[MITgcm-devel] following on testing

[Dimitris Menemenlis]

Jean-Michel, not sure if following is related to recent
global_ocean.cs32x15 breakdown but there is something
strange going on with advection in latest code.

FIG_visca4grid.jpg shows tile-edge problem when biharmonic
viscosity is used.  It shows the difference after 3 time steps
between a 6-tile and a 12-tile integration, based on
global_ocean.cs32x15, but with biharmonic viscosity
turned on and with non-conformal mapping: cs-tan-c-32-512.
That's fine and well and as expected since dxf and dyf have
wrong values in the halos.

The surprising figure is FIG_tempadv.jpg, which is exactly the same
experiment, but with biharmonic viscosity turned off and with
"temp/saltAdvScheme=33".  Something has gone horribly wrong in cube
faces 4-6.

With default advection scheme and horizontal viscosity the 12-tile
and 6-tile solutions are exactly the same after 3 time steps at the
Real*4 output accuracy.

Instructions for reproducing these experiments on top of
global_ocean.cs32x15 are also appended.

D.

-- 
Dimitris Menemenlis <menemenlis at jpl.nasa.gov>
Jet Propulsion Lab, California Insitute of Technology
MS 300-323, 4800 Oak Grove Dr, Pasadena CA 91109-8099
tel: 818-354-1656;  fax: 818-393-6720
-------------- next part --------------
A non-text attachment was scrubbed...
Name: FIG_visca4grid.jpg
Type: image/jpeg
Size: 97254 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-devel/attachments/20040519/f079527c/attachment.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: FIG_visca4grid.jpg
Type: image/jpeg
Size: 97254 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-devel/attachments/20040519/f079527c/attachment-0001.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: edge_problem.tar.gz
Type: application/x-compressed-tar
Size: 228291 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-devel/attachments/20040519/f079527c/attachment.bin>