MOM6 1deg parameterization 1 - mixed layer restratification

[minghangli-uni]

In MOM6 for the 1deg setup, the frontal-length scale MLE_FRONT_LENGTH of MLE is chosen to be 1000.0m and the MLE_MLD_DECAY_TIME is 345600.0s. Is there any relevant reference for this parameter selection?

In the same 1deg setup of OM2, the MLE_FRONT_LENGTH was previously 5000m.

However, in paper The GFDL Global Ocean and Sea Ice Model OM4.0: Model Description and Simulation Features, MLE_FRONT_LENGTH was set to 500m / 250m for 1deg and 0.25deg configurations, while MLE_MLD_DECAY_TIME was chosen as 30days, equivalent to 2592000s, the authors demonstrated that, this time filtering,

allows the parameterized eddies to be seasonally modulated but ensures that restratification of the deepest mixed layer is persistent for several weeks. This time scale of 30 days is longer than the time scales of mixed‐layer instabilities but was empirically found to help counter a cold bias in the coupled model for which this ocean model was developed.

[aekiss]

Thanks for spotting these differences @minghangli-uni.

The short answer is we didn't test different choices of these parameters in ACCESS-OM2. front_length_const=5e3 is the default in MOM5. Note that this value is actually a floor because we also use front_length_deform_radius=true (also the default).

This is a lot larger than the values of mle_front_length =1000 in MOM6-CICE6_1deg_jra55do_ryf or 500 in mom6-panan/ MOM_input - see this table. Perhaps some difference in the implementation of this scheme in MOM5 vs MOM6 could account for the difference (e.g. maybe there's a different definition of mixed layer depth?) Both these values are in turn larger than expected from 500m / 250m for 1deg and 0.25deg in the GFDL OM4 paper.

Although MLE_MLD_DECAY_TIME is just a tuning parameter for the heuristic filter applied to the MLD seen by the mixed-layer restratification scheme, MLE_MLD_DECAY_TIME = 345600 s = 4 days is an order of magnitude shorter than the 30 days used in the GFDL OM4 paper, so we should use a larger value, e.g. our other MOM6 configs use 2592000.0s = 30 days - see this table.

Some experimentation may be worthwhile once the configuration is running. The GFDL OM4 paper provides examples of the impact of different parameter choices - as expected, increased mixed layer restratification shoals the mixed layer depth, which in turn impacts SST.

Note that we're currently using KPP (unlike the GFDL OM4 paper which uses EPBL) - this might also have an impact.