Adding Surface QG dynamics?

[BrodiePearson]

Are there plans to include a SurfaceQG model in GeophysicalFlows, to complement the other 2D and QG systems that can be modelled?

I started making an SQG model for GeophysicalFlows and I'd be happy to share/collaborate on it if there is not one already being developed. A very rough example code is here, and the SurfaceQG.jl file is also in that branch. It produces flow fields that look reasonable but I am still working to reproduce the results of a specific paper. I'd be happy to put together a doc of the equation set I'm trying to solve if that would be helpful (it is essentially the Equations of Section 2.1 in this paper).

[navidcy]

We don't have any such plans at the moment so feel free to continue work on SurfaceQG.jl. Ping us if you run into trouble or you want to chat about anything!

[glwagner]

I have no plans!

I'd suggest opening a PR sooner rather than later so we can collaborate on the new module + docs!

The main challenge is coming up with dynamics / physics unit tests. For example, for two-dimensional turbulence we use the Lamb dipole:

GeophysicalFlows.jl/test/test_twodnavierstokes.jl

Lines 1 to 21 in 5f7795c

	function test_twodnavierstokes_lambdipole(n, dt, dev::Device=CPU(); L=2π, Ue=1, Re=L/20, ν=0.0, nν=1, ti=L/Ue*0.01, nm=3)
	nt = round(Int, ti/dt)
	prob = TwoDNavierStokes.Problem(dev; nx=n, Lx=L, ν=ν, nν=nν, dt=dt, stepper="FilteredRK4")
	zeta₀ = lambdipole(Ue, Re, prob.grid)
	TwoDNavierStokes.set_zeta!(prob, zeta₀)
	xzeta = zeros(nm) # centroid of abs(zeta)
	Ue_m = zeros(nm) # measured dipole speed
	x, y = gridpoints(prob.grid)
	zeta = prob.vars.zeta
	for i = 1:nm # step forward
	stepforward!(prob, nt)
	TwoDNavierStokes.updatevars!(prob)
	xzeta[i] = mean(@. abs(zeta)*x) / mean(abs.(zeta))
	if i > 1
	Ue_m[i] = (xzeta[i]-xzeta[i-1]) / ((nt-1)*dt)
	end
	end
	isapprox(Ue, mean(Ue_m[2:end]), rtol=rtol_lambdipole)
	end

Perhaps there's a simple analogue for SQG, or we can come up with something else. Often such tests are nice sanity checks to have in place before publishing anything, too. @BrodiePearson have you come up with any tests?

It'd be fun to make an example out of a script that reproduces the results of the paper, too.