
[Science] [BAS home] [Met home] [Beowulf home]	Antarctic Meteorology

Verification of yabce - std job, plus fix to fourier

This stuff is rather preliminary (and I have no 64-bit results; perhaps I should have...). Read if you like, but with caution.

Note also: these are with correction to the level-by-level constants, *plus* fixes to the fourier routines. See yabba for without lbl and yabcc for lbl but no fourier fix.

Note that the IDL code is for my info for regeneration of the piccies.

MSLP

I shall compare against aawei, a job run at the Hadley Center which I believe was a standard HADam3 job.

; nb: picture at top right *is* ce-cc, even if not labelled as such!
@verify-pstar
gettwogifs,out='pstar_jja'

Produces:

You see: TR: aawei. TL: yabcc - yabce (ie effects of including fourier mod). BR: aawei - yabcc. BL: aawei - yabce.

Note:

that this is pstar.
this is JJA
I have added an attempt at significance testing of the difference drawn on, using t-test of means. Areas unshaded are "significant". This isg testing can't be fully trusted... its indicative only.

We see: adding the fourier mod appears to increase similarity to aawei (which, as pointed out by the wise WJI, is not what we should expect, since the fourier mod *isn't in* aawei. So I think we'll put it down to Variability until further studies shed more light).

Meanwhile, back in DJF:

m=12
@verify-pstar
gettwogifs,out='pstar_djf'

Once again, 32 and 64 bit are very similar. This time, its the NH that shows largest differences between them (winter hemisphere variance is always largest, so this could indicate that the difference is just variability).

Now, however, differences between yabcc and aawei are not huge, as they were in DJF in yabba. Good. Is this chance, or because of fixing the problem with the level by level constants? Note that its still worst in DJF... perhaps the flow is intrinsically harder then?

Past last modified: 1/5/2002 / wmc@bas.ac.uk
© Copyright Natural Environment Research Council - British Antarctic Survey 2001