ERF vs WRF

The following comparison is based on the WRF Version 4 Technical Report, titled “A Description of the Advancd Research WRF Model Version 4”

Similarities

Equations: both ERF and WRF solve the fully-compressible, Eulerian nonhydrostatic equations, and conserve dry air mass and scalar mass. ERF does not have a hydrostatic option.

Prognostic Variables: velocity components (u,v,w); perturbation moist potential temperature. Optionally, turbulent kinetic energy and any number of scalars such as water vapor mixing ratio, rain/snow mixing ratio, cloud water / ice mixing ratio.

Horizontal grid: both ERF and WRF use Arakawa C-grid staggering.

Time Integration: Time-split integration using 3rd-order Runge-Kutta scheme with smaller time step for acoustic and gravity wave modes. Variable time step capability. Vertically implicit acoustic step off-centering.

Spatial Discretization: 2nd- to 6th-order advection options in horizontal and vertical. In addition, several different WENO schemes are available for scalar variables other than density and potential temperature.

Turbulent Mixing: ERF and WRF have the same sub-grid scale turbulence closures with the Smagorinsky or 1.5-order TKE (Deardorff) model, in isotropic or anisotropic forms, for large-eddy simulation (LES); planetary boundary layer (PBL) schemes (MYNN, YSU) are available. ERF also has support for RANS turbulence modeling.

Diffusion: In WRF and ERF, constant diffusion coefficients may be specified (\(K_h\) and \(K_v\) for horizontal and vertical diffusion). Constant dynamic viscosity may also be specified in ERF. Variable diffusivity is provided in 3-D through LES modeling and in 1-D through PBL modeling. For mesoscale applications, 3-D diffusion is provided by combining a PBL scheme with the Smagorinsky model. Prandtl and Schmidt numbers are used to derive diffusivities of heat or other scalars from the diffusivity of momentum.

Initial conditions: both ERF and WRF have the ability to initialize problems from 3-D “real” data (output of real.exe), “ideal” data (output of ideal.exe), and from 1-D input soundings.

Lateral boundary conditions: Periodic, open, symmetric and specified (in wrfbdy* files).

Bottom boundary conditions: Frictional (Monin-Obukhov Similarity Theory) or free-slip

Earth’s Rotation: Coriolis terms in ERF controlled by run-time input flag (2-D or 3-D, constant or spatially varying for real-data cases)

Mapping to Sphere: ERF supports the use of map scale factors for isotropic projections (read in from wrfinput files).

Nesting: One-way or two-way. Multiple levels and integer ratios.

Wind Energy Modeling: Wind farm parameterizations and a generalize actuator disk are available.

Key Differences

ERF provides performance portability on different computing architectures including GPUs from all major vendors (NVIDIA, AMD, and Intel).

Vertical Coordinates: Unlike WRF, ERF uses a height-based vertical coordinate, with vertical grid stretching permitted.

Governing Equations: ERF supports both fully compressible and anelastic equation sets.

Time Integration: ERF supports using a 3rd-order Runge-Kutta scheme with explicit acoustic substepping or no substepping (in addition to the implicit acoustic substepping in WRF).

Representation of Surface Features: Terrain and urban geometries may be simulated with immersed forcing or embedded (immersed) boundary techniques, in addition to the terrain-fitted coordinates approach.

Interface with AMR-Wind: ERF may be tightly coupled with AMR-Wind, an incompressible ABL solver with integrated turbine aeroservoelastic dynamics modeling and two-phase flow capabilities.

Particles: ERF can be compiled with support for particles, for flow visualization or Lagrangian physics modeling.

User-Defined Functions: ERF provides templates to customize initialization and/or impose spatiotemporally varying source terms.

ERF does not have the capability for global simulation