...
E3SM V1. V1 Topography: GLL grids
V1 uses topography on a GLL grid with a laplacian smoothing. The smoothing reduces the effective resolution of the topography from 2dx to 6.4dx. (effective resolution is determined by the first wavelength to be damped by 50% from the unsmoothed topography, computed via spherical harmonic spectra)
V1 only needs data on the GLL grid.
Tool chain: Requires cube_to_target, homme_tool, and a SCRIP file for the grid.
V1 has poor treatment of topography and compensates in part with increased divergence damping.
E3SM V2:V2 Topography: GLL/PG2 grids
V2 needs topography on the GLL grid, and then a consistent topography on the PG2 grid (created with the same GLL->PG2 algorithm used internally when running EAM)
V2 uses the same smoothing as V1: 50% damping of the 6.4dx wavelength.
V2 has improved topography treatment over V1. V2 configurations do not need to run with increased divergence damping.
E3SM V3 and EAMxx/SCREAM: V3 Topography: GLL/PG2 grids
E3SM V3 and EAMxx have further improved topography treatments ( namelist varaiables pgrad_correction=1, hv_ref_profiles=6, see ( /wiki/spaces/NGDNA/pages/2411397876 ) which allow for rougher topography
The V3 tool chain is similar to V2, with minor modifications to in the smoothing algorithm. We further recommend less smoothing, producing rougher topography.
The rougher topography datasets are the ones with “x6t” in the name - 6 iterations of the tensor-laplace operator. EAM V3 has further improved topography treatments over EAM V2, supporting significantly rougher topography ( /wiki/spaces/NGDNA/pages/2411397876 ) These improvements are enabled by namelist variables ( pgrad_correction=1, hv_ref_profiles=6) which should be made defaults in V3.