[MITgcm-support] Uniform diffkrfile

[Geoffroy Aubry]

Dimitris Menemenlis a écrit :
> Geoffroy, when you specify a diffkrfile with uniform 1.E-5 m2.s-1 
> diffusivity, the situation would (should!) be equivalent to your case 3 
> below, that is, diffKzT=diffKzS=1.E-5.  Do your case 1 and case 3 below 
> give equivalent results?
> 

Yes you're right, the outputs are exactly the same (I did a diff -q for 
all the files in the input directory, and the only differences are the 
data files).
But I have a uniform salinity map (I don't give any hydrogSaltFile in 
the data file). So if I put any diffusivity coefficient for salt, I 
should have the same results isn't it? But it is not the case... I did 
an experiment with diffKzS at 1.E-5 and 5.E-5 and I dont get the same 
results (all the Eta.n, n>2, PH.n, n>3, T.n, n>48, U.n, n>2, W.n, n>2 
differ).

For the moment, I try to understand why with a uniform vertical 
diffusivity, with a temperature profile of θ = −1.10^−3 z + 5.5 °C (z>0) 
and with a topography everywhere -1500m but at one place a wall (see 
enclosed picture), without doing any forcing, I get everything mixed.
If the initial situation is stable (which I think it is), it should stay 
the same over the time.
But after 50000s, I get some huge velocities (W~4m.s^-1).
If I remove the wall, then I get a stable situation again (test2 here : 
http://perso.crans.org/gaubry/dossier/stage/index.html#bug ) and if I 
put the vertical diffusivity for salt to 0, the situation is also stable 
( http://perso.crans.org/gaubry/dossier/stage/bug.html )

Do you have any clues?
Thank you for your support,
-- 
Geoffroy
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image_50000.png
Type: image/png
Size: 32563 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20070807/e9b6027e/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image_00000.png
Type: image/png
Size: 13441 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20070807/e9b6027e/attachment-0001.png>