Running error with WRFV4.3 & WUDAPT LCZ

[Z_wendy]

Hi, all.

I am running WRFV4.3 with WUDAPT LCZ land use data (41 LU_INDEX). When I ran WRF with sf_urban_physics=2, there was a bad termination message.

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

Backtrace for this error:
#0 0x2B9C9A8046D7
#1 0x2B9C9A804D1E
#2 0x2B9C9C5F53FF
#3 0x13DD7B8 in __module_sf_bep_MOD_long_rad

I have tried many ways (e.g., adjust the time step, check the input data, and change the physics schemes and processor numbers) but failed to solve it. I also wonder how to read the above error.

Appreciated with any helps!

Thank you,
Wendy

 

[Z_wendy]

Update:

Now I have run the model successfully. There might be two useful measures to solve it:
1. set "feedback=0" in namelist.input.
2. use High performance computing (HPC) to run the model in case of memory problems.

Thanks for any other ideas.

 

[fipoucat]

I have a similar error using version 4.4.1 using wudapt-lcz with surface physics 3 . I saw you post and trying to run again changing feedback to 0but as always have the me error at time step 6 hours.

d01 2022-05-30_06:00:00 module_io.F: in wrf_write_field

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

Backtrace for this error:
#0 0x14e1f3e23ad0 in ???
#1 0x14e1f3e22c35 in ???
#2 0x14e1f324251f in ???
Wonder if you have a hint or share some settings of your namelist?

 

[kwerner]

@fipoucat
Your error may not be related to the same issue. Can you attach your namelist.input file, as well as your rsl.error* files (package the rsl files together in a single *.TAR file)? Thanks!

 

[Z_wendy]

@fipoucat
Of course, I can share the namelist.input file here. Hope it helps.
Actually, I also tried many other ways before, including adjusting the time step and physical schemes, but they didn't work. I guess this problem may be mainly related to the machine performance.

 

[fipoucat]

attached the namelist I am using. THe WRF version is 4.4.1,there is a problem using version 4.4.2 with wudapt-lcz. tar file from rsl.error too big (74Mb).

 

[fipoucat]

just did not like writing rst files for an unknown reason

 

[kwerner]

@fipoucat
One issue may be that you're probably not using enough processors for the size of your domain (specifically d02). Take a look at this FAQ that discusses choosing an appropriate number of processors, based on the size of your domain.

I also notice you are running WRF-Chem. If increasing the number of processors doesn't help, I would recommend posting your question in the WRF-Chem support section of this forum so their group can assist you.

 

[fipoucat]

my understanding, as stated previously, the model for an unknown reason issues the error message when writing wrfrst file. I increased the number and ovoid them. The model passed this step.

 

[kwerner]

Great, I'm glad to hear that!