INFRA - Dom

- JCSDA-internal and JCSDA github organizations will require Github 2FA beginning of October
- https://github.com/orgs/JCSDA-internal/people?query=two-factor%3Adisabled 
- https://github.com/orgs/JCSDA-internal/discussions/64
- Does only affect logging into github and performing security-relevant options, not day to day work
- Users who don't have 2FA enabled by then will automatically be removed from the organization
- Heads up: we will do the same for our AWS accounts in the future
- ioda parallel reader PR ready for testing
- https://github.com/JCSDA-internal/ioda/pull/1071
- Caveat: can't write to single file (hangs; investigating), workaround is to switch to write to multiple files, see description in PR
- Work continues on the ingest suite and the new CI system (onboarding more repositories this/next week)
- Resumed work on spack-stack to prepare for the release Sep 9+/-
- Includes new bufr@12 library


ALGO - Anna

Nate Crossette joined JCSDA last week to work on the background error covariances.

Upcoming updates: 

Development:

  • 4D covariances (4DEnVar and weak constraint 4DVar): refactoring of saber block chains to be merged this week (shouldn’t affect model interfaces, although small effect on lfric-lite-jedi). Coming up: passing 4D fieldsets to the block chains. Next week: handling cross-time covariances and localization in saber.
  • Merged: option to write out eigen vectors in lat-lon format.
  • Merged: capability to read and write HTLM coefficients to and from disk (with MetOffice)
  • Kate is making good progress on the control Pert EDA app, we have been finding and fixing issues this week. The workflow is ready for when we have a reasonable app to test.
  • Clementine revived 3DFGAT-UFS branches with latest develop / changes
  • Looking into how we can use the new Atlas changes to simplify and speed up HTLM (with MetOffice)
  • Ongoing work on mpas-skylab, EDA comparisons, atlas interfaces, B training workflow.


OBS - Hui

Ingest:
(1) UFO has been updated with "sensor scan" being an integer (previously float). This update is consistent with the obs naming convention. This update also led to updates in multiple repositories related to data files used by the relevant code. Now develop is up to date with this change.
(2) OBS team plans to enforce the naming convention in the following month in coordination with in-kinds. Currently, the naming convention check only gives a warning at this point.
(3) The OBS team is working with GSD and EMC to convert snow cover and SMAP soil moisture data. Some initial tests were done, but we need more discussions/development.
(4) EMC PR for bufr converters is dependent on the BUFR lib v12, which will be available in Sept (as Dom announced) as part of the new Speck-stack release
UFO:
(1) VarBC code sprint is planned for the week of Sept 12. Active discussions are ongoing with EMC, JEDI ALGO, and UKMet. The primary purpose is to allow current VarBC to be applied to non-radiance (channel-based) data. Three PRs (two in UFO and one in IODA) are now in review to allow the coefficients to be used for records (previously being constant cross records). 
(2) Discussion with EMC regarding radiance to brightness temperature variable change inside UFO. It brought us to the attention that new metadata, "wavenumber" or "frequency," needs to be a function of ichan instead of using it as a function of ichan and ilocs. UFO#2981 has discussions among a few groups.
(3) ROMEX skylab is being set up, which will test some new features from the ALGO development (GSI B and FGAT)


CRTM - Cheng

CRTM v3 PR with snow emissivity coefficients

Notice need to run CRTM standalone, so working on that.

Ben is testing coefficients.

Trying to hire to backfill Patrick Stegman’s position


Compo - Jerome

- Working on perturbations within fv3-jedi
- Working on adding tracers in fv3-lm
- Working on adding a forecast score task in skylab 


SOCA - Travis


- Kriti has been adding insitu T/S QC filters into skylab. After 1 or 2 more sprint cycles we should be close to a feature set that should be "good enough" for EMC operations
- Doruk: potential=>insitu variable changes have been added to VADER pending PR approval. We will be adding the linear variable change in the near future
- I have been working on a small-scale correlation operator using an iterative gaussian kernel filter. This will be useful in the ocean at high latitude where the correlation lengths are very small and BUMP struggles. This code is SOCA specific for now, but after the atlas mesh connectivity is implemented, I'll try to find time to generalize this into SABER


EMC -

Tuning offline hybrid EnVar

Encouraging results in cycling with JEDI compared to operational system. “On same planet”


GMAO - Ron


Met Office - Phil Underwood

JOPA processing of obs look promising

HTLM PRs have been merged


NRL - Sarah

Having model failures with long (6 month) cycling runs

Forecast model updates

New Navy HPC Nautilus is in service


ROMS - Hernan

Debugging 4dvar

Trying Advanced features of BUMP

BUMP can be slow

Using hybrid solution for West/East coasts of US.

Problem on West coast


NCAR/MMM - Jake

Robert testing 4Denvar with MPAS-jedi. Time localization made a good improvement.

  • No labels