Re-code ocean model in pure python

[sdat2]

The model will be much easier to use if it were written in python. To deal with Python slowness, I will need to try and use numba and/or xarray. Ideally xarray should be used as much as possible, as it is very popular in the field. pint and pint-xarray could be used so that unit are propagated alongside values, as this is a form of error correction.

• Calculate baroclinic mode and baroclinic modes.
• Simulate thermocline with baroclinic modes alone.
• Add INC (Israeli, Naik, Cane) integration scheme.
• with the calculation of upwelling.
• Add Ekman mode and test if the result sensibly propagates wind stress.
• Add full temperature propagation.
• Add surface fluxes.
• Add geometry to allow land blocks to be added.
• Add more physical tests.
• Make sure units propagate through the whole model by default (if this does not significantly negatively impact the model).
• [Once thoroughly tested] Add python model as pip package.

Questions to answer:

• Does a realistic coupling period affect the long term trends?
• Does a realistic coupling period produce reasonable magnitudes for ENSO.
• Does the integration scheme affect the results?
• Can the python model be made to have similar computational efficiency to the Fortran model?