[MITgcm-devel] seaice code beyond checkoint61j blows for ECCO-GODAE

[Dimitris Menemenlis]

Martin,

> Could you clarify for me:
> 1. Free slip with latest code fails for WedSea and Arctic, but not for
> CS510

Yes, free slip with latest code fails for WedSea and Arctic but (not  
yet)
for CS510.  The longest we've run forward CS510 with latest code
is one year.  The Arctic integration with salt-plume and free-slip ice  
crashes
after only a few dozen time steps, Arctic without salt-plume requires  
a few
days of integration before crashing, and WedSea requires about a month.

> 2. masking of etaMeanZ as suggested in my previous email is meant only
> for the free slip case (SEAICE_no_slip=.false.), so your statements
> are confusing me: What did you actually do?
> a. SEAICE_no_slip=.true. with suggested masking (== no slip without
> suggested masking, as masking should have no effect)
> b. SEAICE_no_slip=.false. with suggested masking


We did "b. SEAICE_no_slip=.false" and the model did not crash.
With SEAICE_no_slip=.true, we now have multi-decade runs on both
the Arctic and the Weddell domains with no crashes.

Has Patrick tried no-slip with his ECCO configuration test?
Does "SEAICE_no_slip=.true," stop the ECCO configuration from crashing?

I will try to look at code but don't hold your breath for miracles.
Is there any specific part of the code that you suspect?  Have you
ever verified that for free slip: "sigma_12 = 0, and e_12=0 (no lateral
stress) on the boundaries (that is on Z-points)"?

D.

Dimitris Menemenlis <menemenlis at jpl.nasa.gov>
Jet Propulsion Lab, California Institute of Technology
MS 300-323, 4800 Oak Grove Dr, Pasadena CA 91109-8099, USA
tel: 818-354-1656;  cell: 818-625-6498;  fax: 818-393-6720