AGU Fall Meeting 2006

Session U12 of the AGU Fall meeting (Session organizers: Gerald North, Bette Otto-Bliesner, John M. Wallace, and Ian Kraucunas[ of NRC]) has the following session description

Numerous studies concerning large-scale surface temperature reconstructions have been published over the last decade. The National Research Council recently completed a report summarizing the status of these efforts. The session invites contributions on the various types of proxy evidence and methods used to derive large-scale climate reconstructions for the past 2,000 years. Of special interest are updated proxy records or new proxy data from under-sampled regions, the spatial and temporal extent of past temperature anomalies as well as their central locations and timings, and reconstructions of solar and volcanic forcing, precipitation, or other climate variables over the past two millennia.

Abstracts are due by Sept 7. Here is my draft. Suggestions are welcome.

The report of the National Research Council (NRC) panel made several important recommendations about reconstruction methodology, including concluding that strip-bark samples (bristlecones/foxtails) should be avoided in temperature reconstructions and that non-robustness to subset selection should be considered in confidence estimation. Inconsistently, their report relied on and illustrated studies that are affected by these problems. We demonstrate the impact of implementing NRC panel recommendations on prominent reconstructions, showing that none of them provide reliable information on relative medieval-modern temperature levels.

28 Comments

  1. Steve Sadlov
    Posted Sep 5, 2006 at 11:30 AM | Permalink

    That would be sure to stir up some lively discussion and maybe even some follow on papers by others who are interested in this topic. Would be great to see some groups from the geological community proper take on a few of them. ;)

  2. John Hekman
    Posted Sep 5, 2006 at 2:44 PM | Permalink

    Instead of “none of them provide reliable information”, it might have more chance of acceptance if you said “the conclusions drawn from these studies are not supported by the statistical results, especially given the recommendation of the NAS panel to avoid the use of strip-bark samples.”

  3. TAC
    Posted Sep 5, 2006 at 3:39 PM | Permalink

    The recent report of the National Research Council (NRC) panel made includes several important many welcome recommendations about related to climate reconstruction methodology. Two key conclusions are including concluding that strip-bark samples (bristlecones/foxtails) are unreliable and should be avoided in temperature reconstructions and that methods for quantifying reconstruction uncertainty need to account for non-robustness to subset selection should be considered in confidence estimation. Inconsistently, their report relied on and illustrated studies that are affected by these problems. Here we consider the impact that the NRC recommendations, if they had been followed, would have had on several prominent climate reconstructions. Applying the NRC-recommended methods, we find that We demonstrate the impact of implementing NRC panel recommendations on prominent reconstructions, showing that none of them provide reliable information on relative medieval-modern temperature levels the statistical results no longer support the conclusions originally drawn from these studies. are not by the statistical results, especially given the recommendation of the NAS panel to avoid the use of strip-bark samples.

  4. Pat Frank
    Posted Sep 5, 2006 at 3:41 PM | Permalink

    “including concluding” is a little awkward. As alternatives, you could eliminate “including” entirely without changing the message, or else write, ‘… methodology, including the conclusion that…”

    On a more general note, I’d be interested to know after the fact, whether anyone at the conference raised the issue that there is no quantitative theory of tree-ring formation that allows the extraction of temperature numbers from tree ring widths. Perhaps you could comment on that when you return, Steve M.?

    The only way I currently see to possibly extract that information is by way of C12/C13 ratios related to the isotope effect of cellulose/lignin formation.

    Good luck, in any case.

  5. Mark H.
    Posted Sep 5, 2006 at 4:20 PM | Permalink

    What the heck, here is my draft based on TAC’s draft:

    The recent report of the National Research Council (NRC) panel includes many welcome recommendations regarding climate reconstruction methodology. Two key recommendations are that strip-bark samples (bristlecones/foxtails) are unreliable and should be avoided in temperature reconstructions, and that current methods need to account for non-robustness to subset selection in confidence estimation. Therefore, we propose to consider the application of those recommendations to several prominent climate reconstructions. Applying the NRC-recommended methods, we demonstrate the impact of implementing NRC panel recommendations on prominent reconstructions, showing that none of them provide sufficiently reliable comparisons on relative medieval-modern temperature levels, and the statistical results no longer support the conclusions originally drawn from these studies. We will also propose certain methodological approaches and additional data that could reduce those uncertainties.

  6. Gary
    Posted Sep 5, 2006 at 5:51 PM | Permalink

    I agree with the intent of the last sentence in #5. Criticizing the failures of past reconstructions only goes half way. Setting out the criteria necessary for statistical reliability in future reconstructions is where this discipline needs the most help.

  7. fFreddy
    Posted Sep 6, 2006 at 7:20 AM | Permalink

    If the format permits, it might be useful to conclude your presentation with “Some other questions that have occurred to me in the course of this research…” or some such.
    You could then mention some of the non-statistical issues that you have talked about here, such as the “divergence problem”. This might help to encourage some work on these issues by people in the relevant fields.

  8. Steve McIntyre
    Posted Sep 6, 2006 at 9:50 AM | Permalink

    Thanks for the feedback. I’ve incorporated most of the suggestions, but I’m not going to suggest new methodologies in a presentation that’s 10 minutes long.

  9. BradH
    Posted Sep 7, 2006 at 12:10 AM | Permalink

    Hey Steve,

    Whatever this meeting is like, I’ll bet it’s nothing compared with the climate change conferences they throw Downunder!

    Climate conference strip show storm

  10. Steve McIntyre
    Posted Oct 25, 2006 at 4:17 PM | Permalink

    My submission has been accepted. Making progress at AGU -poster session 2004; PP oral session 2005; All-Union session 2006.

  11. bender
    Posted Oct 25, 2006 at 5:19 PM | Permalink

    Truth will prevail. That is what makes the nation great.

  12. welikerocks
    Posted Oct 25, 2006 at 6:45 PM | Permalink

    #10 Right on Steve! thank you again for what you do!

  13. jae
    Posted Oct 25, 2006 at 6:49 PM | Permalink

    Congrats again!

  14. Posted Oct 25, 2006 at 7:27 PM | Permalink

    Congratulations Steve. I just hope more of the press will listen and report your findings. The truth just does not seem to fit their global warming templet.

  15. TAC
    Posted Oct 25, 2006 at 8:09 PM | Permalink

    Steve: Congratulations!

    It is interesting to read the abstracts of your session-mates (here). Mann et al. claims to have “produced a revised set of global and hemispheric mean surface temperature reconstructions for past centuries using a newly developed network of long-term climate proxy data.”

    Should be a lively session, anyway ;-)

    I look forward to seeing you there.

    p.s. Al Gore will be speaking later in the week (here) on “Climate Change: The Role of Science and the Media in Policymaking”.

  16. Ken Fritsch
    Posted Oct 25, 2006 at 8:51 PM | Permalink

    Congratulations Steve M. Just keep being Steve M. I look forward to your observations from the conference and the personality sidelights.

  17. bender
    Posted Oct 25, 2006 at 8:55 PM | Permalink

    I like the sound of the Wilson et al. abstract:

    No current tree-ring (TR) based reconstruction of Northern Hemisphere (NH) temperatures, that extends into the 1990s, captures the full range of late 20th century warming observed in the instrumental record. Over recent decades, a clear divergence between large-scale reconstructed (colder) and instrumental temperatures (warmer) is observed. We hypothesize that this problem is partly related to the fact that some of the constituent chronologies used for previous reconstructions show divergence against local temperatures in the recent period. In this study, we compiled TR data and published local-regional reconstructions that show no divergence against local temperatures. These data have not been included in other large scale temperature reconstructions. Utilizing this data-set, we developed a new completely independent reconstruction of NH temperatures (1750-2000).

  18. Steve McIntyre
    Posted Oct 25, 2006 at 9:34 PM | Permalink

    Rob is certainly trying. A 1750-2000 reconstruction sounds like even further retrenchment than the NAS Panel’s 1600.

    Speaking of divergence, let’s not forget that the Hegerl reconstructions ends in 1960 (!?!) along with the Briffa reconstruction. They don’t breathe a word about the divergence problem.

  19. Rob Wilson
    Posted Oct 26, 2006 at 1:29 AM | Permalink

    Dear Steve,
    My new study is not a “further retrenchment than the NAS Panel’s 1600″. I am purposely focussing on the recent period and the issue of divergence at large scales.

    Also – there is much criticism of the non independence between different studies describing NH temperature reconstructions. The reality is that there are few ‘relevant’ millennial length series. If we are focussing on the last 1000 years, it is impossible for the studies to be independent from a data point of view. So essentially, the differences come down to method.

    See you in SF

    Rob

  20. Willis Eschenbach
    Posted Oct 26, 2006 at 2:08 AM | Permalink

    Rob, your paper sounds fascinating. If you have time, could you comment on how you have dealt with

    1) The “upside-down U” shaped growth response of plants to temperature/moisture, which leads to a “many-to-one” one-way transform of the temperature data in to the tree ring data.

    2) The separate problem of removing the effects of moisture (rainfall, humidity, and soil moisture) as a confounding variable in the tree-ring records.

    Thanks,

    w.

  21. Rob Wilson
    Posted Oct 26, 2006 at 2:43 AM | Permalink

    Dear Willis,
    at this stage, as the paper has not yet been submitted (hopefully next week), I would prefer not to comment on specifics.

    Essentially, there has been a lot of unreasonable negative publicity towards tree-ring data and their use as temperature proxies. The divergence issue is a serious problem, but it is not as bad as some would have everybody believe. There are many recent TR based reconstructions that show no divergence in the recent period, yet much attention has been focussed on a few studies that do show it. There are many possible reasons for divergence (more than the two you have noted) and the phenomenon needs to be examined at the site level.

    My paper is essentially a descriptive study using new data that come into the 1990s and beyond (new reconstruction covers the period 1750-2000) and discusses the difficulties of modelling the high temperatures in the late 1990s.

    I hope you can wait

    Rob

  22. Chris H
    Posted Oct 26, 2006 at 3:41 AM | Permalink

    #21 Rob,
    You state that there are TR reconstructions that show recent divergence due to a variety of factors. Does this means that these reconstructions acted as a good temperature proxy for a period of time but then stopped being a good proxy for some reason? If so, what’s to say that other TR reconstructions that act as good temperature proxies during the instrumental period don’t have periods of divergence in the past for similar reasons?

    For example, I can imagine that a particular TR reconstruction might have diverged recently because of the increase in atmospheric CO2 but I can also imagine a reconstruction breaking down for several decades at some time in the past due to an exceptionally hot dry period. Is there any method for excluding such divergences?

    Thanks

    Chris

  23. Rob Wilson
    Posted Oct 26, 2006 at 4:14 AM | Permalink

    yes – that is a valid statement. However, palaeoclimatology must rely on James Hutton ‘s principle of uniformitarianism. If we did not, then we would develop no reconstructions.

    So – worry about problems if they are noted in the calibration period, but if calibration/verification is robust, we have no choice but to trust the transfer function back in time.

    Of course, this is where comparison between different independent reconstructions comes in to try and address your statement.

  24. Stan Palmer
    Posted Oct 26, 2006 at 8:31 AM | Permalink

    Would Prof. Wilson please comment on these two statements extracted from two his recent postings. On their face, tey seem to be saying quite different things. Would you please supply more context.

    In response to crticism that temperature reconstructions are not indepndendent

    “The reality is that there are few “relevant’ millennial length series. If we are focussing on the last 1000 years, it is impossible for the studies to be independent from a data point of view.”

    In reponse to a query as to whether the tree ring transfer function is stable

    “Of course, this is where comparison between different independent reconstructions comes in to try and address your statement.”

  25. bender
    Posted Oct 26, 2006 at 10:16 PM | Permalink

    Re #21

    The divergence issue is a serious problem, but it is not as bad as some would have everybody believe.

    I’d be interested to hear more on this subject. What exactly is the problem, and, if it’s serious, then where exactly are the critics going too far in their concern? (e.g. Is it serious, but solvable?)

    I have a specific question on bristlecone pines too that is somewhat related. I would like to know about the possibility that drought during the MWP led to bristlecone dieback & mortality. I understand pinyon pines are currently getting hammered by drought. How are the bristlecones faring? Has their growth rate declined any?

  26. Steve McIntyre
    Posted Oct 26, 2006 at 10:48 PM | Permalink

    #25. bender, that is the view of Lloyd and Graumlich 1997, which I’ve posted up here http://data.climateaudit.org/pdf/lloyd.graumlich.1997

    This article is in the news recently, mentioned in both Hegerl et al and Juckes et al as the citation for the tow foxtail series, Boreal and Upper Wright Lakes. Needless to say, the citation by the gang that can’t shoot straight is erroneous. There is no mention of either series in Lloyd and Graumlich. They are discussed in Bunn et al 2005, where they are the two most HS shaped series.

  27. Posted Oct 26, 2006 at 11:48 PM | Permalink

    Mann, Ritson, Ammann, …

    Small talk can be a big challenge, but a little preparation and confidence is all you need :)

  28. Rob Wilson
    Posted Oct 27, 2006 at 2:30 AM | Permalink

    Dear Stan,
    you are taking both quotes a little out of context.

    the first was related to the none independence between the swath of NH reconstructions.
    It would be nice to develop independent NH reconstructions, but there is simply not enough millennial long series to do this at this time

    the second comment was related to a comment about how to check for nonlinearities in the past prior to the calibration period. I said that we must relay on the Uniformitariansm principle – i.e. if the calibration is robust, we must assume that the reconstructed signal is robust back in time. However, I then said that one can get some further validation of a reconstruction by comparison against an independent reconstruction. This was a general idealistic statement and is unfortunately not possible with the NH reconstructions (see comment above). However, it is a relevant statement at local/regional scales. For example see the strong coherence between Alpine temperature reconstructions.
    Rob

Follow

Get every new post delivered to your Inbox.

Join 3,246 other followers

%d bloggers like this: