CCPP Framework Meeting Minutes 2019 05 30

November 5-6 meeting between NOAA, NCAR and NRL

• organizers Stan Benjamin, Chris Davis, Jim Doyle
• interoperability and physics, more scientific and not just technical

Can Dave give his WRF workshop presentation to DTC?

• WRF workshop in two weeks, after that
• Ligia will organize a seminar

Update of automated regression testing scripts for NCAR folks required?

• Dom to check and fix

List of standard names

• Email conversation May 29
  • From Dave: Is there an official list of standard names that are being used in CCPP?
  • Dom: Steve has been working on this lately. The best you can get from GMTB at the moment are the unsorted GFS_typedefs.F90 metadata tables in the NCAR FV3 repository, branch gmtb/ccpp (https://github.com/NCAR/FV3/blob/gmtb/ccpp/gfsphysics/GFS_layer/GFS_typedefs.F90)
  • Dave: Thanks for the pointer. For an internal NCAR model group (WRF, MPAS, CESM), I grabbed this file and sent it off via email. People are interested in how we are going to construct a uniform name collection for these diverse models (NCAR + NOAA). Perhaps one of the items that we can discuss at the Thursday meetings (that I don't attend) would regard an organized method to produce a single source (likely some sort of online list), and of course complete with governance, etc.
• Parallel meetings at NCAR regarding this topic
• How to avoid that people are using different names for the same things, or the same name for different things?
• udunits approach (not having to have unique names, but ways to construct names where different names can be identified as being "the same")? Would that help the community?
• Too big of a job to standardize all the names in a dictionary?
  • Some groups are highly concerned about certain aspects of the names, others are not (e.g. mixing ratio dry or not, ...)
  • MMM has designated some non-CPF people to do some of the porting to CPF, this can be challenging since they have not been involved in the design and may have different ideas
• NUOPC field dictionary group (semi-active), how close will we be working with them?
• Chemistry has a standard way to construct names describing all the complex chemical species
  • they will soon be contributing hundreds of names to CCPP
  • Steve to get this list from them and add to the current list of standard names (see below)
• Steve has a tool to create an XML document of standard names for ccpp-physics (GFS physics)
  • Need to publish the XML document, right now it is sitting in Steve's fork
• There must be a clear, sanctioned process for how to find or construct standard names
  • Important to identify which parts of variable properties should be in the standard name, which a variable attribute; this must make sense to physics developers
  • Standard name based on the unit
  • Example: if it is a temperature then all flavors of it (soil, interface, ...) still have the same standard name but different attributes; a temperature tendency has a different unit and should have a different standard name?
  • Variable families, build relation between air_temperature and tendency_of_air_temperature, also total_tendency_of_air_temperature as the sum of tendencies from different groups in a process-split group
    • Diagnostic output still wants the individual tendencies, not just the total tendency
  • Dave has given this some thought and will write it up to discuss it with the group
  • Vector quantities such as wind fields are used differently between the models and will be tricky
• Create a database from the XML, write a front end that makes it easier to search or create variable names (i.e. deterministically construct a name)
• Are we going to try sticking with the CF conventions or move away?
  • Our syntax of standard names is the same as CF
  • CF doesn't have variable attributes
  • Maybe need capability to generate a CF name from our standard names and attributes