[MITgcm-support] 2d pressure solver bdy condition

[Jean-Michel Campin]

Hi Aurelien,

I don't disagree with Martin ; and as I sais yesterday, just providing
and applying eta OBC value will not do much since the 2-d pressure 
solver ignore (matrix coeff set to zero in ini_cg2d.F) values
of eta at the OB.

What I would propose to try (in this order):
1) without changing 2-d solver, compute a correction (in obcs_calc)
 to the prescribe velocity that is function of the pressure
 gradient accross OB. You can put a tunable parameter, and
 try to see if this improve something, with you current time-step
 and also with a smaller time-step. If you reduce the time-step,
 the 2-d solver solution will become closer to an explicit method 
 solution.
2) once you have a solution that works with a small time-step,
 you could try to put this correction into the solver (so that
 it would be solved implicitly).

Cheers,
Jean-Michel 

On Fri, May 13, 2011 at 08:49:01AM +0200, Martin Losch wrote:
> Aurelien,
> 
> my feeling is that the implicit free surface is "sub-optimal" for tidal simulations. See here:
> <http://mitgcm.org/public/r2_manual/latest/online_documents/node41.html>
> Maybe your aspired boundary conditions are also easier to implement in an explicit scheme for the barostropic mode?
> 
> Martin
> 
> On May 12, 2011, at 6:38 PM, Matthew Mazloff wrote:
> 
> > Hi Aurelien,
> > 
> > Yes, I  believe that the bc on the obc for eta is eta = div u where div u is only from tangential velocity -- normal velocity is made divergenceless.  Though what might happen on the obc is that tangential velocity divergence is overwritten and eta is then set to zero -- don't remember exactly.  You can check in, I think, solve_for_pressure.   However, through some niceties of the c-grid I don't think any of this really comes into play in the dynamics though.....at least, when sponge layers are off it doesn't really come into play
> > 
> > Anyway, after a talk with JMC about a similar issue, I feel that the way to solve your issue is to not use obcs sponge layers, but instead apply sponges with rbcs.  Thus obcs will prescribe the obcs you want, and rbcs will restore to the smooth solution you want.....
> > 
> > Im at SIO if you wanna talk today
> > 
> > -Matt
> > 
> > 
> > On May 12, 2011, at 9:22 AM, Aurelien L Ponte wrote:
> > 
> >> hi,
> >> 
> >> I am running a tidal simulation in realistic domain.
> >> The simulation uses a linear implicit free surface pressure method.
> >> 
> >> u/v are prescribed at the open boundaries but I would like to prescribe
> >> the sea level as well.
> >> The reason is that the sea level is a sensitive function of the flow
> >> at the boundaries and hence difficult to get right.
> >> 
> >> I do not see any simple way of doing that.
> >> 
> >> The only solution I can think of would be to modify the boundary
> >> conditions of the 2d pressure solver.
> >> Would anybody know what the current boundary condition are ( deta/t=div u ?)  ?
> >> Could I use instead boundary conditions of the Dirichtlet type (eta=eta_prescribed) along open boundaries ?
> >> 
> >> Thanks
> >> 
> >> Aurelien
> >> 
> >> 
> >> _______________________________________________
> >> MITgcm-support mailing list
> >> MITgcm-support at mitgcm.org
> >> http://mitgcm.org/mailman/listinfo/mitgcm-support
> > 
> > 
> > _______________________________________________
> > MITgcm-support mailing list
> > MITgcm-support at mitgcm.org
> > http://mitgcm.org/mailman/listinfo/mitgcm-support
> 
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support