Technical note (without namelist options, check chapter 8 to learn about the dissipation options):
https://github.com/NOAA-GFDL/GFDL_atmos_cubed_sphere/blob/main/docs/fv3_technical_2021.pdf
FV3 namelist options (somewhat outdated from 2017):
https://www.gfdl.noaa.gov/wp-content/uploads/2017/09/fv3_namelist_Feb2017.pdf
Horizontal transport is defined by the namelist parameters beginning with hord. There are five main options, which are all variants of the piecewise parabolic method, in either an unlimited (fewer constraints) or monotonic (more constraints to ensure monotonicity -- not generating additional maxima and minima) configuration.
hord = 5: Least inviscid unlimited scheme. hord=-5 adds a positive-definite constraint.
hord = 6: Another unlimited scheme with some monotonicity constraints.
hord=7: like hord = 6 with a positive-definite constraint.
hord = 8: The fastest monotonic scheme
hord = 9: A slower monotonic scheme, due to more constraints, but should be less diffusive.
hord = 10: A monotonic scheme that only applies constraints in certain circumstances. hord=-10 adds a positive-definite constraint.
Namelist entries that determine the limiters for the PPM subgrid distribution:
fv3_hord_mt, fv3_hord_tm, fv3_hord_vt: Horizontal transport option for the horizontal momentum fluxes, potential temperature, and vorticity, respectively. These should be set to the same value.
fv3_hord_dp: Horizontal transport option for the pressure thickness of each level, which is a proxy for the mass within the cell. This should be set equal to the previous hord values, unless a positive-definite option is instead used.
fv3_hord_tr: Horizontal transport of tracers, should be positive definite (e.g. -5, -10) or monotonic. Default is 8.
Vertical transport in FV3 is performed through evolving the Lagrangian layers. These need to be periodically remapped. There are six options for remapping:
kord = 4,6,7 use the same piecewise parabolic method as the horizontal transport
kord = 9,10,11 use a high-order reconstruction with cubic splines
The related namelist setting for the vertical remapping step are:
fv3_kord_mt: Vertical remapping scheme for horizontal momentum fluxes. Default is 9 and a cubic spline method is recommended.
fv3_kord_tm. Vertical remapping scheme for potential temperature. It is recommended to use -9, which is option 9 with a positive-definite constraint.
fv3_kord_tr: Vertical remapping scheme for tracers. Default is 9. It is recommended that this is the same option as fv3_kord_tm (barring the positive-definite constraint).
fv3_hord_wz: Vertical remapping scheme for vertical velocity in a nonhydrostatic configuration. Default is 9. It is recommended that this is the same option as fv3_kord_mt.
fv3_k_split: Number of vertical Lagrangian layer remappings in each dynamics timestep. Default is 2.
fv3_q_split: Number of substeps for the tracer transport.
fv3_n_split: Number of acoustic substeps between each vertical Lagrangian layer remapping. Default is 6.
FV3 uses divergence and vorticity damping instead of directly damping the prognostic horizontal velocities.
fv3_d2_bg: Dimensionless coefficient of Laplacian divergence damping in the main region of the domain (not the sponge layers). Default is 0. The maximum stable value (from linear stability theory) for Laplacian damping on the equi-edge grid is 0.289. Recommended range is 0 to 0.02.
fv3_d2_bg_k1: Dimensionless coefficient of Laplacian divergence damping in the upper sponge layer, nearest the model top. Default is 0.15. Recommended range is 0. to 0.2
fv3_d2_bg_k2: Dimensionless coefficient of Laplacian divergence damping in the second sponge layer, between the top sponge layer and lowest subdomain. The diffusion in this layer should be weaker than the upper-most sponge layer. Default is 0.02.
fv3_d4_bg: Dimensionless coefficient of hyperdiffusion (fourth-, sixth-, or eighth-order) in the main region of the domain (not the sponge layers). The maximum coefficients that can be used (for linear stability) are: 4th-order: 0.204, 6th-order: 0.182, 8th-order: 0.172. Note, that these are for the equi-edge (default) grid. Default is 0.15.
fv3_dddmp: Dimensionless coefficient for Smagorinsky-type Laplacian diffusion. Note that this is not the true Smagorinsky diffusion used in MPAS, as it scales with the flow divergence and vorticity instead of the tension and shear. Default is 0. Value of 0.2 (the Smagorinsky constant) is recommended for noisy simulations.
fv3_do_vort_damp: Logical variable to determine whether to use vorticity damping. Default is .false., which relies upon sufficient diffusion of vorticity from transport.
fv3_nord: The order of divergence damping, which also determines the order of vorticity damping:
For divergence damping, fv3_nord=0 for Laplacian damping, 1,2,3 for fourth-, sixth-, and eighth-order hyperdiffusion.
Only fourth- and sixth-order vorticity damping is available; fourth-order is used when nord = 0 or 1, and sixth-order when nord = 2 or 3. The default for nord in CAM-FV3 is 3, although 2 is recommended by the DCMIP organisers.
fv3_n_sponge: Number of layers (counted from the top) to apply a 2dz filter; this is different to Laplacian sponges set with fv3_d2_bg_k1 and fv3_d2_bg_k2.
fv3_d_ext: coefficient for external (barotropic) mode damping. Proper range is 0 to 0.02. Default is 0. Seems to be faulty in CAM-FV3.
fv3_vtdm4: Dimensionless coefficient for vorticity damping. This should be much smaller than fv3_d4_bg, due to implicit diffusion of vorticity in the transport; linear instability will occur at roughly half the maximum value for divergence damping. A good first guess for fv3_vtdm4 is about one-third the value of fv3_d4_bg. Requires fv3_do_vort_damp to be .true. Disabled for values less than 1.e-3. This damping should not be used if a monotonic horizontal advection scheme is used.
fv3_rf_cutoff: Pressure level above which to apply Rayleigh friction: p < fv3_rf_cutoff. This value is in Pa.
fv3_tau: Timescale for the Rayleigh friction, in days.
fv3_grid_type: The choice of gnomonic projection for the cubed-sphere grid. 2 for equiangular (this is what SE uses) and 0 for equi-edge, which is FV3's default.
fv3_layout: Specify the domain decomposition for parallel computing. Set with two comma-separated numbers (L1, L2), ensuring that L1*L2*6 equals the number of processors.
fv3_hydrostatic: Whether to use the hydrostatic solver. Default is true, and set to false for nonhydrostatic simulations.