NEMO 4.2.0 namelist: Questions about top/bottom fricition & horizontal advection/diffusion schemes

Dear all,

I am working with NEMO 4.2.0 on a regional configuration of the Mediterranean Sea (≃ 7km; 75 z-vertical levels with partial steps). Based on the NEMO 4.2.0 reference manual (NEMO ocean engine | Zenodo), I almost finalized my namelist, but I am still unsure about some parameters. To be more specific:

Regarding the top friction (namdrg_top):

  • Since I represent tides but not waves, I wonder if I should set the background kinetic energy at the surface to zero (rn_ke0).

  • With the logarithmic boundary layer formulation (ln_loglayer=.true.), how much is the model solution sensitive to the surface drag coefficient? To be more specific, I have been doing some previous simulations using rn_Cd0 = 2.5e-3 instead of 1e-3 in the reference namelist. Is it worth decreasing this value to match the reference namelist, or is the difference negligible?

Regarding the bottom boundary layer scheme (nambbl): The NEMO documentation clearly describes the scheme and its added value in z-coordinate configurations. However, it is unclear to me whether I should choose only the diffusive, the advective scheme, or both.

Horizontal tracer advection and diffusion schemes: Based on the namelist of other NEMO simulations from my laboratory, I have selected the following:

  • For the advection (namtra_adv): the FCT scheme (ln_traadv_fct=.true.), with nn_fct_h=4 and nn_fct_v=2.

  • For the diffusion (namtra_ldf): a laplacian operator (ln_traldf_lap=.true.) applied in the iso-neutral direction (ln_traldf_iso=.true.).

However, it is unclear to me if this is the right choice. From what I understand from the reference manual, it depends on the configuration, but would you have some advice or references to make a sound choice?

Horizontal momentum advection and diffusion schemes: I wonder the same as for the tracers. So far, I have chosen the following:

  • For the advection (namdyn_adv): the vector form (ln_dynadv_vec=.true.), with the een scheme for the vorticity (ln_dynvor_een=.true. in namdyn_vor)

  • For the diffusion (namdyn_ldf): a bi-laplacian operator (ln_dynldf_blp=.true.) applied in the geopotential direction (ln_dynldf_hor=.true.)

If you have any advice or recommendation on these points, I would be very glad to read it.

Thank you in advance,

Nicolas Gonzalez

Hi Nicolas,

I can not answer to all your questions say, I will try, but some of my answers may reflect my own experience only.
Let start with the easy one. “Top friction” refers to the stress induced by the solid ice cavities. No need to worry about this in the Med Sea for a couple of decades :slight_smile:
BBL. If you consider that, for instance, shelf cascading is clearly ill resolved in your setup (which is likely with z-coordinates), you could git it a try in its full form (advective+diffusive). That could be part of a sensitivity test depending on the processes/timescales of interest. Otherwise, I would not use it at all.
Tracer diffusion. Clearly the practice today is to use isopycnal mixing (which reverts to horizontal mixing in the surface boundary layer btw). it’s a matter of reducing diapycnal mixing in the interior of course. The price to pay is a slight loss of monotonicity. I would suggest that you use the 4th order vertical advection scheme to reduce diapycnal mixing, with tides in particular. The expense is a small reduction of the maximum CFL allowed (and it’s slightly more expensive of course).
For horizontal advection, vector form + biharmonic dissipation makes sense even though more and more users have recently switched to UBS for relatively high resolution such as yours. The built in dissipation is appealing. However, noise issues with that scheme have recently become a matter of concern, and will soon lead to a revision of the scheme.

Hi Jerôme,

Thanks for your advice and suggestions. It helps a lot. I will do some sensitivity experiments with the BBL scheme and try the 4th-order scheme for the vertical advection scheme.