[MITgcm-support] L4rdt
Martin Losch
Martin.Losch at awi.de
Fri Jul 31 09:57:25 EDT 2009
Hi David,
I do think that your fixing of L4rdt (in revision 1.38 of
mom_calc_visc.F) makes sense, but I thought I'll tell you a funny
story about the consequences: I just spent a day trying to figure out
why suddenly my usually stable runs started to explode after my recent
update.
Since I am still fiddling with the (formerly very instable) seaice
solver, I was initally really frustated that apparently the solver
still doesn't produce stable results. But then I realized that the
blow-up started far away from the ice in the deep ocean (and near an
open boundary *and* near a tile edge/corner, so there were many things
that I tried to track down).
I was not following the discussion about L4rdt (or forgot about it),
but then I found out that the change of L4rdt reduces the size of L4
(I am using a constant viscA4grid) by more than 37% near the equator
(and even moreso for the vorticity points. This is a good explanation
for my problems; I am currently checking if reverting to the old L4rdt
makes the problem go away (won't see the results today).
I think it's funny that I stumbled over this, but I know of at least
one user who will stumble over this too (and will never know what
happened), so shouldn't there be a statement about this change on the
MITgcm-support list? (or maybe there is one and I missed it?)
Martin
More information about the MITgcm-support
mailing list