Are all ensemble members expected to be in the same file format?

[lindsayblank]

Software Version

METplus 6.2.0

Text

Hi METplus folks,

I am running GenEnsProd and EnsembleStat on Hera. I am evaluating 3-hourly APCP.

8 of the ensemble members do not have 3-hourly APCP natively, so I calculate it via PCPCombine. The results are netCDF files.

2 ensemble members do have 3-hourly APCP natively, so I pass the files in their original grib2 format.

When I run GenEnsProd, I get the following error:

DEBUG 2: OpenMP running on 1 thread(s).
DEBUG 1: Default Config File: /contrib/met/12.2.0/share/met//config/GenEnsProdConfig_default
DEBUG 1: User Config File: /contrib/METplus/METplus-6.2.0/metplus/parm/met_config/GenEnsProdConfig_wrapped
DEBUG 1: Ensemble Files[10]:
DEBUG 1:    /scratch4/BMC/amb-verif/ensemble_workflow_2026/ensemble_work/HRRR/20260605-12z/HRRR_2026060512_03_APCP03.nc
DEBUG 1:    /scratch4/BMC/amb-verif/ensemble_workflow_2026/ensemble_work/HRRR/20260605-06z/HRRR_2026060506_09_APCP03.nc
DEBUG 1:    /scratch4/BMC/amb-verif/ensemble_workflow_2026/ensemble_work/HRW_ARW/20260605-12z/HRW_ARW_2026060512_03_APCP03.nc
DEBUG 1:    /scratch4/BMC/amb-verif/ensemble_workflow_2026/ensemble_work/HRW_ARW/20260605-00z/HRW_ARW_2026060500_15_APCP03.nc
DEBUG 1:    /scratch4/BMC/amb-verif/ensemble_workflow_2026/ensemble_work/HRW_FV3/20260605-12z/HRW_FV3_2026060512_03_APCP03.nc
DEBUG 1:    /scratch4/BMC/amb-verif/ensemble_workflow_2026/ensemble_work/HRW_FV3/20260605-00z/HRW_FV3_2026060500_15_APCP03.nc
DEBUG 1:    /scratch4/BMC/amb-verif/ensemble_workflow_2026/ensemble_work/HRW_NSSL/20260605-12z/HRW_NSSL_2026060512_03_APCP03.nc
DEBUG 1:    /scratch4/BMC/amb-verif/ensemble_workflow_2026/ensemble_work/HRW_NSSL/20260605-00z/HRW_NSSL_2026060500_15_APCP03.nc
DEBUG 1:    /scratch4/BMC/amb-verif/ensemble_workflow_2026/ensemble_work/NAM_CONUS/20260605-12z/NAM_2026060512_03_APCP03.grib2
DEBUG 1:    /scratch4/BMC/amb-verif/ensemble_workflow_2026/ensemble_work/NAM_CONUS/20260605-00z/NAM_2026060500_15_APCP03.grib2
DEBUG 1: Ensemble Control: None
DEBUG 2:  
DEBUG 2: --------------------------------------------------------------------------------
DEBUG 2:  
DEBUG 2: Processing ensemble field: APCP_03/A3
DEBUG 2: Using named grid (/scratch4/BMC/amb-verif/ensemble_workflow_2026/ensemble_work/HRRR/20260605-12z/HRRR_2026060512_03_APCP03.nc) as the verification grid.
DEBUG 1: Regridding field "APCP_03/A3" to the verification grid using NEAREST(1).
DEBUG 1: Regridding field "APCP_03/A3" to the verification grid using NEAREST(1).
DEBUG 1: Regridding field "APCP_03/A3" to the verification grid using NEAREST(1).
DEBUG 1: Regridding field "APCP_03/A3" to the verification grid using NEAREST(1).
DEBUG 1: Regridding field "APCP_03/A3" to the verification grid using NEAREST(1).
DEBUG 1: Regridding field "APCP_03/A3" to the verification grid using NEAREST(1).
ERROR  :   
ERROR  : MetNcMetDataFile::open(const char *) -> unable to open NetCDF file "/scratch4/BMC/amb-verif/ensemble_workflow_2026/ensemble_work/NAM_CONUS/20260605-12z/NAM_2026060512_03_APCP03.grib2"
ERROR  :   
ERROR  :   
ERROR  : Met2dDataFileFactory::new_met_2d_data_file() -> error opening file "/scratch4/BMC/amb-verif/ensemble_workflow_2026/ensemble_work/NAM_CONUS/20260605-12z/NAM_2026060512_03_APCP03.grib2"
ERROR  :   

I am trying to figure out why GenEnsProd is expecting a netCDF file. Hence, my question in the title, are all members expected to be in the same file format?

Thank you,
Lindsay

[CPKalb]

Hi Lindsay:

For Ensemble-Stat, the file type is determined using the first file in the list of input files. If you are curious, here is the code that determines this file type.

As you suspect above, all members are expected to be in the same file format for Ensemble-Stat. It can't read data from different input file types. One reason for this restriction is that a single input variable name and level is used in the configuration file for each variable in all the files. Since levels are specified differently for netCDF vs grib2 files, there isn't configuration options to support reading multiple data types for the same variables. I'm guessing (without seeing your files), that you would actually also need two different configuration settings for the two different file types:
'name="APCP_03"; level="(*,*)";' for the netCDF files
and
'name="APCP"; level="A03";' for the GRIB2 files

Anyway, all files have to be in the same format. There are 2 ways to get around this for the setup you have:

1. Use PCP-combine with the add option and a single input file for the 2 members that already have 3 hourly APCP computed. This will effectively dump the 3 hourly precipitation data to a netCDF file in the same format as the other members.
2. Read the data from a Python embedding script that knows how to read both the netCDF and GRIB2 data

Let me know if you have questions
Tina

[lindsayblank]

Thanks, Tina!

[CPKalb]

You're welcome. I'm going to go ahead and close and lock this discussion. If you have additional questions, feel free to start up a new discussion.

tina