cc: Kevin Trenberth <trenbert@cgd.ucar.edu>, "Raymond S. Bradley" <rbradley@geo.umass.edu>, Keith Briffa <k.briffa@uea.ac.uk>, Tom Crowley <tcrowley@duke.edu>, Caspar Ammann <ammann@ucar.edu>, Phil Jones <p.jones@uea.ac.uk>, Michael Oppenheimer <omichael@princeton.edu>, mann@virginia.edu, jto@u.arizona.edu
date: Thu, 05 Jun 2003 22:48:03 -0400
from: "Michael E. Mann" <mann@virginia.edu>
subject: Re: EOS text
to: Tom Wigley <wigley@ucar.edu>

   Hi Tom et al,
   Wanted to comment briefly on some of the specifics below, just to make sure we don't get
   too sidetracked. This is a very interesting  and worthwhile discussion. In fact, these are
   precisely the kinds of issues that Phil and I are trying to sort out w/ the review paper
   we're writing for ROG [we'll probably be soliciting comments from many of you on different
   sections of that paper in the near future]. But I think its useful at this juncture to make
   a make a distinction between these sorts of scientifically interesting issues,  and the
   nonsensical arguments that SB03 are actually making.
   We can quibble, for example,  over the nature of the relationship between past variations
   in the surface temperature field, the atmospheric circulation, and the types of proxies
   that might inform our knowledge of each of these. I agree with Tom's point that in many
   case precipitation indicators don't tell us much at all about the surface temperature
   field, certainly  in the 'local' sense. In a sort of 'state space' sense, however, they may
   in some instances be quite helpful.  Winter drought-sensitive tree-ring chronologies
   provide us some of our best proxy information with regard to winter synoptic-scale
   variability in semi-arid regions like the desert southwest or the mediterannean. There
   appears to  been some success (i.e., demonstrated statistical skill) in reconstructing
   patterns of anomalous atmospheric circulation related to the usual suspect sorts of indices
   (PNA, NAO, etc.) from those sorts of proxies. To the extent that much of the regional
   winter season variability in the extratropical surface temperature field is related to
   these sorts of atmospheric circulation anomalies, one expects some skill in using these
   predictors to reconstruct features of the cold-season atmospheric circulation and, thus,
   regional temperature anomalies related to those features. I think a good case has been made
   that we can, perhaps, understand a good detail of the structure of the extratropical winter
   temperature anomalies during parts of the 'LIA' in terms of, e.g., the behavior of the
   NAO--a lot of evidence now seems to be pointing in that direction.   A similar argument can
   be made, for example, that a precipitation proxy in the western tropical Pacific may be an
   excellent predictor of SST variability in the eastern and central tropical Pacific, for the
   obvious reasons. So, in this larger-scale sense, there are some potentially useful
   relationships, and I agree with what Kevin says in this regard. Of course, it is also true
   that there are some obvious stationarity assumptions implicit in this sort of reasoning,
   and in  the use of any proxy precip/drought/atmospheric circulation information to infer or
   help reconstruction features in the surface temperature field. There are, however, similar
   stationarity assumptions implicit in the idea that a modest network (say, of a dozen) proxy
   surface temperatures over, say, the Northern Hemisphere, can be used to reconstruct
   hemispheric mean temperature. The implicit assumption is that the relative importance of
   each of a small number of locations in estimating the large-scale temperature field remains
   constant over time. As the number of regions sampled approaches the number of degrees of
   freedom in the surface temperature field, this because a better and better assumption. If
   were only talking about a handful of locations, it may be a pretty bad assumption. This
   sort of stationarity assumption is potentially just as, or even more (depending on the size
   of the network used) suspect than the former stationarity assumption, but is much more
   rarely discussed or acknowledge. Of course, there are ways to test these sorts of
   assumptions in a modeling context, and there are several studies now published, and others
   in the works, , that suggest the situation probably isn't as bad as we might have feared
   (again, something Phil and I will touch on in our ROG paper). See for example, these:

          Mann, M.E., Rutherford, S., Climate Reconstruction Using 'Pseudoproxies, Geophysical
          Research Letters, 29 (10), 1501, doi: 10.1029/2001GL014554, 2002.
          Rutherford, S., Mann, M.E., Delworth, T.L., Stouffer, R., Climate Field
          Reconstruction Under Stationary and Nonstationary Forcing, Journal of Climate, 16,
          462-479, 2003.
          Zorita, E., Gonzalez-Rouco, F., and Legutke, S., Testing the Mann et al. (1998)
          Approach to Paleoclimate Reconstructions in the Context of a 1000-Yr Control
          Simulation with the ECHO-G Coupled Climate Model, J. Climate, 16, 1378-1390, 2003.

   But these are all legitimate caveats, and interesting points, that would be great to
   discuss over some beers sometime, and which will be given more than adequate treatment in
   e.g. the review paper mentioned above.
   Unfortunately, that's not the task at hand. SB03 have no appreciation whatsoever for these
   sorts of subtle, legitimate considerations,  which involve thinking in  a much higher
   sphere  than the one they are thinking in, and certainly,  the one that they are playing
   to. Their logic is much more basic, and immensely less reasonable, than anything we're
   talking about here.
   Their logic, in essence, literally EQUATES hydroclimatic and temperature anomalies, since
   they hold that the existence of a large extreme in precipitation/drought in a particular
   region is as good as evidence of anomalous warmth, in support of the proposition of e.g. a
   "medieval warm period". So, in a very roundabout way, what I'm saying is, lets definitely
   not give these bozos more credit than they deserve!
   Unfortunately, we have precious little space in this Eos piece. Phil and I have a lot more
   space in our ROG article, and this sort of discussion will help us in making sure that
   these issues are adequately addressed there. I suspect that this longer review, and others
   that Ray and folks are working on, will be helpful in e.g. the next IPCC report. But for
   the time being, we have to keep things simple and to the point here. What we say of course
   needs to be rigorously defensible and we would like to educate the readers as much as we
   can in the short space available, but most of all we really have to do, in as simple terms
   as possible, is explain why the SB03 stuff is so fundmentally flawed. And, to boot, we have
   to do so in such a way that it seems more a casual consequence of what we say, than (as it
   is in fact) the central motivation of the article.
   So there is a real balancing act here, and thats what we're coming up against.  Let me do
   my best to strike this balance, and see if I can come up with a revised version that
   strikes the right balance between everyones concerns here. Again, I still need comments
   from several more people before I can attempt a revised draft. So responses (e.g. in the
   next day or so) would be greatly appreciated from those I haven't heard back from...
   thanks in advance,
   mike
   At 05:08 PM 6/5/2003 -0600, Tom Wigley wrote:

     Dear all,
     Re AGU's position, this is something I must have overlooked if it was in an earlier
     email. One way around this is to make the scientific error points and quote SB as an
     example of how not to do it (which one would have to do for at least three specific
     points).
     Re Kevin's suggestion, his text could be misinterpreted. It implies that one might be
     able to use wet/dry as a T proxy if the right statistical analysis were done first. I
     agree with what Kevin says, but I have looked at these sort of physically meaningful
     relationships and they are invariably too weak to use in a paleo context. For example,
     if the paleo indicator explains 50% of the precip (seasonal) variance (and such a high,
     independently validated value is rare), and if the r**2 for precip vs temp were similar,
     then we are left with 25% (at most -- the above assumptions are very optimistic). This
     is weak. Worse still, this assumes no paleo atmos circulation changes, also doubtful.
     The bottom line is that proxy precip data *cannot* be used as a T indicator except in
     the rarest of circumstances. Even in high latitudes there are problems -- see, e.g.,
     Bradley and England, late 1970s report (Ray, I'm sure you will remember this about the
     rareness of precip events).
     I think it is extremely dangerous to leave SB any loopholes here. In my view, what Kevin
     says does just this.
     Tom.
     _____________________________-
     Michael E. Mann wrote:

     Thanks Kevin,
     I've already made some revisions in response to your earlier comment about explicitly
     discussing the spatial variability issue with regard to the LIA/MWP. The prospective
     Figure 2 should help in this regard--looking forward to hearing back from Ray/Phil on
     that...
     I'll do my best to come up w/ a revised version that reflects everyones suggestions and
     wishes once all the comments are in,
     mike
     At 02:53 PM 6/5/2003 -0600, Kevin Trenberth wrote:

     Tom
     I agree with Mike that it is not possible to directly confront their methods in this
     way.  It can be confronted by stating clearly that cold periods that are not
     contemporaneous at different locations do not make for a cold hemispheric value:
     currently the article already makes this point to some extent but it can be made more
     directly relevant to SB. In fact it may be worthwhile pointing out that the LIA is
     defined by different authors to be in different periods precisely because they were
     looking at a different part of the world (like blind men exploring the elephant).
     And we can also say that it makes no sense to equate wet or dry period with cold or warm
     universally (ref SB).
     In fact what is found generally in mid lats is that warm in winter goes with wet
     (through moist and warm advection) and with dry in summer (drought and heat waves).  So
     seasonality matters a lot.  Maybe we can say womething like this:
     It is well established in current climate studies that warm conditions tend to accompany
     wet conditions in the extratropics in winter owing to the dominant role of the
     atmospheric circulation so that southerlies are warm and moist in the northern
     hemisphere while northerlies are cold and dry.  But in summer, the weaker atmospheric
     circulation means that moist thermodynamics is more important so that dry conditions
     favor warm spells and heat waves, as heat from the sun no longer evaporates moisture and
     instead  increase temperatures.   In the Tropics, during El Nino events, droughts occur
     in one part of the world (e.g. Australia) while wet conditions and floods occur in other
     parts (e.g. Peru), and the wet spots tend to switch with the dry spots during La Nina.
     Accordingly, there is no unique link between wet or dry with warm or cold conditions
     (such as erroneously assumed by SB).
     Not sure if this is useful but I offer it anyway.
     Kevin
     Tom Wigley wrote:

     Mike et al.,
     I will send tracked editorial suggestions later. In the meantime, what is lacking in my
     view is a clear statement at the start of the SB method. At present, the context of your
     later comments is a bit unclear to those who have not read the papers -- which will be
     the case for most readers. I suggest adding the attached before your point (1). What I
     say here overlaps with some things you say later, so minor changes are needed (which I
     will send later) to avoid clear duplication.
     We are using this to educate people about the good paleo work, but a key motivation is
     to demolish the bad stuff. I think, therefore, that the criticism of SB must be more
     focussed and specific -- which is why a statement of their work is essential. This
     suggested new material also provides a balance, and makes what we now have appear less
     self serving (which I know you are not trying to do, but there is still a hint of this).
     Tom.

     --
     ****************
     Kevin E. Trenberth                              e-mail: trenbert@ucar.edu
     Climate Analysis Section, NCAR                  [1]www.cgd.ucar.edu/cas/
     <[2]http://www.cgd.ucar.edu/cas/>
     P. O. Box 3000,                                 (303) 497 1318
     Boulder, CO 80307                               (303) 497 1333 (fax)
     Street address: 3080 Center Green Drive, Boulder, CO  80301

     ______________________________________________________________
                         Professor Michael E. Mann
                Department of Environmental Sciences, Clark Hall
                           University of Virginia
                          Charlottesville, VA 22903
     _______________________________________________________________________
     e-mail: mann@virginia.edu   Phone: (434) 924-7770   FAX: (434) 982-2137
              [3]http://www.evsc.virginia.edu/faculty/people/mann.shtml

   ______________________________________________________________
                       Professor Michael E. Mann
              Department of Environmental Sciences, Clark Hall
                         University of Virginia
                        Charlottesville, VA 22903
   _______________________________________________________________________
   e-mail: mann@virginia.edu   Phone: (434) 924-7770   FAX: (434) 982-2137
            [4]http://www.evsc.virginia.edu/faculty/people/mann.shtml

