[MITgcm-support] Issues with walls/periodic model
jklymak at uvic.ca
Mon Jul 9 17:56:15 EDT 2012
Wild guess w/o seeing any of your setup, but do you have Olx and Oly setup appropriately for your advection scheme? These specify the tile overlap, and thus the message passing across your periodic boundary.
Where does the integrations *start* to blow up? I always find that useful in chasing down these sorts of problems.
Good luck Cheers, Jody
On Jul 9, 2012, at 10:17 AM, Chris Horvat wrote:
> I'm running a model with wind and temperature forcings which I'm trying to force to be a channel run (periodic in x) through setting the topography. The thing is, the integration blows up after 4 steps any time I allow perioidicity zonally.
> If I place the walls on the east and west, the model runs as desired. Similarly if I place walls on all four sides, it will run. Only when I remove the walls on the east and west will the model blow up, whether I have walls on the north and south or not.
> I set up my files from verification/deep_convection originally, so perhaps there is something set up there that will affect my runs in this way. Is there anyone who might be able to point me as to what I've screwed up?
> Chris Horvat
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
More information about the MITgcm-support