Tutorials

In the previous tutorial you used COMPS to produce forecasts for several different configurations. In this tutorial we will see how we can set up COMPS to run new configurations.

The forecasts that COMPS produces are governed by a set of files called namelists. There are 4 types and work in a hierarchy: At the bottom is a namelist that defines all post-processing schemes available in COMPS. The level above (the configuration namelist) defines which schemes are combined together to form configurations.

In part 1 of the tutorial, we ran 4 configurations for temperature. In an operational setting you may want to produce forecasts for several variables, such as precipitation and wind speed. Different variables often use different configurations. Therefore COMPS has a way to assign configurations to variables. The varconf namelist (short for variable-configuration) defines which configurations go together with which variables.

Finally, the runs namelist defines which variable-configurations to use. Here the input datasets are specified as well.

All namelists are read at runtime. Therefore COMPS does not need to be recompiled after changing these files.

In addition to the 4 hierarchical namelists, there is a 5th namelist, which stores definitions of variables. In most cases you will not change this file, as it merely contains information such as minimum and maximum expected values for each variable.