METplus Use Case: EnsembleStat with multiple variable and leads

METplus Use Case: EnsembleStat with multiple variable and leads

This use case takes the PB2NC tool and combines it with the Ensemble-Stat tool, analyzing multiple forecast fields and producing ensemble relative frequencies.

Optional: Refer to the MET Users Guide for a description of the MET tools used in this use case.
Optional: Refer to the METplus Config Glossary section of the METplus Users Guide for a reference to METplus variables used in this use case.
Change to the ${METPLUS_TUTORIAL_DIR} directory:
cd ${METPLUS_TUTORIAL_DIR}
  1. Review the use case configuration file: EnsembleStat_fcstHRRRE_obsHRRRE_Sfc_MultiField.conf

Open the file and look at all of the configuration variables that are defined.

less ${METPLUS_BUILD_BASE}/parm/use_cases/model_applications/short_range/EnsembleStat_fcstHRRRE_obsHRRRE_Sfc_MultiField.conf

Note that in this use case, the PB2NC GenEnsProd, and EnsembleStat tools will be called:

PROCESS_LIST = PB2NC, EnsembleStat, GenEnsProd

Another important aspect of this use case is its use of lead times. While INIT_BEG and INIT_END are the same, there are three leads listed:

LEAD_SEQ = 0,1,2

And that controls what files are looked at in the EnsembleStat call:

{init?fmt=%Y%m%d%H}/postprd_mem0001/wrfprs_conus_mem0001_{lead?fmt=%HH}.grib2,
    {init?fmt=%Y%m%d%H}/postprd_mem0002/wrfprs_conus_mem0002_{lead?fmt=%HH}.grib2

Given the lead times and INIT time, the tools will loop over the following valid times:

  • 20180709 at 12z

  • 20180709 at 13z

  • 20180709 at 14z

The run time for this use case is expected to be longer, due to the multiple lead times, two MET tools, and 6 variables at multiple thresholds being analyzed.

  1. Run the use case:
run_metplus.py \
${METPLUS_BUILD_BASE}/parm/use_cases/model_applications/short_range/EnsembleStat_fcstHRRRE_obsHRRRE_Sfc_MultiField.conf \
${METPLUS_TUTORIAL_DIR}/tutorial.conf \
config.OUTPUT_BASE=${METPLUS_TUTORIAL_DIR}/output/Ensemble

METplus is finished running when control returns to your terminal console and you see the following text:

INFO: METplus has successfully finished running.

 

  1. Review the output files:

You should have two directories of interest:

ls ${METPLUS_TUTORIAL_DIR}/output/Ensemble/model_applications/short_range/EnsembleStat_fcstHRRRE_obsHRRRE_Sfc_MultiField
  • EnsembleStat
  • rap

These two directories correspond to the output directories for two of the tools in the configuration file:

PB2NC_OUTPUT_DIR = {OUTPUT_BASE}/model_applications/convection_allowing_models/EnsembleStat_fcstHRRRE_obsHRRRE_Sfc_MultiField/rap
ENSEMBLE_STAT_OUTPUT_DIR = {OUTPUT_BASE}/model_applications/convection_allowing_models/EnsembleStat_fcstHRRRE_obsHRRRE_Sfc_MultiField/EnsembleStat
Note the GenEnsProd output directory is set to the same as EnsembleStat (GEN_ENS_PROD_OUTPUT_DIR = {ENSEMBLE_STAT_OUTPUT_DIR})

Take a look at the output from PB2NC first:

ls ${METPLUS_TUTORIAL_DIR}/output/Ensemble/model_applications/short_range/EnsembleStat_fcstHRRRE_obsHRRRE_Sfc_MultiField/rap/20180709

You'll see 3 output files, corresponding to the three lead times that were run. These files were read in by the EnsembleStat call and processed for statistics. Taking a look at those files we see a lot of output:

ls ${METPLUS_TUTORIAL_DIR}/output/Ensemble/model_applications/convection_allowing_models/EnsembleStat_fcstHRRRE_obsHRRRE_Sfc_MultiField/EnsembleStat/201807091200

Each lead time is responsible for producing 8 files, for a total of 24 files. These were all requested with the ENSEMBLE_STAT_OUTPUT_FLAG options, with the exception of the .stat and .nc files.

  1. Compare the netCDF output to the .stat output:

To better understand the difference between the FCST and ENS entries in the configuration file used by the EnsembleStat and GenEnsProd tools respectively, take a look at the netCDF file for the first lead time (0):

ncview ${METPLUS_TUTORIAL_DIR}/output/Ensemble/model_applications/short_range/EnsembleStat_fcstHRRRE_obsHRRRE_Sfc_MultiField/EnsembleStat/201807091200/gen_ens_prod_HRRRE_F000_ADPSFC_20180709_120000V_ens.nc

You'll notice that the majority of the extensive variables listed have the words "ENS_FREQ". This is because those variables were called with the ENSEMBLE_FLAG options: GenEnsProd will provide a summary of all fields requested across the ensemble for those ENS_VAR<n> variables.

Now look at the .stat file for the same lead time:

less ${METPLUS_TUTORIAL_DIR}/output/Ensemble/model_applications/short_range/EnsembleStat_fcstHRRRE_obsHRRRE_Sfc_MultiField/EnsembleStat/201807091200/ensemble_stat_HRRRE_F000_ADPSFC_20180709_120000V.stat

In the VX_MASK column, the four mask files requested are listed, along with the various line types that were requested (listed in the LINE_TYPE column). What's important to note is the lack of variable variety that was present in the netCDF: in fact, the only variable listed is TMP at the Z2 level. That's because in the configuration file, only 1 variable was requested with the FCST_VAR<n> options, and only those variables will be used for verification.

jopatz Fri, 03/11/2022 - 11:37