Thank you for uploading namelist.input.
Your case is a high-resolution run with dx = dy =1km. I have a few concerns regarding your namelist options:
(1) At the resolution of 1km, it is inappropriate to turn on cumulus scheme, but in your case you set cu_physics = 93
(2) lightning_option 11 is recommended to be activated when dx> 10km and dx < 50km. For your case with dx = 1km, 11 is not a good option.
Also, note that neutral buoyancy from convective parameterization are affected by many factors. From WRFV3.8 to WRFV4.5, various updates and bug fixes may affect neutral buoyancy, leading to different model behavior.