CCPP Framework Meeting Minutes 2019 05 02

NEPTUNE progress

• now runs with CCPP, Matus starting to experiment with Thompson MP
• Alex's help request: control for tracers, where are the variables allocated; this is not yet possible with our code
  • this is something we (CGD) are developing to some extend at least
  • current status: allocation of variables is part of auto-generated code
  • how to advect tracers that are allocated/defined by auto-generated code?
    • will have to be implemented in a per-host-model basis
  • needs a module or some other way to transfer information with number of tracers and which ones to advect, mix horizontally, deposit, ...

AIR WORLDWIDE

• will visit NOAA GSD end of July and meet with Ligia and Dom

Progress with CCPP transition to EMC VLab master

• currently held back by the fractional landmask commit, which faces b4b issues on WCOSS IBM platforms
• Dom to check if CCPP can jump the queue and go in first, since CCPP port of fractional landmask is done by GMTB anyway

Transition to new metadata parser:

• July timeframe
• use ccpp_prebuild.py with new metadata parser first, later capgen
  • capgen requires getting rid of the dynamic builds
• concern by program managers that pace of changes is too quick
• need to understand how new, internal data structure looks like with the new metadata parser

CCPP release with SCM in June, another release with NEMSfv3gfs around October

• need to define which developments go into these releases

New cap_gen:

• missed the target to have the upgraded cap_gen running by end of April
• new cap_gen will be much more powerful

Other developments:

• GMTB needs to demonstrate capability to flip arrays and convert units by June this year; either do this as a proof of concept with new cap_gen (won't be useable by NEMSfv3gfs/SCM until switch to cap_gen is made) or engineer it in current ccpp_prebuild.py system (potentially using some of the new Python modules from cap_gen)