[MITgcm-support] nonhydrostatic w boundary condition

[Jean-Michel Campin]

Hi David,

Just to add a precision: 
The reason why there is a call to OBCS_APPLY_W is because wVel in 
the OB region is used to compute the tendency of vertical velocity (gW)
in the interior near the OB (horizontal advection of wVel, viscosity ...). 
I don't think it will work if you try to add a relaxation (sponge) on wVel
in the interior near the OB (because it has to satisfy the continuity eq.).

Cheers,
Jean-Michel

On Mon, Apr 19, 2010 at 05:37:28PM +0200, Martin Losch wrote:
> Hi David,
> 
> the obcs package is not failsafe at all, it's more like a do-it-yourself-kit, as you may have noticed. Some standard things work, others do not work too well. Initially, as far as I remember, it was intended to be just a framework for the user to plug in his/her own code. So you'll have to try a few things yourself. E.g. you could comment out the call of obcs_apply_w (or the code within this routine, probably safer, that way you'll catch all instances), or do something that is similar to what is done in orlanksi_*.F
> 
> But whatever you do, the open boundaries will always be the source of trouble, and noise (and not only in the MITgcm). We just do not have implemented all possible schemes to damp the noise.
> 
> 
> Martin
> 
> On Apr 19, 2010, at 3:08 PM, David Hebert wrote:
> 
> > Thanks Martin for your response. So it looks like whenever INTEGR_CONTINUITY is called, w is computed via INTEGRATE_FOR_W, then OBCS_APPLY_W applies boundary values for W.  
> > 
> > I have been trying to use sponge layer to relax interior solutions to the boundary without much success. I am wondering the following: I have been running model in nonhydrostatic mode WITHOUT specifying W open boundary values. If I am not mistaken, in this configuration all OB?w will be set to zero in OBCS_INIT_VARIABLES. Then when OBCS_APPLY_W is called the boundary value for W obtained via continuity is replaced by zero. Then zero would always be used for boundary values when w is timestepped, would it not? And since there is no sponge code for W as of yet, would this be conducive to reflections/boundary noise? 
> > 
> > Thanks,
> > 
> > David
> > 
> > 
> > Thanks 
> > 
> > On 04/19/10 02:15, Martin Losch wrote:
> >> In nonhydrostatic mode, vertical velocities are still computed from "div U = 0". If the 3D pressure solver has converged perfectly, this wVel is the same as the one obtained from stepping the w-equation. If not (in practice you have only finite accuracy), the differences are small, and you have to make a decision which equation you want to satisfy, dw/dt = Gw, or delU=0. In the MITgcm, delU=0, as far as I know. However, W at the open boundary is reset to the prescribed value (see integr_continuity).
> >> 
> >> If (div U) is not zero, you'll generally get extra sources in the tracer equations. I bet that this will also happen along the open boundaries.
> >> 
> >> Martin
> >> 
> >> On Apr 16, 2010, at 7:42 PM, David Hebert wrote:
> >> 
> >>   
> >> 
> >>> In nonhydrostatic mode it looks like it is possible to prescribe all components (u,v,w) of velocity at the boundaries. Is that right? Are there consequences if continuity ("del dot U") is not strictly matched?
> >>> 
> >>> Thanks,
> >>> 
> >>> David
> >>> 
> >>> _______________________________________________
> >>> 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
> >> 
> >>   
> >> 
> > 
> > -- 
> > David A. Hebert
> > Naval Research Lab
> > Stennis Space Center, MS 39529
> > david.hebert at nrlssc.navy.mil
> > Phone: (228) 688-5846
> > _______________________________________________
> > 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