” “experts” were worthy enough to handle this publication; so Trenberth, Raypierre, and Steig just happened to be the only ones around who returned emails…

    • MikeN
      Posted Jun 22, 2011 at 10:27 AM | Permalink | Reply

      AMac, I;m not sure it matters to your analysis, but the CPS and EIV have a distinction where the EIV method is blind to the sign of the predictor. It will flip the Tiljander series to match the correlation, if Tiljander is fed in correctly. In the case of upside-down input, it still leaves you with upside down usage because the late stages show cold as higher values, and no flipping will occur.

  8. mpaul
    Posted Jun 21, 2011 at 5:41 PM | Permalink | Reply

    I can hear the NPR biographical tribute to Mann now:

    Like all great men of history, Mann’s genius is expressed in his highly recognizable and unique style. Similar to Johann Sebastian Bach’s repeated use of the musical cryptographic idiom (comprised of B flat, A, C, B natural), Mann’s repeated use of the Tiljander proxy reveals a deep and impressive intellect…

    /sarc

    • CV
      Posted Jun 28, 2011 at 3:01 PM | Permalink | Reply

      Except that Bach was a genius. They are similar in some ways regarding creativity.

  9. Posted Jun 21, 2011 at 5:42 PM | Permalink | Reply

    Yesterday, Kemp et al. 2011 was published in PNAS, relating sea-level variation to climate over the past 1,600 1,500 years (UPenn press release).

    My bad.

  10. KnR
    Posted Jun 21, 2011 at 5:50 PM | Permalink | Reply

    Tiljander data series, rubbish in reality but so very useful in practice .

  11. kim
    Posted Jun 21, 2011 at 7:52 PM | Permalink | Reply

    The Revolt of the Ash Heap of History.
    ==============

  12. Posted Jun 21, 2011 at 9:18 PM | Permalink | Reply

    I sure hope someone is going to compose a rebuttal paper, or at least a letter.

    • Doug in Seattle
      Posted Jun 22, 2011 at 12:04 AM | Permalink | Reply

      This has been tried before, but unfortunately the editors found the reply too long, then too vague, then too mundane and didn’t find it worthy of publication.

  13. Geoff Sherrington
    Posted Jun 21, 2011 at 10:08 PM | Permalink | Reply

    The part of Mia Tiljander’s paper that draws much attention is -
    “20th century. – In the 20th century the Lake Korttajarvi record was strongly affected by human activities. The average varve thickness is 1.2 mm from AD 1900 to 1929, 1.9 mm from AD 1930 to 1962 and 3.5 mm from AD 1963 to 1985. There are two exceptionally thick clay-silt layers caused by man. The thick layer of AD 1930 resulted from peat ditching and forest clearance (information from a local farmer in 1999) and the thick layer of AD 1967 originated due to the rebuilding of the bridge in the vicinity of the lake’s southern corner (information from the Finnish Road Administration). Varves since AD 1963 towards the present time thicken because of the higher water content in the top of the sediment column. However, the gradually increasing varve thickness during the whole 20th century probably originates from the accelerating agricultural use of the area around the lake.”

    Earlier in the paper, she notes the relative historical importance of alternations of mineral sedimentation – from ground yet to be grassed – and organic-rich varves, from either natural grass or agriculture. (The whole story of weather changes is more complex than my summary here). Unless the calibration period has had similar land use to eralier, the calibration cannot be fully valid, and not just because of disturbance or recent water retention in samples. The author says that land use changes have happened, as above, since older times. Therefore, there is limited potential for the inclusion of her findings in a wider reconstruction. In particular, the last few decades have essentially no information of use to palaeoclimate reconstructions, whether they are inverted or not. The earlier observations can be related qualitatitively to each other, but a quantitative derivation is tenuous.

    This comment is not dismissive of the hard work of the author; it merely notes that in subsequent uptakes, it’s worse than we thought.

  14. EdeF
    Posted Jun 21, 2011 at 11:01 PM | Permalink | Reply

    You would think that by now anyone even slightly interested in climate science would have the word Tiljander burnt into their brain. Tiljander….yes, there was something about that data, need to do a quick check on Google.

    • Derek Walton
      Posted Jun 22, 2011 at 2:11 AM | Permalink | Reply

      And there’s the rub. There is no mention in the paper of the series by this name (at least I have not been able to find it). So unless there has been a multi-layer review, through the paper to the SI and possibly onto a second paper and SI, then you would not know that these sediments have been used. No alarm bells…… The packaging of the data is as much a concern as it obscures the real sources.

  15. Ian Blanchard
    Posted Jun 22, 2011 at 4:37 AM | Permalink | Reply

    Pax
    “When reading about paleo I always end up having to decide between epic incompetence or deliberate manipulation. Still undecided.”

    I think the words you are looking for are ‘confirmation bias’. Mann et al know what result they are now looking for (i.e. something that is relatively flat to gently declining through much of the series but rising at the end to align with the real temperature measurements – i.e. something that supports the original ‘hockey stick’ shape), and will select and ‘torture’ the supposedly proxy data they have available until it gives the pattern they want.

    They, and Steve Mc, understand the datasets they have available and know which ones do what when fed into the various statistical processes, but they do a less good job of demonstrating whether these ‘proxies’ actually are reliable indicators of temperature beyond (or even within, in the case of strip bark bristlecones) the calibration period – hence the use of Tiljander data sets inverted compared with the interpretation of the original authors, because that’s the way round the calibration period puts it owing to the contamination of the sequence in the later years. The argument that the statistical processing (PCA or whichever other approach is used) makes way-up irrelevant is an absurd argument – the way the proxy (supposedly) responds to temperature variations is a physical or biophysical phenomenon (e.g. hot weather = more sedimant accumulation), and so inverting the sequence simply demonstrates it is not a reliable proxy during the calibration period and therefore its use makes no physical sense.

    • MikeN
      Posted Jun 22, 2011 at 10:34 AM | Permalink | Reply

      >hence the use of Tiljander data sets inverted compared with the interpretation of the original authors

      This doesn’t matter for the EIV result used here. Whether inverted or right side up, the result is the same. The problem is that Tiljander shows ‘cold’ in recent times for non-climate reasons.

  16. Hu McCulloch
    Posted Jun 22, 2011 at 7:42 AM | Permalink | Reply

    Ian Blanchard
    Posted Jun 22, 2011 at 4:37 AM
    The argument that the statistical processing (PCA or whichever other approach is used) makes way-up irrelevant is an absurd argument – the way the proxy (supposedly) responds to temperature variations is a physical or biophysical phenomenon (e.g. hot weather = more sedimant accumulation), and so inverting the sequence simply demonstrates it is not a reliable proxy during the calibration period and therefore its use makes no physical sense.

    When using proxies directly (as in Kemp 2011), it is easy to check signs to make sure they are not inconsistent with theory.

    This is harder when using PCA with a big data set (like NAm treerings), however, since PCA might come up with important PCs that have upside down signs on a few of the contributing series.

    As I pointed out in the recent thread on Smith’s PC Retention rule, however, a simple procedure allows one to prevent wrong signs in the final reconstruction without discarding mixed sign PC’s altogether:

    Simply divide each PC into two series, the part with positive signed weights onthe underlying series and the part with negative signs. Then calculate the total variance of each signed PC, and resort them by their variance. Then apply whatever pre-calibration inclusion rule you are using (eyeball scree, Preisendorfer N or whatever) to these signed PCs, and then enter the ones that pass the pre-calibration screening into the calibration in the same order.

    To prevent acceptance bias in such a “data-rich” environment (to use Bernanke and Boivin’s phrase in a recent econometrics paper), the discipline should be imposed on the calibration that PC5(plus), say, may not be included despite a significant t-stat on itself, unless it and all higher-ranked signed PCs together pass an F-test (or something similar) for joint significance.

    Since the null hypothesis is that all proxies have zero coefficients and therefore an equal chance of an accidentally positive or negative sign, I don’t see that it would affect the validity or critical values of the test to just toss proxies that happen to come in with the “wrong” sign.

    the way the proxy (supposedly) responds to temperature variations is a physical or biophysical phenomenon (e.g. hot weather = more sedimant accumulation),

    One thing I learned from a paper being discussed on CA a couple of years ago is that lake sediments can go either way, depending on whether they are glacier-fed or snow-fed. Glaciers are there all year round, so a warm year results in lots of melt in the summer and lots of sediment. But if there’s no glacier, a cold winter leads to lots of snow buildup and therefore a massive spring thaw with lots of sediment, so the correlation is the other way around. I believe Tiljander’s Kortjavarti is in the latter category, while many Canadian Rockies lakes are the in the former category. The researcher should indicate the expected correlation with temperature, depending on the siting of the lake.

    • tvin
      Posted Jun 22, 2011 at 9:12 AM | Permalink | Reply

      Hugh and Ian,

      For M08, I don’t believe the null hypothesis was that all proxies have zero coefficients and therefore an equal chance of an accidentally positive or negative sign.

      The paper states: “The screening process requires a statistically significant (P  0.10) correlation with local instrumental surface temperature data during the calibration interval. Where the sign of the correlation could a priori be specified (positive for tree-ring data, ice-core oxygen isotopes, lake sediments, and historical documents, and negative for coral oxygen-isotope records), a one-sided significance criterion was used. Otherwise, a two-sided significance
      criterion was used.”

      So M08 assumed ex ante that there was a physical meaning to the sediment series and tested accordingly. However (for reasons they have not yet explained), they assumed a physical meaning opposite to that believed by the series creator.

      • Hu McCulloch
        Posted Jun 22, 2011 at 11:27 AM | Permalink | Reply

        tvin —
        In a 1-sided test, the null is still that the coefficient is zero. All that changes is the alternative, as well as the critical value.
        I suppose that when evaluating the joint significance of a number of proxies under a 1-sided alternative, one would use something like a “1-sector F test” that only uses the non-negative components. I’m not sure how that would work, however, even if proxies caused global temperature so that multiple regression was the way to go.

    • Philip Mulholland
      Posted Jun 25, 2011 at 7:12 PM | Permalink | Reply

      Hu,

      You say:

      But if there’s no glacier, a cold winter leads to lots of snow buildup and therefore a massive spring thaw with lots of sediment, so the correlation is the other way around. I believe Tiljander’s Kortjavarti is in the latter category

      and

      The researcher should indicate the expected correlation with temperature, depending on the siting of the lake.

      Good point, but we don’t need to guess where the lake in question is.
      Here is a Google map link showing the location of Lake Korttajärvi in Finland. The country has a generally low elevation and there are no nearby mountains with glaciers sourcing major rivers that cross Western Finland, so you are right, the lake is in the latter category.

      • Posted Jun 26, 2011 at 10:16 AM | Permalink | Reply

        Philip Mulholland & Hu McCullogh —

        On assigning a meaning of “warmer” or “colder” to increased varve sediment accumulation, Hu wrote,

        ..lake sediments can go either way, depending on whether they are glacier-fed or snow-fed. Glaciers are there all year round, so a warm year results in lots of melt in the summer and lots of sediment. But if there’s no glacier, a cold winter leads to lots of snow buildup and therefore a massive spring thaw with lots of sediment, so the correlation is the other way around. I believe Tiljander’s Kortjavarti is in the latter category, while many Canadian Rockies lakes are the in the former category. The researcher should indicate the expected correlation with temperature, depending on the siting of the lake.

        Tiljander measured total varve thickness (in millimeters) as “thickness”.

        She then determined the contribution of inorganics (minerals, in millimeters) as “lightsum”.

        She then deduced the contribution of organic matter (in millimeters) by subtracting “lightsum” from “thickness”, to arrive at “darksum”.

        The discussion on glacier-fed lakes concerns mineral silt (e.g. rock flour) carried from the glacier to the lake by summer meltwater. This would be Tiljander’s “lightsum”, if Lake Korttajarvi were a glacial lake (as you note, it isn’t). Since there is generally little or no deposition of organic materials in this situation, “darksum” would be negligible. “thickness” would be the same as “lightsum” alone.

        I don’t know of a physically-plausible reason for assigning “higher lightsum” to “warmer” and assigning “higher darksum” to “warmer,” as is the case in Mann08.

        Using “thickness” as a separate, third proxy appears to be a case of double-counting.

        It seems to me to be a case of an honest mistake, due to confirmation bias. I would guess that the authors of Mann08 didn’t think very carefully about these things.

        So far as I know, none of the authors have addressed these points, preferring to discuss the similarity of “with Tiljander” and “without Tiljander” reconstructions. Even this not-very-compelling line of reasoning argues against the validity of Tiljander as used in Mann08. This defense was raised again this week, in RealClimate‘s post on Kemp11. I discussed it here.

  17. HaroldW
    Posted Jun 22, 2011 at 7:51 AM | Permalink | Reply

    Just another Mannic mud day.

  18. Posted Jun 22, 2011 at 8:07 AM | Permalink | Reply

    If the proxy isn’t temperature, is it really misuse to flip it?

    There are some absolutely gorgeous quotes in the paper and SI. I don’t believe Kemp wrote it because it has that same mashmatic styling we’ve come to expect from Mann. They certainly don’t let statistics or reasoning stand in their way.

    • j ferguson
      Posted Jun 23, 2011 at 4:29 PM | Permalink | Reply

      “mashmatic style” Are any of the style analyzers which attempt to discover style commonality among different screeds public domain or otherwise available.

  19. JPeden
    Posted Jun 22, 2011 at 8:31 AM | Permalink | Reply

    So now maybe I know what went flipping wrong with the flipping models backing up the use of derivitive products with their own flipping derivitive products in the case of the creation of the otherwise “bad paper” market that created the flipping gov’t inspired Housing Mortgage Debacle, where the loan originators also then flipped their loans upward to Freddy and Fanny, and to the rest of us and left a lot of individuals upside down?

  20. Posted Jun 22, 2011 at 8:38 AM | Permalink | Reply

    Here is PNAS’s Link
    Climate related sea-level variations over the past two millennia
    http://www.pnas.org/content/early/2011/06/13/1015619108.full.pdf?with-ds=yes

  21. Posted Jun 22, 2011 at 9:42 AM | Permalink | Reply

    Here is the interesting thing with the divergence

    The only thing they consider changing is the Mann reconstruction.

    a. the semi empirical model could be off
    b. the sea level recon could be off
    c. mann 08 could be off

    now, think about reconstructing temperature from sea level. That has to be waaay better than reconstructing temperature from tree rings.

    • Posted Jun 22, 2011 at 5:06 PM | Permalink | Reply

      Mosh, maybe that’s the MannPlan. Use the sea level reconstructed from the mishmash to then reconstruct temperature. Voila, Hocky Stick vindication.

  22. Dr. Jay Cadbury, phd.
    Posted Jun 22, 2011 at 12:01 PM | Permalink | Reply

    @AMac

    Realclimate is working fine, they just only allow their biggest fans to post comments. I’ve been banned for several months, my comments are automatically blocked. Also, you know I’ve been complaining about issues like this for an extremely long period of time, you and I have had correspondence at Lucia’s site. Do you have any thoughts as to why she ignores the important topics?

    @AMac and SteveMc

    What can be done about this? It is nice that you are writing about it on the internet but it isn’t going to accomplish anything. Realclimate is tax payer funded, correct? If so, how in the hell can Schmidt pick and choose who he will allow to comment? I think the science behind climate is compromised. I don’t think you can win anymore in the literature because you simply won’t be published. I think the only thing left is legal action. Steve, I’ve noticed you never made the link between Michael Mann’s fake MWP graph and William Connolley’s deliberate collusion with him, by way of deleting the MWP from wikipedia.

    • theduke
      Posted Jun 22, 2011 at 10:02 PM | Permalink | Reply

      Jay Cadbury: what can be done is being done. You don’t just discredit a 30-year scientific movement in one fell swoop. Some of the science supporting AGW may be legitimate. Most appears to be dubious. The value in what Steve, Ross, Anthony, Amac and all the others are doing accrues over time.

      Your assertion that what Steve and others are doing “isn’t going to accomplish anything” is your uninformed opinion only. If you’d been following this blog for the past four years like I have, you would know that the opposite is true. Much has been accomplished and is being accomplished every day.

    • Dean
      Posted Jun 23, 2011 at 9:07 AM | Permalink | Reply

      Realclimate is not directly taxpayer funded. There may be some question as to whether Gavin works on RC while he’s doing his day-job (works for NASA), but that’s a separate issue.

  23. MikeN
    Posted Jun 22, 2011 at 5:30 PM | Permalink | Reply

    Hu McCullough, how do you interpret this portion of Mann’s response to McIntyre in PNAS?

    Screening, when used, employed one-sided tests only when a definite sign could be a priori reasoned on physical grounds.

    • Geoff Sherrington
      Posted Jun 23, 2011 at 1:59 AM | Permalink | Reply

      MikeN
      My point above is that quantitative physical grounds do not exist, so the error was to use the series, not merely to flip it.

  24. Posted Jun 22, 2011 at 8:00 PM | Permalink | Reply

    Michael Mann’s new graph of sea level rise looks like a hockey stick?! Gee, I never saw that comin’.

  25. Armand MacMurray
    Posted Jun 22, 2011 at 9:44 PM | Permalink | Reply

    Dr. J, realclimate.org is not taxpayer-funded. It was set up and (presumably) still funded by “Environmental Media Services,” closely associated with Fenton Communications. According to SourceWatch.org,
    “Fenton Communications is a public relations firm which specialises in PR for not-for-profit organizations, liberal advocacy groups and companies with a claim for social responsibility. They are probably best known for their represenation of MoveOn. “

    • Mark F
      Posted Jun 23, 2011 at 12:42 AM | Permalink | Reply

      Well, except that Gavin seems to spend a lot of normal “business hours” doing RC work.

  26. Dean
    Posted Jun 23, 2011 at 9:11 AM | Permalink | Reply

    Can someone clear up Mann08 from Mann09(?) for me…

    Mann08 used the upside-down Tiljander data and never showed a temperature reconstruction without either Yamal or bristlecone pines. It shows a reconstruction dating back to the first millenium and “passes” the validation tests.

    Mann09 (or whatever the “corrected” version is called) removed the upside-down tiljander and then removed Yamal and BCPs and the reconstruction failed before 1500 AD. Is that basically correct?

    If so, which version did the Kemp et al 2011 use? Mann08 or Mann09? If it used 08 it shouldn’t it be immediately withdrawn?

    • MikeN
      Posted Jun 23, 2011 at 10:40 AM | Permalink | Reply

      Mann 09 also has Tiljander. I’m not sure if either paper has Yamal.

      Steve: Yes, they have Briffa’s Yamal, but not in the place that you’d expect it.Their “Tornetrask” series is an average of Tornetrask, Yamal and Taimyr. More importantly, Mann et al 2008 has a heavy supply of Graybill bristlecone chronologies, as previously observed.

    • Posted Jun 23, 2011 at 11:13 AM | Permalink | Reply

      Re: Dean (Jun 23 09:11),

      the difference between Mann et al 2008 (PNAS) and Mann et al 2009 (Science) is NOT in the data used. Essentially the difference is in target (“calibration”) instrumental temperature. In the 2008 paper, proxies are “calibrated” against a single temperature series (e.g. NH mean temperature) whereas in the 2009 paper the target is a temperature “field” (spatially gridded set of temperature series). To make things a little more complicated, 2008 paper also had another method (“CPS”) which is not employing RegEM (unlike the other method, “EIV”, of 2008 paper and 2009 paper) and is also using “gridding”.

      In the 2008 paper, they essentially found a small MWP, and this was further elaborated in the 2009 paper by showing its “spatial extent”. They found that MWP (called “MCA” there) was a phenomenon in both NH and SH (see figures 2A and 3A in the paper). Now, Kemp et al, which
      utilizies a reconstruction from the 2008 paper, finds a contradiction between their sea level reconstruction and Mann 2008 reconstruction in 500-1000 period, and suggest that this is because Mann 2008 is “too high” there due to poor SH resulution! From SI:

      Lowering reconstructed temperature by 0.2 K for the period
      AD 500–1100 produced good agreement with the North Carolina
      sea-level reconstruction (Fig. S4). We studied the sensitivity of
      this fit to a range of temperature corrections (−0.1 K to −0.3 K).
      As shown in Fig. S5, the best agreement was for a −0.2 K correction.
      An error of this magnitude is not implausible as we used
      the global Mann et al. (34) reconstruction prior to AD 1100
      and not the Northern-Hemisphere-only reconstruction in which
      Mann et al. (34) had greater confidence. For the period prior
      to AD 1100, availability of proxy temperature reconstructions
      is poor for the Southern Hemisphere and this is necessarily
      reflected in greater uncertainty for global estimates which can
      accommodate a 0.2 K reduction in temperature within their
      uncertainty. This reduction in reconstructed temperature would
      make the Medieval Climate Anomaly globally less pronounced
      than Mann et al. (34) suggested, and reduce by a half its temperature
      contrast with the Little Ice Age.

      And this passes “peer review” in PNAS!

  27. Scott Brim
    Posted Jun 23, 2011 at 10:01 AM | Permalink | Reply

    A question for AMac: From your initial look at the paper, does this latest Mannian-style reconstruction — this version having been adapted for use in reconstructing a climate-driven sea level chronology — appear to employ the concept of climate signal telecommunication; i.e., a process in which salt marsh sediment layers in North Carolina might somehow communicate with lake varve sediment layers in central Finland to establish a recognizable climate signal in space and in time, doing so across very long distances?

    • Posted Jun 23, 2011 at 12:05 PM | Permalink | Reply

      Scott Brim –

      > does this latest Mannian-style reconstruction — this version having been adapted for use in reconstructing a climate-driven sea level chronology — appear to employ the concept of climate signal telecommunication…?

      My understanding is “no.”

      I believe that one group of authors (including Kemp and Rahmstorf) worked on creating a reconstruction of sea level, over the past ~2,000 years. They were then interested in relating that to a reconstruction of global temperature. Rahmstorf says (RC post, Response to comment #18) that the best and perhaps only temp recon they could find was “Composite EIV global land plus ocean global temperature reconstruction,” which was first presented in Mann08. This was one of the contributions that earned Prof. Mann a spot on the author list.

      Unfortunately, the uncalibratable Tiljander data series were used as some of the proxies in this temp recon. The farther back in time one goes, the greater the dependence of the temp recon on Tiljander.

      To be clear, the sea level reconstruction, which is the major focus of this paper, does not involve the Tiljander data series in any way.

      None of this involves telecommunication, as far as I can tell.

  28. stan
    Posted Jun 23, 2011 at 11:38 AM | Permalink | Reply

    Lubos — “if one averages and smoothes the sea level and combines the noise in various proxies so that all the wiggles disappear, then all the wiggles disappear.”
    :-D

  29. Steve McIntyre
    Posted Jun 23, 2011 at 2:23 PM | Permalink | Reply

    AMac has an interesting post http://amac1.blogspot.com/2011/06/tiljander-data-series-appear-again-this.html, pointing out past exchanges with coauthor Martin Vermeer. I plan to cover this topic as well.

    I also suggest that readers revisit the posts of around Aug 1, 2010 e.g http://climateaudit.org/2010/07/30/make-a-stick-make-a-stick/ .

    • MikeN
      Posted Jun 24, 2011 at 10:21 AM | Permalink | Reply

      Funny because I am accusing Martin Vermeer of something fr which I was being accused at wattsupwiththat, using semantics to defend Mann. I argued the program is blind to the sign of the data, and Martin argued the algorithm is blind, but not the screening.

  30. AMac
    Posted Jun 23, 2011 at 7:55 PM | Permalink | Reply

    Prof. Mann has responded to this comment/post, albeit without referring or linking to it.

    Update 2 (June 23) to “2000 Years of Sea Level”.

    There is a new graph of a smoothed “Mann et al (2008) global mean (land+ocean) temperature reconstruction.” It is shown with (blue) and without (red) 7 proxy records, which include the Tiljander data series.

    The graph is largely uninterpretable. It lacks uncertainty intervals for the two curves, as Mann et al (2008) would calculate them. Further, there is no discussion of what portions of the two graphs pass the “validation test” as described in Mann et al (2009), at the 95% or higher level.

    In the paragraph of substantive text, the following question remains unacknowledged and unanswered:

    Are the Tiljander proxies calibratable to the instrumental temperature record, 1850-1995?

    Perhaps the RealClimate bloggers are planning to post additional information in a while.

  31. Posted Jun 24, 2011 at 9:01 AM | Permalink | Reply

    I posted some remarks on whether the four three Tiljander proxies data series can be calibrated to the instrumental temperature record.

    Voldemort’s Question

  32. Alex Harvey
    Posted Jun 24, 2011 at 9:51 AM | Permalink | Reply

    Dear Steve,

    Mann responded to a question I asked on behalf of AMac about the Tiljander issue. He responded with an update that included a link to a (fairly negative) comment he made on a paper by McIntyre & McKitrick in PNAS.

    http://www.meteo.psu.edu/~mann/shared/articles/MMReplyPNAS09.pdf

    I imagine you read that and responded in some way. Do you know of a good summary of this issue and the status of the issue today?

    Best,
    Alex Harvey

    • Posted Jun 24, 2011 at 10:17 AM | Permalink | Reply

      Alex –

      That PNAS Letter by Mann et al. is a rebuttal of a short comment that McIntyre and McKitrick had published in PNAS, critical of Mann08. It is part of a lengthy saga.

      For the paragraph concerning Tiljander, much has been written. Brief comments –

      The claim that ‘‘upside down’’ data were used is bizarre.

      This is an uninformative statement.

      Multivariate regression methods are insensitive to the sign of predictors.

      That may be for EIV and/or for CPS, but the claim is not responsive to the issue at hand: whether lightsum and XRD were employed in Mann08 such that data values assigned by Tiljander03 to “warmer conditions” were inadvertantly assigned by Mann08 to “colder conditions” (they were).

      Screening, when used, employed one-sided tests only when a definite sign could be a priori reasoned on physical grounds.

      Again, that may be, but the claim is not responsive to the issue at hand.

      Potential nonclimatic influences on the Tiljander and other proxies were discussed in the SI,

      This is correct.

      which showed that none of our central conclusions relied on their use.

      Mann et al. are asserting that the material in the SI showed that none of Mann08′s central conclusions relied on the use of the Tiljander proxies. This claim is hotly disputed. Even the definition of “the contents of the SI” has been the subject of dispute. A revised version of one key figure, S8A, replaced Fig. S8A after publication of Mann08. The original is still at pnas.org, to my knowledge, with the revision at Prof. Mann’s website. The revision was then subsequently re-revised. The SI of Mann09 then revisited some claims made in the SI of Mann08, weakening them.

    • MikeN
      Posted Jun 24, 2011 at 10:36 AM | Permalink | Reply

      Alex, it depends on how much weight you give to the graph on Mann’s website in the SI that shows the sensitivity of the results to when you exclude Tiljander(and others), as well as the bristlecones. This is the green line. It extends quite a bit higher, and is dashed to show that it does not pass validation prior to 1500. Mann is saying this does not challenge the conclusions of the paper.

      I disagree with AMac that Mann’s responses were uninformative. They were informative and misleading. Mann is essentially saying that his program would flip Tiljander to the correct orientation so it is not possible to use it upside-down. This is both incorrect in the CPS case, and irrelevant to the point Steve McIntyre made, that the proxy cannot be used because the modern portion is diverging.

  33. pouncer
    Posted Nov 25, 2011 at 4:27 PM | Permalink | Reply

    Riddle me this: What impact does it have to invert a couple of columns of climate measure?

    http://tallbloke.wordpress.com/2011/11/25/data-errata-caper-cru-methodology-inexplicable/#comment-9862

    And who will discover the error if it should happen to exist?

5 Trackbacks

  1. [...] AMac: Upside Down Mann Lives on in Kemp et al 2011 [...]

  2. By The Climate Change Debate Thread - Page 767 on Jun 23, 2011 at 3:04 AM

    [...] [...]

  3. [...] paper here). A number of commenters have pointed out huge shortcomings in the paper. AMac has noted at ClimateAudit that Mann’s oft-noted mistake of the upside-down Tiljander series lives on, [...]

  4. [...] 2011). A number of commenters have pointed out significant shortcomings in the paper. AMac has noted at ClimateAudit that Mann’s oft-noted mistake of the upside-down Tiljander series lives on in [...]

  5. [...] reflect that complex relationship between air temperatures and sea level. And on Climate Audit, AMac points out that the global temperature record used here has been the subject of its own debate and [...]

Post a Comment

Required fields are marked *

*
*

Follow

Get every new post delivered to your Inbox.

Join 3,111 other followers

%d bloggers like this: