[MITgcm-devel] multidim.Advect on CS-grid

[Jean-Michel Campin]

Hi,

I have a new version of gad_advection.F (multidim-advection)
that, I think, should conserve exactly tracer on CS-grid.
Unfortunately, it needs (on CS-grid only) an overlap 2 times
larger than the minimum required by the advection scheme,
(generally=2 for 2nd.O-fluxlimiter, DST3 and DST3_fl).

I run the verification/advect_cs test and get a perfect
conservation of salt and also a nice decrease of the
standard deviation of salt (the frequency seen
before in s_std was due to the passage of the patch near
in the corner regions of the cube, and is gone now).
I attached a postscript.

I will check-in this version and the changes in
advect_cs experiment (mainly Olx=Oly=4).

Jean-Michel
-------------- next part --------------
A non-text attachment was scrubbed...
Name: advect_cs.ps
Type: application/postscript
Size: 20952 bytes
Desc: 
URL: <http://mitgcm.org/pipermail/mitgcm-devel/attachments/20040925/1ccee440/attachment.ps>