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

I'm using WRF to simulate 8 days of data in winter, but the simulation stopped automatically without completion.

S

shanmei

New member
I'm using WRF to simulate 8 days of data in winter, but the simulation stopped automatically without completion. After troubleshooting: 1) There are no error messages in the rsl.error and rsl.out files; 2) The same settings, except for the time, successfully simulated 5 days and 6 hours of data in summer; 3) The time step in the namelist.input file was set to either 27 or 18, and the simulation would stop automatically after producing data for one or two days. For example, when simulating data from 12:00 on January 10th to 18:00 on January 18th, it stopped at 16:00 on January 12th. When simulating data from 12:00 on January 11th to 18:00 on January 14th, it stopped at 15:00 on January 13th. 4) Running with either "./wrf.exe" or "mpirun -np 4 ./wrf.exe" would result in an automatic stop halfway through. 5) Changing the nesting from three layers to one layer allowed the simulation to run successfully.
 

Attachments

  • namelist(复件).input
    4 KB · Views: 5
im not an expert or anything but
try this in rsl file
grep -i FATAL rsl.*
grep -i error rsl.*
grep -i SIGSEGV rsl.*
grep -i cfl rsl.*

and change
w_damping = from 0 to 1
it might be a hardware issue
 
dia said:
im not an expert or anything but
try this in rsl file
grep -i FATAL rsl.*
grep -i error rsl.*
grep -i SIGSEGV rsl.*
grep -i cfl rsl.*

and change
w_damping = from 0 to 1
it might be a hardware issue
Click to expand...
Dear dia:
Thank you so much for your previous guidance and suggestions! I truly appreciate your time and expertise in helping me troubleshoot the WRF runtime error.

I implemented the changes you recommended (enabling w_damping = 1 and checking the rsl files for FATAL/CFL/SIGSEGV errors). Unfortunately, the simulation still terminated unexpectedly midway. To assist further diagnosis, I’ve attached:
  • Relevant rsl.error/rsl.out logs
  • My updated namelist.input file
If there’s any additional information I can provide, please let me know. I’d be incredibly grateful for any further insights you might have.
 

Attachments

  • files.zip
    588.8 KB · Views: 4
shanmei said:
Dear dia:
Thank you so much for your previous guidance and suggestions! I truly appreciate your time and expertise in helping me troubleshoot the WRF runtime error.

I implemented the changes you recommended (enabling w_damping = 1 and checking the rsl files for FATAL/CFL/SIGSEGV errors). Unfortunately, the simulation still terminated unexpectedly midway. To assist further diagnosis, I’ve attached:
  • Relevant rsl.error/rsl.out logs
  • My updated namelist.input file
If there’s any additional information I can provide, please let me know. I’d be incredibly grateful for any further insights you might have.
Click to expand...
can you include the rsl.out files as well

Domain looks good

1746881833743.png
 
William.Hatheway said:
can you include the rsl.out files as well

Domain looks good

View attachment 18022
Click to expand...

William.Hatheway said:
and the latest namelist.input file
Click to expand...
Thank you for your prompt response and guidance! I’ve now attached the following additional files to assist with troubleshooting:

All requested rsl.out.0000 files (from the latest run)

The most recent version of my namelist.input

Please let me know if further details or adjustments are needed. I deeply appreciate your continued support in resolving this issue.

Best regards,
Mei
3e6312b06850ae030942be57b18bb0ba.png
 

Attachments

  • files.zip
    588.8 KB · Views: 5
have you try run using only 2 domains? if there's also got successfully, maybe the trouble come from domain 3 setting, and maybe it from the setting that the system not detect it as an error but still give force close/killed for the program, maybe try check every met_em d03 data
 
Ioga Lazuardi said:
have you try run using only 2 domains? if there's also got successfully, maybe the trouble come from domain 3 setting, and maybe it from the setting that the system not detect it as an error but still give force close/killed for the program, maybe try check every met_em d03 data
Click to expand...
Dear Ioga Lazuardi,

Thank you very much for your suggestions! I have started running the simulation with only 2 domains as advised. If this configuration succeeds, I will thoroughly review the Domain 3 settings and verify all met_em.d03 input files for potential inconsistencies or formatting issues.

I will update this thread with the results once the 2-domain run completes. If you have further recommendations for debugging Domain 3 (e.g., specific parameters to check or tools to validate the meteorological data), please let me know. Your expertise is greatly appreciated!

Best regards,
Mei
 
shanmei said:
Thank you for your prompt response and guidance! I’ve now attached the following additional files to assist with troubleshooting:

All requested rsl.out.0000 files (from the latest run)

The most recent version of my namelist.input

Please let me know if further details or adjustments are needed. I deeply appreciate your continued support in resolving this issue.

Best regards,
Mei
View attachment 18027
Click to expand...
I think it has to do with the urban parameter turned on for domain 3.

@kwerner or @Ming Chen correct me if I am wrong but for urban canopy model, since it is a "Max Domain" parameter it would need to be changed in the namelist from

Code: 
 sf_urban_physics                    = 0,     0,     3,

to

Code: 
 sf_urban_physics                    = 3,     3,     3,


1747150398229.png
 
You must log in or register to reply here.
Top