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

Running domain 3 and 4

jmcabrera

New member
Hi đź––!

Currently I'm simulating for domains 3 and 4 correctly. However, I would like to reduce the runtime, especially for domain 4. I'm using the recommended number of processors, which is 16.


I have shared the namelist.input and namelist.wps files, so if anyone could give me some advice, I would appreciate it.

Thank you!
 

Attachments

  • namelist.input.txt.input
    4.7 KB · Views: 4
  • namelist.wps.txt.wps
    891 bytes · Views: 2
Why are you using adaptive time stepping? Did you have CFL errors without it?

Additionally, there are numerous namelist settings that need to be, or should be modified:
1) radt should have the same value for all 4 domains
2) mp_physics should be set to the same physics option for all 4 domains
3) it's possible 60 levels may be a bit too much. What happens if you change e_vert = 45 for each domain?
4) you have interval_seconds = 900 - so you have input data that's available every 15 mins?
 
Last edited:
Thanks Kwerner,

I try the changes you suggested. I chose that option because I saw it on the internet; however, the information was a bit confusing.
If you don't mind, could you share a namelist.input file for a similar situation so I can compare it?
 
Hi,
I don't know if it's normal, but simulating approximately 24 seconds is taking almost 30 seconds when using 25 recommended processors.
 
I'm attaching a namelist.input file that includes your domain set-up, your dates/times, etc., but it's set up to a more basic namelist - without all the extra adaptive time-stepping, etc. included. I didn't set the physics options to the ones you have in your namelist. I just have it set to the CONUS suite, with all the default settings for that. If you change the physics, make sure you have the same setting for all domains for each option, with the exception of cu_physics - it's okay to set the high-res domains to 0 for cu_physics. Try this one to see if it works any better. You'll need to re-run real.exe before running wrf. I'm also attaching the default namelist.input file, in case you want to start with that one and modify it with just your settings.
 

Attachments

  • namelist.input
    4 KB · Views: 1
  • namelist.input.default.txt
    3.7 KB · Views: 1
Top