date: Thu Apr 24 15:02:43 2008
from: Tim Osborn <t.osbornatXYZxyz.ac.uk>
subject: Re: Abnormal normals!
to: Ian Harris <i.harrisatXYZxyz.ac.uk>
smoothing out the peaks might account for it, but (for pre) it seems too large a
reduction. Can you calculate the mean of each of the normals files, since presumably the
mean should be unaltered by such smoothing.
At 12:09 24/04/2008, you wrote:
Sit down.. it's the 'scaling' business again!
I've examined the rd0 and pre normals at half degree and two-half
degree binary, and half-degree ascii (the clim files we publish).
Here are the results and my interpretations:
FILE MIN MAX UNITS
glo25.rd0.6190 0 303 days*10
glo25.pre.6190 0 391 ???
glo.rd0.norm 0 310 days*10
glo.pre.norm 0 1244 mm
clim.6190.lan.wet.grid 0 3090 days*100
clim.6190.lan.pre.grid 0 12430 mm*10
As you can see, there is a big difference between the precip normals
over all three versions! The best interpretation I can place on the
2.5-deg binary normals ('???') is that the much larger area is
'softening' the impact of individual high-recording stations.. what
do you think?
We can see from the rd0_gts_tdm.pro program that they are treated as
days*10 (rd0) and mm (pre), assuming natural units are in play:
I wonder if the squashing of variability in the 2.5 degree grid is
causing the low variability you're seeing in the output?
I will try running with half-degree synthetics to give us a comparison.
Ian "Harry" Harris
Climatic Research Unit
School of Environmental Sciences
University of East Anglia
Norwich NR4 7TJ