Bogusing of vortex in WRF run

[samudra]

I have been trying to bogus a vortex in WRF, and so I made the necessary changes in the namelist.input file and got an output file for my parent domain as auxinput_d01_... . After that I renamed the file to my met_em filename. I went through the queries regarding bogusing and found out that I need to interpolate all the variables to my inner domain file of the same time stamp as d01. Am I stating that right? And also how can I do that?
I am attaching my namelist.input file here. Thank You.

 

[Ming Chen]

Hi,
A relatively easy way to do so is to set the namelist option

input_from_file = True, False,

By this option the initial data of D02 will be interpolated from that of D01.

Please try and let me know whether it works for your case.

 

[samudra]

Hi,
I made the changes, but it didn't work, the model broke down after running wrf.exe. I have attached the rsl.error file and namelist.input file here

 

[Ming Chen]

Hi,

Your namelist.input looks fine.

This case crashed after ~30 minutes of integration, and the error message in your rsl file shows "Segmentation fault - invalid memory reference".
Such an error message often indicates that something goes wrong in the physics, although we don't know exactly when and where the model crashed first.

Can you compile WRF in debug mode, then rerun this case? By running in debug mode, you can get more information that may be helpful to figure out what is wrong.

 

[samudra]

Hi,
I compiled WRF in debug mode and attaching the rsl.error and namelist.input file here. kindly look into it.
Thank You.

 

[Ming Chen]

Please look at your code

/home/samudra/WRF_MODEL/WRF/phys/module_sf_noahlsm.f90

At line 1513, some variables went wrong, probably divided by zero. Print all the variables in this line and then trace back. Hope this can give you some idea what is wrong.

By the way, if you run this case without TC bogus, can you get it done successfully?

 

[samudra]

Hi,
I looked at the code but I am unable to print the variable from the fortran file, Issue with the modules used there. I tried specifying their paths too but unable to resolve so far. Can you help me regrading this? (I am very new to WRF, and fortran too.)
Also, running the case without TC bogus works fine for me.
Thank you.

 

[Ming Chen]

There are many "print" and "write" lines in WRF codes. Please just follow the same style to print variables. The most straightforward way is:
print*, 'variable'

Note that you need to look at the line 1513 in your module_sf_noahlsm.f90, find all those variables involved, then modify the code module_sf_noahlsm.F by adding "print" or "write", then please recompile WRF by issuing the command ./ocmpile em_real in top WRF directory.

 

[samudra]

Hii,
So after you asked to check if it runs properly without bogusing, I made some runs and it turns out the run isn't successful. I am getting some errors in the modules only. I have tried it for different time steps and also changing my domain size, but the same error comes. I also have modified the code in the files from where the error was coming, based on your suggestion, and recompiled, but still it persist. I am guessing the problem is present for all cases now, irrelevant to my configuration. I am attaching my namelist and rsl.error files from the latest configuration that I adopted. Kindly check.
Thank you in advance.

 

[Ming Chen]

I guess the rsl.error.0000 file you sent is from the run without TC bogus. Please let me know if I am wrong.

This case crashed immediately after wrf.exe started, indicating that something could be wrong with your input data.

What data did you ungrib?

 

[samudra]

Hi,
Yes, the last run is without bogusing of vortex, just to make sure it worked. But the issue has been resolved for the case without bogusing. Turns out changing my micro-physics and cumulus scheme worked for single domain case without TC bogus. But again when I tried to add a bogus vortex. error still comes. I am using ERA5 reanalysis data as my input. I am attaching the error files here.
The idea was to first check if bogusing works for a single domain, then extend it for sub-domains, hence I decided to start with this.

Thanks.