WRF fails with ExitCode 66

[chunchih]

Hi WRF Support,

I am running a real-case WRF simulation (WRF v4.2.1) with four nested domains. My simulation initialized correctly with four wrfinput files, one for each domain, produced. However, when I executed wrf.exe, it failed pretty quickly while still writing the wrfout files at the initial time. Only the initial wrfout files for the first two domains were created. I was able to run the same simulation with just three domains (the initial wrfout files were created properly for all three domains), but as soon as I activate the fourth domain, only the first two initial wrfout files were created and the simulation failed with ExitCode 66.

I have attached the rsl files that were generated during the WRF execution, with the debug_level=100. In one of the rsl files (specifically, rsl.error.0026. FYI:I requested 48 cores to run my simulation), there is an error statement saying:

forrtl: severe (66): output statement overflows record, unit -5, file Internal List-Directed Write
Image PC Routine Line Source
wrf.exe 0000000003071E9E for__io_return Unknown Unknown
wrf.exe 00000000030C547A for_write_int_lis Unknown Unknown
wrf.exe 0000000002C8A676 Unknown Unknown Unknown
wrf.exe 0000000002506980 Unknown Unknown Unknown
wrf.exe 00000000024FE125 Unknown Unknown Unknown
wrf.exe 0000000001EF81D3 Unknown Unknown Unknown
wrf.exe 0000000001A6B871 Unknown Unknown Unknown
wrf.exe 00000000014CA5D1 Unknown Unknown Unknown
wrf.exe 00000000014BD78C Unknown Unknown Unknown
wrf.exe 000000000056D712 Unknown Unknown Unknown
wrf.exe 000000000056DE8E Unknown Unknown Unknown
wrf.exe 000000000056DE8E Unknown Unknown Unknown
wrf.exe 0000000000410F31 Unknown Unknown Unknown
wrf.exe 0000000000410EEF Unknown Unknown Unknown
wrf.exe 0000000000410E8E Unknown Unknown Unknown
libc-2.24.so 00002B8ADE0722E0 __libc_start_main Unknown Unknown
wrf.exe 0000000000410DAA Unknown Unknown Unknown

and it seems to come after the land_use initialization:

INPUT LandUse = "MODIFIED_IGBP_MODIS_NOAH"
LANDUSE TYPE = "MODIFIED_IGBP_MODIS_NOAH" FOUND 33 CATEGORIES 2 SEASONS WATER CATEGORY = 17 SNOW CATEGORY = 15
d03 2007-06-22_00:00:00 returning from of landuse_init
INITIALIZE THREE Noah LSM RELATED TABLES

However, no fatal statement was called and the wrfinput files looked OK for the most part. My namelist files, for both WPS and WRF, are also attached. I have looked into my namelist files several times but nothing really jumped out.

Any suggestions on the root cause of this issue and how to resolve it would be greatly appreciated!

Thank you,

David

 

[chunchih]

Hi WRF support,

Just some update.

I found a workaround to this problem by setting 'input_from_field' in my namelist.input to false for the fourth domain. Thus, the initial static and meteorological fields for the fourth domain are interpolated down from the third domain (I believe). Now, the simulation runs fine with the outputs for all four domains produced. Although I do lose some benefits of higher resolution static datasets, but I think this is an acceptable solution for my application for now.

I suspect that it might be a problem with the input geography dataset that finally shows up when I go to a sub-kilometer model resolution. However, any other thoughts are welcomed.

David

 

[Ming Chen]

Hi David,
Thanks for the detailed description of the issue and the solution. I am suspicious that the failed case is probably due to landuse type in D04, which has a very high resolution and somehow the interpolation conducted in WPS might give wrong information at this resolution.
Would you please rerun WPS but replace your previous option by the one below:
geog_data_res = 'default', 'default','default','default',
Then rerun this case and let me know whether it works. However, I completely understand if you don't feel like to repeat the process. That is totally fine.

 

[chunchih]

Hi Ming,

Yes. I agree. That is what I thought too.

I can try your suggestion. I will let you know once I have done that. In the end, I would still like to keep using the higher resolution geographic dataset though.

Thank you and bests,

David