GOSI10: TRIADS formulation for the isoneutral mixing

[oceandie]

Two formulations for the iso-neutral diffusion are available in NEMO v4.x.x - see manual:

1) Modified Cox 1987 (COX): with a non-linear EOS gives wrong evaluation of the neutral slopes when used with Generalised Vertical Coordinates (GVCs) -> introduces diapycnal mixing (NEMO manual, pag 120)

2) Griffies et al. 1998 (TRIADS): correct with GVCs, but the discrete version cannot guarantee that no new extrema are produced (e.g., Beckers et al. 1998)

Our GOSI9 configurations use the COX formulation. However, Bruciaferri et al. 2024 shows that with local ME s-coordinates the COX scheme introduces numerical diapycnal mixing that almost cancels out all the improvements obtained from using TRIADS with local terrain-following GVCs.

Therefore, if we want to have the capability of implementing not geopotential GVCs we need to switch to the TRIADS formulation.

The aim of this issue is to assess the impact of switching on the TRIADS scheme in GOSI9p8.0 forced with JRA and using standard $z^*ps$ GVCs.

[oceandie]

Two different types of experiments will be conducted:

1) @gnurser and @oceandie will replicate the idealised experiments of @gnurser to check that NEMO 4.2.1 is bugfree;

2) @ukmo-cguiavarch will run GOSI9p8.0 forced with JRA and with TRIADS switched on - to be compared with the run of @AlexM62 using GOSI9p8.0 with JRA (suite id = u-cn082).

3) @AlexM62, @DaveStorkey. @ukmo-cguiavarch and @oceandie will analyse the results.

[ukmo-cguiavarch]

Two different types of experiments will be conducted:

1. @gnurser and @oceandie will replicate the idealised experiments of @gnurser to check that NEMO 4.2.1 is bugfree;
2. @ukmo-cguiavarch will run GOSI9p8.0 forced with JRA and with TRIADS switched on - to be compared with the run of @AlexM62 using GOSI9p8.0 with JRA (suite id = u-cn082).
3. @AlexM62, @DaveStorkey. @ukmo-cguiavarch and @oceandie will analyse the results.

2. Configuration GOSI9p8.0 eORCA025 with JRA55 forcing and TRIADS with initialisation from EN4 set up and currently running (for the years 1976-2021): u-db914

[oceandie]

The experiment has finished and results are availble on mass at moose:/crum/u-db914

[oceandie]

Validation notes after running MARINE_ASSESS can be found here: u-db914_nemo_vs_u-cn082_nemo

[AlexM62]

Thanks, Diego - very interesting! First impressions are that you haven’t found much sensitivity to enabling the triads with JRA-55 forcing.

Could you do a VN between u-cn081 and u-cq734? That is the corresponding pair of CORE2-forced runs where I have seen the strong dependence.

Alex

On 18 Jan 2024, at 11:17, Diego Bruciaferri @.***> wrote:

Validation notes after running MARINE_ASSESS can be found here: u-db914_nemo_vs_u-cn082_nemo https://gws-access.jasmin.ac.uk/public/jmmp/valid_ocean/u-db914_nemo_vs_u-cn082_nemo/assess.html — Reply to this email directly, view it on GitHub https://github.com/JMMP-Group/GO_coordination/issues/14#issuecomment-1898283935, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGEPKYI5D5CHJJUYZQEG2E3YPEAC5AVCNFSM6AAAAABAGKU5ZGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQOJYGI4DGOJTGU. You are receiving this because you were mentioned.

[oceandie]

HI @AlexM62 ,

The validation notes after running MARINE_ASSESS for u-cq734 (GOSI9p0 with CORE2 and COX) and u-cn081 (GOSI9p0 with CORE2 and TRIADS) can be found here: u-cq734_nemo_vs_u-cn081

[AlexM62]

Hi @oceandie,

Here are the annual global temperature and salinity drifts with respect to the first year for my CORE2-forced GOSI9 ensemble with parameter changes selected to increase or decrease numerical mixing. These now include the z* simulation u-cq734 with Griffies triads enabled. Although these integrations ended in 2005, the time axis goes to 2022 to enable them to be compared directly with the corresponding figures from my coupled runs.

The temperature drift in the simulation with Griffies triads is the strongest of the whole ensemble - even larger than the two where the horizontal tracer advection was degraded from 4th to 2nd order (H2V2 and H2V4).

The salinity drift is less sensitive to the parameter changes, although again the enhanced salinity increase in the upper 500 metres is comparable to that in the simulations with 2nd order advection.

From these figures, one might conclude that the drifts in the Griffies simulation are consistent with higher numerical mixing, although I will have to recalculate the effective diffusivity for the ensemble to confirm one way or another.

My feeling is that once we are convinced that the Griffies scheme is working without any bugs we may have to re-tune the mixing in the configuration.

Alex

[ukmo-cguiavarch]

Here are the MARINE_VAL metrics for the following integrations (45y integrations forced by JRA and initialised from EN4):

• u-cn082: GOSI9
• u-db914: GOSI9 with TRIADS
• u-db949: GOSI9 with logarithmic bottom drag coefficient https://github.com/JMMP-Group/GO_coordination/issues/13
• u-db981: GOSI9 with TRIADS and logarithmic bottom drag

VALNA: With TRIADS, increase in AMOC and SPG heat content plus northward shift of the NAC associated with SST warming off Newfoundland.

VALSO: With TRIADS, reduction in the Weddell Gyre and Ross gyre transport but little impact on ACC

VALTRANS: With TRIADS, limited impact on transport