Specifying oddly named GRIB2 field for MET tools

[Jeff-Duda]

I am attempting to generate a verification mask using gen_vx_mask. My chosen method of masking is to use values from a GRIB2 data file. The data file is from MRMS. In fact, it is an MRMS file containing the MRMS product called RadarQualityIndex, a scalar ranging from 0.0 to 1.0. My goal is to create a mask on the HRRR grid using thresholded values of this "RQI".

I tried doing all of this in a single command using:
[Jeffrey.Duda@ufe01 PMP]$ gen_vx_mask "lambert 1799 1059 21.138123 -122.719528 -97.5 3.0 6371.2 38.5 N" /scratch4/BMC/zrtrr/jdduda/MRMS/2025/MRMS_RQI_to_use_2025.grib2 HRRRv4_RQI_mask.nc -type data -mask_field 'name="RadarQualityIndex"; level="Z0";' -thresh ">=0.7"
But it gave me the error below.

DEBUG 1: Start gen_vx_mask V12.2.0 by Jeffrey.Duda(20121) at 2026-03-27 21:58:29Z with command: gen_vx_mask lambert 1799 1059 21.138123 -122.719528 -97.5 3.0 6371.2 38.5 N /scratch4/BMC/zrtrr/jdduda/MRMS/2025/MRMS_RQI_to_use_2025.grib2 HRRRv4_RQI_mask.nc -type data -mask_field name="RadarQualityIndex"; level="Z0"; -thresh >=0.7
DEBUG 2: OpenMP running on 1 thread(s).
DEBUG 1: Input Grid: lambert 1799 1059 21.138123 -122.719528 -97.5 3.0 6371.2 38.5 N
DEBUG 1: Mask File: /scratch4/BMC/zrtrr/jdduda/MRMS/2025/MRMS_RQI_to_use_2025.grib2
DEBUG 2: Parsed Input Grid: To (lambert) (1799 x 1059)
DEBUG 2: Parsed Mask Grid: LatLon (7000 x 3500)
ERROR :
ERROR : process_mask_file() -> The input grid and mask grid must be identical for "data" masking.
ERROR : Data Grid -> Projection: Lambert Conformal Nx: 1799 Ny: 1059 Lat_LL: 21.138 Lon_LL: 122.720 Lon_orient: 97.500 Alpha: 4203.494 Cone: 0.623 Bx: 899.1693 By: 3198.9984
ERROR : Mask Grid -> Projection: Lat/Lon Nx: 7000 Ny: 3500 lat_ll: 20.005 lon_ll: 129.995 delta_lat: 0.010 delta_lon: 0.010 wrapLon: false
ERROR :

Oh, okay. This grid error message implies I'm going to have to do this in two steps instead of one. So let me create the first mask on the MRMS grid, as follows;
[Jeffrey.Duda@ufe01 PMP]$ gen_vx_mask "latlon 7000 3500 20.005 -129.995 0.01 0.01" /scratch4/BMC/zrtrr/jdduda/MRMS/2025/MRMS_RQI_to_use_2025.grib2 MRMS_RQI_mask.nc -type data -mask_field 'name="RadarQualityIndex"; level="Z0";' -thresh ">=0.7"

DEBUG 1: Start gen_vx_mask V12.2.0 by Jeffrey.Duda(20121) at 2026-03-27 21:59:40Z with command: gen_vx_mask latlon 7000 3500 20.005 -129.995 0.01 0.01 /scratch4/BMC/zrtrr/jdduda/MRMS/2025/MRMS_RQI_to_use_2025.grib2 MRMS_RQI_mask.nc -type data -mask_field name="RadarQualityIndex"; level="Z0"; -thresh >=0.7
DEBUG 2: OpenMP running on 1 thread(s).
DEBUG 1: Input Grid: latlon 7000 3500 20.005 -129.995 0.01 0.01
DEBUG 1: Mask File: /scratch4/BMC/zrtrr/jdduda/MRMS/2025/MRMS_RQI_to_use_2025.grib2
DEBUG 2: Parsed Input Grid: To (latlon) (7000 x 3500)
DEBUG 2: Parsed Mask Grid: LatLon (7000 x 3500)
DEBUG 2: Data Threshold: >=0.7
DEBUG 1: Output File: MRMS_RQI_mask.nc
DEBUG 1: Finish gen_vx_mask V12.2.0 by Jeffrey.Duda(20121) at 2026-03-27 21:59:43Z

No problem! The data looks plenty good (see image)

Now for applying that data to the HRRR grid to create the final mask. I'm not exactly sure how to do this, but all of the below attempts have failed to get the result I want:
[Jeffrey.Duda@ufe01 PMP]$ gen_vx_mask "lambert 1799 1059 21.138123 -122.719528 -97.5 3.0 6371.2 38.5 N" MRMS_RQI_mask.nc HRRRv4_RQI_mask.nc -type grid
This actually completed successfully but gave me an undesired result:

This is just the portion of the HRRR grid that is within the MRMS domain No, I want the RQI mask above interpolated to HRRR grid points.

This doesn't work; it just gives me the mismatched grid error again:
[Jeffrey.Duda@ufe01 PMP]$ gen_vx_mask gen_vx_mask "lambert 1799 1059 21.138123 -122.719528 -97.5 3.0 6371.2 38.5 N" MRMS_RQI_mask.nc HRRRv4_RQI_mask.nc -type data -mask_field 'name="data_mask"; level="(*,*)";' -thresh "eq1.0"

It seems like from the examples in the documentation on gen_vx_mask that I may have to spatially interpolate the MRMS radar quality index file to the HRRR grid before doing step 1. Is that correct? If so, it seems strange that you can create a grid file based on an input grid and a polygon mask (ASCII) but not on a different grid (MRMS -> HRRR). Am I missing the point of gen_vx_mask?

For SnGs, I went ahead with regridding the MRMS file first and tried to create the mask I wanted, with success:
wgrib2 MRMS_RQI_to_use_2025.grib2 -new_grid_interpolation neighbor -new_grid lambert:262.5:38.5:38.5:38.5 237.280472:1799:3000.0 21.138123:1059:3000.0 MRMS_RQI_2025_on_HRRR.grib2
gen_vx_mask "lambert 1799 1059 21.138123 -122.719528 -97.5 3.0 6371.2 38.5 N" MRMS_RQI_2025_on_HRRR.grib2 HRRRv4_RQI_mask_ge0p5.nc -type data -mask_field 'name="RadarQualityIndex"; level="Z0";' -thresh ">=0.5"

[JohnHalleyGotway]

@Jeff-Duda thanks for sharing your experiences using the data masking option in Gen-Vx-Mask!

I read through your steps and am glad you landed on your desired result... defining a RadarQualityIndex on the 1799 x 1059 HRRR domain. I did want to mention a couple other options.

In this call to Gen-Vx-Mask:

gen_vx_mask "latlon 7000 3500 20.005 -129.995 0.01 0.01" \
/scratch4/BMC/zrtrr/jdduda/MRMS/2025/MRMS_RQI_to_use_2025.grib2 \
MRMS_RQI_mask.nc -type data \
-mask_field 'name="RadarQualityIndex"; level="Z0";' -thresh ">=0.7"

By default, Gen-Vx-Mask will name the output variable as data_mask:

	float data_mask(lat, lon) ;
		data_mask:coordinates = "lat lon" ;
		data_mask:long_name = "data_mask masking region" ;
...

But I'd recommend using the -name command line option to give the output variable a more meaningful name, like -name RQI_GT0.7:

	float RQI-GT0.7(lat, lon) ;
		RQI-GT0.7:coordinates = "lat lon" ;
		RQI-GT0.7:long_name = "RQI-GT0.7 masking region" ;
...

That way, when you use this to define a masking region in other MET tools (like Grid-Stat, Point-Stat, and Ensemble-Stat), the VX_MASK column will be set to RQI-GT0.7 rather than the very generic data_mask string.

And after this call to Gen-Vx-Mask, you have an RQI mask defined on the 7000 x 3500 MRMS grid.
With that output file, you have 2 choices:

1. You could actually pass it directly to the other MET tools through the mask.poly config option to define the masking region over which stats should be computed. So you'd be verifying forecasts on the 1799 x 1059 HRRR domain using a mask define on the 7000 x 3500 MRMS grid. When the vx grid and masking grids don't match, MET automatically regrids the masking grid to the vx grid using nearest neighbor interpolation... and it prints this log message.

Honestly though, I didn't find this detail clearly described in the MET User's Guide.

2. While (1) does technically work, in practice, it's not a very good idea. You'd be wasting time in each run regridding that MRMS RQI mask to the HRRR domain. It'd be much more efficient to do that regridding just one. And fortunately, you can use the Regrid-Data-Plane tool to do exactly that. Something like this:

regrid_data_plane MRMS_RQI_mask.nc \
"lambert 1799 1059 21.138123 -122.719528 -97.5 3.0 6371.2 38.5 N" \
HRRR_RQI_mask.nc \
-field 'name="RQI-GT0.7"; level="(*,*)";' \
-method NEAREST -width 1

And that'll produce the RQI mask on your desired HRRR domain.

Hope that helps!