INFRA (Evan):
CI migration for space stack containers, etc. completed.
Ashley completed script to retrieve TEMPO data (from S3), TROPICS, GMI ingest work as well.
R2D2 ingestion for NASA ocean biochemical model and some fixing of Var-BC files.
ALGO (Clementine):
Upcoming coordinated merges:
- BUMP update: https://github.com/orgs/JCSDA-internal/discussions/136, affecting all model repos that use BUMP; target merge date: June 10th
- Improvement to variable change test (no announcement yet, but there may be one by tomorrow am): https://github.com/JCSDA-internal/oops/pull/2662, requires updates from model interfaces and improves on testing. Tentative merge date: June 12th (unless Mayeul makes an announcement and proposes a later date)
- Improvement to oops::Variables (update from Steve V): work in progress, and given how invasive these changes are we would like to merge them as soon as possible. We'll communicate as soon as model interfaces can start to adapt and provide examples, tentative announcement some time next week. (@Steve V please comment/add/correct on this if needed).
List of updates from Steve H and Anna:
- ioda: work in progress on adding functionality to append ioda ObsGroups which will be useful for the continuous DA as well as for reading multiple input files into one ObsSpace
- ioda/ufo: work in progress on adding functionality to reduce ObsSpace as a filter "action" that can be used with pre-filters (e.g. to reduce memory footprint after the thinning filters were run). OBS team has been testing this functionality and reported first successes with H(x) application; more testing and rigorous comparisons under way.
- work on improving oops::Variables class in a mini code sprint (I'll defer this update to Steve V).
SteveV:Working on a code sprint relating to the variable renaming project. The variable renaming project has taken a detour into a refactoring of the oops::Variables class. At a high level, we are changing it from basically vector of strings to a vector of Variable objects that contain well-defined metadata. This will allow us to have generic JEDI code be generic enough to handle different model practices, without ambiguity about what variables are being used. But it is an impactful change to JEDI since Variables are used all over.Nate:
We merged an update that now requires interpolators used in SABER block chains to explicitly specify forward and inverse interpolation methods, and a bug fix so that backgrounds & first guesses are passed directly in the block chain. I've also been spinning up on spectral covariances.Christian:
Christian and Clementine presented JEDI related work at the adjoint workshop in Lake George, NY last week. Clementine presented on JEDI in general as well as Block EDA methods. Christian presented on the HTLM in JEDI as well as some new results using it with GEOS related to atmospheric chemistry. Results show good reduction in linearization error in most all cases and for those, improved o-a when doing 4dvar as well.Two PRs are open allowing the reading of an ensemble while using the full nonlinear model as the nonlinear control member to calculate update coefficients. One from Tom Hill at the met office is the likely choice to merge and expand the modes one can use the HTLM with.
Clémentine:
Work on GEOS build with NASA modules and spck stack modules, able to build with JEDI now too
Hope for release of GEOS-gcm modules+all submodules component soon (common between us and NASA)
COMPO (Jerome):
Worked with Ashley on the acquisition of TEMPO data & sample demo (3DFGAT ~25km spacing)
Maryam finalized HTLM experiment with chem/atmos
GEOS-CF 4dvar cycling; next step to cycle ens
NO-NAME (Francois H):
Generic interpolators in OOPS; also SABER
Optimization work and handling of edge cases
Marine (Hamden):
Bug in HofX (multi-channel data); R2D2 needed a fix; Skylab Exp PR associated with it.
Generic diffusion operator in SABER.
Kriti working on HTLM and fixing a Skylab experiment.
OBS (Hui):
a) Ingest:
- Satbias converter for some MPAS files needed a bug fix.
- TEMPO (v3 updates) and SMAP converters completed recently.
b) UFO:
- SOCA bug fix
- ATMS hydrometeor check also had a bias correction applied (which needed updating for naming convention changed)
- Saturation specific humidity filter had to be changed; allow optional since not all model interfaces have GeoVaLs with this variable.
- Discussion #135 needs participation by others - observations near model edge or outside a limited area domain; also 2 PRs already created.
- New Space Weather effort (Navy is major in-kind) - will require a new operator for total electron count
CRTM (Ben):
Some work to improve how UFO interfaces to the older CRTM versions within develop.
Requests for aircraft platform observation altitudes. Currently no TL/AD support built in.
Reminder S4 account updates - contact Ben.
Improve land-surface emissivity particularly for UV/Vis - asks others in the land DA groups to provide up to date info.
EMC (Dan):
GFSv17:
- LETFK marine work, QC ocean filters
- B-matrix work for aerosols
- Ensemble centering and infrastructure for snow
- Time-series plotting with EVA
- JEDI cycling on WCOSS2 now working
Regional (RRFS v2)
- MGBF efforts in SABER
- PR adding VADER to MPAS-JEDI
- Rejecting obs external to domain (discussion #135)
Atmos DA:
- LETKF memory issues (CRTM simobs)
- Full cube-sphere data pipeline (can we remove FMS I/O from fv3-jedi?)
AI/ML:
- Interface to MAGICK model
- Adjoint sensitivities for GraphCast
UFO:
- Some OBS work with going upstream of prepBUFR for conventional data
- thinning SatWinds
- GOES ABI
ROMS (Hernan):
Working on diffusion operator mostly
UKMO (Phil U.):
When comparing Obs-Analysis between JEDI and existing UKMO results, need more diagnostic outputs.
GMAO (Jianjun):
Emily L. helped with tropopause pressure problem, which turned out to be a bug in GMAO version of GSI.
Navy (Sarah):
Wrapping up some Skylab-v6 testing; some model physics changes prior to moving towards more recent Skylab releases
LETKF efforts
MMM (Jake):
MPAS v8.1 - good progress migrating into JEDI (multi-week effort expected).