D’Arrigo et al. on Bristlecone Calibration

D’Arrigo, Wilson and Jacoby [2006] represents state-of-the-art in dendrochronology and is hot off the press. It is unique among such studies in using a considerable amount of up-to-date data and is relatively candid about its results. I’ll try to discuss it in more detail. Here I want to pick up on one issue that featured strongly in my review of Osborn and Briffa – my inability to replicate the claimed correlation of bristlecones to gridcell temperature. Remarkably DWJ06 make the same point: thus we have the remarkable situation of two dendrochronological studies published in the same week, making completely opposite claims about the correlation of bristlecones and foxtails to gridcell temperature.

Osborn and Briffa provide their claims to quality control in connection with their comments on Soon and Baliunas [2003], who they criticize for not verifying that proxies were positively correlated with local temperature, thereby supposedly confusing precipitation and temperature proxies. O&B:

This study [Soon and Baliunas, 2003] has been criticized [Mann ...Briffa, Osborn...et al. 2003] for its lack of rigour in assessing whether the proxies used are useful indicators of temperature, for not distinguishing between regionally restricted anomalies and hemispheric-scale warmth, and for providing no calibration or uncertainty estimates that would enable comparison with late 20th-century temperatures. Here, we investigated whether a more carefully designed assessment of proxy records on an individual basis supports the conclusion that recent NH temperatures are unusual in the context provided by these records. We only used proxy records that are positively correlated with their local temperature observations

We twitted O&B in our review for not applying similar criticisms to, say, MBH98, but more substantively we asserted that we were unable to replicate the claimed O&B correlation of (decadal) 0.56 between the MBH98 PC1 and the relevant gridcell temperatures. We did not review the separate O&B claim about the supposed correlation between foxtail proxies (Boreal Plateau/"Upperwright" [Upper Wright?]) and gridcell temperatures due to lack of data on the foxtail series actually used by O&B.

DWJ06 consider bristlecones in general and Boreal Plateau/"Upperwright" in particular, also stating that they screened their chronologies against local temperature data, as follows:

The regional chronologies were also screened by comparisons with instrumental (local and larger scale) temperature data to ensure that the temperature signal in the final reconstructions was as strong as possible and relatively unmuddied by precipitation effects. In so doing, some potential data sets were discarded due to ambiguous signals. For example, we did not utilize the long bristlecone pine data sets from Colorado and California as many appear to portray a mixed precipitation and temperature signal (in addition to a purported CO2 fertilization effect [LaMarche et al., 1984]). We also did not use the Mackenzie Mountains, Boreal, Upperwright and Gotland data sets utilized by Esper et al. [2002a] for similar reasons, specifically that these records either (1) did not demonstrate a significant temperature signal on the local to regional scale, (2) displayed significant correlations with precipitation, or (3) were located at lower latitudes than those compiled for the present analysis.

A few comments.

First, it seems a bit churlish for D’Arrigo and Jacoby to raise the issue of the validity of bristlecones as a proxy, citing Lamarche et al 1984 (which is fine), without acknowledging our recent critique, which brought the matter to the fore. In fact, readers might well ponder the question of whether there might even have been some difference of opinion on using bristlecones between the authors of this excellent study and whether their ultimate decision to exclude their use was made to avoid an obvious criticism – a prudential measure not taken by O&B.

Second, how did Osborn and Briffa (or Mann and Jones for that matter) get the supposed positive correlations between bristlecones, bristlecone PC1s and foxtails and gridcell temperature? I said in my review that I couldn’t replicate the results for bristlecones; D’Arrigo et al, 2006 say te same thing. I have archived the bristlecone chronologies www.climateaudit.org/data/osborn06/tree.collation.AD200.txt and annualized versions of the three HadCRU2 gridcell series at www.climateaudit.org/data/osborn06/bristle.gridcells.txt so anyone who doubts the lack of correlation can test for themselves. We have here a pretty black-and-white difference of result: Osborn and Briffa say that there is a positive correlation between the sites and a decadal positive correlation of the PC1 to the gridcell temperatures of 0.56 – if anyone has the faintest idea of how they come to that conclusion, I’d sure like to see it.

Third, we have the conundrum of the foxtails. While I’m interested that DWJ06 did not use the Boreal and Upperwright sites used by Esper et al [2002], there is still a conundrum in the provenance. When I was looking for the Boreal and Upperwright data (now apparently lost by Graumlich !?!) , I asked Wilson whether he had it; he said that he didn’t. Wilson was working in conjunction with D’Arrigo and Jacoby at Lamont-Doherty and presumably had access to all the data that Cook had; Cook was a coauthor of Esper et al [2002] (and actually shows as the corresponding author). So do Esper and/or Cook have the lost Graumlich data (but not make it available to DWJ) or did they actually use Cirque Peak and something else and mislabel it? Why won’t Science make Esper produce his data?

But the take-home point here is: we now have black-and-white contrary views about the correlation of bristlecones and foxtails to gridcell temperature in two articles in the same week. Of course, neither study provides data by which readers can reconcile the contradiction. No one on the Hockey Team seems to have noticed. But there’s going to be more on this and this might even be a nice wedge point to force Science to require O&B (and Esper) to disgorge their data.


8 Comments

  1. jae
    Posted Feb 11, 2006 at 9:15 AM | Permalink

    WOW. They sure can’t both be right! Bet you won’t see a discussion of this over at realclimate.

  2. JerryB
    Posted Feb 11, 2006 at 10:50 AM | Permalink

    Just putting the full citation into this thread:

    D’Arrigo, R., Wilson, R. and Jacoby, G. 2006. On the long-term context for late 20th century warming. Journal of Geophysical Research, Vol. 111, D03103, doi:10.1029/2005JD006352,

  3. hans kelp
    Posted Feb 11, 2006 at 11:20 AM | Permalink

    As I see it it is now becoming crystal clear that the work of
    Steve McIntyre has begun to make it´s impact on the community. Respect, respect!. Two contradictionary research results published within the same week and certainly from people of within the very same community, that must spell trouble for the Hockey Teams (and Science). Now people start to take the work of S.M. seriously. About time!

  4. Pat Frank
    Posted Feb 11, 2006 at 4:05 PM | Permalink

    Steve, neither one of your climateaudit bristlecone URLs worked for me. Looking about, I did find this NOAA site that seems to archive lots of tree ring data, and that offers a very explicit search regime:

    http://www.ncdc.noaa.gov/paleo/ftp-treering.html

    Now, to figure it out.

    Steve: I fixed the directory. The links should work now. I think that one of the reasons that these guys don’t archive their data sets is that once you see them, the analyses that they carry out are so trivial. The only heavy lifting, and it’s laborious rather than complicated, is mastering the data sets.

  5. Steve McIntyre
    Posted Feb 11, 2006 at 7:05 PM | Permalink

    Pat, I’m very familiar with this site. One of the irritating aspects to their tree ring data is that the formats sometimes are inconsistent between authors. In 2003, when I was collating data for the MBH principal components calculation, I made consistent R-objects of all the crn (chronology) and rwl (measurements) files and made a table of details – site, location, altitude, start, end, author, species, etc. as an R object for each of the regions. I use these all the time. WDCP also has irritating use of 999 to end lines – it’s very irritating when there are values in the same range. However, on balance, WDCP is very good. I’ve corresponded frequently with their data manager, who has been very cooperative.

  6. Pat Frank
    Posted Feb 12, 2006 at 2:15 PM | Permalink

    Thanks Steve — I’ve accessed the fixed links. Thanks also for the heads-up about those “999”‘s. I had no idea what they were for, but they sure do screw up a plot. :-)

    But a question: in the NOAA.gov tree ring data, e.g., the “White Mountains B” compilation, the first entry is presumably a tree-code, the second is year, and is the rest of each year-line of numbers just repeated measurements of widths at different points on a ring? Should one then just average these numbers to get a net tree ring width for a given year, +/- SD?

    Why, by the way, in the WMs B data set, do all the 1960 strings end early? Hmmm — maybe I should just consult Schulman and Fritts. Apologies for bothering you with picayune questions.

  7. Steve McIntyre
    Posted Feb 12, 2006 at 2:23 PM | Permalink

    Pat, I’ll post up a function which I use to convert WDCP data to a time series – it works for most WDCP records, but there are a few oddball formats and I spent quite a bit of time a couple of years editing and fiddling with this.

    They report data in an antique format left over from punch cards and memory problems. the first entry is the tree code (but sometimes the last digit is a core# within the tree). The date is usually the first year of a decade, but if it’s not a full decade present is the first available year. The values are by year for the decade. I’ll post up some functions a little later at http://data.climateaudit.org/scripts/tree.functions.txt.

    You can use them in an R program by just
    source(“http://climateaudit.org/scripts/tree.functions.txt”)

    R is really quite magical for this sort of thing.

    The White mountains B is an old compilation.

  8. Pat Frank
    Posted Feb 13, 2006 at 12:47 AM | Permalink

    Thanks, Steve — the format clarification helped a lot. I have an application called “Kaleidagraph” that can help me sort out the rest. Now that I know what it is. :-)

    Pat

One Trackback

  1. By Wilson on Yamal Substitution « Climate Audit on Apr 16, 2011 at 3:05 PM

    [...] away from it. It’s too bad. D’Arrigo et al 2006 is a vastly superior paper. I published some first comments on Feb 11 focussing on bristlecones. In response to that note, Rob wrote to say that he used [...]

Follow

Get every new post delivered to your Inbox.

Join 3,301 other followers

%d bloggers like this: