I am running a wrf-les ideal simulation to generate some convections with stratus already existing. I modify the source code "/dyn_em/module_initialize_ideal.F" as attached file to initialize stratus. When it comes to about 45 mins, the convections begin to generate. However, the result is completely different when changing the grid number in namelist.input. When I changed the gird number from "e_we = 400, e_sn = 400" to "e_we = 401, e_sn = 401", there is no convections during the whole simulation time (2 hours). And the convections regenerated again when "e_we = 402, e_sn = 402" and "e_we = 403, e_sn = 403". Did anyone meet the similar question? What are the reasons?
Grid number destory the WRF-LES ideal case of stratus with embedded convection
It's strange in deed, and thank you for your advice. By the way, I would appreciate it if you could tell me where I can find the detailed differences between each parameters in namelist.input except the README.namelist.
You can find a lot of the definitions in the Namelist chapter of the Users Guide. You can also look at the Registry/Registry.EM_COMMON file, which gives good explanations for most variables. The links from this Best Practices page also gives some additional details about some of the variables - specifically good practices for setting up a domain (seen in the WPS namelist link)
Thanks a lot. I have found somthing NEW. Actually the convections were initialized by some "non-physical" waves. When e_we is not equal to 401, there will be some perturbation generated from the x boundary, which will perturb the whole domain and initialize the convections. However, there are no same waves in y direction, which makes the convections perturbed from the wave are 2d and uniform in the y direction. The speed of the wave is about 1km/s, therefore it is a numerical error probably. Do you have some idea about this numerical boundary effects and the asymmetry about x and y direction?
Many apologies for the long delay. If you're still experiencing this issue, do you mind attaching the namelist.input file you use for these cases (I assume they are identical except for the e_we and e_wn settings)? Will you also let me know which specific variables you're looking at? Thanks.
Here is one of the namelist.input. I am sure they are identical except for the e_we and e_sn, but i can not find other parallel namelist.input now. I have looked the max_pressure(like mdBZ), mdBZ and max_qcloud.
Attachments
Thanks for sharing that. I'm not able to find variables "max_pressure," "max_qcloud," or "mdBZ" in the WRF code. Are these variables you added to the code? In your initial post, you mentioned that you modified the code. Can you verify whether you still see these issues if you use WRF code that has not been modified?
Sorry to confuse you. I did not add these variables to the code. The variables I mentioned above (max_qcloud,max_pressure,mdBZ) are which I have ploted to research the strange noises generated at the x boundary, as you can see in 1.png above. Besides, I have run an em_quarter_ss ideal case using the original WRF code but still have this problem as long as the wind is zero to avoid the effects of wind, which will mix the pertubation soonly.
Thanks for your reply. I used mp_physics = 28, which can output QCLOUD directly. As for mdBZ and pressure, I simply used the function getvar(wrfncfile, "mdbz", time_idx = idx) (idx = 10 for example) in wrfpython to calculate mdBZ/pressure. Then I used numpy such as np.max(pressure, axis = 0) to get the max_pressure and same for max_qcloud.