Failed to run real.exe - Error "pal, linear is about as low as we go"

[davidl2000]

I'm trying to run WRF with RRFS-A input data processed through WPS using the Vtable.raphrrr variable table (besides a line of text saying "unknown model from NCEP 134" while doing ungrib.exe there was no indication that it did not work) and when I check the variables with "ncdump -h met_em.d01.2024-04-01_22_00_00.nc | grep -i num_metgrid" the output is almost the exact the same as the past HRRR data I used (except for num_metgrid_levels being 46 for RRFS instead of 41 like HRRR) (First image below is HRRR metgrid data, second image is RRFS-A metgrid data)
When I run real.exe the error "pal, linear is about as low as we go" is what comes up, and no changes I do seem to affect this. The only reason I can think this wouldn't work is because the Vtable.raphrrr is meant for RAP/HRRR as I was testing RRFS since its variables are very similar to RAP/HRRR's so that would mean I'd need a new Variable Table and I can't even begin to fathom how to make it and would need to seek someone else to make it or share an existing one I am not aware of.

I've included my namelist.wps, namelist.input and Vtable. For my input I used the RRFS prslev CONUS files (rrfs.t00z.prslev.f022.conus_3km.grib2).

 

[kwerner]

Hi,
Apologies for the long delay in response. Can you share a couple of your met_em* files so I can take a look and test this out? Please also attach your WRF error log that shows the error message (e.g., rsl.error*). The met_em* files will likely be too large to attach. If that's the case, see the home page of this forum for information on sharing large files. Thanks.

 

[davidl2000]

Hi,
Apologies for the long delay in response. Can you share a couple of your met_em* files so I can take a look and test this out? Please also attach your WRF error log that shows the error message (e.g., rsl.error*). The met_em* files will likely be too large to attach. If that's the case, see the home page of this forum for information on sharing large files. Thanks.

Sure thing, I shared the met_em files in a zip file on Nextcloud, called "RRFS_met_em_files.zip". I've also included the rsl.error file."

EDIT: I got this error using Vtable.HRRR.bkb on WPS and that's what was used to ungrib the data, it yielded more files in metgrid but resulted in the same error as shown in the rsl.error.0000 file attached.

 

[kwerner]

Thanks for sharing those files. I am able to repeat the error you're getting, but I'm still not sure why it's happening. I'm reaching out to a colleague to see if they are able to help. I'll keep you posted. Thank you for your patience.

 

[davidl2000]

Thanks for sharing those files. I am able to repeat the error you're getting, but I'm still not sure why it's happening. I'm reaching out to a colleague to see if they are able to help. I'll keep you posted. Thank you for your patience.

I definitely wish you the best of luck on that, hopefully it can be figured out because I would love to use RRFS-A for WPS input.

 

[kwerner]

Thank you so much for your patience. We have found what we think to be a solution. Somehow the variable 'linear_interp' becomes undefined in dyn_em/module_initialize_real.F, which we don't yet understand, but I'm attaching a modified file. Place this in your dyn_em directory and then you'll need to recompile the code. Since you aren't making any registry changes, you DO NOT need to clean the code, or to reconfigure. Just simply recompile, using the same compiler/libraries you used when you originally compiled, and it should be much faster than a full compile. This code is for WRFV4.5.

real.exe should work for your data with these changes. Let me know!

 

[davidl2000]

Thank you so much for your patience. We have found what we think to be a solution. Somehow the variable 'linear_interp' becomes undefined in dyn_em/module_initialize_real.F, which we don't yet understand, but I'm attaching a modified file. Place this in your dyn_em directory and then you'll need to recompile the code. Since you aren't making any registry changes, you DO NOT need to clean the code, or to reconfigure. Just simply recompile, using the same compiler/libraries you used when you originally compiled, and it should be much faster than a full compile. This code is for WRFV4.5.

real.exe should work for your data with these changes. Let me know!

So I replaced the module in dyn_em and now compilation does not complete. I'm not too sure why so here's the full log of what the terminal shows when I attempt to compile.

 

[davidl2000]

So I replaced the module in dyn_em and now compilation does not complete. I'm not too sure why so here's the full log of what the terminal shows when I attempt to compile.

Scratch that, I did a "./clean -a" then reconfigured and recompiled and it successfully recompiled em_real. Now will attempt to run with RRFS-A data.

 

[davidl2000]

Thank you so much for your patience. We have found what we think to be a solution. Somehow the variable 'linear_interp' becomes undefined in dyn_em/module_initialize_real.F, which we don't yet understand, but I'm attaching a modified file. Place this in your dyn_em directory and then you'll need to recompile the code. Since you aren't making any registry changes, you DO NOT need to clean the code, or to reconfigure. Just simply recompile, using the same compiler/libraries you used when you originally compiled, and it should be much faster than a full compile. This code is for WRFV4.5.

real.exe should work for your data with these changes. Let me know!

Okay so WPS and such proceeded as normal and real.exe worked just fine and wrf.exe started normally but then I got these cfl errors then shortly after, segmentation fault. Here's a text file containing the full log of the terminal (essentially what would be the rsl.out.0000).

 

[kwerner]

I'm so glad to hear that you were able to get past your initial issue - that's great news.

Since the new problem is different than the original one posted here, will you please post it as a new thread? We like to keep threads specific to a single issue. Thanks!

 

[davidl2000]

I'm so glad to hear that you were able to get past your initial issue - that's great news.

Since the new problem is different than the original one posted here, will you please post it as a new thread? We like to keep threads specific to a single issue. Thanks!

I figured out the CFL errors were due to too high time steps, I fixed it and now I have RRFS-A data input and running (also to confirm a friend of mine also did this exact same stuff (./clean -a then reconfigured then recompiled em_real) and it also worked for them so this is definitely a fix. Thanks very much!
(below is an image of the dbz output using the RRFS-A data)