cc: Susan Solomon <ssolomon@al.noaa.gov>,  Kevin Trenberth <trenbert@ucar.edu>, Matilde Rusticucci <mati@at.fcen.uba.ar>, Phil Jones <p.jones@uea.ac.uk>,  Brian Hoskins <b.j.hoskins@reading.ac.uk>, Peter Lemke <plemke@awi-bremerhaven.de>,  Jurgen Willebrand <jwillebrand@ifm-geomar.de>, Nathan Bindoff <n.bindoff@utas.edu.au>, zhenlin chen <cdccc@cma.gov.cn>,  Melinda Marquis <Marquis@ucar.edu>
date: Tue, 09 Jan 2007 16:00:36 +0000
from: Brian Hoskins <b.j.hoskins@reading.ac.uk>
subject: Re: IPCC WG1 Observations Conference Call
to:  Martin.Manning@noaa.gov

<x-flowed>
Dear All

I kept quiet over the discussion about trend lines as I think the final 
result of not putting them on the 3 curves in the SPM is the right one. 
However I certainly defend the use of the 25, 50,100, 150 year trend 
lines on the temperature curve as in the TS and Chapter 3 as being 
better than the alternatives. A linear fit may not be a good one but 
when one is trying to make the smallest number of assumptions it is more 
defensible than for example putting in seemingly arbitrary break-points. 
The picture also gives a visual impression of how representative the 
average rates of change numbers are.The fact that the trend lines for 
different time-wondows are all different itself shows that the linear 
fit is not good for the longer time-scales.  The monotonically 
increasing slopes as one moves from the longer to the shorter 
time-scales is a strong indication of acceleration, but I would not put 
this in the same sentence as the word "unequivocal".

Best wishes

Brian


</x-flowed>
