Strange perturbation development in idealized simulation with open lateral boundary (ARW 3.9.1.1)

[Taka Iguchi]

I'm working on idealized simulation using WRF-ARW version 3.9.1.1, actually NASA-Unified WRF based on that version. My final goal is simulation of pyro-cumulus and tracking tracers emitted from wildfire. Now I'm doing sanity check simulations with an idealized configuration without giving any forcing for convection. The configuration is similar to test/em_les, but I would like to use an open lateral boundary condition rather than a periodic condition to avoid a loop of tracer variables. When using an open lateral boundary under an input sounding, I found the development of noisy patterns in dynamics and thermal variable fields. The perturbation is developed to several m/s in the z-wind component, which is not neglectable. This is not developed when using periodic lateral boundary condition. I have tested several input soundings. This perturbation is developed when the lower atmosphere is nearly unstable around PBL. If there is an inversion layer, this is not developed.

Please let me know if there are any suggestions and comments about namelist options or methods to prevent this type of perturbation in an idealized simulation when using open lateral boundary condition. Thank you so much for your help.

Taka Iguchi
iguchi@umd.edu

 

[kwerner]

Hi,
Apologies for the delay in response and thank you for your patience. I have relayed your question to one of our dynamics experts. They are questioning why you don't want to use the periodic condition, which is generally going to be better constrained for LES domains, especially if there is a mean wind. Alternatively, if you don't want downstream effects to come upstream, nest the LES domain in a large outer domain which is also like an open condition in some ways.

 

[Taka Iguchi]

Thank you so much for your comments and suggestion. As I mentioned above, my final goal is simulations of pyro-cumulus and tracking tracers. This is similar to test/em_quarter_ss but with higher spatial resolution. I want to avoid having a loop of advection of pyro-convection and tracer variables through periodic lateral boundaries in the simulation. An open boundary condition can be applied only to the tracer variables by hard coding, while it cannot be to thermo-dynamics field variables. If there are no suitable methods to suppress this downstream effect, nesting is another good option. I would like to discuss your suggestion with project members and maybe have another reply post again.