[MITgcm-support] the weird Temperature diffusion

Chun-Yan Zhou c.zhou at dundee.ac.uk
Sun Dec 18 06:14:34 EST 2011



Thanks for all your suggestions. 

run 9-7-4 with an unstable stratification, but no KPP. I indeed used
nonHydrostatic=.TRUE. I am also surprised without seeing huge changes in
temperature, I supposed I had to turn KPP=TRUE in order to get sensible
result while dealing with unstable stratification.  In your suggestion,
the convection should be very fast even without KPP=TRUE. 
I took Jody’s advise to add a bit of noise to the initial T to give the
convective instability, and tried the very very small noise of
±0.00001(run9-7-6, without KPP), the result seems very reasonable.
The temperature changes very fast in the trend to a stable state. It
surprised me such a tiny noise can make the result totally different. 
 Cheers 
 chunyan 
% This is a matlab script that generates the input data 
% variable x resolution 
 prec='real*8'; 
ieee='b'; 
 % Dimensions of grid 
nx=700; 
ny=1; 
nz=60; 
% Nominal depth of model (meters) 
H=-90; 
% Size of domain 
% Temperature profile 
t=zeros([nx,ny,nz]); 
for k=1:nz 
   for i=1:nx 
   t(i,:,k)=15+(k-1)*0.2+rand(1,1)*0.00001; 
   end 
end 
fid=fopen('T.init','w',ieee); fwrite(fid,t,prec); fclose(fid); 


> 
>
> For the other runs, if you have nonHydrostatic = .FALSE. your only
temperature change will be due to the diffusion KzT=1e-5, so your time
scale will be more like 9x10^8 s, i.e. it'll take a long time to mix! 
(or, 80 tides = 6m, which agrees w/ your plots). 
>
> What I don't understand is your run 9-7-4 with an unstable
stratification, but no KPP. If you actually did have
nonHydrostatic=.TRUE. as you indicated, this run should have had no
trouble convecting, and you should see huge changes in the temperature. 
Are you sure you had nonHydrostatic =.TRUE. for this run?  If you did,
and this is how it behaved, you might consider adding a bit of noise to
the initial T to give the convective instability something to grow off
of.  Though it'd sure be a testament to the numerics if 80 tidal cycles
of running on an unstable stratification didn't set off some sort of
instability!
>
> Cheers,   Jody
>
> --
> Jody Klymak   
> http://web.uvic.ca/~jklymak/
>
>
>

The University of Dundee is a registered Scottish charity, No: SC015096
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20111218/f224d255/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: run9-7-6good onlyTreversenoobcsnokpp.jpg
Type: image/jpeg
Size: 160151 bytes
Desc: JPEG image
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20111218/f224d255/attachment-0002.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: run9-7-6fine onlyTreversenoobcsnokpp.jpg
Type: image/jpeg
Size: 209401 bytes
Desc: JPEG image
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20111218/f224d255/attachment-0003.jpg>


More information about the MITgcm-support mailing list