An Esper Spaghetti Graph

It’s obviously been pulling teeth to get data from Esper. After only two years of trying, I’ve recently obtained all but one site chronology and 10 of 14 measurement sets; and gobbledy-gook about methodology. It’s a very Mannian process. While I continue to try to get the rest of the information, I thought that it would be interest just to do some simple plots and averaging of what I have. First, here is a spaghetti graph of the 13 site chronologies that have been sent so far (all smoothed), together with an unweighted average. The average inter-series correlation of the smoothed series is 0.10.

Figure 1. Spaghetti graph of Esper site chronologies (except Mongolia). Black- unweighted average.

You will notice that there are only two series out of the 13 which have strongly elevated closing values. Guess what they are? You should know already. They are both foxtails (interbreeding cousings of bristlecones) both from sites very close to Sheep Mountain, California. Even if you decided to permit bristlecone/foxtails as a proxy – questionable in terms of robustness – don’t you think that two out of 14 is overkill? Here are plots of the individual sites, again as delivered in February 2006:

Figure 2. Plots of 13 of 14 Individual Esper Site Chronologies

Now these series have not been re-centered or re-scaled. My impression is that Esper’s method does not involve re-centering or re-scaling at a site level. Indeed, my impression of Esper’s method is that site chronologies are not used – so while I appreciate getting these chronologies, I’m not entirely sure of how they were used by Esper (but soem of them were used by Osborn and Briffa after re-scaling.)

When you see these series, you really have to wonder what some of them are doing in a millennial chronology – what’s the point of the Zhaschiviersk segment, or, for that matter, the relatively short Tirol series?

Now take a look at these series and guess which ones were carried over into Osborn and Briffa?

1) an average of the two foxtail series (Boreal and Upperwright) except the earliest portion. (Osborn and Briffa also use Mann’s PC1, which is realy just the Sheep Mountain bristlecone in this period – so they also have 2 bristlecone/foxtail series out of 14, although the method of getting them is a little different.

2) the Ath(abska) series under the name Alberta or Jasper, in a different version with a lower MWP;

3) Taymir in a very slightly different version;

4) Tornetrask in a considerably different version with more HS;

5) Mangazeja is used in the same version;

6) Quebec (Bonif) is used in the same version;

7) Mongolia is carried over – probably in the same version, but I don’t have the Esper version;

8) Tirol is carried over in the same version, but the provenance in Osborn and Briffa (germ21) is wrong;

9) Campbell Dolomite/Mackenize Mts; Gotland; Jaemtland; Zhaschieviersk are not used in Osborn and Briffa.

10) the updated Polar Urals series shown here with a strong MWP is discarded in favor of Yamal. (The earlier Briffa version of Polar Urals had a very cold 11th century – Briffa said that 1032 was the coldest year of the millennium.) See posts on Yamal from Jan-February.


  1. TCO
    Posted Apr 23, 2006 at 4:22 PM | Permalink

    1. Is there a PDF of the Esper paper available?

    2. Is there a good overview of what is said in Esper in one of your posts? (I did look.)


    I’m trying to understand the context of where this is all going.

  2. Steve McIntyre
    Posted Apr 23, 2006 at 5:22 PM | Permalink

    Look here for pdf’s.

    The Esper category is here:

    Because I’ve only started getting data in the last month, I’ve never been able to make much progress with it. So I’ve mostly just pecked away whenever I got a little snippet of information here and there. The data from Science, even though incomplete, has been a HUGE help.

  3. John A
    Posted Apr 23, 2006 at 5:31 PM | Permalink

    What causes the big downspike at the end? Is that the “Divergence Problem”?

  4. TCO
    Posted Apr 23, 2006 at 5:39 PM | Permalink

    Steve, I did look at the “Esper catagory” already. Without rereeading every single word, I did not see one post which just gives an overview of the paper, nor did I see a post that describes your suspiciouns/concerns with the paper (your audit plan). Mostly just stuff about the data battles. That’s fine. Just wondered if there was an overview.

    Reading the paper itself is helpful, thanks. Although I often think you’ve already got some suspicions and overview feelings and sometimes even implicitly refer to them, when we’ve never really been briefed. That’s fine, it’s your blog.

    Was just trying to encourage more science posts, by having some comments here. I get so sick of the he-said/she-said Lambert arguments.

  5. Dave Dardinger
    Posted Apr 23, 2006 at 6:09 PM | Permalink

    I get so sick of the he-said/she-said Lambert arguments.

    I do too, but you can’t avoid it unless you want to risk complaints of censorship. What would be nice would be someone, maybe not even on this site, rating various threads as to whether or not they’re worth reading. An individual thread rating could even include particular parts of it which are of interest. I know Steve does have some favorite posts listed in the sidebar by that name, but something more comprehensive might be of value. Who knows, maybe I’ll take it on myself someday if nobody else steps up to the plate.

  6. TCO
    Posted Apr 23, 2006 at 6:48 PM | Permalink

    I read the Esper paper. Do you want my comments?

  7. TCO
    Posted Apr 23, 2006 at 7:22 PM | Permalink

    I agree. We should not censor the he-said/she-said. I just want to keep some volume of good stuff coming. I can sift through the chaff fine on my own. Just want some good science discussion and discovery…

  8. Posted Apr 23, 2006 at 8:57 PM | Permalink

    The downspike at the end is most likely caused by some of the elevated series ending slightly before the last series to end. Thus the mean (which is not robust, as you will remember from previous discussion) suddenly jumps down as there are less series to average at that point and they happen to be those with a lower average. I would chop it off if possible.

  9. Michael Jankowski
    Posted Apr 24, 2006 at 7:46 AM | Permalink

    Now if the mean suddenly jumped UP as there are less series to average at that point, what do you think people wanting to show 20th century warming would want to do? Cut it off? Only the honest ones!

  10. TCO
    Posted Apr 24, 2006 at 7:56 AM | Permalink

    It would be good to annotate the chart with the places where series are dropped/added/changed. shows the regime changes.

  11. Douglas Hoyt
    Posted Apr 24, 2006 at 8:42 AM | Permalink

    Let’s see a spaghetti graph with the following changes:

    1. Remove both of the bristlecone/foxtail time series. They are not temperature proxies.
    2. Use the alternate versions of the Athabska, Taymir, and Tornetrask series.
    3. Use the updated Polar Urals time series.

    That will eliminate most of the dubious time series. A plot of the difference of the mean of these series and the mean of the series plotted above would be helpful as well.

  12. Steve McIntyre
    Posted Apr 24, 2006 at 9:25 AM | Permalink

    #9. In the other multiproxy studies when there are only a couple, the last survivors tend to go up and they are sometimes (although not always) left in charts. In Briffa’s MCD study where there a lot, but they go down after 1960, the post-1960 values are truncated in spaghetti graphs e.g. IPCC TAR.

  13. jae
    Posted Apr 24, 2006 at 9:31 AM | Permalink

    It’s amazing; these guys do anything they can to eliminate the MWP. They did not do that back in the 80s I think. All of a sudden the MWP became a real obstacle to showing AGW, so they simply cherry picked those series that don’t show it. LOL.

  14. Jaime Arbona
    Posted Apr 24, 2006 at 7:14 PM | Permalink

    Changing the subject somewhat, are you all aware that M&M reconstructed records of the Northern Hemisphere temperatures, among others (Jones et al., 1998; Mann et al., 1999; Briffa, 2000; Esper et al., 2002; McIntyre and McKitrick, 2003; and Moberg et al., 2005) have been used to show that

    the hypothesis that at least part of the recent warming cannot be solely related to natural factors, may be accepted with a very low risk, independently of the database used.

    What? M&M have not done any temperature reconstructions? Well, it’s all here:

  15. TCO
    Posted Apr 24, 2006 at 7:32 PM | Permalink

    MM did a simple variation of the MBH reconstruction. I would not really call that a seperate reconstruction unless you want to call all 64 of the Burger and Cubasch MBH variants reconstructions also.

  16. Greg F
    Posted Apr 24, 2006 at 8:44 PM | Permalink

    Changing the subject somewhat, are you all aware that M&M reconstructed records of the Northern Hemisphere temperatures…


    I would suggest you spend some time on the site before repeating the misrepresentation of their work. M&M have never claimed to have done a reconstruction. What they did is show that Mann’s reconstruction is statistical garbage.

  17. Ed Snack
    Posted Apr 24, 2006 at 11:17 PM | Permalink

    Uh, guys, I think Jaime maybe being a little ironic, are you Jaime ?

    The link is to a recently published paper, Rybski, Bunde, Havlin, & Von Storch, GRL Vol 33. There is a cost for the pdf so I haven’t read it. They use 6 reconstructions as quoted above. Steve, you may want to communicate with VS as I think this may be a misrepresentation of M&M 03.

  18. Louis Hissink
    Posted Apr 24, 2006 at 11:30 PM | Permalink



    It is called an April Fool’s joke on AGU – note the date? 31 March! And the gobbledygook in the abstract?

    No wonder Mann et al get away with so much their climate papers.

  19. Steve McIntyre
    Posted Apr 24, 2006 at 11:32 PM | Permalink

    Here was my letter to von Storch last July in repsone to a request about our”reconstruction” for study in long-memory analysis. I’m not sure how I could have made the point more clearly.

    dear Hans,
    I am extermely interested in long-memory phenomena in time series. It is a topic that I have studied in some detail. I always try to make clear that our work has been entirely critical and that the “reconstructions” that we have graphed are in effect sensitivity studies showing the non-robustness of Mann’s work (or possibly reductiones ad absurdum).

    I would encourage Armin Bunde to modify his focus to include long tree ring series (especially bristlecones) and tree ring PC1s. The Mann NOAMER PC1 has extraordinary persistence properties ( a positive autocorrelation up to 250 years) and a Hurst parameter > 0.9. One of the under-appreciated aspects of our GRL simulations was our attention to long-memory properties. Interestingly, Mandelbrot himself studied and reported on tree ring site chronologies, with calculations of Hurst parameters, including earlier versions of some sites used in MBH98-99. The Nile River series is a prototype series in fractional processes, but seldom discussed in climate studies; it would be interesting to restore it to climate studies. It is readily accessible.

    All kinds of interesting issues arise in regression of series with very high Hurst parameters against other series with high autocorrelations (these problems would still exist under the various scaling methods). Spurious significance occurs easily.

    These issues are very active areas of research in economics.It’s also tough to distinguish between long-memory and trends and again there is much interesting work in econometrics.

    You can use this reconstruction which is at column 5. I’ll recommend the tree ring PC and tree ring site chronologies if this interests him/you.

    Cheers, Steve

  20. john lichtenstein
    Posted Apr 24, 2006 at 11:59 PM | Permalink

    Louis that abstract looks lucid. Maybe I should check again after coffee. Although considering M&M 2003 as independent of MBH 1998 is even worse than considering it as a reconstruction.

  21. Willis Eschenbach
    Posted Apr 25, 2006 at 12:16 AM | Permalink

    Re 19, Steve, many thanks for your corrected MBH98 data. You have three columns, which are headed MBH98, emulation, intermediate, and corrected.

    I understand MBH98, emulation, and corrected, but what is "intermediate"?


    Steve: this is the digital version of a diagram in EE 2005 which showed the effect of PCs and Gaspe, both of which affect the AD1400 step; the intermediate is varying only one at a time.

  22. Louis Hissink
    Posted Apr 25, 2006 at 12:41 AM | Permalink

    Re #20

    John, well, John Ray on his Greenie Watch blog also picked it up as a spoof I just discovered. Lucid it may be but that is the hall mark of a good April Fool’s joke – I engineered one in 2001 and caught quite a few but these things don’t work to well if it does not appear lucid. By memory I caught Miningnews, some govt geophyscists and some big Canadian diamond explorer.

    Of course I could be quite wrong on this and it’s for real but if Jaime and John Ray spotted it, then I suspect its what it is, a really good 1 Apriller.

  23. IL
    Posted Apr 25, 2006 at 12:58 AM | Permalink

    Well I looked over this paper. The first line is interesting:

    “It is well accepted that the temperature of the Earth has been in the rise in the last hundred years, with a more pronounced increase in the last 25 years (Figure 1, red curve). The open question is how much of this increase can be attributed to natural fluctuations, and how much is of anthropogenic origin caused, for example, by the enhanced greenhouse gas (GHG) emissions in the last century.”

    They set their stall out! They clearly accept the hockey stick graph and look to see whether the recent warming from the instrumental record (pace, questions about heat islands etc!) can be natural by comparing with variations from the past from the reconstructions. As far as I can tell, they assume that the temperatures from the reconstructions are indeed temperatures in a meaningful sense of global temperatures and can be analyzed and checked for autocorrelation etc. They look at the scatter of temperatures in the reconstructions and find that they are Gaussian distributed (isn’t random noise also distributed this way!?) and derive a standard deviation from this. Then when they look at the present rise from the instrumental record the rise is more than 2.5 standard deviations so it is statistically significant

    They conclude:

    “We conclude that the previous claim that the most recent warming, observed by quality controlled instrumental data, would be inconsistent with the hypothesis of purely natural dynamics [Hasselmann, 1993; Hegerl et al., 1996; Zwiers, 1999; Barnett et al., 2005] is supported by our long-term persistence analysis of different proxy-based reconstructions extending over many centuries and even up to two millennia.”

    Seems to me that its as good as the assumptions built into it. Garbage in, garbage out.

    Some other little interesting points: It does refer to the reconstruction (sic) of McIntyre and McKitrick and what is more says that the 5 other reconstructions used are archived at NOAA World Data Centre for Palaeoclimatology whereas M&M is not (almost implies that M&M are very naughty in this respect!).

    On their figure 1 it refers to MCINTRYE 03

    In their final conclusion they say:

    “An interesting detail is that the two fiercely arguing groups around Mann and McIntyre lead both to very early detections, while the most conservative detection result is obtained when the more “”bumpy” reconstruction by Jones and coworkers is used.

    The detection referred to here being the detection of significant recent anthropogenic warming and the significance of the comment is that the “reconstructions’ of M&M and Mann et al. have much lower scatter than the other reconstructions so it is possible to see the recent warming signal above that statistical noise much earlier than those reconstructions with much higher scatter. Since at the NAS panel, Mann was the ONLY person who thought that his reconstruction could meaningfully obtain past temperatures to high precision and M&M is not a reconstruction but a variant of Mann et al. to show that Mann et al. is garbage then it is not surprising that they share similar characteristics. As I said, garbage in, garbage out. Who referees this stuff?

  24. IL
    Posted Apr 25, 2006 at 1:00 AM | Permalink

    ps its definitely published in GRL, its no spoof!

  25. Louis Hissink
    Posted Apr 25, 2006 at 2:10 AM | Permalink

    Re #24

    Just because its in GRL (and published 31 March) does not ensure it is not a spoof but maybe the spoof spotters could explain how and why they recognise it as such? Anything published on or around 1 April is a likely candidate for spoofiness unless incontrovertible evidence exists that its not.

  26. Ed Snack
    Posted Apr 25, 2006 at 4:06 AM | Permalink

    Ask the expert, Steve, do you think it is a spoof ? If it is, it is a good one, it looks real enough to me.

  27. Dave Dardinger
    Posted Apr 25, 2006 at 7:21 AM | Permalink

    Hey guys, these are academics, right? As such they must buy into modern critical analysis in literature. And a hallmark of such analysis is that an object of literature isn’t to be judged on what the person/people who produced it intended. So on the one hand, if we here think it was a cute Aprils fool joke, then for us it is. Likewise, if the Manniacs think Ross and Steve produced a climate reconstruction, then they must have. So now all we have to decide… errr; what is it we have to decide? I haven’t quite gotten that part of lit. anal. down yet.

  28. Doug L
    Posted Apr 25, 2006 at 7:36 AM | Permalink

    I expect an april fool joke to say so when you click on the “purchase” or “add to cart” button. such as:

    the link is about wireless extention cords. 🙂

  29. Paul Penrose
    Posted Apr 25, 2006 at 7:58 AM | Permalink

    It does seem unlikely that they would charge you for an article which is an April-fools joke. It must be real, but it’s telling that we can’t discern the difference by reading it!

  30. Jaime Arbona
    Posted Apr 25, 2006 at 9:14 AM | Permalink

    When I fist saw it, I thought it was some sort of joke, given the publishing date and that M&M ’03 was included in the same team as all the other hockey stickers (now that’s a joke!). I couldn’t judge on the content of the abstract, not my gig. So it’s up to you climate experts to decide if the joke is on us or on them.

  31. Gerald Machnee
    Posted Apr 25, 2006 at 11:04 AM | Permalink

    It is difficult to tell if it is serious as there are April 1 papers published any time of year.

  32. SC
    Posted Apr 25, 2006 at 6:09 PM | Permalink

    Click to access rybski-etal.2006.pdf

    Above is the refernced Von Storch paper.

  33. James Lane
    Posted Apr 25, 2006 at 8:32 PM | Permalink

    OK, I’ve read the paper, it’s not a joke, and the treatment of MM03 is bizzare to say the least. Actually, it’s amazing given that Steve warned von Storch that MM03 was not proposed as a valid reconstruction (in the context of this very paper) last year. Even New Scientist managed to register that point.

    And you have to wonder about this line at the end of the paper:

    “An interesting detail is that the two fiercely arguing groups around Mann and McIntyre lead both to very early detections, while the most conservative detection result is obtained when the more “”bumpy” reconstruction by Jones and coworkers is used.”

    What’s the point of the phrase “two fiercely arguing groups” if not to somehow imply that MM03 is supporting Mann’s reconstruction? Weird.

  34. Greg F
    Posted Apr 25, 2006 at 9:27 PM | Permalink

    Re: 33

    “An interesting detail is that the two fiercely arguing groups around Mann and McIntyre lead both to very early detections…”

    What’s the point of the phrase “two fiercely arguing groups” if not to somehow imply that MM03 is supporting Mann’s reconstruction? Weird.

    The casual reader might assume that the “fiercely arguing groups” are in a juvenile argument over some minor detail. It would be an “interesting detail” as the 2 are alleged to be basically in agreement. It seems everyday is April 1st in climate science.

  35. Steve Sadlov
    Posted Apr 27, 2006 at 11:54 AM | Permalink

    Regarding the down spike, many indicators do in fact suggest a probable downturn in – pardon the use of this questionable attribute – global mean temperature, since the peak during the ’97 – ’98 strong El Nino.

    For me, the question would be is it recovery from that El Nino or something worse? Again, putting on my inevitable EE hat, whenever you see a step function in a signal that is in a high state (which, generally speaking, we are in and have been for some 10K years) there is always the possibility of the step being the onset of a state transition. So, if it’s a glitch, all well and good. If not, then woe unto Northern Europe, Canada and the Northern tier of the US. This is not alarmism, it is a simple statement of fact. All evidence, and much current knowledge regarding glaciation behavior, points to the inevitable eventual end of this interglacial.

    Instead of wasting money, time and human capital on “stopping global warming” the only rational focus of climatologists, physicists, politicians and others who care about things like national viability ought to be what will be the exact continency management plans not if but when the continential glaciers begin their inexorable return.

%d bloggers like this: