Generalises gas transfer velocity and allows surface_value to interpolate FieldTimeSeries

[jagoosw]

Closes #368 @lgloege

I'm a little concerned about how deep the parameter space is for these boundary conditions and how that might effect peformance.

[jagoosw]

@lgloege As an example of how this could now be generalised, if you wanted a transfer velocity dependant on wind speed, temperature, and significant wave height, $k(u, T, H)$, you could make a struct like:

struct WindWaveTempTransferVelocity{FT, WS, WH}
    a_constant :: FT
    wind_speed :: WS
   wave_height :: WH
end

and define a function:

using OceanBioME.Models.GasExchangeModel: surface_value

@inline function (k::WindWaveTempTransferVelocity)(i, j, grid, clock, model_fields)
    T = @inbounds model_fields.T[i, j, grid.Nz]
    u = surface_value(k.wind_speed, i, j, grid, clock, model_fields)
    H = surface_value(k.wave_height, i, j, grid, clock, model_fields)

    return k.a_constant * u * H
end

Then you can make an instance like:

k0 = WindWaveTempTransferVelocity(0.1, wind_speed, wave_height)

where wind_speed and wave_height are number, arrays (in space not time), fields, or field time series (on the same grid as the model or not). When you actually make the boundary condition you do need to wrap this in the Schmidt number scaling:

k = SchmidtScaledTransferVelocity(
         schmidt_number = CarbonDioxidePolynomialSchmidtNumber(), 
         solubility = MolPerKgPerAtmToMMolPerCubicMPerMicroAtm(carbon_chemistry.solubility, carbon_chemistry.density_function),
         base_transfer_velocity = k0
     )

which goes in the boundary condition constructor:

CarbonDioxideGasExchangeBoundaryCondition(; transfer_velocity = k)

This isn't the best interface but we can work on that.

Perhaps there is a way NumericalEarths atmosphere might carry around the waves (or someone might write a wave model in Julia that could be coupled), but for now it seems most straight forward to directly put the wave info in this way.