Stat Analysis NBRCTC Aggregation Issue

[lindsayrblank]

Hello all,

I'm having a strange problem with Stat_Analysis. I am trying to aggregate the NBRCTC line type. I am running MET version 10.0.0 Docker as a Singularity image on an HPC. When I run the following command, I get the following error:

singularity exec /path/to/singularity_image/met.sif stat_analysis -job aggregate -line_type NBRCTC -lookin /path/to/data/directory -out_stat ./test.stat -by FCST_VAR,FCST_THRESH,FCST_LEAD,FCST_INIT_HOUR -fcst_init_hour 00  -set_hdr DESC 'NA' -rank_corr_flag FALSE -n_boot_rep 0 -out_bin_size 1 -v 3

 Creating STAT output file "./test.stat"
terminate called after throwing an instance of 'std::logic_error'
  what():  basic_string::_S_construct null not valid
Aborted

When I run the same command with -line_type CTC, it works perfectly. There are both CTC and NBRCTC lines in the input files.

Any help would be greatly appreciated!

Best,
Lindsay

[JohnHalleyGotway]

@lindsayrblank thanks for letting us know about this runtime error. @georgemccabe found a pretty obvious bug in the code that is the source of this issue starting on this line of code.

         case stat_nbrctc: write_header_row       (nbrctc_columns, n_sl1l2_columns, 1,      stat_at, 0, 0); break;
         case stat_nbrcts: write_header_row       (nbrcts_columns, n_sl1l2_columns, 1,      stat_at, 0, 0); break;
         case stat_nbrcnt: write_header_row       (nbrcnt_columns, n_sl1l2_columns, 1,      stat_at, 0, 0); break;

It's a copy/paste error and the fix is easy.

FYI, I'm able to replicate this error by running:

MET-main_v10.0/met/bin/stat_analysis -lookin grid_stat_APCP_12_120000L_20050807_120000V.stat -job aggregate -line_type NBRCTC -out_stat out_stat.txt

Produces this:

terminate called after throwing an instance of 'std::logic_error'
  what():  basic_string::_M_construct null not valid
Abort trap: 6

But if I change -out_stat to just -out, the error goes away:

MET-main_v10.0/met/bin/stat_analysis -lookin grid_stat_APCP_12_120000L_20050807_120000V.stat -job aggregate -line_type NBRCTC -out out.txt

So this is obviously a bug that we need to fix (see dtcenter/MET#2271), but I just wanted to point out a temporary workaround.

I assume you'll want a bugfix for MET version 10.0, or would providing on for MET version 10.1 be sufficient?

Thanks,
John

[lindsayrblank]

Thank you, @JohnHalleyGotway. I tested your work around and it worked on my instance. A MET version 10.0 bugfix would be preferable to 10.1, if you're able to.