GOSI11: Numerical mixing in the ocean interior

[oceanaguiar]

Ongoing work to reduce spurious numerical mixing. Mainly focused on testing the z~ scheme in the context of tidal models in the near future.

• Work on z-tilde. Alex is working on creating a z~ version of GO8p8 on Monsoon and once he is convinced that is working, will commit that to a repository. The code is working and has been tested in several combinations of forcing etc, but I haven’t saved it to a repository yet. I need to find out how to do that!
• Diagnose mixing in ORCA12 and compare with ORCA025 mixing. Alex will use his ACSIS runs on Archer2 to look at this, once it is running stably at 1/12°. I completed a satisfactory run of GO8p7 with z* on Archer2 last year, but I haven’t done mixing analysis yet (see next item).
• Write NEMO code to calculate transports in density space at runtime. AM, MB Possible alternative strategy involving generating 5-day output at runtime and doing diagnostics using this, but I haven’t run on Archer2 since the recent shutdown.
• Diagnose mixing in coupled models? Alex has done this for GO8p3 version of GC5. Have done this for C20 control GC5. Mixing is much lower than in CORE2-forced run, and similar to that with ERA5-forced simulation.
• Once the first step is done, he will start work on adding z~ to a GC5 suite. Need to consider minor code changes for coupled configuration I have preliminary version of this, but need to eliminate rapid instability at high latitudes. Originally I thought this was linked to initial conditions in Arctic (which remains a significant issue), but it seems connected to the divergence calculation of the interface smoothing fluxes in sshwzv.F90.
• Analysis of mixing in tidal runs: see SE-NEMO work plan. Paper on tidally forced GO8p0 is under review at Ocean Modelling. Need to discuss strategy for implementing tides in GOSI9 and GC5, building on results from SE-NEMO: I have a GC5 run with the same tidal setup as in my GO8p0 configuration (only 5 constituents, and no modification of bottom friction), but this isn’t optimum. My main goal is to be able to run GC5 with z~ and tides.
• Try to diagnose mixing of AABW in Southern Ocean? I have started work on diagnosing density transformation rates from cabbeling and thermobaricity in the Southern Ocean, which in the past have complicated mixing analysis, in an ERA5-forced GOSI9.
• Z-tilde blows up in ORCA12, John Thuburn (U. Exeter) is investigating.
• Switch from Cox to Triad scheme DB. I still have doubts about whether this is working properly in my GOSI9 runs – I am in discussions with Diego.
• Embedded sea ice?