cc: p.jones@uea.ac.uk
date: Thu, 12 Oct 2000 21:47:38 +0100
from: Tim Osborn <T.Osborn@uea.ac.uk>
subject: RE: seasonaliy
to: Keith Briffa <k.briffa@uea.ac.uk>

Hi Phil & Keith,

you should find (on the colour laser) some more plots.  Maps same as before 
but with better colours.
Grey means near to zero.  The oceans are often grey due to their lower 
variability, but I decided
that pre-normalisation of summer & winter data wasn't the right way to go.  
Also, new timeseries
of Apr-Sep, Oct-Mar and the difference, averaged over NH, NH-land, NH-ocean, 
and also some
quasi-frozen grids (essentially removing any boxes without data in either 
1860s or 1890s and
then computing land-only or NH averages.

>The winter warming over land in recent
>decades is clear, though the large area of slightly warmer ocean
>compensates remarkably.

Yes - the separate land & ocean curves now show this better.

>In fact, I am surprised at the general concurrence
>between the winter and summer curves. One wonders at what level of spatial
>coverage this begins to break down.

The correlations decrease somewhat when looking at land only, or when taking a 
fixed grid
(with smaller coverage).  No doubt the correlation will be smaller at the 
individual grid-box
scale.  Therefore, in the maps, many of the summer-winter differences will be 
real, even in
decadal means, and so it is hard to tell if there are any artificial biases 
given that we expect
some real variability.

>The anomalies in the first two decades
>are not that obvious as to indicate a clear regional bias , and it could
>well be that your suggestion of the poor spatial coverage causing the
>difference might be true. Phil suggested you do the frozen grid ( coverage
>of 1870s/80s) to redo the time series and I suggest you could do it also
>for the land only . The work by Chenoworth is pretty convincing that there
>is indeed bias in early western US records but this is not where you see it
>- in fact, your early summer warmth is as much a marine phenomenon.( and I
>for one am sceptical about the lack of exposure problems in the very early
>Scandinavian summer temperatures that Phil has assembled. We  note also
>that areas of ocean coverage appear relatively green (summer warmer) right
>through to the present. It is possible that there is a problem with the
>marine temperature records?

I suggest you re-examine this with the new colour plots and land/ocean time 
series.
They're much clearer.  There is an ocean difference, but not as large as the 
land one.
Look also at the big change from the 1890s to 1900s.  I'm not sure that any
external forcing change is going on then that could explain it - so if it's 
real then
one would expect a sustained (i.e. multi-decadal) change in atmospheric 
circulation
between the pre- and post-1900 periods.

>We don't think that you can resolve this easily

True.  I shall leave it for now.   The few plots I've done might even make a 
small GRL or
Atmospheric Science Letters paper, but I wouldn't want it to be too critical 
of the
instrumental data - I ought to read what Mike Chenoweth (is that the guy who 
was
also doing the analysis of the 1805-1824 period?) has done sometime.  Has he\
published his "north-facing wall" bias work at all?

> and we wonder why the extra
>20 years will make such a difference ?

Because how can we be critical of Crowley for throwing out 40-years in the 
middle of
his calibration, when we're throwing out all post-1960 data 'cos the MXD has a
non-temperature signal in it, and also all pre-1881 or pre-1871 data 'cos the
temperature data may have a non-temperature signal in it!  If we write the
Holocene forum article then we'll have to be critical or our paper as well as 
Crowley's!

>Do you have a phone where I can ring you?

001 804 924 4669 (after 2.30pm your time, 9.30am here)

Cheers

Tim

