Implementation of spectral truncation in Eulerian dycore

[thakur.abubakar]

Dear All,

I'm trying to understand how the CAM Eulerian dycore implements spectral truncation. I already know what parameters control the truncation (ptrm, ptrn and ptrk) but it appears that the truncation is quite harsh.

The trunc() subroutine in components/cam/src/dynamics/dyn_grid.F90 does consistency checks on the spectral parameters etc. I'm interested in where the truncation is being done and how it is being done.

I would prefer to damp out the wavenumbers beyond 'ptrm'.

Any guidance in this regard would be greatly appreciated.

Thanks and regards,
Abu Bakar

 

[nusbaume]

Hi Abu Bakar,

Here is a quote I received from our (now-retired) scientist who developed the Eulerian dycore, in response to your questions:

The truncation is a byproduct of the spectral transformations from grid point space to spectral space associated with the time stepping algorithm. There is no truncation of coefficients after any are calculated; only those coefficients implied by the truncation parameters are calculated.

One can include damping of individual coefficients with the diffusion calculation in spectral space. One can increase the truncation parameters above the default values which are consistent with a quadratic unaliased Gaussian grid, up to values consistent with a linear unaliased Gaussian grid, but there may be stability issues if the damping is not strong enough.

They also recommended that if you would like any additional information then you can check out Section 3.3 (Eulerian Dynamical Core) and the associated subsections in the scientific description of the CAM5 model, which can be found online here:

https://www.cesm.ucar.edu/models/cesm1.2/cam/docs/description/cam5_desc.pdf

Hope that helps, and have a great day!

Jesse