[MITgcm-support] [Fwd: MITgcm-support Digest, Vol 21, Issue 21]

[Tom Haine]

Alistair, 

Yes, I understand that no special pickup files are required for Leith.
However, the code clearly diverges due to instability associated with
the variable viscosity. For instance, setting: 
 viscC4Leith=0.0,
 viscA4GridMin=0.005,
 viscA4GridMax=0.050,
instead of:
 viscA4Grid=0.,
 viscC4Leith=1.0,
 viscA4GridMax=0.005,
allows the picked up calculation to survive for many more steps (before
ultimately failing again). It seems that some aspect of the internal
model state is different when picked up. This difference seems to cause
a rapid divergence in the Leith scheme. Any ideas?

Tom.