[MITgcm-devel] Re: What to do!

[Jean-Michel Campin]

Hi Andy,

Here is what I suggested yesterday:
keep the same timestep (mom: 300s , temp: 900s)
and run 10y with tempImplVertAdv=T but momImplVertAdv=.F.

the goal was to see how the results are (zonal wind, zonally averaged)
since you find big differences (with identical time-step)
between the fully explicit case and the fully implicit one.
I think also that there is no need to do other test until
we know what are the reason of such large differences
since it's likely to be a bug in the implicit implementation.

Did you try this test ?
Does it blow-up ?
If not, how the solution looks like ?

See you,

Jean-Michel