Regional Grid-to-Grid (g2g) Retro Verification (against ECMWF analyses)

Step 1: Pull the relevant grids from the model output and analysis files¶

1. Copy /home/role.amb-verif/GSL_verif-regional/xml/retro/regional_g2g_init_ecmwf_retro_template.xml to a working directory of your choice.

2. Open it in vi or another editor.

3. String replace MODELNAME for your model name. In vi, this is done with :%s/MODELNAME/<your_model_name>/g

   1. Note: you can verify more than one model in the same workflow. For an example of how this works, you can view (but please don’t edit!) /home/role.amb-verif/GSL_verif-regional/xml/regional_g2g_init_rt.xml.

4. String replace MODEL_LOWERCASE for your model name in all lowercase. In vi, this is done with :%s/MODEL_LOWERCASE/<your_model_name_lowercase>/g

5. Change the path on line 12 from “/PATH/TO/MODEL/DATA” to the actual path to your model output files.

   1. For RRFS models, you want the path to the directory where all the dated output directories are (like rrfs.20240501).

      1. If you’re using an older version of UPP that results in “postprd” directories, you want the path to the directory where all the dated output directories are (like 2024050100). You will also need to change the script name on line 84 from “RRFS_init_with_calc.py” to “RRFS_init_with_calc_old_UPP.py”.

6. Change the path on line 16 from /PATH/TO/ANL/DATA to the actual path to your ECMWF analysis files.

   1. The path should be to the directory that contains the ECMWF oper and scda subdirectories.

7. On lines 69 and 70, set the date strings for your start and end dates. These should be in the format YYYYMMDDHHMM.

8. Submit and run the workflow in rocoto. Upon completion, you should have new pre-processed grib2 files in /home/role.amb-verif/GSL_verif-regional/verif/g2g_retro/MODELNAME/realtime and /home/role.amb-verif/GSL_verif-regional/verif/g2g_retro/ECMWF_anl/realtime

Step 2: Run the METplus verification to generate statistics¶

1. Copy /home/role.amb-verif/GSL_verif-regional/xml/retro/metplus_ecmwf_g2g_retro_template.xml to a working directory of your choice.

2. Open it in vi or another editor.

3. String replace MODELNAME for your model name. In vi, this is done with :%s/MODELNAME/<your_model_name>/g

   1. As before, you can modify the workflow to verify more than one model at once. For an example of how this works, you can view (but please don’t edit!) /home/role.amb-verif/GSL_verif-regional/xml/metplus_ecmwf_g2g.xml.

4. String replace MODEL_LOWERCASE for your model name in all lowercase. In vi, this is done with :%s/MODEL_LOWERCASE/<your_model_name_lowercase>/g

5. On line 73, set the date strings for your start and end dates. These should be in the format YYYYMMDDHHMM.

6. Submit and run the workflow in rocoto. Upon completion, you should have new .stat files in /home/role.amb-verif/GSL_verif-regional/verif/g2g_retro/MODELNAME/realtime/METplus

7. Repeat steps 1-5 for /home/role.amb-verif/GSL_verif-regional/xml/retro/metplus_ecmwf_g2g_anom_retro_template.xml if you also want to calculate ACC.

Step 3: Load your stat files into the METexpress database¶

1. Copy /home/role.amb-verif/GSL_verif-regional/xml/retro/mv_load_gsl_retro_g2g_template.xml to a working directory of your choice, and rename it something descriptive.

2. Change the database on line 9 from mv_gsl_regional_g2g_retros to something to describe your work. For example, if this retro is part of a certain case study, the database name could be mv_your-name_case-study-name. YOUR DATABASE NAME MUST START WITH “mv_” OR METEXPRESS WILL IGNORE IT. You can use the same database for multiple retro loadings (and you should load all related retros into one database), but please don’t load data into other people’s databases without their permission!

3. On line 26, replace “/PATH/TO/STAT/FILES” with the absolute path to the top-level directory where your stat files exist.

4. Use lines 29-31 (you can add more or take some away if necessary) to denote subdirectories under your top level path. These subdirectories should contain the stat files.

5. On line 3, enter your email.

6. If this is your first time using the database on line 9 (i.e., neither you nor anyone else has ever loaded data to it before), you will need to initialize the database. To do this, enter the following on the command line as amb-verif (replacing <your_db_name> with the actual name of your database):

   mysql --defaults-file=/home/role.amb-verif/GSL_verif-global/config/gsd_met_admin_my.cnf -e'create database <your_db_name>;'
   mysql --defaults-file=/home/role.amb-verif/GSL_verif-global/config/gsd_met_admin_my.cnf <your_db_name> < /home/role.amb-verif/GSL_verif-global/config/mv_mysql.sql

7. Run the following commands:

   conda activate avid_verify_py3
   
   setenv PYTHONPATH '/home/role.amb-verif/GSL_verif-global/bin/METdataio-3.2.0/'
   
   python /home/role.amb-verif/GSL_verif-global/bin/METdataio-3.2.0/METdbLoad/ush/
   met_db_load.py <path to your XML load file>

8. Your data will appear in METexpress the following morning, when the metadata refreshes.

9. If you need to view your data sooner than this, email mats.gsl@noaa.gov.