Add salinity restoring file generation

[ezhilsabareesh8]

This script processes and smooths sea water salinity data from the initial conditions NetCDF files generated using https://github.com/COSIMA/initial_conditions_access-om2 . It applies a uniform smoothing filter to the surface layer (0m depth) of the salinity for each month and concatenates the smoothed data into a single output NetCDF file.

[anton-seaice]

Thanks @ezhilsabareesh8 - I will review soon.

I assume this uses the hh5 conda/analysis module ?

If you run black on the script and commit the change, then the formatting test should pass.

From the Readme:

Python scripts are formatted using the black formatter. This is enforced using a github action running on pull requests. You will need to have pre-commit and black installed in your python environment. It is included in the conda/analysis enviornments in hh5, otherwise it can be installed through pip or conda. Then run pre-commit install once to ensure your commits have black run on them before committing.

[ezhilsabareesh8]

Thanks @anton-seaice, I just reformatted it using black.

[anton-seaice]

Thanks @ezhilsabareesh8

If you keep the data as an xarray object, it will save having to recreate all the attributes and dimensions etc. See the suggestions

[anton-seaice]

If you use xarray suggestions above, the only extra attributes you need are these (i think):

    salt_var.standard_name = "sea_water_salinity"
    salt_var.long_name = "Smoothed sea water salinity at level 0m"
    salt_var.units = "1"

you should just be able to set those and then say

salt_smoothed_da.to_netcdf()

[anton-seaice]

Have a look now @ezhilsabareesh8 - this should work if we fix the time:units attribute in the source data

[anton-seaice]

This change will contribute to https://github.com/COSIMA/access-om3/issues/207

[anton-seaice]

Lets park this until

• Fix coordinates COSIMA/initial_conditions_access-om2#6
• Update to WOA23 COSIMA/initial_conditions_access-om2#7
• numba not supported for numpy masked array COSIMA/initial_conditions_access-om2#3

are completed, and then we can finish this review.

[ezhilsabareesh8]

@anton-seaice I noticed that the time variable was changed to match the initial conditions, but the salinity restoring should actually begin at the middle of each month. In my original code, I had overwritten the time variable with:

times[:] = np.arange(0.5, 12.5, 1)  # Time values from 0.5 to 11.5
times.units = "months since 0001-01-01 00:00:00"

Could you please revert the changes made to the time variable in this commit 8b0c207?

[anton-seaice]

@ezhilsabareesh8 - I think this script should go in https://github.com/ACCESS-NRI/initial_conditions_access-om3/ ?

[anton-seaice]

Hi @ezhilsabareesh8

I pushed a couple of refinements to go with COSIMA/initial_conditions_access-om2#6

This is the time coordinate now, which should be correct by cf-conventions and be the middle of the month

 time = 15.5, 45, 74.5, 105, 135.5, 166, 196.5, 227.5, 258, 288.5, 319, 349.5 ;

 double time(time) ;
                time:_FillValue = NaN ;
                time:standard_name = "time" ;
                time:long_name = "time" ;
                time:axis = "T" ;
                time:climatology = "climatology_bounds" ;
                time:units = "days since 1955-01-01" ;
                time:calendar = "gregorian" ;

I am running some tests now, can you see any issues ?

This is the diff between /g/data/vk83/prerelease/configurations/inputs/access-om3/mom/surface_salt_restoring/global.25km/2025.01.30/salt_sfc_restore.nc and my new one in January. Its bigger than expected, can that be explained by the changes we made to the initial conditions (e.g. 75 levels instead of 150 ?)

[ezhilsabareesh8]

Its bigger than expected, can that be explained by the changes we made to the initial conditions (e.g. 75 levels instead of 150 ?)

We are using the surface salinity for restoring and smoothing it horizontally, as discussed here., so I don't think that the difference is due to the change in vertical levels.

[ezhilsabareesh8]

There's no surface salinity difference between the old initial conditions /g/data/vk83/prerelease/configurations/inputs/access-om3/mom/initial_conditions/global.25km/2025.03.19/ocean_temp_salt.res.nc and new initial conditions /g/data/vk83/prerelease/configurations/inputs/access-om3/mom/initial_conditions/global.25km/2025.06.17/ocean_temp_salt.res.nc.

[anton-seaice]

Thanks - in the 25km we are using the 2025.01.30 version of the salinity restoring still.

Looking at the 2025.01.30 initial conditions, it has 38 levels, and the top layer is at 1.083m, however in the new initial conditions, the top layer is at 0m. That's why we are seeing the difference.

New-old, top layer:

New-old, 1.083m layer:

If you are happy, feel free to merge this :)

[anton-seaice]

Oh I also ran the 25km for a year with the new salinity restore file and it appears to run ok

[ezhilsabareesh8]

Thanks @anton-seaice. It looks good to me.