CCPP Framework Meeting Minutes 2019 12 12

Agenda:

• discussion/update on outputting diagnostic variables from physics (no time left...)
• discussion on how to handle redundancy between suite definition file and model namelist

Suite definition files and model namelists:

• choosing physics schemes is responsibility of the suite definition file, but changing the number of tracers (e.g. turning graupel on or off) is a runtime configuration that affects both the host model and the physics
• namelists specific to schemes only will be difficult (some values may affect multiple schemes)
• to address most pressing issues for UFS, ccpp_prebuild.py could translate the choice of scheme into the integer/logical flag that sets the value instead of reading it from a namelist; generate an error if that parameter is still in the namelist
• a lot of this will need to be coordinated with EMC and GFDL (e.g. the way tracers are managed using tracer_manager and field_table)
• is it possible to enforce standards on how schemes handle runtime options? For example, schemes can not read namelists but gets its values through the host model or the suite? This would be a big change for CESM, less so for UFS (MPAS? WRF?). Need a survey to see what physics developers are saying.

Misc:

• different suites for parent and nest? need to check where we are with this at the moment
• move towards using the project board for ccpp-framework