date: Mon, 19 Mar 2007 21:49:35 -0000 (GMT)
subject: Re: SPM figures
to: "Leonard Smith" <lennyatXYZxyzhs.ox.ac.uk>
Im Seoul this week - back at the weekend. All the detail
is in the chapters which will be available in May.
This doesn't help you now, I know. All the time series
in Ch 3 are decadally smoothed in the same way, which
is described in an Appendix. I can send you this when
back. Only the global and NH/SH series extend to the
end of 2006. The other series end whenever we got them
or could extend them to.
Smoothing goes to the end (and beginning) of the series
and is done using a method by Mann (2004? in GRL). This
reflects the appropriate number of years to get the
smoothed series to the ends. With temperatures likely
warmer in future years this is conservative at the
If you need a quicker response then email Kevin Trenberth
(firstname.lastname@example.org ?) and he can likely send you the
If you can wait till Friday then I can send it. Piers
probably has Appendix 3.A which describes the smoothing.
Why can't people wait till May !!
> dear both,
> brian hoskins suggested i contact you directly regarding figures SPM-1 and
> SPM-3 which are taken from your chapters; i am fielding various questions
> from the public, including some deniers, questioning why some graphs "end
> early". i expect this is clear in your chapters but it is not clear from
> the figure captions in the SPM.
> ending well before 2006 makes sense if the graphs are decadal averages
> (which of course always end half-a-window-width before the last data
> point; this is clearly the case in Fig SPM-4.).
> some confusion is caused by the small insets in fig SPM-1, where it is not
> clear from the caption what is being plotted (annual values, longer term
> averages?); could you tell me if these are averaged values and when the
> series end?
> also, i notice that in fig SPM-3 the smooth line ("decadal averages")
> extend to the end of the record! no one has flagged that yet, but i am
> sure they will and i would rather preempt with an explanation, esp if i
> am explaining why decadal averages end 5 years before the last data point
> in other figures.
> thanks much for your help.
> Prof Leonard A. Smith
> Director Senior Research Fellow
> Centre for the Analysis of Time Series (Mathematics)
> Department of Statistics Pembroke College
> London School of Economics Oxford OX1 1DW
> Houghton Street England
> London WC2A 2AE
> 020 7955 7626 (voice) 01865 270 517
> 020 7955 7416 (fax) 01865 270 515