[Mitgcm-support] Re: is this a problem (ini_Cori.F)

[mitgcm-support at dev.mitgcm.org]

If you have walls at all the computational boundaries then not
having "f" vary continuously across the boundary shouldn't
be a problem. In the original ini_fCori.F I made sure
it was continuous by not doing an exchange and calculating
proper values in the overlap. In your case, since you're
reading data then the exchange is necessary for it to work in
the interior and it will have to be assumed you have a wall
around the model domain.

A case where you wouldn't have a wall is a global model with
the Artic. This would fail for a vector invariant formulation
because the absolute vorticity at the poles is active.

A.

> For the rotated grid, I've redone a little of ini_cori.F to
> read in coriolis fields. I check the values by text output
> of fCoriG and fCoriC from calculated as is in useSphere;
> and those from my readin files.
> Two attachements give those results
> 
> The current code doesn't wrap coriolis values -- see top and
> bottom of each test fCori* output --
> Could this be a problem?