seg fault running OBSGRID with hourly data from MADIS

[mkharkey]

I have hourly data from MADIS, converted to little_r format, and I'm running obsgrid v3.8 (CHANGES file says 3.8 even though the log text says 3.7) for eventual use with WRF v3.8.

Obsgrid compiled fine with intel19, and I get a runtime seg. fault with the attached namelist. I assume there isn't a problem with the executable (compile log shows nothing out of place) and that the input data are correctly formatted. The obs. data were linked in to my input_data directory as obs_METAR:2017-MM-DD_HH (with approprite time/date info.). The met_em* files were linked in to the OBSGRID directory where I'm running the executable, and were generated with 6h GFS data. I'm not sure if I've read the user guide correctly for setting up the namelist, and would appreciate any suggested changes and/or things to test. Thank you!

-Monica


[...]
3d analysis dates to be processed by this program (excluding SFC FDDA times):
Time period #00001 is for date 2017-05-12_00:00:00
Time period #00002 is for date 2017-05-12_06:00:00
Time period #00003 is for date 2017-05-12_12:00:00
Time period #00004 is for date 2017-05-12_18:00:00
Time period #00005 is for date 2017-05-13_00:00:00
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
obsgrid.exe 00000000004A0E03 for__signal_handl Unknown Unknown
libpthread-2.28.s 00007F54E958EB20 Unknown Unknown Unknown
obsgrid.exe 000000000040C1C9 Unknown Unknown Unknown
obsgrid.exe 000000000041D6F5 Unknown Unknown Unknown
obsgrid.exe 000000000040B822 Unknown Unknown Unknown
libc-2.28.so 00007F54E91DC7B3 __libc_start_main Unknown Unknown
obsgrid.exe 000000000040B72E Unknown Unknown Unknown
srun: error: r815-307-33: task 0: Exited with exit code 174
srun: Terminating job step 15849029.0

 

[kwerner]

Hi Monica,
I first want to make sure that you built OBSGRID serially and that you are only running with a single processor. If that is all true, then because you did make a lot of modifications to your namelist.oa file, I'd like you to run a test with the default namelist file just to see if the problem is related to any of your settings. I'm attaching a namelist that I *think* is set up for your case. Give it a try. If you get an error related to a specific field in the namelist, you may need to tweak it a bit to work, but try to not change too many options. If you find that the default namelist works, you may need to run several tests to track down the option causing the issue.

 

[mkharkey]

I assume I've built it serially (attached compile log). There is an error compiling plot_soundings but I thought that was safe to ignore.

Thank you for the default namelist--when I tried it, I got the same error. I put some print statements in main.F90 and driver.F90 and recompiled, it seems like everything being passed in to the driver subroutine is okay (?) but it fails right at the start of the driver subroutine; it won't get to the print statement I put in after all the declarations.

 

[kwerner]

Thanks for trying that out. Do you mind sending a couple time frames of your obs* data and your met_em* files? Since it stops at the first time step, just the initial time should be okay. I want to try to run this on my end. If the files are too large to attach, see the home page of this forum for details on sharing large files. Thanks!