O2.5.4: Run coupled ClimaAtmos + ClimaLand soil/canopy/snow model and ensure stability comparable to bucket model

[juliasloan25]

The Climate Modeling Alliance

Software Design Issue 📜

Purpose

We want to couple with a more complex land model than the bucket, with the goal of providing more realistic boundary conditions to the atmosphere. The full land model contains 3 main subcomponents: soil, canopy, and snow. It also contains a Soil CO2 model, but this will be largely uninvolved in the flux exchanges when coupling with the atmosphere.

Cost/Benefits/Risks

costs: developer time, potentially unseen changes to ClimaCoupler
benefits: more accurate model, will test ClimaCoupler capabilities in new setup
risks: model may be less stable and need to be debugged in Atmos, Land, and Coupler

People and Personnel

• Lead: @juliasloan25
• Collaborators: @imreddyTeja @kmdeck @szy21
• Reviewers: @Sbozzolo

Components

• Add a new option to select between the complex land model or the bucket for the land component
• Extend Coupler-required adapter functions for the new complex land model type
• Update partitioned fluxes to handle the new land model type (including each of the sub-components of the land model)

Inputs

• Existing coupler codebase
• Coupled runs using bucket
• ClimaLand LandModel longruns
• O2.5.2: Generalize coupling and surface fluxes to accommodate full land model #1031
  • Will need some form of generalized flux calculation for coupling full land model, but can be simplified for now (see update PartitionedSurfaceFluxes for full land model #1180)
• O2.5.3: Run single column model coupled runs of ClimaAtmos + ClimaLand soil model #1032
  • Could be a nice stepping stone to global coupled runs, but not strictly a prerequisite
  • We'll skip this step and go straight to coupling the full land model (soil+canopy+snow). I don't think coupling the soil model alone would be very informative in terms of changes required in the coupler

Results and Deliverables

• Global runs of coupled atmos + full land, with stability similar to bucket runs

Scope of Work

Changes needed in ClimaCoupler.jl

turbulent fluxes

• extend partitioned fluxes calculation to compute for each land component
  • for now, call functions defined in ClimaLand, which are already specified for each component
• land update_field! for turbulent_energy_flux, turbulent_moisture_flux: provide tuple of fields for each land component rather than one field
• atmos update_field! for turbulent_energy_flux, turbulent_moisture_flux: provide one area-weighted field
  • Q: do we want to store all 4 TF fields in the coupler, or store 3 and compute the area-weighted field on the fly
• land get_field for roughness_buoyancy, roughness_momentum, surface_temperature, displacement_height: return tuple of fields for each land component rather than one field
• after computing turbulent fluxes in coupler, need update_field!s to provide turbulent_energy_flux, turbulent_moisture_flux to both atmos and surface models

radiative fluxes

• land get_field functions for emissivity, albedo: 1 field each for entire land model, not one per component
  • also return T_eff from land surface_temperature get_field so we can use this for radiative fluxes in atmosphere
• add atmos get_field functions for SW_d, LW_d, cos(zenith_angle), diffuse_fraction
• add land update_field! functions for SW_d, LW_d, cos(zenith_angle), diffuse_fraction

Changes needed in ClimaLand.jl

turbulent fluxes

• functions to compute q, \rho from land, atmos inputs, return land thermo state
  • this will be called by coupler during flux calculation
  • move inside land turbulent flux calculation function? already has access to atmos thermo state

radiative fluxes

• move diffuse_fraction update out of update_aux!, into function that dispatches on radiation driver type
• use cos(zenith angle) instead of zenith angle to avoid converting back and forth

conservations

• functions to compute total water, total energy
  • integrate soil variables over column

Changes needed in ClimaAtmos.jl

• provide diffuse light fraction, SW_d, LW_d, cos(zenith_angle), LHF, SHF to coupler via get_field functions

Accommodate full land model and fields required for flux calculations

Preview Give feedback

1. Dispatch on component model types to construct coupler fields #1206
   🍃 leaf +1
   
2. add default get_field methods #1210
   +0
   
3. AtmosSimulation changes needed to couple with full land #1200
   🍃 leaf +1
   
4. Changes needed for full land model coupling with ClimaAtmos ClimaLand.jl#1038
   enhancement +1
   
5. Update coupled drivers for coupling with ClimaAtmos ClimaLand.jl#1041
   +0
   
6. add ClimaLandModel struct and interfacing functions #1182
   🍃 leaf +1
   

Compute fluxes in ClimaCoupler for each component of land model

Preview Give feedback

1. Extend ClimaLand turbulent_fluxes! methods for CoupledAtmosphere #1202
   🍃 leaf +1
   

Run simulation with full land model

Preview Give feedback

1. Add coupled atmos/full land model run to CI
2. Debug stability of coupled run

SDI Revision Log

CC

@tapios @sriharshakandala @charleskawczynski @cmbengue

[tapios]

Looks good. It'll be exciting to see the full land model coupled to the atmosphere.