Error running wrf.exe after the usage of tc.exe

[DeebakVijay]

Greetings,
I am using wrf 4.4 , ERA5(0.25degs , hourly ) reanalysis data for simulating a case by removing a cyclone in bay of Bengal region at16 - 10 -1999:00:00:00 UTC.I am attaching the namelist.wps , domain , namelist.input files. After processing tc.exe I renamed the aux_file as met_file and tried running real.exe , which produced the following error :

"
-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 332
---- ERROR: The input file appears to be from a pre-v4 version of WRF initialization routines
-------------------------------------------
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
"

So I copied the only the necessary fields (RH,TT,UU,VV,GHT,PRES,PMSL,PSFC) from aux file to met file with the attached ncl code(WRF_bogusing - Google Drive was able to run real.exe successfully , but wrf.exe is producing an error every time after integration up-to 16 - 10 -1999:00:22:12 UTC.

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

Backtrace for this error:
#0 0x2ac5632483ff in ???
#1 0x2a3b20d in ???
#2 0x2a3f08c in ???
#3 0x2a441f4 in ???
#4 0x21adaeb in ???
#5 0x21cfe97 in ???
#6 0x19fb98e in ???
#7 0x147c72d in ???
#8 0x132681a in ???
#9 0x47d9a1 in ???
#10 0x406403 in ???
#11 0x405dbc in ???
#12 0x2ac563234554 in ???
#13 0x405df3 in ???
#14 0xffffffffffffffff in ???
"

I used with multiple no of cores (24,48,50).I tried 2 way nesting and one way nesting. Every time the error is produced after the same number of integrations. I am attaching the necessary files here"WRF_bogusing - Google Drive", Kindly help me to sort out the problem.

Note: While using the original met file , wrf runs without any error for the same namelist.input . I noticed something contradicting between name and size of dimensions of original met files and aux files. The dimensions of fields in aux files , though named as staggered in a particular direction , it still has the same size as if it was not staggered.

Regards,
Deebak Vijay

 

[kwerner]

Hi Deebak,
Can you send a packaged .tar file with all the rsl* files so I can take a look? Thanks!

 

[DeebakVijay]

Hello Kwerner,
I have attached a .tar file containing all rsl* files. I have used 24 cores.thank you.

 

[kwerner]

Thanks for sending that. Instead of trying to reconfigure your met_em* file, can you first try to run real.exe again, but before running, add

"force_use_old_data = .true." into the &time_control section of the namelist. That will get you past the version issue. After that, however, we are aware of an ongoing issue with the TC bogus program. It will mess up the input soil data (moisture and temperature), causing a seg-fault error (and that may be what was happening to your simulation anyway). To fix this, you can write over the data in your auxinput* file to match the met_em* file for the fields that were corrupted.

Let me know if that is beneficial to your run.

 

[DeebakVijay]

Thanks for sending that. Instead of trying to reconfigure your met_em* file, can you first try to run real.exe again, but before running, add

"force_use_old_data = .true." into the &time_control section of the namelist. That will get you past the version issue. After that, however, we are aware of an ongoing issue with the TC bogus program. It will mess up the input soil data (moisture and temperature), causing a seg-fault error (and that may be what was happening to your simulation anyway). To fix this, you can write over the data in your auxinput* file to match the met_em* file for the fields that were corrupted.

Let me know if that is beneficial to your run.

Hello Kwerner ,

Thanks for the reply. I tried running after renaming the auxinput* file to met_em*file and with "force_use_old_data = .true." . Real.exe ran successfully , while wrf.exe produced segmentation error.

So I tried replacing the soil data in auxinput* file with original met_em* file's fields. The only common soil Variable I could find in auxinput* file and met_em*file is "SOILTEMP" , so I replaced it and tried running . WRF.exe produced segmentation error. Kindly let me sort this out.


"""
The Variables and their description in auxinput* file and met_em* file are in the attached files named as bogused_description.txt and original_description.txt.
The Variables and their full metadata are in auxinput* file and met_em* file are in the attached files named as bogused.txt and original.txt.
"""

KINDLY LET ME KNOW IF I HAVE TO REPLACE ANY OTHER SOIL DATA (OTHER THAN SOILTEMP) FROM MET_EM * FILE TO AUXINPUT * FILE.

Thank you,
Regards,
Deebak Vijay

 

[DeebakVijay]

Hello,
The problem is solved by following steps:

1. After running WPS , "tc.exe " is ran with "force_use_old_data = .true." in namelist.input
2. The first "MET_EM file" is replaced with generated "AUX_file"
3. "real.exe " is ran to produce WRFINPUT file
4. Another "real.exe " is ran without replacing the original "MET_EM file" (2nd step). This produces another WRFINPUT file
5. The " H2O MIXING RATIO , SURFACE TEMPERATURE , SURFACE MOISTURE " from WRFINPUT file of step 4 is used , Rest of the variables are used from WRFINPUT file of step 3.
6. With the modified "WRFINPUT file " , "wrf.exe" is able to run smoothly.

So in my case not just the surface data but also the H20 mixing ratio (3D , prognostic state variable , scalar) is also changed for wrf to run without segmentation error. I wonder what could be the reason? I am attaching the 2D H20 mixing ratio fields of both WRFINPUT files(before and after removing vortex) at the lowest model pressure level. I am also attaching the absolute difference.Kindly let me know what could be going wrong?

I would like to know if there is any documentation for bogusing (tc.exe) in WRF?

Thank you,
Regards,
Deebak Vijay

 

[kwerner]

Hi Deebak,
I'm so glad to hear that you were able to get past the issue and figure out a solution! I'm not sure exactly what the issue is, and I wasn't absolutely sure of the exact variables that caused the issue, but the variables are all related and when interpolated vertically (from met_em* files to wrfinput* files), a "bad" variable can make another variable bad, as well. Unfortunately it's an older program that hasn't been looked at in a while, due to lack of resources. If you were able to get it to work and things look reasonable, I'd say you're good to go! The only documentation we have for TC bogusing is in this section of the WRF Users' Guide. It may be possible to find more resources (e.g., publications, or posts on other websites about it) by searching on the web.

 

[Francesca]

Hello,
The problem is solved by following steps:

1. After running WPS , "tc.exe " is ran with "force_use_old_data = .true." in namelist.input
2. The first "MET_EM file" is replaced with generated "AUX_file"
3. "real.exe " is ran to produce WRFINPUT file
4. Another "real.exe " is ran without replacing the original "MET_EM file" (2nd step). This produces another WRFINPUT file
5. The " H2O MIXING RATIO , SURFACE TEMPERATURE , SURFACE MOISTURE " from WRFINPUT file of step 4 is used , Rest of the variables are used from WRFINPUT file of step 3.
6. With the modified "WRFINPUT file " , "wrf.exe" is able to run smoothly.

So in my case not just the surface data but also the H20 mixing ratio (3D , prognostic state variable , scalar) is also changed for wrf to run without segmentation error. I wonder what could be the reason? I am attaching the 2D H20 mixing ratio fields of both WRFINPUT files(before and after removing vortex) at the lowest model pressure level. I am also attaching the absolute difference.Kindly let me know what could be going wrong?

I would like to know if there is any documentation for bogusing (tc.exe) in WRF?

Thank you,
Regards,
Deebak Vijay

Hello Deebak,

Could you please tell me how to change informations of variables of different .nc files cause I am not familiar with that and I've got the same problem as you posted. I would appreciate it if you can share it with me, thank you so much!

Regards,
Francesca

 

[DeebakVijay]

Hello Francesca,
Editing the values of variables within an .nc can be done with ncl or any suitable packages of python . I generally use xarray or netCDF4 or wrf-python. These are well documented and easy to use.Hope it helps.


Regards,
Deebak Vijay

 

[Francesca]

Hello Francesca,
Editing the values of variables within an .nc can be done with ncl or any suitable packages of python . I generally use xarray or netCDF4 or wrf-python. These are well documented and easy to use.Hope it helps.


Regards,
Deebak Vijay

Hello Deebak,

Thank you for your reply, it helped me a lot. Good day!

Regards,
Francesca