Running WRF with high resolution -- 300m

[seti]

Hello everyone,

I am planning to run the WRF model over a mountainous region with a very high horizontal resolution of approximately 300 meters. Before starting the simulations, I would like to better understand the important considerations required to ensure model stability and avoid numerical problems.

In particular, I would appreciate guidance on the following points:

• How should the physical and dynamical stability of the model be handled at such high resolution?
• Which physics schemes are recommended for a 300 m setup in complex terrain?
• Should cumulus parameterization be turned off completely at this resolution?
• Which PBL, surface layer, and microphysics schemes are more suitable for convection-permitting or near-LES simulations?
• What additional options should be activated to improve numerical stability (e.g., damping, diffusion, adaptive time step, turbulence options)?
• Are there specific options that should be disabled to avoid instability or unrealistic behavior?
• How important is terrain smoothing in mountainous regions at this resolution?
• Are there any recommended settings for vertical resolution, time step, or nesting strategy?

I would also appreciate any advice based on personal experience running WRF at sub-kilometer resolution over complex terrain.

Thank you in advance for your help.

 

[haiqingsong]

Hello everyone,

I am planning to run the WRF model over a mountainous region with a very high horizontal resolution of approximately 300 meters. Before starting the simulations, I would like to better understand the important considerations required to ensure model stability and avoid numerical problems.

In particular, I would appreciate guidance on the following points:

• How should the physical and dynamical stability of the model be handled at such high resolution?
• Which physics schemes are recommended for a 300 m setup in complex terrain?
• Should cumulus parameterization be turned off completely at this resolution?
• Which PBL, surface layer, and microphysics schemes are more suitable for convection-permitting or near-LES simulations?
• What additional options should be activated to improve numerical stability (e.g., damping, diffusion, adaptive time step, turbulence options)?
• Are there specific options that should be disabled to avoid instability or unrealistic behavior?
• How important is terrain smoothing in mountainous regions at this resolution?
• Are there any recommended settings for vertical resolution, time step, or nesting strategy?

I would also appreciate any advice based on personal experience running WRF at sub-kilometer resolution over complex terrain.

Thank you in advance for your help.

set epss = 0.9

 

[kwerner]

I'll provide responses to your questions below. Note that I combined some questions that warranted the same response.

How should the physical and dynamical stability of the model be handled at such high resolution?
What additional options should be activated to improve numerical stability (e.g., damping, diffusion, adaptive time step, turbulence options)?
Are there specific options that should be disabled to avoid instability or unrealistic behavior?
How important is terrain smoothing in mountainous regions at this resolution?
Are there any recommended settings for vertical resolution, time step, or nesting strategy?

Make sure you set time_step to no larger than 6xDX, and if you run into CFL errors, you may need to make some adjustments. See
Segmentation Faults and CFL Errors.

Which physics schemes are recommended for a 300 m setup in complex terrain?
Which PBL, surface layer, and microphysics schemes are more suitable for convection-permitting or near-LES simulations?

This is hard to say without knowing the full picture. If you're running over a tropical region, we recommend starting with the "tropical" physics suite, but if over a mid-latitude land mass, we recommend starting with the "CONUS" physics suite.

Should cumulus parameterization be turned off completely at this resolution?

Yes.



You may also need to consider some best practices for domain setup. For e.g., depending on the resolution of your input data, you may need to include a parent domain (or more than one) surrounding the 900 m resolution domain. See Best Practices for guidance - in particular, see the best practices for namelist.wps, which specifies rules for domain setup.