[MITgcm-support] problem in pkg/seaice

[chris hill]

Tom,

  Martin had a run with pretty much identical similar symptoms in the 
Southern Ocean. We spent some time trying to figure it out, assuming it 
was an ice related problem, but then we reduced the time step slightly 
and everything was fine. The solutions then looked great. So while its 
possible that there is an issues with the sea-ice (its undergoing lots 
of development), its also possible you are right at the limit of 
stability. Can you try simply halving the timestep.

Chris
Thomas Haine wrote:
> Hi, 
> 
> I've been having a problem in pkg/seaice that causes a sudden divergence
> of T & S adjacent to the ice edge under strong winds:
> 
> The resolution is 2km horizontal with 2m thick upper level. I've run
> fine with NCEP 6-hrly forcing (atmospheric state) and have now switched
> to a 1/4 deg-resolution 6-hrly satellite-derived 10m wind product (only
> change). This has caused the trouble. At the time of the crash the wind
> speed is ~40m/s with strong gradients (an intense polar low - the
> numbers are realistic!).
> 
> Code is: checkpoint57y_post.
> 
> Symptom: Both S & T go wild (+/- 10^6) in 2 cells at the sea floor in a
> single step. The ~80m deep column is adjacent to ice with thickness of
> ~40cm. 
> 
> Switching off pkg/seaice allows the code to carry on fine for at least
> several more steps.
> 
> Switching off seaice dynamics makes only small changes (survives 1 more
> step, symptoms the same).
> 
> So, it looks to me like this is a seaice thermodynamics issue.
> 
> Any suggestions?
> 
> Thanks, Tom.