[MITgcm-support] Open boundaries and sponge layers

[Dimitris Menemenlis]

Stanislav, excellent that your simulation setup is working.  The obcs_add_tides.F tides are implemented as a depth-independent (barotropic) transport at the boundaries, so sponges (or nudging as you say) will have a tough time stopping this flow, although some artifacts are possible (and probable) in the sponge region.

May I ask what region and period of time you are attempting to model and in a sentence or two the purpose of your study.  You are aware, I assume, that obcs_add_tides.F will only add barotropic tides, it will not include baroclinic tides.  I am pretty sure that neither does GLORYS12 and even if it did, you will not be capturing this forcing with daily obcs fields.  So if you are studying internal waves or their impact, or you need to include them for whatever reason, you need a different strategy.

Finally, for the reflection issue at the boundaries, I think that a little reflection, especially in downscaled simulations is not bad, as long as it does not lead to runaway amplification of kinetic energy in your domain.  There will be high-wavenumber signals that will be missing in your boundary conditions that could be compensated in part and in a statistical sense from these reflections.  If you have observations of kinetic energy in your simulation domain, say a current meter mooring, you should be able to adjust your sponge layer parameters to give you just the right amount of kinetic energy in your domain.  A Green’s Function optimization would be perfect for that.  If you decide that you need this kind of adjustment, I would be very happy to walk you through it, and very curious to see if it works.

Cheers, Dimitris


> On Sep 19, 2026, at 1:39 PM, Stanislav Martyanov <martyanov.sd at gmail.com> wrote:
> 
> Hello Dimitris!
> 
> >> If they are daily velocities, the largest tidal components would either be aliased (if snapshots) or heavily damped (if daily mean). I have not been following this thread, but can’t you just use: https://github.com/MITgcm/MITgcm/blob/master/pkg/obcs/obcs_add_tides.F ?
> 
> Yes, I do use this subroutine to set tidal forcing (through phases and amplitudes at the boundaries).
> 
> Having looked at this subroutine once more, I've just realised that I might have been wrong about the nudging mechanism in MITgcm. Tidal oscillations ARE added to the boundary non-tidal currents in this subroutine, and AFTER that, if the sponge is ON, the internal model's solution is pulled towards these new boundary values (inside the sponge layers) that already contain both daily-mean currents and tidal oscillations. So tides do not get smoothed out inside the sponge layers due to the use of GLORYS's daily-mean non-tidal currents, because the tidal component has already been added to boundary values.
> 
> Thank you for the help!
> 
> Regards,
> Stanislav
>