wrf.exe error:forrtl: severe (174): SIGSEGV, segmentation fault occurrederror

[amemori]

Hi, dear administer. My WRF version is 4.4.1. My outermost domain is east China and running time is three months. Amounts of processors for running is 48 (I think it may be enough).
At first, I follow the rule to set time_step=54 (6 x dx ,my dx is 9000). However, CFL error happened after 1 hour running. I reduce time_step to 36 (4 x dx), but, another segmentation fault occured after 15 minutes running. I found the rsl.error.00*, but there were no clear instructions for help. The files and my namelist.input are uploaded.
Another question is that when I use default parameters of &pyhsics of 'CONUS'(all are set to =-1) and turn off urban physics, no segmentation (CFL or any other) fault occurred. However, the same segmentation fault occurred again when set sf_urban_physics =1.

 

[kwerner]

Hi,
Can you package all of your rsl.* files into a single *.tar file and attach that so I can take a look? There are sometimes errors hidden in those files. And yes, for your domain size, 48 processors should typically be enough. Thanks!

 

[amemori]

Hi,
Can you package all of your rsl.* files into a single *.tar file and attach that so I can take a look? There are sometimes errors hidden in those files. And yes, for your domain size, 48 processors should typically be enough. Thanks!

Dear administrator,
Thanks for your reply! The error I mentioned before has been solved by setting sf_urban_physics = 2, and this urban physics scheme is acceptable for my research. I keep this configuration (namelist.input below) for analysis of 2016,2017,2018 and 2019, and all successfully complete.
However, the same question (segmentation fault but not CFL) occurred when I used the FNL dates of 2020 after about 10s running. These errors are contained in 0023, 0029, 0033, 0034, 0035, 0039, 0040, and 0053 respectively, and all rsl* files are upload. I suspect something is wrong with the data of 2020 specially. I try to use default configuration but seems to be noneffective.
Looking forward to your reply.

 

[kwerner]

Hi,
Since this is a new issue, with a new namelist setup, will you post this as a new topic? We want to keep the threads clean so they will be easier to read through for any other users experiencing similar problems in the future. Thanks!