...
GFS v17: Retrospective carrying on. Thanks to the core team for release branches; will tag when stable.
Configuration: Adopting JCB paradigm for YAMLs (HAFS and RFS systems) to separate configuration YAMLs from workflows.
ROPP Operator Comparison:
Running a series of $HH(x)$'s comparing RMS and Bias.
ROPP 1-D is not as good as others.
ROPP 2-D looks the same; need to check increments.
Discussion (Ron & Dan): Ron noted that at GMAO (ROPP vs. NDAM in OSSE), they usually didn't find differences in similar tests (no cycling). Dan noted that 2D wouldn't work in a GSI context but might show differences in cycling (which hasn't been done yet).
...