WRFJoiner

A collection of Python scripts to join WRF output patches (io_form_history = 102) into a single NetCDF file.

Description

When io_form_history or io_form_restart in the WRF's namelist.input is set to 102, then the WRF produces output patches assigned to each MPI process. To join these patches into a single NetCDF file, you can use the Fortran program called as "JOINER" is provided at the WRF Users' Page.

Since the JOINER program was developed years ago, it needs a slight modification for the WRFV4 outputs. As an alternative, I made some Python scripts that do the same job. A simple verification with my WRF outputs showed that these scripts work OK. Still, please use with caution if you are going to use these scripts.

1. WRFJoiner_Parallel.py

This script has a feature of fetching patch files in parallel using Python's multiprocessing module. Note that writing the output NetCDF file is in serial for now.

2. WRFJoiner_Serial.py

This script works using only a single process. One patch file is loaded and then pushed to the output NetCDF file, so this script uses less memory than the parallel version.

3. namelist.ini

This is a sample configuration file for this script. You need to define the following configurations. You can set most of them same as your namelist.input. Here are descriptions of some entrie.s

• loc_in : Location of the patch files
• loc_out : Location of the output file to write
• interval: Interval of WRF outputs in minute (= history_interval in minute)
• verbose : Whether to print more lines
• nproc : Number of processes to use (only valid for the parallel version)

Contact

Jeonghoe Kim (jeonghoekim.14@snu.ac.kr)