"Resting" ocean-only scenario for MOM6 in CESM

[paulhall]

I have been trying to create a "simpler models" scenario along the lines of the ocean-only "resting" example from MOM6-examples for MOM6 within CESM. I am running into issues with temperature and salinity instabilities forming very rapidly along one boundary (see image of surface layer T and S at 1 hr and 12 hrs into the run below).

The basic setup is as follows:

CESM version: cesm2_3_beta11
compset: 2000_DATM%SIMPLE_SLND_SICE_MOM6_SROF_SGLC_SWAV
ocn model domain: 0.5° x 0.5° x 75m depth
ocn grid: custom, with 100 grid points in each direction in the horizontal (dx~dy~500m), 30 layers

Wind speed is set to 0 over the entire domain (u = v = 0.0 in user_nl_datm) and the air temperature is set to match the initial ocean temperature (5° C).

I am able to run essentially the same scenario, on the same grid, in MOM6 outside of CESM without any issues. I have tried varying a number parameters (e.g., initial TS configuration, equation of state, using a zonally re-entrant model domain), but none of them seem to help. I was hoping someone might have an idea of why things are going wrong, or how best to go about trouble shooting this case.

Thanks!
Paul

 

[gmarques]

I wonder if the problem is with the T&S initialization via TS_CONFIG = "linear" in combination with your custom vertical grid. What happens if you use the same vertical grid settings as the "resting" test case?

 

[paulhall]

Thanks, Gustavo. For the vertical grid I've been running with REGRIDDING_COORDINATE_MODE = "Z*" and have tried both NK=9 (as in the original resting case) and NK=30. Both cases show the same pattern of instability.

I also wondered about the T&S initialization, so in addition to TS_CONFIG="linear" I also tried TS_CONFIG="fit" with T_REF=air temperature and FIT_SALINITY=True. They both result in the same pattern of instability (the instability is somewhat more pronounced with linear than with fit).

I've also tried these various configurations using MOM6 on its own, outside of CESM, including using the same custom grid, as a sanity check. I have not been getting the instabilities in cases with MOM6 outside of CESM.

I have a bunch of test cases set up on Cheyenne if a closer look at the user_nl_mom and user_nl_datm would be helpful.

 

[milicak]

Could you save the initial conditions? Are T and S uniform horizontally? If so I guess no problem with the initilization.
Then you might want to save surface fluxes in your diagnostics. In salinity, the instability is in the south-west corner, if the forcing there changes maybe due to a halo update or something that can create an asymmetry in the forcing.

 

[paulhall]

Thanks Mehmet. I have saved the initial conditions and T and S are completely uniform in the horizontal, as would be expected. Given this, I agree, it does not appear that initialization of the ocean model is an issue. I will look at surface fluxes.