making LOBSTER easier to extend with external plankton communities

[nanophyto]

This is related to @BWBlack's 'FrankenLOBSTER' model, which I am hoping to turn into something more generic on the Agate.jl side:

using Agate.Models: FrankenLOBSTER

# Agate is used to define arbitrary community of n phytoplankton, n zooplankton and n bacterioplankton
plankton = FrankenLOBSTER.construct()   

# this is then passed to LOBSTER directly:
biogeochemistry = LOBSTER(
    plankton = plankton,
    [...]
    )

Thus more broadly, this would be about extending LOBSTER/NutrientsPlanktonDetritus to accept arbitrary externally-owned plankton tracers without OceanBioME needing to know the structure of the external community.

It would require two PRs (currently in prep):

1. a more generic 2-band light function that does not assume a single phytoplankton tracer named P

2. a LOBSTER/NutrientsPlanktonDetritus refactor which includes:

• formalizing and document the existing aggregate plankton/NutrientsPlanktonDetritus exchange hooks
• providing a supported way for externally owned tracers to declare the stoichiometric metadata needed by NutrientsPlanktonDetritus bookkeeping and conservation
• formalizing/documenting the existing grazing(...) contract for consumption of NutrientsPlanktonDetritus-owned detritus

Maybe this will make sense once the PRs are ready, but I figured this could be a good starting point and perhaps @BWBlack can also weight in.

Hopefully this is also useful for OceanBioME beyond the Agate use case, since it would remove a few assumptions about plankton structure/naming and make the LOBSTER/NutrientsPlanktonDetritus interfaces a bit more generic.

More broadly, I see this as a targeted way to improve Agate x OceanBioME compat which we can hopefully extend to PISCES in the future.

[jagoosw]

Hi @nanophyto,

That all sounds like a good plan.

For the light I think we could straightforwardly make a function like chlorophyll(i, j, k, grid, fields, biogeochemistry) that returns a value, so for default LOBETER it would just return fields.P[i, j, k] times chlorophyll ratio.

We could definitely do with documentation of the new structure. For the ratios you can import the carbon_ratio etc functions and add methods for other planktons.

I'm going to be working on MARBL and want to keep everything as generic as possible and keeping everything interoperable, and this will definitely be helpful there too.

[nanophyto]

Great! I've opened two related PR's #391 and #393 hopefully they both make sense