from: Tim Osborn <t.osbornatXYZxyz.ac.uk>
subject: Re: O&B 2006
to: Gerd B�rger <gerd.buergeratXYZxyz.fu-berlin.de>
Hi again Gerd,
regarding the filter settings, I thought the easiest thing was to
send you the code I used -- it's in IDL language but probably simple
enough to extract the key points even if you aren't familiar with
IDL! I used it with "thalf"=20 years. Please don't forward this
code on to any one else.
Regarding correlations with temperature, we used whatever had
previously been published, whether annual or decadal. But for those
series (mostly from Esper) for which correlations weren't already
published, we calculated our own. For these we used annual
resolution, because that is obviously preferable for sample-size
reasons. (In most cases, at least, annual time scale is preferable,
but if you have a proxy that is not capable of resolving variations
on that time scale, it would be unfair to expect it to do so). We
could have calculated our own correlations for all the proxies we
used, using annual time scale where appropriate. But our main
purpose was to try our different analysis method with a similar set
of proxies to those used by other studies, rather than to select a
very different set of proxies and change the analysis method at the
same time. Given that purpose, we accepted previously published
correlations as evidence of probably temperature sensitivity of the proxies.
Hope that clarifies things,
At 08:33 16/01/2007, you wrote:
>I can now almost replicate the analysis, except for the exact filter
>settings (esp., how many adjacent points for the ends?).
>With respect to proxy selection, why did you choose annual
>correlations in some cases and decadal in others, the latter not
>being easy to replicate.
Attachment Converted: "c:\eudora\attach\filter_cru.pro"
Dr Timothy J Osborn, Academic Fellow
Climatic Research Unit
School of Environmental Sciences
University of East Anglia
Norwich NR4 7TJ, UK