OSVAS

Step 3: Configure and run the simulations

For each experiment defined in OSVAS_steps.Expnames:

• The launcher loads SURFEX installation settings from config_files/SURFEX/surfex_configuration.yaml. Shared settings are defined in general, with optional overrides in the selected runner’s platforms.linux or platforms.atos block.
• Optional top-level blocks named exactly after an experiment can override individual settings for that experiment. For example, a MY_EXPERIMENT block overrides matching keys from general and the platform block; keys omitted there continue to use the inherited settings. These overrides belong in surfex_configuration.yaml, not the station YAML.
• A run directory is created at RUNS/${STATION_NAME}/${EXPNAME}/run/.
• Forcing files are linked from forcings/${STATION_NAME}/ into the run directory.
• Required physiography files are linked from the SURFEX setup and local physiography paths.
• The station-specific template namelist namelists/${STATION_NAME}/OPTIONS.nam_${EXPNAME} is copied into the run directory as OPTIONS.nam.
• The launcher updates NYEAR, NMONTH, NDAY, XTIME, and CFORCING_FILETYPE in OPTIONS.nam to match the configured forcing period.

The SURFEX steps listed in OSVAS_steps.Surfex_steps are then executed in the run directory.

In the shared configuration, surfex_exe and paramfiles may be relative to surfex_home or specified as absolute paths. physiography_paths is an optional list of additional directories linked into each run directory. See How to make your SURFEX installation known to OSVAS for the configuration layout and an example.

If Initialization_data.Init_to_namelist is true, soil initialization profiles are applied to OPTIONS.nam before the first SURFEX step. If Initialization_data.Init_to_prep is true, profiles are applied to PREP.txt/PREP.nc immediately after PREP completes. These initialization operations are implemented by scripts/python_scripts/apply_soil_initialization.py.

After execution, important output files are moved or copied from the run directory into RUNS/${STATION_NAME}/${EXPNAME}/output/.

This current workflow therefore supports:

• reusing validation-derived initialization profiles for SURFEX runs,
• running only part of the SURFEX chain and then applying initialization before continuing,
• keeping both raw run directory contents and a cleaned output directory.