Thursday, December 29, 2011


date: Tue, 25 Feb 2003 15:10:09 -0000
from: "Max Beran" <>
subject: Tree rings and the Mann hockey stick
to: <>

Dear Keith

I deliver courses on global change in Oxford and area and one of the matters
that comes up is the Mann hockey stick and its implications (Mann-made
climate change:-). It has been given enormous prominence both in terms of
its message about the recent and "deep" past, and in terms of its portents.
Its use as the take-home message from the policymakers summary of the
IPCC-TAR demonstrates this clearly.

I am aware that the detailed form of the curve conflicts with what is known
about well attested features of the millennial climate (weak, if any,
signatures of medieval warming and the little ice age), but what is
exercising me more is what it says about trees themselves (I know it is
multi-proxy but as I understand it, dendrochronology rules).

That tree-ring contribution to the temperature reconstruction obviously uses
a numerical expression of the sensitivity between temperature and tree ring
width/density. I don't have any numbers but if these are anything like the
ones you show in Figure 2 of your 1998 Nature paper, then there is an
approximate one-to-one between the standardised departures of
April-September temperature and tree ring width or density.

Two areas of concern are (a) the situation up to the present, and (b) the

The present

Given that the standard deviation of the yearly values of summer average
temperature is of the order of 0.5 degree, this is coincidentally about the
same difference as between pre-industrial times and now. This implies that
there ought to have been a similar one standard deviation growth in tree
rings. (Again I've no access to real numbers but I guess we are talking a mm
or 3). At an individual site and year this is doubtless well within the
noise level, but would be expected to shine through when maintained over a
number of years and sites. I tried to compare this with Figure 7 of your
1998 Royal Society paper but got mixed up with whether this shows the annual
values of the BAI (as implied by the text) or the annual values of the
year-on-year "change-in-BAI" as in the caption. If the former, one might
have expected some sort of compound interest pattern, if the latter an even
faster growing pattern (compound compound). Perhaps the modesty of the rise
is indicative of the reversal of the sensitivity between tree rings and
temperature that is visible in the post-1940 data on Figure 6 of the Royal
Society paper.

How do you reconcile this reversal in the sign of the relationship between
tree growth and temperature, and Figure 6 in general, with the statements
elsewhere in the paper saying there has been a non-climatic "enhancement" of
tree growth?

If there has indeed been a reversal in the sign of the sensitivity this
would imply a very large reduction in NPP as a result of the conspiracy
between ring width and wood density. One might then ask why this post-1940
sensitivity is not a more reliable basis for backward reconstruction? I know
you tend towards non-climatic explanations (notwithstanding my confusion
over the direction) but for my money this explanation could be at least as
legitimately aimed at the period from 1880 to 1940. Huge proportional
changes in land use and industrial pollution in that era make current
incursions look relatively speaking benign. Just look at population,
agricultural area, industrial outputs and emissions data to see this.

The future
The climate models, bless'em, indicate a temperature increase of the order
of less than 5 to more than 10 standard deviations by the 2080s. Accepting
the robustness of the sensitivities implicit in the Hockey stick
reconstruction (much used to tune and confirm GCM behaviour), that suggests
to me that we can anticipate a similar order of growth in tree ring width
and density? I can't picture what the standard deviation of the density
series might be in relation to the mean, but I would hazard a guess that
applying this to the tree ring width alone would lead to a more than
doubling of today's BAI. The overall effect on NPP of such a dramatic shift
in growth behaviour would surely turn the current 60-ish Gt to well over 100
Gt. If only a modest fraction was turned into NBU this could make a mighty
hole in emissions and would be good news at least over the lifetimes of the
trees. And all this is would put the benefit of CO2 fertilisation completely
in the shade.

Seems to me we have a classic checks-and-balances situation here. The
climate modellers (and the policy makers) implicitly accept the tree-ring to
climate sensitivity as far as the past is concerned. This bolsters their
belief in the forward projections of temperature with all that that implies
for impacts and policy. By their own logic, they should then also accept
that trees (far and away the dominant living carbon pool) would continue in
their positive temperature-driven response and provide a hefty negative
feedback acting via the land carbon cycle. In all seriousness though, does
anyone really believe trees would respond so dramatically. We'd know about
it from physiology and see some signal in latitude clines - as far as I know
they don't exist, but you'd know for sure.

So at what point does the tree ring to temperature sensitivity break down?
And what might its impacts be on the hockey stick and through that the GCM
tuning? Have there been other periods when your post-1940 reversal occurred
perhaps due to macroclimatic forces? Could these also account for the
discrepancy between the hockey stick and what we thought we used to know
about the climate since 1000 AD?

Any thoughts on any of the above would be delightedly received. You may even
save a soul from falling into the embrace of the sainted Lomborg!

Max Beran

1 The Croft
East Hagbourne
Didcot OX11 9LS
Tel: 01235 812493
Fax: 0870 054 7384

No comments:

Post a Comment