Reg. velocities at the open boundary

[Afroosa]

Dear MOM6 community,

I am running MOM6+SIS2 for the Indian Ocean regional model with a 1/4-degree resolution and 75 vertical hybrid (Hycom) layers. The model is forced with climatological wind, heat and freshwater fluxes. The WOA13 Temperature and Salinity climatology is provided as open boundary conditions for the eastern and southern boundary of the Indian Ocean domain. And the other fields, such as U, V, and eta, are provided as zero at the boundary. The boundary condition is specified in the MOM_input as shown below:

OBC_NUMBER_OF_SEGMENTS = 2
OBC_FREESLIP_VORTICITY = False
OBC_COMPUTED_VORTICITY = True
OBC_FREESLIP_STRAIN = False
OBC_COMPUTED_STRAIN = True
OBC_ZERO_BIHARMONIC = True
OBC_RAMP_TIMESCALE = 2.0
OBC_SEGMENT_001 = "J=0,I=0:N,FLATHER,ORLANSKI,NUDGED,ORLANSKI_TAN,NUDGED_TAN" !
OBC_SEGMENT_001_VELOCITY_NUDGING_TIMESCALES = 0.3, 360.0
OBC_SEGMENT_002 = "I=N,J=0:N,FLATHER,ORLANSKI,NUDGED,ORLANSKI_TAN,NUDGED_TAN" !
OBC_SEGMENT_002_VELOCITY_NUDGING_TIMESCALES = 0.3, 360.0

I'm writing to confirm my understanding of how this open boundary condition works. In the above setup, even though I'm providing zero velocities, I observed non zero velocities at the boundaries in model simulations. I guess these non-zero velocities are the geostrophic velocity computed from the WOA Temperature and Salinity climatology provided at the boundary and the same are used for nudging. Is this how it works? Please correct me if I'm missing something.

 

[kshedstrom@alaska_edu]

You are not clamping the velocity boundary conditions, but rather allowing the model to radiate out signals at the boundary. Do things look more or less well behaved at the boundary?

Note that in my experience, providing boundary conditions from a larger model more frequently than monthly has given me better results. That way, you get velocities consistent with your T, S, SSH. There are global model reanalyses that you can download daily from both HYCOM and Mercator simulations, at least for some years.

 

[Afroosa]

Thank you for your response

Yes, it is decent at the boundary. And, as you suggested, we plan to change the boundary condition from climatology to global model reanalysis data once we successfully implement the radiation boundary condition.

I'm particularly interested in knowing how models compute the velocities in this radiating boundary condition, such as flather. The velocities are set to zero at the open boundary in the configuration described above. However, based on the model simulations, the velocity appears non-zero at the boundary and is evolving. Is this because the model computes geostrophic velocity from the WOA Temperature and Salinity provided at the open boundary?

 

[milicak]

Hi,

First thing I would suggest you to read some fundamental OBCs paper such as Marchesiello et al.

Neverthless, if you need a crash course related to the open boundary conditions especially radiating versions such as Flather or Orlanski,
the model will have a solution (i.e. values) at the velocity points at the boundary depending on your option.
Flather is only for 2D, and Orlanski is generally for 3D.
These kind of boundary conditions rely on what is called "group velocity" or something like that indicating if the signal is leaving the domain or entering into the domain.

First you need to find a way to compute this group velocity. For instance in Orlanski it is a discrete form computed internal velocity values close to the domain, and for Flather it is anlaytical such as sqrt(g/H).

Then depending on your outside velocity (and ssh) dataset, these two parameters will blend to compute the velocities at the boundary.
Again, I am extremely simplifying.

If you are not using what is called "clamp" boundary conditions where you overwrite the velocity values at the boundary with OBC dataset, generally outside velocities will "guide" the model to the dataset, but it will not be the same values, espcially depending on the sign of the group velocity.

I hope this helps.

Cheers,
Mehmet

Marchesiello, P., J. C. McWilliams, A. F. Shchepetkin, 2001: Open boundary conditions for long-term integration of regional ocean models, Ocean Modelling, 3, 1-20.

 

[Afroosa]

Thank you for your response.