CCPP Framework Meeting Minutes 2019 08 01

Metadata conversion

• all typedefs in FV3 now converted, also one scheme
• uses new metadata parser and converts to old metadata format
• workaround to compose the full name of the variable as it is available from the host model, this information is now in the ccpp_prebuild_config.py
• next to convert schemes (should be straightforward)
• what about dimensions Model%levr+1? convert to Model%levrp1 (= Model%levr+1)
• for the moment new metadata parser is used only for parsing the metadata file
• a follow-up step will be to also parse the Fortran code and fix the inconsistencies between metadata and Fortran
• need to bring in updates of metadata parser code and make sure it still works
  • this metadata parser will help creating metadata templates from plain Fortran
  • do this as step 2 after step 1 (convert all schemes to new metadata standard) is completed

CCPP-CPF migration at NCAR:

• MPAS folks now working on one MPAS-WRF suite

Standard names:

• ... 5 minutes of trying to figure out presentations in goto meeting ...
• extracted all standard names from GFS_typedefs.F90 to see how those can be used for the YSU PBL scheme in MPAS
  • targeting the 2D version of the scheme, mapping from current standard names (derive from others if needed)
• the community needs a defined list of names, and how this list will be governed
• the standard names should also be used for netCDF output etc, hence follow-up processes will rely on these
  • this is a requirement for CCPP informing the host model about diagnostics to output
• may need the capability to convert a standard name into a netCDF variable name if not provided by user
• standards for netCDF output are CMIP/CORDEX names, but these are for climate and don't really make much sense for other applications
• CCPP to provide a library that translates CCPP standard names into output names for different applications/standards (CMIP, CORDEX, MET, UPP, ...) - can be an external repository/tool?
• HDF5 had the capability of aliases - did netCDF incorporate this?
• ...
• need to keep these issue separate: (1) standard names and their governance, (2) if and if so, how to standardize output names of diagnostic variables coming from the CCPP that the host model does not know about
  • haven't agreed yet on whether to keep expanding standard names or adopt families of standard names in conjunction with attributes