Running wrf.exe using parallel computing, which quickly terminates with the message "forrtl: severe (174): SIGSEGV, segmentatio fault occurred."

[范小勤]

I have tried the following methods without success:
1. grep -rn "cfl" rsl.*
2. w_damping = 1
3. ulimit -s unlimited
The meteorological data should be fine, as I have successfully run a case previously using this dataset.
This time, I increased the resolution and added sst_update = 1, but it failed to run successfully. I seek guidance from experts, and I have attached my namelist.input, and rsl.* files for reference.
The content of wrf.sh is as follows
:#!/bin/bash#SBATCH -p v6_384
#SBATCH -N 1
#SBATCH -n 30
ulimit -s unlimited
module purgemodule load zlib/1.2.7
module load pnetcdf/1.12.1-ips17
module load hdf5/1.10.5-ips17
module load mpi/intel/17.0.7-thc
module load netcdf/4.7.1-ips17-fortran4.5.2
export PATH=/public5/home/******/new/WRF/main:$PATH
mpirun -np 2 /public5/home/******/new/WRF/main/wrf.exe

 

[kwerner]

Hi,
There are potentially two problems I see right away.
1) You're probably not using enough processors for the size of the domain you're using. See Choosing an Appropriate Number of Processors.
2) The resolution of your domain should not be any larger than a 5:1 ratio between it and the input data you're using. What is the resolution of your input data? If it's larger than that, you'll need to include a parent domain with a nested 5km domain inside to get around this problem. See best practices for namelist.wps for suggestions on setting up a reasonable domain.

 

[范小勤]

Thank you very much for your suggestion！A workaround has been found, and the reason for the error is that there was a problem with the initial field data during the interpolation process. I will pay attention to your suggestions in the future, thank you very much!