cc: Malcolm Hughes <mhughes@ltrr.arizona.edu>
date: Mon, 20 Mar 2006 13:55:44 -0500
from: "Michael E. Mann" <mann@meteo.psu.edu>
subject: [Fwd: Re: Trees]
to: Jonathan Overpeck <jto@u.arizona.edu>, Keith Briffa <k.briffa@uea.ac.uk>

   FYI, Mike

From: "Michael E. Mann"
Reply-To: mann@psu.edu
Organization: Dept. of Meteorology, Penn State University
To: Richard Alley
Subject: Re: Trees References:


   Hi Richard,
   Thanks, that sounds reasonable. Let me respond to one point, though. I rebuked Cuffey for
   asking the wrong question.
   I pointed out to him that we certainly don't know the GLOBAL mean temperature anomaly very
   well, and nobody has ever claimed we do (this is the question he asked everyone). There is
   very little information at all in the Southern Hemisphere on which to base any conclusion.
   So I told him that of course the answer to that question is *no* and it would be surprising
   if anyone answered otherwise. But, as I proceeded to point out, that's the wrong question.
   I pointed out that a far more sensible question is, "do we know the relative temperature
   anomaly for the NORTHERN HEMISPHERE to within that accuracy, and that we almost certainly
   do know that. I pointed to a comparison of about a dozen different reconstructions, at
   least one of them based entirely on non-tree ring information, all of which agree within
   about 0.2 C on average temperature of the 11th century relative to the late 20th century,
   and suggested that this reflects a reasonable estimate of the uncertainty, though some of
   the reconstructions are not entirely independent, so the non-independence needs to be taken
   into account in estimating the true uncertainty--but that this is almost certainly smaller
   than about 0.2C given the multiple estimates that roughly agree.
   I then pointed out that we don't need the global temperature to draw conclusions about the
   relative roles of different natural and anthropogenic forcings, because the models tell us
   that the NH alone is a good proxy for the response of the climate system to estimated
   forcings over the past 1000 years, and that the simulations (also about a dozen now) look
   very much like the reconstructions, within estimated uncertainties.
   I could be wrong, but I thought I sensed that the panel was quite satisfied with my answer
   (and my correction of Cuffey) and that they also probably recognized, in the context of my
   explanation, that Cuffey had been asking speakers the wrong question!
   I must confess I was not impressed by Cuffey's questions. I thought the other panel members
   asked more insightful questions. But that's just my view.
   talk later,
   mike
   The question by Richard Alley wrote:

   Mike--Thanks.  Comments embedded, I hope.
   On 3/19/06 4:53 PM, "Michael E. Mann" [1]<mann@meteo.psu.edu> wrote:

     Hi Richard,
     Thanks for your email, and for your earnest views. There was indeed considerable
     discussion of thes issues on friday, the day after your talk.  Both Malcolm Hughes and I
     discussed these issues in some detail with the committee.  Please feel free to take a
     look at the presentation I gave to the committee:
     [2]http://www.meteo.psu.edu/~mann/Mann/lectures/lectures.html
     Quite nice.  Thanks.
     There is no doubt that there are issues with the potential non-stationarity of tree
     responses to climate, and this introduces caveats. As I pointed out to the committee,
     these issues were actually stressed in our '99 article which produced the millennial
     temperature reconstruction, the title of which was (emphasis added) "Northern Hemisphere
     Temperatures During the Past Millennium: Inferences, Uncertainties, and Limitations
     <MBH1999.pdf> ". The underlying assumption of our own work has always been that each of
     the proxies have their own potential problems, and "multiproxy" approaches are probably
     the most robust. I don't have a particular axe to grind about any particular proxy, and
     recognize  that there are some pretty serious potential problems with all proxies,
     including ice core delta o18 (as you're aware, these are not clean paleotemperature
     proxies at all), and Sr/Ca or o18 from corals. There is a good discussion of the
     strengths and weaknesses in all of the proxies in Jones and Mann (2004): Jones, P.D.,
     Mann, M.E., Climate Over Past Millennia,  <JonesMannROG04.pdf> Reviews of Geophysics,
     42, RG2002, doi: 10.1029/2003RG000143, 2004.
     Agreed completely on value of multiproxy. And yes, a lot of my earlier work was on
     figuring out how much of the isotopic signal in ice cores is temperature and not other
     things.  The reassuring result was that all the big stuff is temperature, although with
     a rather bizarrely unexpected calibration.  Of the little stuff, stack several cores and
     you get up toward order of half of the variance being temperature with the rest left for
     something else.  The devil is in the details of when big meets little, as well as what
     calibration to use.  But, there is a pile of data from the 3% of the globe that is ice
     sheets that have not been assembled properly.  I've been trying in a quiet way to do
     what is possible so that the recommendations are forward-going; if the recommendations
     are appropriate, who knows, maybe money will become available to get someone to pull the
     data together.
     I won't try to defend Rosanne D'Arrigo's analysis, because frankly many in the tree-ring
     community feel it was not very good work.You should be aware that her selection criteria
     were not as rigorous as those used by other researchers, and the conclusions she comes
     to reflect only the data and standardization methods she used--they don't speak for many
     other, in my mind, more careful studies.  If you want the views of the leading experts
     in this community,  I would refer you to my colleagues Malcolm Hughes and  Keith Briffa,
     who have been carefully researching these issues for decades. With your permission, I'd
     like to forward your email to them for a more informed response--would that be ok?
     My comments may not be ideally suited for Keith.  When I sent inquiry to Overpeck (who
     coincidentally happens to be one of the convening lead authors of the IPCC paleoclimate
     chapter), he promptly sent it to several others including Keith.  Keith's reassurances
     were among those that I found less than reassuring--Rosanne probably screwed it up,
     there are some unpublished data that make it look better. (Unfortunately, in my switch
     from my old Sun to my new G4, I can't find that message and a couple of related ones.  I
     fear that I will repeat this statement a few times in the near future--what was supposed
     to be an afternoon changeover took three weeks, and a lot of things were in limbo during
     the interim.  I'm now trying to adapt a file system that developed on Suns since 1988 to
     a different computer...).  Ed Cook was also on the list; I can't for the life of me
     remember whether Malcolm was or not.  So fine to send there, and to Keith if you'd like,
     but it will probably insult him, and I would hate to do that.
     >From the questions asked by the community, I really only sensed from one individual the
     sort of extreme tree-ring skepticism that you describe. And I frankly think the
     individual proved himself to be not especially informed. The committee appeared to be
     convinced by the responses I provided to that individual. In short (and please see my
     presentation for further information) I made the following points: 1) multiproxy
     reconstructions that don't use tree-ring information at all for the long-term
     variability (Moberg et al, 2005) agree w/ all other (roughly a dozen now)
     reconstructions that late 20th century warmth is anomalous in the context of the past
     2000 years at the hemispheric scale. This is one key point (i.e, the take home
     conclusion doesn't depend on tree-rings at all!).  Another point I made is that the
     criticism (by some) that tree-rings underestimate the low-frequency variability is
     seriously challenged by the fact that temperature reconstructions based on data such as
     northern hemisphere glacier mass balance inversions (i.e. Oerlemans et al, 2005) show
     less hemispheric LIA cooling than many of the purely tree-ring based reconstructions.
     Another point I made in response to this line of criticism is that many of the long-term
     tree-ring series used in these reconstructions (see e.g. the recent Science article by
     Osborn and Briffa) show late 20th century conditions that are unprecedented in at least
     a millennial context. That is to say, if there is some upper temperature threshold in
     the past beyond which trees do not record, they do not appear to have encountered that
     threshold prior to the late 20th century, because the most positive anomalies in more
     than a thousand years are encountered in the last 20th century for most regions. The
     Osborn and Briffa science paper (attached) shows that the conclusion of anomalous 20th
     century warmth is spatially robust in a pan-hemispheric data set, it does not just
     reflect one region (in fact, they show that their conclusion of anomalous late 20th
     century warmth is robust to the elimination of any three data series used). Of course,
     Lonnie Thompson comes to the same conclusion using composites of his tropical ice cores.
     i.e., that the late 20th century behavior is anomalous in a greater-than-millennial
     context. Of course, tropical ice core delta o18 is difficult to defend as a pure
     paleothermometer too, but in this case it is difficult to see where non-climatic impacts
     could enter into the anomalous late 20th century behavior.
     I had read Osborn and Briffa with great interest when it came out. I attach my talk;
     Lonnie's data figured prominently.  His isotopic ratios have always worried me; he
     personally is coauthor on a paper pointing out that whatever they are, they probably
     aren't temperature.  I am more optimistic than that, but not much.  And his beautiful
     stack that looks so much like the instrumental record is almost entirely a couple of
     cores.  But what he does have are i) the loss of annual layering in Quelccaya after more
     than a millennium; ii) the appearance of melt at the top of Kilimanjaro after more than
     10 millennia; and iii) the appearance of organic matter from one of the outlets of
     Quelccaya after more than 5 millennia.  Unfortunately, (iii) is not published yet, (ii)
     is not as well published as it should be (he noted the appearance of melt in the paper,
     but the slide I showed is from him and not published, and the details of the physical
     nature of the cores are not out there), and (iii) is also not published--his papers note
     the existence of the annual layers and then their loss in isotopes, but the actual
     deepest level reached is not in the papers but given as a personal communication.  The
     difficulty of spending your life at high altitude or publishing Science papers is that
     the mundane parts don't come out so easily.
     So, in short, while the issues you mention are real (and have been emphasized by those
     actually working in this area for decades, as well as by us in all of our key
     publications), the primary conclusions (i.e. that late 20th century warmth is robust in
     at least a millennial context) appears robust, and is common to reconstructions whether
     or not they use tree-rings to reconstruct the low-frequency variability. . There is yet
     another study (embargoed right now in Science) that comes precisely to this same
     conclusion yet again.
     I'll actually be quite surprised if the committee comes to a *different* conclusion from
     that. Nonetheless, I appreciate your comments and your concerns, and your message does
     highlight a few issues which would be useful for us to clarify for the committee in case
     there is still any misunderstanding of the key points I have raised.
     I surely hope you're right, but I would wager a beer that they are less favorable than
     you'd like.  I don't know how you answered, but all the people I heard were asked the
     same thing: Do we know the temperature of a millennium ago within 0.5 C?  All gave some
     qualified version of "no".
     As I said, I'd like to be able to forward your message to Keith and Malcolm, if this is
     ok w/ you, so that they can provide a perhaps even better informed response to the
     criticisms you raise. So please let me know if that would be ok...
     As noted above, I want to get the science right, and if you think appropriate, go ahead.
      But, I fear that Keith will view them as insulting, and I don't want to do that.
     thanks,
     mike
     Thank you.  Also as noted above, I don't have time for this... But it is interesting.
     --Richard

--
Michael E. Mann
Associate Professor
Director, Earth System Science Center (ESSC)

Department of Meteorology              Phone: (814) 863-4075
503 Walker Building                    FAX:   (814) 865-3663
The Pennsylvania State University      email:  [3]mann@psu.edu
University Park, PA 16802-5013

[4]http://www.met.psu.edu/dept/faculty/mann.htm
