WRF nested run keeps crashing

[edougherty]

Hi there, I am running WRF on Derecho with two domains with the feedback turned on. Everything works fine until wrf.exe, where I receive this error message in rsl.out:

forrtl: error (78): process killed (SIGTERM)

My namelist is here:

/glade/derecho/scratch/doughert/WRF_SA/nov29/wrf_run/namelist.input

And my run script is here:

/glade/derecho/scratch/doughert/WRF_SA/nov29/wrf_run/runwrf.csh

Do you know why this happens? Is this a segmentation fault that could be resolved by using more nodes?

Thanks!

 

[kwerner]

Hi,
When I search for 'cfl' in your rsl* files, I find several CFL errors. This means that the model has become unstable. It looks like you've probably already tried to address the issue with some of the solutions mentioned in "What is the most common reason for a segmentation fault?" If you're running over really complex terrain, you may need to increase epssm even more, and if the CFL errors occur near the boundary, you may need to set "smooth_cg_topo" to true, prior to running real.exe.