GENERAL
- Quarterly review coming up in a few weeks, please send relevant information to JCSDA leads
COMPO
- Finalizing 4dvar with tracers, working on a few bugs
- Evaluation with metplus, prototype of scorecard
- Technical debt sprint: figuring out ingest of geos ensemble with meteorology and aerosol for coupled analysis (low-res ensemble), for 4denvar/html testing
SOCA
- Debugging hybrid 4DEnVAR capabilities, before we launch regional soca experiments
- last project outside of Skylab, then switch over
- Wrapping up explicit diffusion correlation operator in SOCA, it seems to be working, serves as an alternative to BUMP for small length scales
- Currently specific to soca grid, after atlas updates go in (need mesh information) can generalize and add in saber
OBS [INGEST]
- Met with PI on tropics mission at MIT lab, currently 5 tropics sensors in orbit, soon 10+ (tomorrow.io)
- GSI-like implementation of GEOS GNSSRO configuration is now part of the develop branch and used at NASA
- Snow depth from in-situ stations (3 networks: ghcn feed climo (skylab v6), real-time from GTS (ADP surface snow, snow cover file)
- GMI and AMSUA-2 new data types in skylab v6 (converters from raw data files)
- Python API to NCEP bufrlib - PR to move them to a different place in the code
- GNSSRO updates, verified by NASA (final piece needed by NASA for now), sorting out a few issues with NRL
- With MPAS (low model top - 30km); n-band operator (pass in switches to account for low model top to make bending angle work)
- Time window object PRs (meeting next week Tuesday where this will be introduced)
- qc flags in oops, passing flags to all UFO operators (PRs in oops, ufo; touch all operators) - forces it to not compute h(x)
- Code sprint on VarBC: concluded, finish up final I/O bits (coefficient I/O)
INFRA
- Technical debt sprint
- Fixing ctest issues
- Fixing a few spack-stack issues for upcoming bug fix release 1.5.1 beginning of next month (has bufr 12.0.1 and ecmwf-atlas 0.35.0)
- Data download PRs - planning a merge on Oct 20, please update your bundles similar to the PR for jedi-bundle
- Final ioda parallel reader PR to go in this week, a few bugs need to be sorted out afterwards (note that the old reader is still and will remain the default for now)
- Working on R2D2 compute host / data store logic updates, more CI testing (added soca and coupling), Terraform Github management
CRTM
- Converting v3 build from ecbuild to cmake
- Finalize feature for netcdf interface for RT solution
- JEDI can try for CRTM tests - output all test results in netCDF format
- Output TOA and reflectance/albedo for solar channels
- Will enable DA of more remote sensing data
- JEDI Interface for CRTM
- Allow users to input their own optical properties
ALGO
- Upcoming changes:
- Oct 18: Bugfix to saber diagnostics output that may affect test references: https://github.com/orgs/JCSDA-internal/discussions/76
- Beginning of November (after spack-stack/1.5.1 release): begin to require the eckit/1.24.4 and atlas/0.35.0 https://github.com/orgs/JCSDA-internal/discussions/74
- Updates merged:
- Updates to oops and other repositories to silence compiler warnings coming from boost headers, and address other warnings, obsolete tests, etc.
- Bugfix for 4DEnVar without parallel time decomposition and using linear variable changes for B.
- Skylab experiment with the GFS 3-hourly data; next is analyzing and improving on the setup
- Vader algorithm improvement: not calling recipes multiple times when not needed (should cut down on time in applications that use vader)
- Vader log improvements
- Work in progress:
- Better log management: env variable OOPS_INFO controlling which processors to output from https://github.com/JCSDA-internal/jedi-docs/pull/671
- Fv3-jedi test cleanup: goal is faster testing, better coverage
- Updates to the documentation: adding documentation on saber block chains
- Adding task on computing forecast scores in the ewok and coordinating with Met+ work in the Compo team.
- Looking into reducing the amount of plots that are published on experiments.jcsda.org
- Bugfixes to ensure that JEDI code works both with atlas develop and atlas 0.34.0.
- Handling of D-grid and A-grid winds in fv3-jedi: working towards not exposing D-grid winds to the generic interface layer (saber/vader/geovals)
- Looking into refactoring the random number generator use in jedi (defining requirements)
- Updates relevant to Algo project from Met Office:
- Tom Hill from Met Office for HTLM tests in oops and fv3-jedi
- Diagnostic computation of quadratic cost function in DR minimizers (Tsz Yan Leung)
EMC:
- Preparing for end-to-end (bufr to fv3-jedi) code sprint end of the month
- What about mom6 / cice5 / gsw forks at JCSDA - plan to abandon those? Dom: related to UFS work, need to merge updates into authoritative code first
NRL: n/a
GMAO
- Merged RO capabilities into swell
- Implementation of channel-correlated error
- Getting cycled version from Ricardo to perform (JEDI-based version of NASA database in swell env)
- Working on tropics in JEDI
UKMO
- Global obs: achieved KPI for operational implementation
- Finishing odb write by using a second template
- 3dvar cycling suite: fixed saber problem, still seeing large different of current system with JEDI 3dvar
ROMS
- Issue with nonlinear trajectory, don't get same checksum when doing offline vs in JEDI
- Afterwards, look into interpolating trajectory to coarser grid so that inner loop can be on coarse grid
MMM
- Porting to Derecho, not fast enough
- Can now build with Intel (as opposed to Cheyenne)
- Preparing annual report for Air Force
- Improvements from 4denvar