access_JEDIAtNCI_BoMMOJEDIMeeting20201209 - ACCESS-NRI/accessdev-Trac-archive GitHub Wiki

BoM/MO JEDI OPS meeting, 9 Dec, 2020

Present: Chiara Piccolo, Stefano Migliorini, David Simonin, Douglas Boyd, Peter Steinle, Fiona Smith, Andy Smith, Jin Lee, Susan Rennie

Aiming to deliver NG-OPS by March 2023
- to be used in 3DVar UM-JEDI and 3DVar LFRic-JEDI
- assumption is that LFRic and NG-DA will be implemented together
- plan to use ODB
The NG-OPS and NG-DA projects are driven by supercomputer changes
- avoid porting UM
- Plan A: 4DVar using LFRic-JEDI and hybrid
- Plan B: 4DEnVar and possibly EnKF for LAM

Best if partners take on workpackages which are not in the critical path
- work undertaken can be considered in stretch packages
3 possible work areas,
1. Currently C++ JOPA calls Fortran. Helpful if all Fortran can be replaced with C++
2. FSOI capability
  - UKMO is not planning to implement FSOI
  - depending on whether the global DA uses 4DVar or 4DEnVar the implementation of FSOI would be radically different: adjoint-based in the case of former and ensemble-based in the case of the latter
3. Currently the 1DVar used in radiance processing is only an interface to Fortran. Useful if Fortran is replaced with C++. This is not entirely a coding exercise,
  - UKMO would like to use OOPS 1DVar
  - implement different minimisation method: e.g. Levenberg-Marquardt
  - Instead of processing column by column there might be a way to process all at once
  - TL and Adj need to written in C++
  - desirable to merge radiance and GPSRO 1DVar processing
Already Jorge Bornemann (NIWA) has taken on a piece of work
- will start in Jan, 2021
- working on
  - IAU
  - LFRic-JEDI interface to Fortran (?)

Chiara to ask Keir if NG-OPS and NG-DA project board meeting can be shared
Adam Maycock in charge of TIDA (?)
To improve communication with partners about NGMS we need to talk to Keir
Meet again in Feb/Mar before Partnership Board meeting
- BOM may be in a position to give available resources