[Mitgcm-support] Exact conservation?

[mitgcm-support at dev.mitgcm.org]

Hi

I am doing some simple lock exchange problems with high resolution (1mm 
resolution). The box is 512*256*250 mm and there is a thin wall at x=25 
cm, with a small opening in the middle of the that wall (2 cm wide 25 cm 
high). Initially, the water is 5C on one side of the wall and 0C at the 
other side.

I use
#define  ALLOW_NONHYDROSTATIC (I did not use nonHydrostatic=.TRUE. in 
the data file, will that matter?)
#define EXACT_CONSERV

However when I run the model for 50 s, I get a mean surface rise (Eta) 
on the warm side of the chamber that is about 1e-04 (m???) but where the 
sum of the velocities through the opening in the wall would give a mean 
surface rise of about 100 times that value!!

I guess my question is: Is Eta in m? Will exact conservation work in 
this small scale case? Any ideas what I migth do wrong?

Sincerely
Göran Broström



Ps. The matlab program to calculate the conservation. Have I done 
something really stupid ion the diagnostics?


   dt=2   % time between outputs 2 s
   nx=512
   for ifil=1:25
     % reading Eta files
     esum(ifil) = sum(sum(Eta(1:nx/2,:)))*0.001^2 % last part is grid size
     % reading U files
     usum(ifil) = sum(sum(U(nx/2,:,:)))*0.001^2 % last part is grid size
   end
now I get       esum(25)= 7.1195e-06
and             -sum(usum)*dt=7.1786e-04
shouldnt they be equal?