Program received signal SIGSEGV + chem-opt 202

[Setareh Rahimi]

Dear all,

I set 5 WRF-Chem experiments using chem-opt 202. (WRF-chem version 4.5.2)

However, once I ran them on an HPC, 3 of them stopped after running for one by showing this error message: 

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:

#0  0x3503635a4f in ???

#1  0x350367eae6 in ???

#2  0x350368053b in ???

#3  0x307d0a5 in ???

#4  0x3090388 in ???

#5  0x309c2ca in ???

#6  0x309c3b7 in ???

#7  0x309c5a3 in ???

#8  0x228555b in ???

#9  0x185def0 in ???

#10  0x16f1c5c in ???

#11  0x42d8ab in ???

#12  0x42dbff in ???

#13  0x406a61 in ???

#14  0x40604c in ???

#15  0x3503621b44 in ???

#16  0x406084 in ???

#17  0xffffffffffffffff in ???

So I wonder why this happens for these 3 jobs and not the other 2 jobs. I set up 1 node and 8 cores to run the jobs.

A few days ago I ran them (the same name lists that I currently use) on the HPC and the simulation lasted for 4 days, and some of them stopped with no error message.

Would you please kindly advise me on how to fix this issue? (name list and rsl.error.0000 have been attached)

Kind regards.

[Gabriele Pfister]

Hi Setareh

I took a quick look at your namelist and there are a number of inconsistencies. E.g.

- when running with any of the MOZART gas phase chemistry options you will not need the gocart background fields (auxinput8)

- increase debug_level to a higher number to achieve more information on possible model errors in your log file

- switch to phot_opt=4 (updated TUV)

I suggest carefully assessing your namelist.input and consulting the MOZCART-MOSAIC and TUV User guides on our website (https://www2.acom.ucar.edu/wrf-chem). 

Gabi

--

================================
Gabriele Pfister

Atmospheric Chemistry Observations & Modeling (ACOM) Lab

NSF National Center for Atmospheric Research (NCAR)

Email: pfi...@ucar.edu

Temporary Home Office Phone: +1 303 241 1329

Work Phone: +1 303 497 2915

Web: https://staff.ucar.edu/users/pfister