cc: "Michael E. Mann" <mannatXYZxyztiproxy.evsc.virginia.edu>, tom crowley <tomatXYZxyzan.tamu.edu>, firstname.lastname@example.org, email@example.com, firstname.lastname@example.org, mhughesatXYZxyzr.arizona.edu
date: Tue, 24 Apr 2001 16:00:44 -0400
from: "Michael E. Mann" <mannatXYZxyzginia.edu>
subject: Re: Some perspective
to: Chick Keller <ckelleratXYZxyzp.ucsd.edu>
Unfortuntely, I don't have time (no or in the forseeable future) to address
your specific question, though I'll assert that they have all been dealt w/
in the peer-reviewed literature by me or my various colleagues. I believe
that Keigwin believes his evidence supports a strong NAO-like fluctuation
in the North Atlantic in past centuries. It should strike people that the
Sargasso Sea record looks nothing like n. hem mean temperature during the
19th and 20th century. Of course, why should it? If Keigwin is right, and
there is a strong NAO overprint in past centuries, then any hemisphere-mean
signal in precip, temperature, or circulation indicator throughout the
North Atlantic basin would be completely overprinted by the NAO and
essentially irrelevant to determining hemisphere mean temperature. You
might refer to Hurrell's work in this area. The correlation between the NAO
and annual-mean, full Northern hemisphere mean temperature is about 0.2 or
so. Based on discussing this w/ Keigwin, I think my and his interpretation
are consistent. Have you discussed this with him?You might want to get in
touch w/ him yourself if you have further questions about his work.
Other than that, I've referred your comments to other people who you might
want to consult w/ on this...
best of luck w/ your pursuits. Unfortunately, I have to sign off from the
discussion at this point,
At 12:35 PM 4/24/01 -0700, Chick Keller wrote:
>Take your time answering this as your schedule sounds really busy, but you
>write, "I hope this answers your questions". It helps but it
>doesn't. There are just too many exceptions and conundrums to say the
>issue is closed.
>I have read carefully all the papers you mentioned and more, and have been
>plowing thru Ray Bradley's book Paleoclimatology (wonder why he never
>mentions borehole methods??). The problem with tree ring
>"standardization" is dealt with in some detail in various Briffa
>publications, most recently in Briffa et al, Feb 2001, JGR also, Briffa,
>Annual climate variability in the Holocene: interpreting the message of
>ancient trees, Quaternary Sci. Rev, 19, 87-105, 2000 see figure 5),
>and Briffa et al, Trees tell of past climates, Phil. Trans. R. Soc. Lond.
>353, 67, 1998, Figs 2&3. Using Age Band Decomposition they show that how
>you do the standardization makes a big difference. So my first
>question--how does your 1999 reconstruction compare with the ABD method
>(can't find a discussion of this in either of your papers)?
>Second, I am fully aware of Keigwin's later paper and have discussed the
>pair on email with him. While he makes an interesting point about NAO
>connections to explain the differences between the Labrador and Sargasso
>records (very different proxies here), if I understood his paper, he is
>saying that the Labrador fan data is an anomalous result of local
>NOA-driven changes in the surface current that make it appear there is
>warming when the larger region is cooling. If that's what he's saying, it
>means we can ignore the Labrador fan in temperature reconstructions
>because it's not a real temperature change, just a minor redirection of a
> Consider that the Sargasso record agrees very well with the
> GRIP/Dye 3 borehole temperatures in showing a broad, multi-century
> warming centered around the year 1000 followed by a drop to cooler (LIA)
> temperature. This is consistent with the fact that temperature
> variations from Greenland (there's something wrong with the GISP II
> isotope record which shows little or no low frequency signal at all) saw
> temperature variations similar to those in the Sargasso Sea region
> regardless of what the foraminifera were doing off Labrador. And so I was
> asking what people are thinking about the Greenland borehole inversions
> since there obviously is no "land use changes" effect there.
>"Summer only" reconstructions show significantly cooler hemispheric
>temperatures in the time interval 1550-1700 as well as slightly higher
>temperatures around 1100.
>Finally, I read recently that the expected cooling from Milankovitch
>cycles is about 0.4�/millenium right now. But that's greater than the
>cooling you show, part of which must be due to solar activity decreases
>and volcanoes. So that also appears to be a problem. (I did a
>not-unreasonable BoE calculation including estimates of temperature change
>due to solar activity changes, volcanic effects, etc, and get that the
>warming around 1000 was about the same as in the 1940-50 peak.)
>Bottom line. I'm impressed with how extensive and carefully done your
>work is, but it seems to be missing some low frequency features.
> Just a personal observation -- Whenever I see a series of multi-proxy
> plots I can almost always pick out the tree ring ones. They're the ones
> with little or no low frequency variation, just low amplitude, high
> frequency variation. Hmmmm.
>In your Earth Interactions web stuff, you show your reconstruction (Fig 4)
>back to 1750 with and without tree rings. Do you have such a graph back
>to 1000, or are there too few non-tree rings records to make that comparison?
>Regardless, consider a disturbing statement from Briffa's Quat. Sci rev,
>p. 94, 2000: "it was not possible....to employ statistical techniques
>that preserve potential long-timescale climate variability for the large
>majority of locations (400 cool and moist sites, circum hemispheric).....
>they represent timescales of growth variability up to centennial, but for
>most sites, not longer."
> In another publication he describes the RCS method (not the same
> as his ABD method) of standardization and says that it helps the problem,
> but "The RCS method potentially provides much more information on
> century-to-century time-scale changes in temperatures than the previous
> approach, but the lack of very long temperature data means that it is
> simply not possible to verify this centennial time-scale information
> without reference to other evidence." ( Briffa et al, Trees tell of past
> climates, Phil. Trans. R. Soc. Lond. 353, 67, 1998)
>Also,in that same article he shows three records from Canadian Rockies,
>North Sweden, and the Urals all back at least 1,000 yrs. They don't
>generally agree, but they all show amplitude variations as large or larger
>than in the 20th century and all have peak to peak (50 year filters)
>amplitude of a degree C or more! (This raises an entire set of questions
>of how regional variations could persist for greater than 50 years yet be
>Crowley and Lowry do a reasonable job of combining multi-proxy records,
>and show it warmer 1000-1300 and cooler 1600-1800. But Tom told me he
>didn't use absolute amplitudes in his reconstruction but normalized all
>the records to one amplitude. I guess I don't understand how he did that
>or how Jones and Briffa, in their Tenerife Paper (ESA SP-463 Dec 2000)
>converted the results into temperature, but it sounds like this approach
>could also reduce total amplitude (for example they include the Keigwin
>Sargasso record but apparently greatly reduce its amplitude by normalization.
>The nagging point here is that there exist a non-negligible number of
>published proxy records that show temperature amplitudes much larger than
>those dominated by tree rings. It would be an important contribution,
>having shown the reconstructions you and others have, to drop the other
>shoe as it were and discuss why these other records individually (a)
>aren't trustworthy, or (b) are fine but were not included (here merely
>saying that they aren't high enough resolution is insufficient since what
>people are taking from your results is the low frequency result, or (c)
>aren't relevant. Without such a discussion people like Wally Broecker can
>say with relative impunity that only "elevation of mountain snowlines and
>borehole records are accurate over centuries to 0.5�C.
>I guess while you may not be the person to do it. I'd love to see a 1000
>yr temperature reconstruction from a large variety of records exclusive of
>tree rings, accompanied by a discussion of the accompanying estimated
>Let me list a few examples of what I'm seeing in the literature.
>Bradley in his Quat. Sci. Rev (2000) article shows two interesting records,
>(1) Fig 6c, Oxy isotope records in calcite (stalagmite from cave Mo i
>Rana in northern Norway--Lauritzen and Lundberg,1998
> delta T (last thousand years = 2.0�C
>(2) Figs 6b & 7, Treeline changes northern Ural Mtns. show temperatures
>1000-1300 were as high or higher than at present, and much colder 1550-1900.
> delta T (last thousand years ~ 0.5�C
>West African SST (deMenocal, et al, Science, 288, 2198-2202, 2000) faunal
>record of SST (20�N, 18�W)--two coolings ~1300 and 1600 with substantial
>warming between delta T (last
>thousand years ~4.0�C
>GRIP and Dye 3 bore hole reconstructions show broad warming centered near
>1000 and two major coolings 1500 and 1900
> delta T (last thousand years ~1.5�C
>>Unfortunately, you catch me just as I return from travel and am getting
>>ready to host a week-long workshop, so can't respond to you in detail.
>>However, I think several of your questions are addressed in the
>>publications i pointed out in the previous message I cc'd you. The issue
>>of sensitivity of our reconstructions to possible loss of low-frequency
>>information is addressed both in our '99 GRL article and, again, in our
>>Earth Interactions (online) article (there is a link to it from my web
>>page). I think this is largely a settled point now.
>>But to the other stuff: There is mounting evidence that European
>>temperature trends over the past millennium have largely reflected (as we
>>might expect) overprints due to changes in the NAO and North Atlantic
>>ocean-atmosphere processes. There is some modeling evidence of just this
>>claim (based on the response to solar forcing) that is in the pipeline,
>>(NASA/GISS group, pers. comm). Basically, the argument is that solar
>>forcing imposes an NAO pattern on top of more modest hemispheric/global
>>mean changes . The phase of the response is such that you get a positive
>>NAO anomaly (warmer European winters) when solar irradiance is low (your
>>"MWP" if you like) and a negative NAO during the "LIA". There is already
>>good empirical evidence (e.g. Keigwin stuff, such as I cite in my ''00
>>perspectives article that accompanied Tom Crowley's Science article).
>>You need to read the recent Keigwin and Pickart paper to see how
>>Keigwin's earlier record (Bermuda rise "Sargasso Sea" record) has been
>> So the point is that certain regions, like Europe, are likely to have
>> experienced considerably enhanced anomalies relative to the more modest
>> hemispheric mean trends. Undoubtedly, effects on ENSO, etc. further
>> complicate the possible spatial structure of temperature variations. But
>> neither the NAO or ENSO project much onto hemispheric or global annual
>> mean temperature. At least, not during the instrumental era, and that is
>> unlikely to have changed...
>>I hope that helps,
>>At 10:50 AM 4/13/01 -0700, you wrote:
>>Hope all is well with you. I'm doing fine here at Scripps immersed in
>>things climate. After having read much of the literature on recent
>>paleo-tempeatures (2,000 yrs ago to present), I am trying to write a
>>summary. The summary is nearly finished but has several nagging problems
>>that I thought I ask you about.
>>When I look at recent reconstructions of temperature over the past 1,000
>>yrs. I get a picture of a fairly low amplitude record which agrees on
>>little warming in the period 1000-1200, but disagrees somewhat on how
>>cold it got during the different so-called LIA episodes. Nevertheless,
>>the picture is looking like total warming since the late 1600s till 1945
>>is about 0.6� �0.1�C, and temperatures in the period 1000-1200 don't
>>exceed those in the middle of the 20th century.
>>So far so good.
>>Nagging problems with that:
>>There exist published records from around the world that suggest much
>>larger amplitude variations (at least regionally) and temperatures
>>900-1200 equal to or higher than at present.
>>The standardization problem with tree rings/density points to loss of low
>>frequency signal the longer back in time one looks. (which would lead to
>>unrealistically low amplitude records).
>>GISPII delta O 18 record is flat but GRIP and Dye II borehole records are
>>of very large amplitude (no land use problems on the Greenland Ice Cap so
>>boreholes might be pretty good there?), which could say you shouldn't use
>>GISP II as a temperature proxy for low frequency purposes.
>>Sargasso sea station S Keigwin) sediment record seems to corroborate GRIP
>>and Dye borehole records of large amplitude warming centered on 1000.
>>Similar other large amplitude records exist from other sites around the
>>Thus, proxy records dominated by tree rings and GISP II give low
>>amplitude, but one wonders how far you can push tree rings.
>>Now one argument is that when you average the large amplitude records,
>>because they are not synchronous, the average amplitude is much reduced.
>>But even then, the average would suggest an amplitude larger than from TR
>>So my questions:
>>Has subsequent work by you and colleagues answered any of these
>>questions? Just how much can we rely on the latest standardization
>>techniques to have preserved low frequency in tree ring?
>>In short, how comfortable are we that the low amplitude, not so warm MWP
>>result is robust, and why?
>>Charles. "Chick" F. Keller,
>>IGPP.SIO.UCSD - Attn: Chick Keller
>>9500 Gilman Drive
>>La Jolla, CA 92093-0225
>>(858) 822-1510 office
>>(858) 534-2902 FAX
>>(858) 456-9002 home
>>Is the noticeable increase in surfers off Scripps Beach a possible
>>indication of global warming?
>> 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
>Charles. "Chick" F. Keller,
>IGPP.SIO.UCSD - Attn: Chick Keller
>9500 Gilman Drive
>La Jolla, CA 92093-0225
>(858) 822-1510 office
>(858) 534-2902 FAX
>(858) 456-9002 home
>Is the noticeable increase in surfers off Scripps Beach a possible
>indication of global warming?
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