cc: tom crowley <tomatXYZxyzan.tamu.edu>, "Michael E. Mann" <mannatXYZxyzginia.edu>, firstname.lastname@example.org, Jonathan Overpeck <jtoatXYZxyzrizona.edu>, Keith Briffa <k.briffaatXYZxyz.ac.uk>, email@example.com, firstname.lastname@example.org, email@example.com, srutherfordatXYZxyzginia.edu
date: Wed, 02 May 2001 11:51:24 -0400
from: "Michael E. Mann" <mannatXYZxyzginia.edu>
subject: Re: hockey stick
to: Edward Cook <drdendroatXYZxyzo.columbia.edu>
A few quick points Ed,
These "Wally seminars" are self-promoting acts on Broecker's part, and I
think the community has to reject them as having any broader significance.
If Broecker had pulled this w/ Ray, Malcolm, Keith, Phil, and Tom around,
he wouldn't get away w/ such a one-sided treatment of the issue. I've been
extremely troubled by what I have heard here.
I'm also a bit troubled by your comparisons w/ glacial advances, etc. and
how these correlate w/ your reconstruction. Malcolm, Ray, Phil, and others
have been over this stuff time and again, and have pointed out that these
data themselves don't support the notion of globally-synchronoous changes.
You seem to be arguing otherwise? And with regard to association w/
volcanic forcing, Tom has already shown that the major volcanic events are
captured correctly in the existing reconstructions, whether or not the
longer-term trends are correct or not...
Re the boreholes. Actually, if Tom's estimates are correct, and it is also
correct that the boreholes have the low-frequency signal correct over the
past few centuries, we are forced to also accept Tom's result that the
so-called "MWP", at the hemispheric scale, is actually even COOLER relative
to present than our result shows! That was clear in Tom's presentation at
the workshop. So lets be clear about that--Tom's work and the boreholes in
no way support Broecker's conclusion that the MWP was warmer than we have
it--it actually implies the MWP is colder than we have it!
Tom, please speak up if I'm not correct in this regard!
We are in the process of incorporating the borehole data into the
low-frequency component of the reconstruction. The key difference will be
that they are going to be calibrated against the instrumental record and
weighted by the spatial coherence within the borehole data rather than what
Pollack has done. I expect the results will be different, but in any case
I'll let Malcolm and Keith respond to the issues related to the
standardization of the Esper chronologies, though it immediately sounds to
me quite clear that there is the likelihood of of having contaminated the
century-scales w/ non-climatic info. Having now done some work w/
chronologies in disturbed forests myself now (in collaboration w/ Dave
Stahle), I know how easy it is to get lots of century-scale variability
that has nothing to do w/ climate. I imagine the reviewers of the
manuscript will have to be convinced that this is the case w/ what Esper
has done. I'm very skeptical. I'm also bothered that Broecker has promoted
this work prior to any formal peer review. There are some real issues w/
the standardization approach and there is a real stretch in promoting this
as a hemispheric temperature reconstruction.
Finally, what is the exact spatial distribution of the sparse data he used.
Scott R. drove home the point regarding the importance of taking into
account spatial sampling in his talk at the workshop. A sparse
extratratropical set of indicators, no matter how
locally-temperature-sensitive they are, will not, unless you're *very*
lucky w/ the locations, be an accurate indicator of true N. Hem temp. In
general it will overestimate the variance at all timescales. The true N.Hem
temperature (ie, weighted largely by tropical ocean SST) has much less
variance than extratrpoical continents. There may be a large apples and
oranges component to the comparisons you describe.
We've shown that are reconstructions in continental extratropical regions
have lots more variance and variability. It is, as we have all shown, the
averaging over many regions that reduces the amplitude of variability. Our
regional reconstructions show far more significant warm and cold periods.
But they cancel out spatially!
If a legitimate argument were to be made that we have significnatly
understiamted, within the context of our uncertainty estimates, the
amplitude of the MWP at the hemispheric scale, I'd be the first to accept
it (note that, as Phil et al pointed out in their recent review article in
Science, we do not dispute that temperatures eearly in the millennium,
within the uncertainty estimates, may have been comparable to early/mid
20th centurys--just not late 20th century temperatures).
Frankly though Ed, I really don't see it here. We may have to let the
peer-review process decid this, but I think you might benefit from knowing
the consensus of the very able group we have assembled in this email
list, on what Esper/you have done?
Comments or thoughts?
At 10:59 AM 5/2/01 -0400, Edward Cook wrote:
> >heard some rumor that you are involved in a non-hockey stick reconstruction
> >of northern hemisphere temperatures. I am very intrigued to learn about
> >this - are these results suggesting the so called Medieval Warm Period may
> >be warmer than the early/mid 20th century?
> >any enlightenment on this would be most appreciated, Tom
> >Thomas J. Crowley
> >Dept. of Oceanography
> >Texas A&M University
> >College Station, TX 77843-3146
> >979-847-8879 (fax)
> >979-845-6331 (alternate fax)
>As rumors often are, the one you heard is not entirely accurate. So, I will
>take some time here to explain for you, Mike, and others exactly what was
>done and what the motivation was, in an effort to hopefully avoid any
>misunderstanding. I especially want to avoid any suggestion that this work
>was being done to specifically counter or refute the "hockey stick".
>However, it does suggest (as do other results from your EBM, Peck's work,
>the borehole data, and Briffa and Jones large-scale proxy estimates) that
>there are unresolved (I think) inconsistencies in the low-frequency aspects
>of the hockey stick series compared to other results. So, any comparisons
>with the hockey stick were made with that spirit in mind.
>What Jan Esper and I are working on (mostly Jan with me as second author)
>is a paper that was in response to Broecker's Science Perspectives piece on
>the Medieval Warm Period. Specifically, we took strong exception to his
>claim that tree rings are incapable of preserving century time scale
>temperature variability. Of course, if Broecker had read the literature, he
>would have known that what he claimed was inaccurate. Be that as it may,
>Jan had been working on a project, as part of his post-doc here, to look at
>large-scale, low-frequency patterns of tree growth and climate in long
>tree-ring records provided to him by Fritz Schweingruber. With the addition
>of a couple of sites from foxtail pine in California, Jan amassed a
>collection of 14 tree-ring sites scattered somewhat uniformly over the
>30-70 degree NH latitude band, with most extending back 1000-1200 years.
>All of the sites are from temperature-sensitive locations (i.e. high
>elevation or high northern latitude. It is, as far as I know, the largest,
>longest, and most spatially representative set of such
>temperature-sensitive tree-ring data yet put together for the NH
>In order to preserve maximum low-frequency variance, Jan used the Regional
>Curve Standardization (RCS) method, used previously by Briffa and myself
>with great success. Only here, Jan chose to do things in a somewhat radical
>fashion. Since the replication at each site was generally insufficient to
>produce a robust RCS chronology back to, say, AD 1000, Jan pooled all of
>the original measurement series into 2 classes of growth trends: non-linear
>(~700 ring-width series) and linear (~500 ring-width series). He than
>performed independent RCS on the each of the pooled sets and produced 2 RCS
>chronologies with remarkably similar multi-decadal and centennial
>low-frequency characteristics. These chronologies are not good at
>preserving high-frquency climate information because of the scattering of
>sites and the mix of different species, but the low-frequency patterns are
>probably reflecting the same long-term changes in temperature. Jan than
>averaged the 2 RCS chronologies together to produce a single chronology
>extending back to AD 800. It has a very well defined Medieval Warm Period -
>Little Ice Age - 20th Century Warming pattern, punctuated by strong decadal
>fluctuations of inferred cold that correspond well with known histories of
>neo-glacial advance in some parts of the NH. The punctuations also appear,
>in some cases, to be related to known major volcanic eruptions.
>Jan originally only wanted to show this NH extra-tropical RCS chronology in
>a form scaled to millimeters of growth to show how forest productivity and
>carbon sequestration may be modified by climate variability and change over
>relatively long time scales. However, I encouraged him to compare his
>series with NH instrumental temperature data and the proxy estimates
>produced by Jones, Briffa, and Mann in order bolster the claim that his
>unorthodox method of pooling the tree-ring data was producing a record that
>was indeed related to temperatures in some sense. This he did by linearly
>rescaling his RCS chronology from mm of growth to temperature anomalies. In
>so doing, Jan demonstrated that his series, on inter-decadal time scales
>only, was well correlated to the annual NH instrumental record. This result
>agreed extremely well with those of Jones and Briffa. Of course, some of
>the same data were used by them, but probably not more than 40 percent
>(Briffa in particular), so the comparison is based on mostly, but not
>fully, independent data. The similarity indicated that Jan's approach was
>valid for producing a useful reconstruction of multi-decadal temperature
>variability (probably weighted towards the warm-season months, but it is
>impossible to know by how much) over a larger region of the NH
>extra-tropics than that produced before by Jones and Briffa. It also
>revealed somewhat more intense cooling in the Little Ice Age that is more
>consistent with what the borehole temperatures indicate back to AD 1600.
>This result also bolsters the argument for a reasonably large-scale
>Medieval Warm Period that may not be as warm as the late 20th century, but
>is of much(?) greater significance than that produced previously.
>Of course, Jan also had to compare his record with the hockey stick since
>that is the most prominent and oft-cited record of NH temperatures covering
>the past 1000 years. The results were consistent with the differences shown
>by others, mainly in the century-scale of variability. Again, the Esper
>series shows a very strong, even canonical, Medieval Warm Period - Little
>Ice Age - 20th Century Warming pattern, which is largely missing from the
>hockey stick. Yet the two series agree reasonably well on inter-decadal
>timescales, even though they may not be 1:1 expressions of the same
>temperature window (i.e. annual vs. warm-season weighted). However, the
>tree-ring series used in the hockey stick are warm-season weighted as well,
>so the difference between "annual" and "warm-season weighted" is probably
>not as large as it might seem, especially before the period of instrumental
>data (e.g. pre-1700) in the hockey stick. So, they both share a significant
>degree of common interdecal temperature information (and some, but not
>much, data), but do not co-vary well on century timescales. Again, this has
>all been shown before by others using different temperature
>reconstructions, but Jan's result is probably the most comprehensive
>expression (I believe) of extra-tropical NH temperatures back to AD 800 on
>multi-decadal and century time scales.
>Now back to the Broecker perspectives piece. I felt compelled to refute
>Broecker's erroneous claim that tree rings could not preserve long-term
>temperature information. So, I organized a "Special Wally Seminar" in which
>I introduced the topic to him and the packed audience using Samuel
>Johnson's famous "I refute it thus" statement in the form of "Jan Esper and
>I refute Broecker thus". Jan than presented, in a very detailed and well
>espressed fashion, his story and Broecker became an instant convert. In
>other words, Wally now believes that long tree-ring records, when properly
>selected and processed, can preserve low-frequency temperature variability
>on centennial time scales. Others in the audience came away with the same
>understanding, one that we dendrochronologists always knew to be the case.
>This was the entire purpose of Jan's work and the presentation of it to
>Wally and others. Wally had expressed some doubts about the hockey stick
>previously to me and did so again in his perspectives article. So, Jan's
>presentation strongly re-enforced Wally's opinion about the hockey stick,
>which he has expressed to others including several who attended a
>subsequent NOAA meeting at Lamont. I have no control over what Wally says
>and only hope that we can work together to reconcile, in a professional,
>friendly manner, the differences between the hockey stick and other proxy
>temperature records covering the past 1000 years. This I would like to do.
>I do think that the Medieval Warm Period was a far more significant event
>than has been recognized previously, as much because the high-resolution
>data to evaluate it had not been available before. That is much less so the
>case now. It is even showing up strongly now in long SH tree-ring series.
>However, there is still the question of how strong this event was in the
>tropics. I maintain that we do not have the proxies to tell us that now.
>The tropical ice core data are very difficult to interpret as temperature
>proxies (far worse than tree rings for sure and maybe even unrelated to
>temperatures in any simple linear sense as is often assumed), so I do not
>believe that they can be used alone as records to test for the existence of
>a Medieval Warm Period in the tropics. That being the case, there are
>really no other high-resolution records from the tropics to use, and the
>teleconnections between long extra-tropical proxies and the tropics are, I
>believe, far too tenuous and probably unstable to use to sort out this
>So, at this stage I would argue that the Medieval Warm Period was probably
>a global extra-tropical event, at the very least, with warmth that was
>persistent and probably comparable to much of what we have experienced in
>the 20th century. However, I would not claim (and nor would Jan) that it
>exceeded the warmth of the late 20th century. We simply do not have the
>precision or the proxy replication to say that yet. This being said, I do
>find the dismissal of the Medieval Warm Period as a meaningful global event
>to be grossly premature and probably wrong. Kind of like Mark Twain's
>commment that accounts of his death were greatly exaggerated. If, as some
>people believe, a degree of symmetry in climate exists between the
>hemispheres, which would appear to arise from the tropics, then the
>existence of a Medieval Warm Period in the extra-tropics of the NH and SH
>argues for its existence in the tropics as well. Only time and an enlarged
>suite of proxies that extend into the tropics will tell if this is true.
>I hope that what I have written clarifies the rumor and expresses my views
>more completely and accurately.
>Dr. Edward R. Cook
>Doherty Senior Scholar
>Lamont-Doherty Earth Observatory
>Palisades, New York 10964 USA
Professor Michael E. Mann
Department of Environmental Sciences, Clark Hall
University of Virginia
Charlottesville, VA 22903
e-mail: mannatXYZxyzginia.edu Phone: (804) 924-7770 FAX: (804) 982-2137