GO-novf

Code to generate localized ME s-coordinates (Bruciaferri et al. 2024) in the Greenland-Scotland ridge region of GOSI configurations to better represent the Nordic overflows.

Quick-start

1. Clone the repository

git clone https://github.com/JMMP-Group/GO-novf.git
cd GO-novf

2. Create and activate conda environment

conda env create -f pyogcm.yml
conda activate pyogcm

3. Ancillary files needed to generate localization masks and envelopes

The localisaztion masks that will be used to define the localised multi-envelope terrain-following vertical levels in the proximity of the Greenland-Scotland ridge region and all the ancillary files needed to generate and optimise the envelopes of GOSI10p3.0-eORCA025 and GOSI10p3.0-eORCA12 can be found at the following zenodo archives:

1. GOSI10p3.0-eORCA025:https://zenodo.org/records/15625102
2. GOSI10p3.0-eORCA12:https://zenodo.org/uploads/15633874

For the bathymetry files, GOSI10p3.0 uses the Storkey et al. 2024 bottom topography files:

1. GOSI10p3.0-eORCA025:https://zenodo.org/records/15494369
2. GOSI10p3.0-eORCA12:https://zenodo.org/records/15495870

4. Create the localisation area

cd src/loc_area/
python generate_loc_msk.py ${loc_area_novf} 

where loc_area_novf=loc_area_novf_gosi10_025.inp in the case of eORCA025 and loc_area_novf=loc_area_novf_gosi10_12.inp for eORCA12.

As we can see from the ${loc_area_novf} input files, in GOSI10 we use a wider localisation area than in Bruciaferri et al. 2024 - here we target the 2930m isobath instead of the 2800m one.

5. Create the envelopes

cd ../envelopes/
python generate_envelopes.py ${input_file}

where inp_file=MEs_novf_gosi10_025_4env_2930_r12_r16-r075-r040-r035_it2-r030.inp for the case of GOSI10p3.0-eORCA025 while inp_file=MEs_novf_gosi10_12_4env_2930_r12_r16-r075-r040-r035-r030-r025-r020-r015_itr3-r010.inp for GOSI10p3.0-eORCA12.

In order to reduce horizontal pressure gradient (HPG) errors, envelopes are smoothed using the iterative preocedure detailed in Appendix C of Bruciaferri et al. 2024, which uses the Martinho and Batteen (2006) smoothing algorithm to reduce the local slope parameter $r$ below multiple user defined $r_{max}$ values, effectively allowing one to apply distinct level of smoothing in different areas of the model domain. In the case of GOSI10, increasingly more severe $r_{max}$ values were applied only in those grid points where spurious currents were $\geq 0.025$ $m s^{-1}$.

The three months long tests to assess the horizontal pressure gradient (HPG) errors were conducted with

1. GOSI10p3.0-eORCA025: u-dl766@312554 suite.
2. GOSI10p3.0-eORCA12: u-dn555@311670 suite.

6. Generate the vertical grid

The vertical grid of GOSI10 with local ME s-levels in the Nordic overflows area is generated using the 423-adding-more-flexibility-to-me-gvcs@17eae5a7 development branch of the DOMAINcfg tool - this has now be merged in the main, see main@8e326e96b546251509d8d5a4db484500817ef784 commit.

The namelist_ref and namelist_cfg used to configure the vertical grid of GOSI10 can be found in namelists.

The output of this step is a domain_cfg.nc file - the files for GOSI10@1/4 and GOSI10@1/12 can be found at the following zenodo archives:

1. GOSI10p3.0-eORCA025:https://zenodo.org/uploads/15634496
2. GOSI10p3.0-eORCA12:https://zenodo.org/uploads/15634622

7. Horizontal pressure gradient error

Left is GOSI10p3.0-eORCA025, right GOSI10p3.0-eORCA12.

8. Comparison with GOSI9

GOSI9 - eORCA025 --------------------------

GOSI10 - eORCA025 --------------------------

The strange canyon that can be seen in the first sections of GOSI9 are due to the fact that in the case of GOSI9 the bathymetry has been modified by hand as shown in the following map:

GOSI10 - eORCA12 --------------------------

9. INITIAL CONDITION FOR GOSI WITH LOCALISED ME s-coordinates

Because of the way NEMO interpolates the T&S fields passed in input when initialising from rest (ln_tsd_init=.true.), the following two important point MUST be considered when initilising a model with localised ME levels (ln_sco=.true. and ln_loczgr=.true.):

1. Since the land-sea mask will be different in the localisation area, The T&S used to initialise the model should be flooded, i.e., continents should be filled with value from the ocean to avoid issue when interpolating in the vertical.

2. When using local GVC s-coord, the last level must be a copy of previous level:

ori_S=woa13v2.omip-clim.abs_sal_gosi10p1-025_flooded.nc
new_S=woa13v2.omip-clim.abs_sal_gosi10p1-025_flooded.MEs.nc
ori_T=woa13v2.omip-clim.con_tem_gosi10p1-025_flooded.nc
new_T=woa13v2.omip-clim.con_tem_gosi10p1-025_flooded.MEs.nc

ncap2 -O -s 'so_abs(:,74,:,:)=so_abs(:,73,:,:)' ${ori_S} ${new_S}
ncap2 -O -s 'thetao_con(:,74,:,:)=thetao_con(:,73,:,:)' ${ori_T} ${new_T}