date: Mon, 19 Dec 2005 13:26:13 +0000
from: Gerard van der Schrier <schrieratXYZxyzi.nl>
subject: Re: "Bjerknes hypothesis" reprints
to: Tim Osborn <t.osbornatXYZxyz.ac.uk>
It would be nice (though not essential) to have them here: I've already
send the pdf and handed out print-outs of the article to quite a few people.
About work: I've been busy with preparing some of the work we promised
to do in the (new) RAPID proposal. However, most of the time has been
spend on the fingerprinting work.
I was not very happy with the relatively low skill of the fingerprinting
method in determining MHT (Meridional Heat Transport) changes. So I
first tried to redo the analysis with two response patterns, hoping to
be able to estimate the *meridional structure* of the oceanic heat
transport. That did not work. So I tried something less ambitous:
estimating oceanic heat transport at 30N using two response patterns
(instead of one). What I did is:
1) estimate the response pattern using regression techniques on MHT and
sea surface height (SSH)
2) you can then orthogonalize the MHT time series and SSH maps by using
ordinary Gram-Schmidt orthogonalization.
3) the second response pattern is determined using regression on the
*reduced* MHT and *reduced* SSH.
Unfortunately, that did not give the improvement I hoped it would. In
fact: using the fingerprinting with two response pattern does not
outperform the single-pattern estimates.
The motivation to search for ways to 'tweak' the fingerprinting method
into giving more reliable estimates are the recent results by Knight et
al (2005, GRL doi:10.1029/2005GL024233) and Bryden et al (Nature) who
an increase and decrease of the overturning over the past decades
respectively. It would be nice to contribute to this discussion with the
TOPEX/Poseidon data. We can still contribute to this, but with the
rather large error bounds.....
We aimed the study to comment on the predictability of trends in MHT
using SSH data. Initially, we wanted to construct a 'predictability
matrix' by making estimates and 95%-confidence intervals for 10, 20 and
40 year trends. I was wondering, wouldn't it be more relevant for the
paleo-perspective, to make a different predictability matrix. One that
gives estimates of error bounds for trends in MHT based on single or
multiple (point) reconstructions of SSH? This would be more closely
associated with your original idea (only based on the marine climate
rather than the terrestrial climate).
The question we might try to answer is then: what is the skill of
estimating trends in MHT based on reconstructions of SSH in key areas
(like west-Iceland or Bermuda).
I know that there are (or will be shortly) reconstructions of SSH from
west-Iceland (I'm not sure of Bermuda though) which would give some
extra relevance to this.
Will you be at CRU this week, and next week?
> Hi Gerard,
> just to let you know that your reprints for this arrived from the
> publishers of Climate Dynamics. I can send them on to you in The
> Netherlands, unless you're happy to wait until you visit next year to
> collect them in person? Please let me know.
> Best wishes and Happy Christmas!
> Dr Timothy J Osborn
> Climatic Research Unit
> School of Environmental Sciences, University of East Anglia
> Norwich NR4 7TJ, UK
> e-mail: t.osbornatXYZxyz.ac.uk
> phone: +44 1603 592089
> fax: +44 1603 507784
> web: http://www.cru.uea.ac.uk/~timo/
> sunclock: http://www.cru.uea.ac.uk/~timo/sunclock.htm
Gerard van der Schrier
Royal Netherlands Meteorological Institute (KNMI)
PO Box 201
3730 AE De Bilt