from: "Rob Wilson" <rob.wilsonatXYZxyzac.uk>
subject: Re: Re: Fwd: RE: FW: divergence problem]
thanks for forwarding your message.
I tried to ignore Marcel's initial e-mails but he phoned me at home so I said I would write
He was really fishing for me to criticise your truncation. Hopefully my answer was
It looks like I will be co-chairing a session on Divergence at the AGU with Mike Evans
(Rosanne probably will not be there) and it would be great if you (or Tom?) could make it
over to give a presentation on your recent work in this regard.
I will contact you when we know for sure that the session will go ahead
----- Original Message -----
Sent: Tuesday, July 10, 2007 1:20 PM
Subject: [Fwd: Re: Fwd: RE: FW: divergence problem]
---------------------------- Original Message ----------------------------
Subject: Re: Fwd: RE: FW: divergence problem
Date: Tue, July 10, 2007 1:18 pm
To: "Keith Briffa" <email@example.com>
I am putting a few thoughts down on paper - they are in noe particular
order . We can talk later also
as for the "innapropriate" response to the suggestion to show the Briffa
reconstruction "after 1960" - there is no reconstruction from these data
after 1960 to show. The authors did not include this as their exploration
of the tree-ring density data used clearly showed a low-frequency
divergence between the chronologies being used and the regional summer
temperature against which they were being compared. There were sufficient
overlaps between the available pre-1960 data to demonstrate strong
associations at high (inter-annual) and medium (decadal scale) timescales
to provide support for the value of presenting the reconstruction based on
The Figure in Chapter 6 of the AR4 report , was seeking to provide a basis
for comparing pre-20th century temperatures with those known to have
occured (on the basis of instrumental measurements) in the last century or
two. The data shown were all as provided by published articles (in the
peer-reviewed literature) . At no time was the issue of the apparent
divergence ever "hidden" or its implications - these were discussed
elsewhere in the Chapter.
The specific reconstruction (of summer temperatures based on Tree-ring
density data ) is explicitly representative of restricted land regions in
the northern high latitudes and some high-elevation regions further south
(- in the western US and the Alps). It is only the "northern" series that
show the divergence. The secondary scaling of the regional series to
represent the best estimate of the Northern Hemisphere (NH) as a whole
clearly depends on the association between the average of of tree growth
with this restricted distribution and the "true" hemisphere temperatures
remaining stable through time. Other reconstructions use wider spaced ,
and different , often more varied types, of proxy records . Some use
actual temperature series (or other climate series e.g. precipitation)
among their predictors. It is not surprising that these do not show the
The data used in this Briffa et al reconstruction sought to provide
independent evidence of past temperures from those. The year-to -year
match with temperatures is seen to be strong , even post 1960, and the
strong warming seen in the northern instrumental records between 1920 and
1940 is generally well matched even in the density data. It is for this
reason that we suspect the divergence might be a problem only associated
with recent decades. In such a situation it would be wrong to ignore this
and calibrate using all the modern data - risking serious biasing of the
calibration relationship used to infer past temperatures.
Subsequentlty other researchers have reported "divergence"
phenomena , but again associated with high latitudes only.
There is as yet, no definitive answer or even concensus that these studies
represent the same phenomenon. Most suggested "solutions" (see Rob's
comments)to the cause are problematic and it is important to study the
nature and possible causes further. At present such studies are hampered
by a lack of recent tree-ring and tree-density data (especially post 1980)
. The answer may lie in a mixture of methodological and biological
However, it is very important to recognise that a number of the
reconstruction studies use tree-ring series that are not affected by this
problem (crucial among these are tree-ring records in Scandinavia, west
and central Siberia). Many of the predictors in the published
reconstructions included in the IPCC Figure were not affected.
Tree-ring based and virtually all proxy reconstructions (including of the
NH) are subject to large statistical uncertainty, arising out of
diminishing quality and coverage of predictors back in time. The methods
used to translate these data into quantitative estimates of past
temperatures also assume uniformitarianism in the relationships between
predictors and predictand. This is hard if not impossible to prove. There
is a myriad of environmental changes occuring as a result of the
activities of human kind that might compromise this assumption :
increasing carbon dioxide, increasing acidification of terrestrial
environments; changing atmospheric clarity; changing ozone and hence uV
levels, to name only some. The ecology of all biological proxies must be
futrther explored in this context.
However, NONE OF THIS IS HIDDEN in the work of researchers or in the IPCC
AR4 report. On the contrary, palaeoclimatologists actively seek to
research and understand the implications of these factors. They also seek
to understand the implications , biases and limitations of the statistical
techniques they use.
What is not acceptable though , is to throw up ones hands and declare that
" nothing can be done because of these problems".
We strive to overcome the problems and in the meantime , make the best
interpretation of the available data that can be achieved , taking into
account the uncertainties. The latest AR4 report tries to do just this -
express the best concensus in the clear understanding that there are these
problems. The issues and uncertainties are made explicit. Look at the text
and the "key Uncertainties" section .
I am not in the office at present - if you need to discuss this further,
please ring me at
+44 1953 851013 before 4 pm my time
>>Subject: RE: FW: divergence problem
>>Date: Mon, 9 Jul 2007 15:22:14 +0200
>>Thread-Topic: FW: divergence problem
>>From: "Crok, Marcel" <firstname.lastname@example.org>
>>To: "Keith Briffa" <email@example.com>
>>X-OriginalArrivalTime: 09 Jul 2007 13:22:15.0116 (UTC)
>>Thanks for your reply. Tomorrow afternoon would be fine. I will call
>>you, let's say at 2 pm your time? Is your number 1603593909?
>>Van: Keith Briffa [mailto:firstname.lastname@example.org]
>>Verzonden: maandag 9 juli 2007 15:02
>>Aan: Crok, Marcel
>>Onderwerp: Re: FW: divergence problem
>>have been away - today busy all day - can read Rob's comments and
>>comment myself and perhaps ring you - but not til tuesday
>>Hope this ok
>>At 09:56 09/07/2007, you wrote:
>> >Dear Keith,
>> >I sent you an email last Friday about the truncation of your
>> >reconstruction after 1960 in AR4. Meanwhile I discussed the topic
>> >with your colleague Rob Wilson (see his answers below). For my
>> >article I would really like to have a reaction from you as well. Rob
>> >explained that things are very sensitive. I understand that, but I
>> >think that makes it even more important for you to give a reaction
>> >to the press yourself.
>> >For me as a journalist the comment that it is "inappropriate" to
>> >show the reconstruction after 1960 makes no sense. The
>> >reconstruction is real and therefore should have been shown. You
>> >then can try to explain why things happen. Could you comment on that
>> >by email or could I phone you today? I would also appreciate your
>> >comments on the questions that I sent to Rob.
>> >My article will be written in Dutch and will be published in the
>> >September issue of Natuurwetenschap & Techniek. I have no plans
>> >right now to make a translation, but if I do, I will sent you a
>> >draft so you can check your own quotes.
>> >Many thanks in advance,
>> >Marcel Crok
>> >Van: Rob Wilson [mailto:email@example.com]
>> >Verzonden: zaterdag 7 juli 2007 10:25
>> >Aan: Crok, Marcel
>> >Onderwerp: Re: divergence problem
>> >Dear Marcel,
>> >some brief answers to your questions are below (in red).
>> >I hope they will be of use for your article.
>> >recently the review comments of the AR4 were released. McIntyre
>> >asked to show the Briffa reconstruction after 1960. This comment was
>> >rejected saying it was "inappropriate" to show. Do you agree with
>> >that or would you have included the section?
>> >2) Briffa et al. have presented limited evidence in their work
>> >to show that the divergence noted in their regional and large scale
>> >reconstructions may be anthropogenic related and therefore
>> >restricted to the recent few decades. If true, then the post 1960
>> >truncation is valid.
>> >How do you explain that Briffa's reconstruction which is based on
>> >hundreds of series is going down after 1960 while most (if not all)
>> >of the other constructions go up?
>> >5) We need to be careful on which reconstructions you are
>> >talking about. Briffa (2000) presented a ring-width (RW) based
>> >extra-tropical reconstruction that shows divergence only after the
>> >mid 1980s. This is similar to other RW based reconstructions
>> >developed since (e.g. Esper et al. (2002), D' Arrigo et al. (2006)
>> >with further discussion in Wilson et al. 2007). In 2001, Briffa
>> >published a set of regional (and hemispheric) scale reconstructions
>> >that were based on maximum density (MXD) data.It is the NH
>> >temperature reconstruction from this study that Macintyre has been
>> >critical about with the post 1960s truncation. The divergence
>> >phenomenon in MXD data seems much worse than for RW. BUT, the signal
>> >in MXD is generally much stronger. Catch 22.
>> >6) The divergence phenomenon, whether for RW or MXD is complex.
>> >To date, no one cause has been implicated and it is likely that the
>> >divergence issue is a phenomenon which is related to multiple
>> >factors over different regions and species. I could quote you as
>> >many studies that show no divergence as those that do.
>> >7) Some possible hypothesised causes are: (1) non linear
>> >effects - e.g. there is a thermal threshold (i.e. it is getting to
>> >warm and some other parameter (e.g. precipitation) becomes limiting
>> >to growth; (2) anthropogenic affects (e.g. effects of pollution
>> >etc); (3) increasing CO2 may not result in fertilisation but the
>> >opposite effect as it may reduce the water use capacity of the tree
>> >and result in moisture stress; (4) issues related to the
>> >statistically processing of the tree-ring data - i.e. end effect
>> >issues when the TR data are detrended to remove biological age
>> >biases in the series; (5) Urban Heat Effects. If there is a positive
>> >bias in the instrumental data, then this could affect the
>> >calibration with tree-ring data; (6) wrong target season - some
>> >studies at large scales target annual temperature when in reality
>> >tree-ring data portray a summer temperature signal. Reconstructions
>> >also calibrate against mean temperatures although the bulk of
>> >cambial activity is in the day time. Day time maximum temperatures
>> >might therefore be a more realistic parameter to target. Increasing
>> >temperatures are observed more at night than the daytime.
>> >8) I could go on.........
>> >In the SPM of AR4 they concluded: "Studies since the TAR draw
>> >increased confidence from additional data showing coherent behaviour
>> >across multiple indicators in different parts of the world. However,
>> >uncertainties generally increase with time into the past due to
>> >increasingly limited spatial coverage." Do you support this
>> >conclusion or do you think the divergence problem is so serious that
>> >this conclusion is premature?
>> >11) Yes, I would agree with this statement in general.
>> >Unfortunately, none of the millennial long NH temperature
>> >reconstructions are entirely independent so this observation is not
>> >completely valid, but many new independent regional reconstructions
>> >(e.g. from NW N. America, the Alps, China etc) clearly show late
>> >20th century conditions to be warmer than the last 1000 years.
>> >Confidence does decrease markedly back in time and I personally
>> >believe that prior to ~1400 we do not have enough proxy data to make
>> >robust comparisons between present conditions and the Medieval Warm
>> >Period (for example). More long data-sets are being generated all
>> >the time and it will not be long before we can further improve on
>> >the present swath of reconstructions. Current evidence, however,
>> >does suggest that the late 20th/early 21st century temperatures are
>> >higher than any time for the last 1200 years.
>> >12) I believe the divergence issue is extremely serious and more
>> >work is needed to explore reasons why it is occurring. I am chairing
>> >a session at the Fall AGU this December on this issue and we are
>> >trying to get all groups studying this phenomenon involved. However,
>> >as I said above, divergence is not noted at all locations in the
>> >Northern Hemisphere and with the inclusion of other proxy types
>> >(e.g. documents, corals, speleothems etc etc), it should be possible
>> >to minimise the limitations of each of the proxy types while
>> >improving the final reconstruction.
>> >14) What are the main causes of the divergence problem and how
>> >certain are you about these causes?
>> >16) see my response above.
>> >If the divergence problem is real, how serious a problem is this for
>> >millennial temperature reconstructions?
>> >19) Again - I think the above reponses covers this question. The
>> >issue is serious but I not think it represents an impossible hurdle
>> >that we cannot get around. If nothing else, it provides
>> >dendrochronologists with an exciting research challenge.
>> >21) I hope these answers will help
>> >22) best regards
>> >23) Rob
>> >Dr. Rob Wilson
>> >School of GeoSciences,
>> >Grant Institute,
>> >Edinburgh University,
>> >West Mains Road,
>> >Edinburgh EH9 3JW,
>> >Scotland, U.K.
>> >Tel: +44 131 650 8524
>> >26) Home Page:
>> >28) ".....I have wondered about trees.
>> >29) They are sensitive to light, to moisture, to wind, to pressure.
>> >Sensitivity implies sensation. Might a man feel into the soul of a tree
>> >for these sensations? If a tree were capable of awareness, this faculty
>> >might prove useful. "
>> >31) "The Miracle Workers" by Jack Vance
>>Professor Keith Briffa,
>>Climatic Research Unit
>>University of East Anglia
>>Norwich, NR4 7TJ, U.K.
> Professor Keith Briffa,
> Climatic Research Unit
> University of East Anglia
> Norwich, NR4 7TJ, U.K.
> Phone: +44-1603-593909
> Fax: +44-1603-507784