Problem with the Grid-Stat tool.

[Victorrani]

I'm trying to better understand how the Grid-Stat tool from MET works, but I've been facing some difficulties.

It seems to be running and generating the .nc file; however, it's not producing any statistics in the .stat output file.

Could someone help me understand how to properly configure the configuration file?

In this test, I'm trying to analyze the u10 variable from two different versions of the MONAN model. The .nc files contain a 5-day forecast.

./grid_stat /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MONAN_v0_AS/MONAN_v0_AS_2024021400.nc /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MONAN_v1_AS/MONAN_v1_AS_2024021400.nc /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MERGE_14_19/GridStatConfig_default -outdir /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MERGE_14_19

./grid_stat /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MONAN_v0_AS/MONAN_v0_AS_2024021400.nc /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MONAN_v1_AS/MONAN_v1_AS_2024021400.nc /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MERGE_14_19/GridStatConfig_default -outdir /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MERGE_14_19
DEBUG 1: Start grid_stat by victor(1000) at 2025-01-31 14:14:17Z  cmd: ./grid_stat /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MONAN_v0_AS/MONAN_v0_AS_2024021400.nc /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MONAN_v1_AS/MONAN_v1_AS_2024021400.nc /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MERGE_14_19/GridStatConfig_default -outdir /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MERGE_14_19 
DEBUG 2: OMP_NUM_THREADS is not set. Defaulting to 1 thread. Recommend setting OMP_NUM_THREADS for faster runtimes.
DEBUG 2: OpenMP running on 1 thread(s).
DEBUG 1: Default Config File: /home/victor/DTC/MET-11.1.1/share/met/config/GridStatConfig_default
DEBUG 1: User Config File: /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MERGE_14_19/GridStatConfig_default
GSL_RNG_TYPE=mt19937
GSL_RNG_SEED=18446744072532696167
DEBUG 1: Forecast File: /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MONAN_v0_AS/MONAN_v0_AS_2024021400.nc
DEBUG 1: Observation File: /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MONAN_v1_AS/MONAN_v1_AS_2024021400.nc
DEBUG 1: Regridding field u10(0,*,*) to the verification grid.
DEBUG 2: Processing masking regions.
DEBUG 2: 
DEBUG 2: --------------------------------------------------------------------------------
DEBUG 2: 
DEBUG 2: Processing u10(0,*,*) versus u10(0,*,*), for smoothing method NEAREST(1), over region FULL, using 72000 matched pairs.
DEBUG 2: 
DEBUG 2: --------------------------------------------------------------------------------
DEBUG 2: 
DEBUG 1: Output file: /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MERGE_14_19/grid_stat_000000L_20240214_000000V.stat
DEBUG 1: Output file: /home/victor/DTC/MET-11.1.1/MET-11.1.1/data/datain/MONAN_v0_v1_MERGE/MERGE_14_19/grid_stat_000000L_20240214_000000V_pairs.nc

Can anyone see what I might be missing?

Thank you very much! - Muito obrigado!

[j-opatz]

Hello,

And thank you for your question. With the configuration file snippet you provided, I see that you're very close to getting output. The current setback is due to thresholds.

To understand why, let's take a look at the output_flag dictionary, where you're requesting the FHO line type. This stands for Forecast Hit Observation rates. As we can see from the documentation, this line type will contain 4 columns of statistics: the total number of matched pairs, the forecast rate, the hit rate, and the observation rate.

If we then take a look at appendix C for how these statistics are calculated, we see that except for matched pairs, all of the rates are based on contingency table counts (forecast rate is F_RATE, hit rate is H_RATE, and observation rate is O_RATE). To create a contingency table for a binary categorical variable, you must establish what binary category the forecast and observation values are divided into via a threshold.

If, for example, you wanted to verify all u10 forecast values less than 10 m/s, that would look like

fcst = {
            file_type = NETCDF_NCCF;

       field = [
           {
            name = "u10";
            level = "(0,*,*)";
            thresh = [<10.0];
           }
           ];
}

obs = fcst;

Note in this example, I've set the obs dictionary equal to the fcst dictionary since the same field settings are used (u10 variable name, level access via (0,*,*), and thresholds of less than 10.0). If you want the thresholds to be different, you should set each dictionary's thresh accordingly. This sets up a contingency table where all u10 forecasts that are less than 10.0 will be an f1 (or forecast yes) and all u10 forecasts that are equal to or greater than 10.0 will be an f0 (or forecast no), and a similar setting for the observation u10 field (but with o1 and o0).

Hopefully this makes sense; you need to make sure the appropriate thresholds are in place for both forecast and observation fields to gain output for the requested line types. If you have any additional questions on thresholds or getting the configuration file to run, please let me know!