Closed milankl closed 7 months ago
@TomBolton Is this ☝🏼 behaviour to be expected if your parameterization is too strong? Note this is an atmospheric shallow water model setup and also I guess in general your $\kappa = -4.87e8$ strength should probably depend on the resolution too? I remember self-amplifying eddies also from backscatter experiments and it also makes dynamically sense to me if you end up putting more energy in than you take out!
No idea sorry! I haven't touched this stuff in years 😅
Haha, fair enough 😉
@swilliamson7 I'm closing this issue as I'm currently not working on this, but if you ever want to feel free to reopen!!
I've just implemented the Zanna & Bolton eddy parameterization
with $D = \partial_x v + \partial_y u$ the shear, and $\bar{D} = \partial_x u - \partial_y v$ the strain in vorticity-divergence formulation as (here now divergence $\sigma$ to not confuse it with Zanna and Bolton's notation)
using
because that replaces the meridional derivative with the already available vorticity and divergence. Because $D,\bar{D}$ have to be available in grid-point space for the products with $\zeta$, we "cheat" and evaluate $\partial_xu, \partial_yu$ in grid-point space with centred 2nd order finite differences (which is easy to generalise to any ringgrid!) in order to avoid 4 transforms . In spherical coordinates with radius scaling we then have
On the grid, do (radius squared scaling omitted)
Divide by cosine squared as we'll be taking two divergences/divergence+curl. Then to spectral $\hat{A},\hat{B},\hat{C}$.
Implemented as
Construction
Initialization
Simulation
And then run with
so about 2x stronger than default at T127 the blowup-test shows a self-amplifying eddy, which seems to inject energy into the flow. No more sophisticated test that the term is actually injecting energy into the system, but that sound somehow legit
@swilliamson7