GOSI-SO HPG test

[ukmo-cguiavarch]

Create HPG test for the GOSI-SO (MEs in Southern Ocean) configuration.

[ukmo-cguiavarch]

To initialize the HPG test we need temperature and salinity profiles representing the stratification in the Southern Ocean. I have prepared profiles for 3 regions: Weddell gyre, Ross gyre and Dumont d'Urvile.
The profiles are calculated from WOA climatology, averaged over the 3 regions for bathymetry deeper than 800m, for the months of February and August.
The profiles have been added to the following NEMO branch: hpg_test_SouthernOcean

[ukmo-cguiavarch]

HPG experiments have been carried out for all 6 profiles using this suite u-dr701.
Experiments were run for 3 months initialized from each profile. Profile representing Ross gyre in February shows the largest errors in the MEs area and is chosen to do the optimization of the envelops.
The results are available /data/users/catherine.guiavarch/GOSI10/SouthernOcean/HPG_SO/u-dr701.

[ukmo-cguiavarch]

Optimisation of the envelopes:

1. Creation of new envelopes targeting 0.1 cm/s HPG with new bathymetry file available here: vdi: /data/users/catherine.guiavarch/GOSI10/SouthernOcean/GOSI-SO/GOSI-SO_inputs/bathymetry.loc_area-antarctica.dep3550_sig3_stn9_itr3.MEs_4env_3550_r12_r12-r07_r05.nc

2. Creation of new domain_cfg with the new bathymetry file using NEMO main@ae67ac8a: exab: /data/users/catherine.guiavarch/GOSI-SO/DOMCFG_tool/main-ae67ac8a/tools/DOMAINcfg
   New mesh_mask and domain_cfg copied in vdi:/data/users/catherine.guiavarch/GOSI10/SouthernOcean/GOSI-SO/GOSI-SO_inputs/

   - domain_cfg.loc_area-antarctica.dep3550_sig3_stn9_itr3.MEs_4env_3550_r12_r12-r07_r05.nc
   - mesh_mask.loc_area-antarctica.dep3550_sig3_stn9_itr3.MEs_4env_3550_r12_r12-r07_r05.nc

3. Set up and run new HPG suite u-dt147 to test the new envelopes with the February Ross Sea profile.

[ukmo-cguiavarch]

Results from HPG experiment with u-dt147 still display errors superior to 4cm/s in the 4th envelope. The 4th envelope can not be smoothed as it is flat. Tests to smooth 2nd and 3rd envelopes did not reduce error in 4th envelope. For those tests, smoothing was applied:

• on the 2nd envelope in location where error in 2nd envelope where superior to 1cm/s.
• on the 3rd envelope in location where error in 3rd envelope where superior to 1cm/s.

As the 1st and 4th envelopes are flat, it make more sense to apply smoothing:

• on the 2nd envelope in location where errors is superior to 1cm/s in the 1st and 2nd envelopes
• on the 3rd envelope in location where errors is superior to 1cm/s in the 3rd and 4th envelopes

@oceandie I have tried to summarize our reasoning but plus correct me if needed.