Feature/535 stratification diag

[jpolton]

WIP

A couple of diagnostics files are separated from the SE-NEMO branch as they are without unit testing. This means they can not be maintained.

These files have been edited for consistency with COAsT conventions.
They also have some COMMENTS within.

Without worked examples and testable code, this will not age well and so is not ready for develop.

AC (for discussion):

• documentation
• unit tests

[soutobias]

Hi @jasontempestholt @jpolton , I am in a conference right now and after that I will be on annual leave. I will return to the office on 04th of december. I will write here some of the updates:

@jasontempestholt , here's a summary of our discussion and the action points:

1. gridded_monthly_hydrographic_climatology:

• Incorporate unit tests (Tobias) -> I wrote it using zarr data. However, I did not have time to update the results of the tests
• Provide a notebook or example demonstrating the code usage -> I wrote it using zarr data. I did not have access to the example files that Jason used, because it is on a Jasmin VM that I did not have access. I can work on this part when I come back
• Optimize the method PEA.calc_pea for improved performance, considering the expensive nature of GSW (Tobias). -> I do not have salinity, temperature and bathymetry data on that zarr files. Because of that, it was not possible to calculate PEA. When I have access to the nc files, I will work on that.
• Share a small dataset for use as example_files, with the data residing on JASMIN; this can be a link to the object store. -> I used zarr files to test the codes. *
• Prepare a notebook with an example (Tobias) -> I prepared a notebook using zarr files
• Documentation (Tobias) -> I will work on that when I come back. I think I can add more information to the example notebook

2. profile_hydrographic_analysis:

• I did not have time to work on this one. I still need to have access to a sample dataset and a sample code on how to use it. *

[jasontempestholt]

I've changed the gridded climatology method so that it doesn't produce a copy of the original data set - this makes saving the output more difficult.

I've check it gets same answer to earlier version - and it does for PEA (bit identical), but not identical SST or SSS. I think because sos_abs ~= so_abs[:,0,:,:] for some reason. But nothign to worry about

[jasontempestholt]

@jpolton I've added the match-to-grid capability and found (and fixed) a very slow bit of the code. So this now produces NWS profile dataset 1978-2019 in ~1 hour. So should be practical cycle through all LME's without further optimisation. I'll need some guidance on finishing the pull to develop.