Login from a terminal (ssh -Y YourUserName@cheyenne.ucar.edu) following the instructions you received (Yubikey, Duo, other) to authenticate and complete the login.

Go to your work space:

or

where $USER is your login/username, or

where ~ is shorthand for your home directory.

To download the Medium-Range Weather Application, with externals pointing to its constituent components, go to your $HOME directory and enter:

Then, to get the components

This will clone the ufs-v1.1.0 branch of the ufs-mrweather-app repository  and check out all of the externals and sub-modules.  The process will take several minutes and show the following progress information in the screen:

The repositories URL, tag number and their location path of external and sub-modules are defined in the text file Externals.cfg. Users can check this file and match the new cloned repositories with their local paths.

Once the externals have been been downloaded, you can proceed to the next step, setting up the environment.

Prior to building the model, the environment for CIME needs to be set up. This requires five environment variables: 

Enter the following commands on Cheyenne.  You may need to create a directory on /glade/scratch.

If you see something like

Then

This directory should persist throughout the tutorial.

For bash:

Note:  You'll need to set the environments (export/setenv) each time you log in to do work with the MR-Weather-App.  You can add these lines to your .bashrc/.cshrc file (found in your home directory) or put in a shell script that you can source each time you log in to cheyenne.  Some exercises in this practical session involve loading different modules or versions of modules and/or editing namelists; you can use this page to reset back to the proper build/run environment for the baseline case.

A case is an instance of a MRW model simulation. It is determined by: 

• component set 
• model grid 
• machine 
• compiler 
• any other additional customizations 

Invoke create_newcase as follows from the cime/scripts directory using the following format:

This will 

• Create a case directory named /glade/scratch/$USER/UFS_training/ufs-mrweather-app workflow.c96 
• Use the GFSv15p2 CCPP physics suite, C96 grid and the UFS weather model with pre and post-processing steps 

The output can be seen here.  When the case directory is created, the build is done.

Click here to move to step 4.

Move to this case directory to set up and build the model:

This step creates the scripts needed to build and run the model along with the namelist:

Note the new bld and run directories under your case directory: 

• /glade/scratch/$USER/ufs-mrweather-app-workflow.c96/bld
• /glade/scratch/$USER/ufs-mrweather-app-workflow.c96/run 

The bld directory will contain the executables ufs.exe, ncep_post and chgres_cube.exe when ./case.build is executed (next step). The run directory will be populated when the case is run.  Note that chgres_cube.exe and ncep_post are links to the NCEPLIBS-ufs-v1.1.0  directory, since they are built with the NCEPLIBS.

You can see more details of the case setup by running:

This step builds ufs.exe in about 8 minutes: 

The last part of the output looks like:

Once the build has finished successfully, check bld and run to see the results of the build process.

Modify the runtime settings in env_run.xml for a low-resolution, 48 hour forecast:

• Reduce run length from 5 days to 48 hours

• Reduce wallclock time from 12 hours to 30 minutes

• Turn off short term archiving 

• For the Tutorial, there is a reserved queue:

Run the case: 

See https://ufs-mrweather-app.readthedocs.io/en/ufs-v1.1.0/quickstart.html for  more information on this command. 

After running ./case.submit, you will see: 

Monitor your job's progress in the queue:

If needed, you can kill your job:

How to tell if your job is completed: 

tail run/chgres_cube.yymmdd-hhmmss.log: DONE. 

tail run/ufs.log.<jobid>.yymmdd-hhmmss: PROGRAM nems HAS ENDED. 

tail run/oi.hhh: PROGRAM UNIFIED_POST HAS  ENDED.

Your outputs will be in the $UFS_SCRATCH/ufs-mrweather-app-workflow.c96/run directory. After a successful run, the directory will look like this:

For visually checking the results of your run, we have provided a NCL script that plots 2-m temperature (tmp2m) and total precipitation (tprcp) at 48 h.

You will need to load NCL first:

Then get and run the script:

To visualize the resulting images in png, use the command:

The sample plots are below. 

Now that you completed this step, the next step is to change a namelist option and run a second test to check your understanding of how results will change.  

You will need to create a new case directory for this experiment (we will use ufs-mrweather-app-workflow.c96_exp - note the "_exp" this time), and setup the new case. 

At this point, you can see the current values for CCN (ccn_o = 100 and ccn_l = 300) by looking at the input.nml file in the control run:

You will set the values through the workflow:

Use your editor to append the following to the entries at the bottom of the file user_nl_ufsatm

Next you will build the case:

Once that completes, change the run duration to 48 hours and reduce the time request. Then submit the case.

Once again, you can monitor your jobs in the queue:

And if needed, you can kill your job:

As before, you should see the output files in the case run directory for the experiment:

tail run/chgres_cube.yymmdd-hhmmss.log: DONE. 

tail run/ufs.log.<jobid>.yymmdd-hhmmss: PROGRAM nems HAS ENDED. 

tail run/oi.hhh: PROGRAM UNIFIED_POST HAS  ENDED.

Changing the concentration of CCN creates a noticeable difference in the total cloud cover. Now that you've completed both runs compare the total cloud cover before and after the CCN change.

We are providing a script that plots the differences between control and test runs for 2m temperature (tmp2m), total precipitation (tprcp), and atmospheric column total cloud cover (tcdc_aveclm) at 48h.

To get and run this NCL script execute the following commands. This script will produce png files for each variable being plotted. The argument "c=" is the control run directory, and "t=" is the test run directory.

View the total cloud cover:

Your NCL output of atmospheric column total cloud cover difference should look like this:

You can also look at the 2-meter temperature and precipitation:

Congratulations!  You've completed Session 1.

If necessary, follow steps 2-7 in Session 1 to rebuild and rerun the baseline case. This should only be necessary if files in the workflow have been deleted.

Data is available for several different weather scenarios.  Additional data may be downloaded and staged in your working directory, as described in the Users Guide.

First, to save any previous runs that you have completed, move that run directory to another location.  Start from your case directory,

Then, set the new run date (and init time if needed), again from your case directory:

For a severe storm case:

For Hurricane Barry: 

And, submit the new run using

Follow Session 1, step 7 to visualize the output.

There are 4 global resolutions supported by the UFS Medium Range Weather App.  For this training session, please use only the C96 or C192, due to computational (and classtime) limits.  The model resolution is specified with the “--res” option to the create_newcase command.

Starting at Step #3 from the Day 1 exercises, select a different resolution.  You can query available grid resolutions by:

Select a C192 grid, and create the new case directory:

Then, continue with the instructions for Steps #4-7 from Day 1.

Physics parameterizations for the UFS Medium Range Weather App are selected with a Suite Definition File, used by the ufs-weather-model.  This is configured as part of the “comp set” for a case.  The physics suite is specified with the “--compset” option to the create_newcase command.

Starting at Step #3 from the Session 1 exercises, select a different physics suite.  You can query available compsets by:

The Medium Range Weather App supports the two listed GFS suites, so select the GFSv16beta for this experiment:

Change to the new case directory:

Then, continue with the instructions for Session 1:  Steps #4-7.

Change input format from “grb2” to “nemsio” or “netCDF”

The MR Weather App requires numerous input files. The input files data format can be GRIB2, NEMSIO, or netCDF, and the input files must be staged by the user.  More information can be found in the Users Guide at: https://ufs-mrweather-app.readthedocs.io/en/ufs-v1.1.0/inputs_outputs.html

The input file data format is set using namelist parameters.

Repeat (or re-use) the case you built yesterday. 

Prior to Step #4, edit the file user_nl_ufsatm and add one of the following:

or

NOTE:  NetCDF formatted input files are ONLY available for the 20200202 case, so be sure to set the run start date if you select the netcdf input option.  Also, the netcdf input processing requires more memory, so adjust the processors-per-node as follows:

And, submit the new run, using:

In this exercise, you will set up and build the baseline case in a new directory, run it for 24 hours, then restart it and run for another 24 hours.

Start by creating a new case.

Note that the directory in $UFS_SCRATCH has the _restart suffix.

Continue, analogous to steps 3-5 of Session 1.

As before, it should take about 8 minutes to build the new case.  Once done, change a few variables:

Set runtime to 24 hours

Reduce wallclock time from 12 hours to 30 minutes

Turn off short term archiving 

For the Tutorial, there is a reserved queue:

Run the model for 24 hours

As before, you can use qstat to monitor your job's progress

If needed, you can kill your job:

When your initial job has completed, change CONTINUE_RUN to True

And the case will run for another 24 hours.

After it finishes, check your results with NCL:

To visualize the resulting images in png, use the command:

The sample plots are below. Click here to compare the restart run to the baseline run.

Exercises in session 2 may have changed variables or output.  To reset, follow steps 2-7 in Session 1 to rebuild and rerun the baseline case.  This is usually not needed, but added her for completeness.

Setup the environment for the case directory location.

For bash:

For csh:

Load the wgrib2 module on Cheyenne.

Move to the directory that contains the configuration files and scripts used for creating a custom flat text file read by UPP.

We recommend creating copies of the original *.xml and *.txt files that will be modified or created by this process so they are not overwritten.

This example describes how to add the field SFEXC (heat exchange coefficient at the surface) to the postcntrl_gfs.xml and postcntrl_gfs_f00.xml configuration files. Using your favorite text editor, edit each configuration file by adding the following parameter to the list at the end of the file, but before the </paramset> and </postxml> tags.

You may also delete an unwanted field by completely removing the parameter block of that field from these files and then continuing with step 5.

* Formatting is important; use other entries in the file as a guide.

** Adding fields requires knowledge on which fields can be output for your specific dynamic core and physics options. Not all fields available in UPP can be output with the UFS MRW application.

This example will describe how to add height agl levels to the SPFH field in the postcntrl_gfs.xml and postcntrl_gfs_f00.xml configuration files. Using your favorite text editor, edit each configuration file by adding the bold highlighted levels to the field ‘Specific humidity on flight levels’.

Leave no whitespace between the opening <level> and 20. and between 100. and the closing </level>

For the list of available levels for each level type, please reference the UPP documentation on ‘Controlling which levels UPP outputs’.

You may also easily remove levels from the <level> tag by simply deleting them from the list.

* Formatting is important; use other entries in the file as a guide.

In order to ensure there are no errors in your modified *.xml files, you may validate them against the XML stylesheets provided in the directory:

If the xml’s are error free, then the commands will return that the xml validates. (e.g. postcntrl_gfs.xml validates)

In the $HOME/UFS_training/my_ufs_sandbox/src/post/parm directory, create the flat text files.

Running make calls the perl program PostXMLPreprocessor.pl, which checks for any modified xml files and re-creates the flat text configuration files where changes were found. The new files generated for this example include:

In order to run the MRW post-processing with these new customized flat text files, you will need to copy or link them to the following location from the top level of your case directory.

Check that the workflow recognizes the flat text files in the new directory location by previewing the namelist before re-running UPP.

When running the workflow, CIME first looks in the ${CASEROOT}/SourceMods/src.ufsatm directory for the UPP parameter file(s) and uses them for the post-processing. If there are no flat text files present in this directory, then the default parameter file(s) in the pre-configured location are used.

UPP output from the initial case run (Session 1) is located in $CASEROOT/run with the naming convention:

If you would like to keep the previous UPP output to compare, copy it to a new name or directory so it does not get overwritten.

Rerun the UPP job to output the new field and/or levels.

You can check the new UPP output using a simple wgrib2 command. For example, to check that the new SPFH height levels were added:

You can also view additional diagnostic output such as min/max values. For example, to just look at the newly added variable SFEXC:

If UPP fails, the UPP log files, $CASEROOT/run/oi.HHH, can be consulted for errors.

UPP grib2 output is on the same grid as the model output, which for the MRW is a Guassian grid. The specifications for a simple lat-lon projection are:

-new_grid latlon lon0:nlon:dlon lat0:nlat:dlat

Move to the run directory and run the following example.

This would interpolate to a 0.25 degree latitude-longitude grid with earth-relative winds (U-wind goes east, V-wind goes north).

You can check the new grid by again using wgrib2. We will print out the info just for the SFEXC field rather than list all fields.

For more examples, you can refer to the wgrib2 site.