Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

Using both epssm and w_damping to stop segmentation faults induced by CFL errors


Hello! As a newbie in the WRF Model, I'm still wondering about the various strategies to stop segmentation faults induced by CFL errors. I am currently experimenting with various cumulus parameterization (cu_physics) schemes, making multiple simulations of the same single-domain real case of my choice (a shear line that affected the Philippines in late December 2022) but with different cu_physics, while keeping all other physics schemes the same (in my case, I'm applying the NCAR Tropical Suite). While doing the WRF simulations, the "wrf.exe" executable terminated multiple times due to some cu_physics schemes leading to segmentation faults induced by CFL errors.

I'm currently attempting to fix the CFL errors in one of the simulations. For a simulation that has cu_physics = 3 (Grell-Freitas), I first tried setting epssm = 0.2 since I thought maybe the complex terrain on some Philippine islands led to the CFL errors, but it did not solve the problem. Next, I increased the epssm to 0.5 since increasing its value may help in preventing model instability and the maximum recommended value for epssm is 0.5 (beyond which inaccuracies are said to ensue in the simulation), but it still did not work.

So, in a rather desperate move, I decided to run the simulation with epssm = 0.5 as in the previous attempt but, this time, I also set w_damping = 1. Thus, I had combined two of the recommended CFL error handling strategies in the same simulation. And fortunately, no CFL error occurred in the run, which was completed successfully.

The question now is that, would it be okay if I used both epssm and w_damping to stop segmentation faults induced by CFL errors? Are these settings supposed to be set one by one, or is it okay to apply them in combination without incurring inaccuracies in the simulation? Was the combination of epssm = 0.5 and w_damping = 1 acceptable in my case?

I'm hoping to read on your comments and recommendations about this matter. Thank you very much, and have a nice day!
W_damping can be combined with larger value of epssm to overcome the CFL violation issue.
w_damping suppresses overly strong vertical motions to keep the model numerically stable
epssm is a coefficient to control slope-generated instabilities due to vertically propagating sound waves.