Archive‎ > ‎Spring 2016‎ > ‎EAPS 434 Spring 2016‎ > ‎

run your own WRF model

run the WRF model

Set up the files that you will need to run the WRF. In a linux terminal, type:

cd /class/hypocenter/a/EAS434/2016
mkdir your_name
cd your_name
mkdir wrfrun
cd wrfrun
tar -xvf /class/hypocenter/a/EAS434/2016/wrfrun.tar
source setup_wrf_ex.csh

2. Follow the NCAR online tutorial for running the WRF, located at:

Follow the tutorial from the introduction through the page where graphics routines are listed.

Do not bother with downloading and compiling the codes, we will use the versions of the code that I have already compiled for you.

Pay particularly close attention to the “namelist.wps” and “namelist.input” files, and follow the tutorial links related to these files in order to learn about the meaning of the namelist parameters. These control everything about the model run, grid spacing, time step, physics parameterizations, etc.

We will run a real data case from last week, starting Mar 03 2016 12 UTC and ending 24h later at 12 UTC Mar 04. The initial and boundary conditions will be supplied by the NCEP NAM model, GRIB2 format data can be found in: /class/hypocenter/a/EAS434/DATA/20160303/nam*

In summary, we need to do several steps to set up and run the WRF. First we need to get data for initial and boundary conditions (link_grib.csh). We need to set up the model domain including terrain and other surface conditions (geogrid.exe). We have to convert the initial/boundary condition data from GRIB format to a format that WRF can use (ungrib.exe). We have to interpolate the data to the model grid, first horizontally (metgrid.exe), then vertically (real.exe). Then we run the model (wrf.exe). Basically, you will need to type the following in order to execute the WRF model run: (make sure that you are in your “wrfrun” directory/folder)

./link_grib.csh /class/hypocenter/a/EAS434/DATA/20160303/nam

Most of these steps take a few seconds to complete and will print a lot of information to your terminal. In the “wrf.exe” step, you are running the WRF interactively. This will take ~30 min to execute this example). Make sure that you have time to do this run in the lab.

3. Assuming that everything worked perfectly (which of course it did!), you should have files with names like: “wrfout_d01_2016-03-04_12:00:00” in your wrfrun directory. The date/time info represents the valid time for the forecast. The “post_wrf” step runs the WRF post-processor, which computes a number of diagnostic fields and converts them to GEMPAK. The GEMPAK files should be in a sub-directory called “wrf”. You should be able to use nmap2 and other GEMPAK programs to analyze the output. The files are located in GRID|local_mesoscale_models|wrf (make sure to source setup_wrf_ex.csh) and in gdplot2 you’ll need to set

GDFILE = $MODEL/wrf/2016030312fXXX_wrf.gem

make sure that you can plot variables from your WRF run using nmap2 and GEMPAK (gdplot2).