Idealized ice-ocean interaction scenarios

[glwagner]

We need a few idealized scenarios to test and explore the dynamics of our coupled ice-ocean system. I think we should try at least 2, but maybe 3 cases (1D, 2D, 3D).

• The 1D case could be a diurnal cycle in fall, when we expect ice growth.
• The 3D case could be to reproduce http://eisenman.ucsd.edu/papers/Si-Stewart-Eisenman-2022.pdf
• (The 2D case could just be the zonal average of the 3D case.)

@jbuffo @jlk9

[glwagner]

@navidcy has set up an Antarctic Shelf Current simulation before... not sure what they did about the surface / sea ice boundary conditions but he might be able to point @jlk9 to some code.

Also @jlk9 do you have access to GPUs?

[glwagner]

As for a column model, @jlk9 check out this somewhat-related example:

https://github.com/CliMA/ClimaOceanBiogeochemistry.jl/blob/main/examples/nutrients_plankton_bacteria_detritus.jl

That example uses CATKEVerticalDiffusivity to model ocean turbulence which I think we should also use for our column model here. Of course, you can ignore the biogeochemistry for now for the ice column model. I think we just want to use temperature and salinity with a linear equation of state.

Another relevant case is

https://github.com/CliMA/Oceananigans.jl/blob/main/examples/ocean_wind_mixing_and_convection.jl

[glwagner]

We can also consider setting up a large eddy simulation (LES) case. For example, we could try to embellish the simulations in this paper:

https://journals.ametsoc.org/view/journals/phoc/51/2/jpo-d-20-0114.1.xml

with a dynamical ice model.

[navidcy]

@navidcy has set up an Antarctic Shelf Current simulation before... not sure what they did about the surface / sea ice boundary conditions but he might be able to point @jlk9 to some code.

Yes, we did something. Quite preliminary... but here it is
https://github.com/navidcy/ASC-idealised

Sea-ice: we ignored it :)
Surface BCs: relax to a T and and S profile

Happy to chat more!

[jlk9]

@glwagner Thanks for the references! I have access to GPUs, Patrick got a Lonestar6 allocation on TACC. It doesn't have Julia modules but I've downloaded it locally and gotten Julia code to run. The queue times for GPU jobs can be kind of long though.