We held a breakout session on Australian suite development in the RNS at the 2024 ACCESS-NRI workshop. This thread is to document the discussion and actions following the session.
The session discussed the means by which we should share suite development between projects. Committing useful changes to the RNS trunk was agreed to be the best way forward.
A table of existing model suites and their respective developments is collated below. Please update to fill in the blanks.
Actions
- Share RMED working practices on committing change to the nesting suite trunk
- Investigate whether KGO tests can be fully run on Gadi
Desirable existing functionality to be shared more broadly/to target trunk
- Chermelle’s updates to ERA5-to-grib (massive speed-ups due to parallelisation)
- BARRA-R nesting as per HM suite
- Self-servicing of BARRA2 analysis dumps & BARPA periodic dumps and hourly LBC frames from NCI Massdata archive
- SST updates via ancillaries as per BARPA/HM suite (ancils created using u-cj399, branched from an ancil suite in 2017 by Simon Tucker)
- SU usage monitoring BARRA/BARPA/HM suite
- Shortstep retry runs as per BARRA/BARPA/HM suite
Desirable functionality to be developed
- SST ancillary generation moved into suite cycling (not yet implemented)
- Netcdf conversion (multiple versions exist, more discussion needed to identify useful
- Land ancillaries under development in Bureau HM team (using land ancillary suite)
- Archiving: including sending ancillaries to joint location on tape
- Log file archiving
- Spectral nudging as per BARPA-C (also requires code to be committed to trunk)