Creating ocean_vgrid file

[anton-seaice]

To run MOM6, we provide an ocean_vgrid.nc as an input file.

At this point, we've only been creating the ocean_vgrid file by deriving / subsetting it from access-om2 files. See #101 (comment)

To ensure complete provenance, we need to be making this from scratch. I think we can write a script to do it, (probably can go in om3 scripts?)

@ezhilsabareesh8 - did you have a plan for this already ?

[AndyHoggANU]

I think that would be worth doing. If you don’t have them already, I think Kial Stewart would have created some scripts for the 75-level case from his 2017 vertical levels paper …

[ezhilsabareesh8]

I used the code here, referenced in Kial's 2017 paper, to replicate the OM2 1/10th-degree vertical grid with 75 levels. The grid starts with a 1.1 m spacing at the surface, increasing to 198.4 m at the bottom, with a maximum depth of 5808.7 m.

While the OM2 vertical grid generated using Kial's code is quite similar to the original OM2 1/10th-degree grid, the maximum depth differs between them. The grid generated using Kial’s code has a maximum depth of 5710.86 m, compared to the OM2 1/10th-degree grid's 5808.65 m. The dz_min and dz_max parameters I used here are identical to those used for the OM2 1/10th-degree grid. I also adjusted the depac parameter here to ensure the number of levels matched the OM2 grid.

Could there be an additional tuning step that accounts for this difference?

Vertical grid spacing (Plotted are difference between the vertices of supergrid):

[aekiss]

1.1 m spacing at the surface, increasing to 198.4 m at the bottom, with a maximum depth of 5808.7 m

These are only approximate values. What happens if you use the the topmost and bottom-most supergrid spacings from /g/data/vk83/experiments/inputs/access-om2/ocean/grids/vertical/global.01deg/2020.05.30/ocean_vgrid.nc, ie 0.541280746459961 and 5808.65673828125-5709.44287109375=99.2138671875?

Or the st_edges_ocean values derived from the top and bottom two supergrid cells in that file, ie 1.08256149291992 and 5808.65673828125-5610.22900390625=198.427734375?

[ezhilsabareesh8]

Thanks @aekiss. I reached out to Kial regarding the vertical grid code, and he provided some valuable insights. Kial mentioned that the vertical grid generated by his code might need to be adjusted to double precision. He also clarified the parameters he used for the KDS75 grid, were not included in the original paper or code:

H: 6000.0
dzd: 200.0
min_dz: 1.0
depfac: 1.101

Using these parameters, I was able to recreate the KDS75 grid, which matches the OM2 1/10th-degree vertical grid.