WRF failing at first time step

[Xochitl H.]

Hi,

I have been attempting to run WRF, but repeatedly experience crashes at the first time step. Real.exe has run successfully and my input data seems to be fine, but I receive a segmentation fault error (forrtl: error (78): process killed (SIGTERM)) when I attempt to run wrf. While I know my parent_grid_ratio is not the ideal, I have had previous simulations with the same domains run successfully. The only change I have made since that simulation was in WPS (now using nlcd2011_9s+modis30s instead of the default). I've read the post about segmentation faults and have tried implementing/checking the suggestions. Mr rsl files do not show a cfl error and I am using the same amount of processors as I did for my previous simulation that ran successfully (select=4:ncpus=128:mpiprocs=96). Attached are my namelist, one of my rsl.error files.

Any suggestions about how to proceed would be greatly appreciated. Thanks!

 

[William.Hatheway]

Hi,

I have been attempting to run WRF, but repeatedly experience crashes at the first time step. Real.exe has run successfully and my input data seems to be fine, but I receive a segmentation fault error (forrtl: error (78): process killed (SIGTERM)) when I attempt to run wrf. While I know my parent_grid_ratio is not the ideal, I have had previous simulations with the same domains run successfully. The only change I have made since that simulation was in WPS (now using nlcd2011_9s+modis30s instead of the default). I've read the post about segmentation faults and have tried implementing/checking the suggestions. Mr rsl files do not show a cfl error and I am using the same amount of processors as I did for my previous simulation that ran successfully (select=4:ncpus=128:mpiprocs=96). Attached are my namelist, one of my rsl.error files.

Any suggestions about how to proceed would be greatly appreciated. Thanks!

is there anyway to go to a 1:5 ratio?

can you also share you're namelist.wps

 

[Ming Chen]

Can you specify how you prepare the time-varying 2D input data of aerosol?

It is fine to do 1:5 nesting ratio. Your time step of 45 is way too small for dx=27km. Please increase it to 150.

 

[Xochitl H.]

Hi,

I can try the 1:5 ratio, but would need to add a third domain to achieve the resolution I want. The same domain configuration and time step ran successfully in the past, but does not work when I try to use nlcd2011_9s+default, which does not make sense to me. As for the time-varying 2D input data for aerosols, I have never had to explicitly prepare it in past runs, so I am unsure of what is happening with that field. Attached is my namelist.wps and rsl.error.0037 (the rank where the error originates, with the debug_level set to 100).

Thank you!

 

[William.Hatheway]

Hi,

I can try the 1:5 ratio, but would need to add a third domain to achieve the resolution I want. The same domain configuration and time step ran successfully in the past, but does not work when I try to use nlcd2011_9s+default, which does not make sense to me. As for the time-varying 2D input data for aerosols, I have never had to explicitly prepare it in past runs, so I am unsure of what is happening with that field. Attached is my namelist.wps and rsl.error.0037 (the rank where the error originates, with the debug_level set to 100).

Thank you!

WRF Domain Wizard

HTML implementation of GUI for the WRF Preprocessor System (WPS) and namelist.input

jiririchter.github.io

That might help you visualize it easier

 

[Ming Chen]

The nesting ratio of 1:15 is not a reasonable setting. Please change it to 1:5 or 1:3, even if this means that you need to add one more child domain.

Also, when you run wrf.exe, please stay with debug_level =0 because higher number of this value doesn't provide more helpful information.

For debugging, please recompile WRF in debug mode, i.e., ./clean -a and ./configure -D. In this case, wrf.exe output log files will tell exactly when and where the model crashes first. This information gives you hints what is wrong.