Vertical viscosity/diffusion schemes and DSW

[PaulSpence]

How sensitive is DSW to the choice vertical parameterizations?

ACCESS-om2 uses KPP v4p1 (i.e. the kitchen sink), but there are others avaiable in MOM5 and MOM6:

MOM5 code options here: https://github.com/mom-ocean/MOM5/tree/master/src/mom5/ocean_param/vertical

There is KPP, GOTM (? www.gotm.net), constant, chen (A hybrid vertical mixing scheme) and PP (vertical mixing coefficient based on the Richardson number) options in MOM5

Lots of MOM6 options are here: https://github.com/mom-ocean/MOM6/tree/main/src/parameterizations/vertical

In particular, the energetic PBL scheme is popular.

[willaguiar]

On this subject - The paper below might be useful, as it suggests that entrainment buoyancy fluxes on KPP depend on vertical resolution (check fig 5). I put on my reading list, but it might be useful if someone with more KPP experience take a look.

The KPP Boundary Layer Scheme for the Ocean: Revisiting Its Formulation and Benchmarking One-Dimensional Simulations Relative to LES, Van Roekel et al (2018)

[willaguiar]

We have now experiments that both increase and decrease dz at surface, and it seems like there is a inverse relationship between the top cell thickness and the DSW formation on the shelf.

The image below is figure 5, from Van Roekel et al, 2018, and shows the relative error of KPP in representing the ocean surface boundary layer (OSBL) in a few sets of experiments with different dz's. The FC experiment is probably the one of more interest, as it simulates free convection generated by −75 W m^-2 cooling at surface, with no winds or pre-existing OSBL. The 0 error line is set relative to a large eddy simulation.

It seems that for the FC Free convection case there is a optimal vertical resolution between 5-10m dz. According to this paper, thinner layers would create a shallow bias in the OSBL, while thicker would create a deep bias.

Speculation: A shallower OSBL that is continuously cooling and gaining salt at the surface would get denser than a thicker OSBL. Denser OSBL on thin dz cases then would likely be more efficient in creating DSW. Not sure if the rationale above is correct (suggestions/corrections more than welcome!)

The FCML case supposes an initial OSBL is set, and in this case the dependence of resolution does not follow a defined pattern, so it is hard to unerstand what that would mean for DSW formation. The other experiments are:

CEW = cooling, evaporation, and wind; CWB = cooling with background shear; DC = diurnal cycle; FCML = free convection with a mixed layer; HW = heating and wind

[adele-morrison]

That sounds reasonable to me. It's surprising that if the relative error in boundary layer depth is only 10% or so between the 1m and 10m cases that it has such a large impact on DSW formation.

Have we looked at mixed layer depths in the different experiments? I'm not sure if we'd see the effect in the standard mixed layer depth diagnostics, or do we need the mixing depth that comes out of KPP? Worth a try to check?

This is good motivation for repeating the tests with ePBL in panan too.

On Thu, 19 Oct 2023 at 11:06, Wilton Aguiar @.***> wrote:

We have now experiments that both increase and decrease https://github.com/willaguiar/DSW-collaborative-project/issues/20#issuecomment-1769579013 dz at surface, and it seems like there is a inverse relationship between the top cell thickness and the DSW formation on the shelf.

The image below is figure 5, from Van Roekel et al, 2018 https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018MS001336, and shows the relative error of KPP in representing the ocean surface boundary layer (OSBL) in a few sets of experiments with different dz's. The FC experiment is probably the one of more interest, as it simulates free convection generated by −75 W m-2 cooling at surface, with no winds or pre-existing OSBL. The 0 error line is set relative to a large eddy simulation. [image: Screen Shot 2023-10-19 at 10 37 57 AM] https://user-images.githubusercontent.com/70033934/276435962-212548a5-4f29-4650-92d4-c6b3b343171b.png

It seems that for the FC Free convection case there is a optimal vertical resolution between 5-10m dz. According to this paper, thinner layers would create a shallow bias in the OSBL, while thicker would create a deep bias.

Expeculation: A shallower OSBL that is continuously cooling and gaining salt at the surface would get denser than a thicker OSBL. Denser OSBL on thin dz cases then would likely be more efficient in creating DSW. Not sure if the rationale above is correct (suggestions/corrections more than welcome!)

The FCML case supposes an initial OSBL is set, and in this case the dependence of resolution does not follow a defined pattern, so it is hard to unerstand what that would mean for DSW formation. The other experiments are:

CEW = cooling, evaporation, and wind; CWB = cooling with background shear; DC = diurnal cycle; FCML = free convection with a mixed layer; HW = heating and wind

— Reply to this email directly, view it on GitHub https://github.com/willaguiar/DSW-collaborative-project/issues/29#issuecomment-1769660419, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACA44UZTJNU45H3S2ZEZ423YABVIFAVCNFSM6AAAAAA5CTOX7GVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTONRZGY3DANBRHE . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[willaguiar]

Good suggestion on the MLD! I'll check how different is the calculation of the MLD output by the model, and the one diagnosed by KPP (or if we already have an output for that).

On ePBL.... Reichl and Hallberg (2018) do similar experiment swith MOM6 ePBL mixing scheme. On Fig1, they diagnose how the integrated Potential Energy changes in the OSBL, as the thickness of the top cell changes. It seems like the dependence here is much smaller (Although is measured in a different metrics).

So if the speculation of the previous comment is correct, maybe the DSW formation dependence on the top cell thickness might be lower or non-existent on MOM6.

[PaulSpence]

The MLD diagnostic is poor/twichy on the shelf. Its hard to tag the delta rho transition with weak stratification. Could look at where the convective mixing scheme is active (e.g. diff_cbt).