Mesoscale Simulations in Library

[sandreza]

This PR adds idealized mesoscale resolving zonally re-entrant channels to the library. The first stage is to define a function whose input is a few control parameters and whose output can be used for informing parameterization development. There are two main classes of simulations being added: flat bottom and hilly

[sandreza]

Tentative to-do list

• Surface Buoyancy Relaxation
• Quadratic Drag
• Check output in simple case

[glwagner]

Do you mind if I commit to this PR?

TODO:

• [x] put eddying_channel_simulation into IdealizedExperiments, get rid of IdealizedMesoscaleExperiments
• [x] run the cases and report results in this PR
• [x] add a simple example analogous to https://github.com/CliMA/LESbrary.jl/blob/805c473b11f517a53e43f40b8c32d94988ecb73e/examples/run_three_layer_constant_fluxes.jl where we demonstrate how to use the function eddying_channel_simulation
• [x] We probably don't need the example examples/eddying_channel.jl

[glwagner]

A few results. I added biharmonic dissipation and removed the Laplacian diffusion, which wasn't having any effect (dissipation was dominated by WENO I guess). With a coefficient Δx^4 / 15days, we get smooth solutions after 1 year of simulation:

Constrast this with

a bit noisier.

With a biharmonic coefficient Δx^4 / 10days we suppress the growth rate of the baroclinic instability, so that after a year we are just about to go unstable:

[glwagner]

Ok, I've also implemented an initial buoyancy distribution that matches the relaxation profile. Here's what we get after a year with Δx^4 / 20days

The simulations also complete faster because we don't get large vertical velocities early on, perhaps because the initial turbulent motions are a bit larger scale?

[glwagner]

Note: we'll have to implement some tracer forcing if we want it to be useful for this equilibrated case. Something for the future.