[MITgcm-support] obcs and linear free surface

[Jean-Michel Campin]

Hi Martin,

> observe that the variable eta along the open boundary ...
> In particular there is a
> strong discontinuity between boundary value and interior

This is what I observed when I run exp4 for instance (if I remember well).

The way I understand the OBCs with linear FS is that the 
ssh responds to the divergence of the flow (that is imposed
at the OpenBC, either prescribed or computed by OBC pkg) 
but the gradient of ssh does not affect the velocity at 
the OpenBC (because it is imposed).
The solver knows about OBCs (some coefficients are reset to zero)
and is consistent with this general idea (there are few details
about using exactConserv or not, that I forgot).

I didn't have time to check that the staggerTimeStepping changes
that I did a month ago was consistent regarding OBC and NonLinFS,
and if you see something wrong, let me know.

Cheers,

Jean-Michel