Feedbacks from biogeochemistry to physics in FABM (light attenuation)

[BeatrizArellano]

Hello,

I would like to ask for advice on best practices for representing biogeochemical feedbacks to physical variables, in particular those related to radiation penetration. More specifically, I am interested in how FABM is typically used to account for light attenuation driven by biogeochemical state variables (e.g. phytoplankton, detritus), and how this information is expected to be passed to the physical model.

From the documentation and some examples, I understand that the biogeochemical modules can define how different variables contribute to an attenuation coefficient of the radiative flux. However, I am unsure about what is considered the intended or most robust approach in practice. Should biogeochemical modules define optical properties via a standard variable (e.g. attenuation_coefficient_of_shortwave_flux), with the expectation that the physical model retrieves and applies this information when computing light penetration and heat fluxes?

Thanks in advance!

[jornbr]

FABM supports such feedbacks through standard variables. These can be predefined, such as attenuation_coefficient_of_photosynthetic_radiative_flux, or you can define your own as described in this section of the wiki. Thus, this is extensible - you can add any feedback to physics that you like. For a feedback to work in practice, the biogeochemical model that provides the variable and the physical model using the variable both have to be talking about the same thing. For FABM, this means that the standard variable you use must have the same name+units on both sides. This is important if you define your own standard variable to implement your feedback, as the biogeochemical and physical model then typically each define the standard variable independently - names and units need to match!

Some feedback examples:
visible light (PAR) attenuation:

fabm/src/models/examples/npzd/phy.F90

Lines 71 to 73 in 0df797c

	! Contribute to light attentuation
	call self%add_to_aggregate_variable(standard_variables%attenuation_coefficient_of_photosynthetic_radiative_flux, self%id_c, scale_factor=self%kc)
	call self%add_to_aggregate_variable(standard_variables%attenuation_coefficient_of_photosynthetic_radiative_flux, self%p0 * self%kc)

density:

fabm/src/models/akvaplan/tracer.F90

Lines 65 to 66 in 0df797c

	call self%add_to_aggregate_variable(non_water_volume_fraction,self%id_c,scale_factor=sp_vol)
	call self%add_to_aggregate_variable(non_water_density, self%id_c,scale_factor=sp_vol*sp_dens) ! converting from kg tracer/tracer_volume to kg tracer/total_volume

salinity:

fabm/src/models/niva/brom/brom_salt.F90

Lines 117 to 118 in 0df797c

	call self%add_to_aggregate_variable(mass_concentration_of_solute, self%id_Na, scale_factor=u_Na*1e-6_rk)
	call self%add_to_aggregate_variable(mass_concentration_of_solute, self%id_Cl, scale_factor=u_Cl*1e-6_rk)

morphology (sediment thickness):

fabm/src/models/niva/brom/brom_salt.F90

Line 119 in 0df797c

call self%add_to_aggregate_variable(volume_fraction_of_particulates, self%id_NaCl_bot, scale_factor=(u_Na+u_Cl)/1000/NaCl_density/(1-self%porosity))

heating due to shortwave absorption:
https://github.com/pmlmodelling/fabm-spectral/blob/cf0a2efa832f5e96d2a6f12d6d124f048b774e65/src/spectral.F90#L265

Most of these examples tend to be specific to particular research questions/processes, but they may serve as inspiration.

On the side of the physical model (host), you'd typically request the value of such standard variables with something like:

call model%require_data(<STANDARD_VARIABLE>) 
pvalues => model%get_interior_data(model%get_interior_variable_id(<STANDARD_VARIABLE>))

The first line is used before model%start to tell FABM that you are interested in this variable (and that it therefore needs to be computed and stored), the second line is used after model%start to retrieve a pointer to the variable data.

This is the general principle. If you are coding a biogeochemical model and are contemplating feedbacks, you'll typically want to find out what feedbacks from FABM are supported by your physical model of choice. Or you'd liaise with the developers of the physical model to implement the one you care about.

Hope this helps!