No special announcements.

Algorithms (Anna):

Upcoming merges: Monday Oct 2nd, removing some Parameters from State class: https://github.com/orgs/JCSDA-internal/discussions/67 (most model interfaces didn’t need an update, but please check).

Merged:

  • Workflow for coupled H(x) in ewok (tested with artificial data; currently waiting on the more realistic data).
  • First skylab-mpas experiment (3DVar at 120km res) merged into develop.
  • 4D covariances for the static covariance component that is defined in the model interface (affects toy models, soca, and maybe NRL) for the 4DEnVar.
  • Support for output to pressure levels in the generic lat-lon grid writer (for the scores)
  • Bugfix in saber (by Chris Thomas, Met Office): added background and first guess copies so that bg and fg can be changed during error covariance calculation (e.g. resolution change or variable change) without affecting minimization.

Work in progress:

  • Adding tasks to ewok to compute scores on pressure levels
  • Read of 4D ensembles in saber for the 4D covariances
  • Looking into GSI block 4D aspects
  • Data for the 4DEnVar and EDA experiments.
  • Next two weeks focussing on technical debt; one of the priorities is revisiting fv3-jedi tests (faster and better testing).

CRTM (Cheng):

  • Chneg worked on a generic interface for crtm in the past 2 weeks. It is going to be finished by the end of October for others to test.
  • Ben worked on creating release branch for crtm.
  • Fabio found an issue with the release branch; issue fixed in release branch.
  • Working on the science part of CRTM.

Q&A for CRTM

Tom: What is the generic interface?

Cheng: it is separate from current traditional crtm interface, won’t require any changes from current users. New feature within crtm which allows users to use their optical profiles as input for crtm. Highly requested by users; want to use their data before using it operationally.

Hernan: What if we are interested in using optical properties for ocean models (biology). How can we do that?

Cheng: Currently only surface emissivity is possible; the best option is to convert it to surface emissivity. 

OBS (Hui):

  • VarBC code sprint 2 weeks ago (allow VarBC to work with other types of data than satellite, e.g. aircraft) and met on Tuesday to follow up. Still some issues to finalize. Discussed applying naming convention to the bias correction predictors. 
  • Testing converters for adpsurf and moving forward with adpair.
  • Sam is working on TOPICS radiance data and has initiated skylab tests with a basic QC filter
  • Hailing and Hui working on ROMEX; setting up jedi-ufs workflow for data impact studies.
  • Pat working on using QC flags in the obs operators.
  • Fabio helped bugfix crtm/v3, which caused a crash during the GEOS ABI moisture channel data assimilation. 
  • Lots of changes coming due to applications validations of obs operators.
  • Francois V: NASA RO validations between GSI and JEDI work. Small differences in QC and obs errors – only differences are in QC currently. Difference mainly due to super refraction treatment.

COMPO (Jerome):

  • Working on release of skylab-v6: increased resolution of skylab aero from C90 to C450 (C270) with stretch grid. Switch to 3D-FGAT with PseudoModel. This will allow COMPO to take advantage of TEMPO data at good resolution. New release will have TEMPO pseudo data, stretch-grid and 3D-FGAT.

SOCA (Hamideh):

  • Hamideh has been working with Lionel Artega from GMAO on a NASA PACE science and application solicitation proposal. Also She has been preparing observations for testing OASIM.

EMC (Cory):

  • Still working on finalizing VarBC code sprint.
  • Longer term planning on end-to-end bufr-ioda direct conversion. 
  • Guillaume and his team working on converting the converters from Python to C++.

Q&A:

Dan: One of the effects of possible shutdown on GMAO is that there will be no one available to update their bundles but the contractors will still be working on development. If possible, could the code remain working with the libraries used in spack-stack/1.4?

Discussion: yes short term, would be harder to do longer term, but JEDI team can work with GMAO and EMC contractors if bundles need updating. 

GMAO (Ron):

  • Completed UFO testing (waiting on the RO validation)
  • Moving to the evaluation of VarBC for satellite radiances, and participating on the work for the code sprint
  • Looking at the channel correlated error for hyperspectral instruments.
  • Ricardo running cycled 3DVar test with near complete obs systems; looks like most things are within the expected range.

NRL (Sarah):

  • Sorting out what to do with the missing values with the geovals. It seems that different operators handle missing geovals differently; and it probably should be standardized. NRL would like to proceed with missing geovals in the obs operators. Working to fix missing geovals long-term, but near term we would like the system to work with them.

ROMS (Hernan):

  • Update to the Parameters to the State class.
  • Using spack-stack; have roms connected to ufs, looking into coupling roms and fv3

MetOffice (Tom Hill):

  • JEDI observation processing is signed off by the JOPA project board, but still needs to be signed off by the Next Generation Modelling Systems program board, expected today.

Infrastructure (Dom):

  • Github 2FA comes into effect on Monday (everyone who didn’t set it up will lose access to jcsda-internal).
  • spack-stack/1.5.0 installed on all platforms, documentation updated, release will be pushed out likely today.
  • On NOAA Parallelworks Azure platform, aws cli doesn’t work (and only there)
  • Switched the CI system to new spack-stack/1.5.0 (both old CI and new CI)
  • New CI now on oops, ioda, vader, saber, soca, fv3-jedi!
  • Release: ready to tag the repositories; run the tests and experiments for the release, and once those are finalized release is out.
  • R2D2 improve handling of different job types, and split compute host/data stores
  • Ongoing work on data and obs ingest suite
  • Ongoing work on parallel reader
  • Technical debt sprint

Discussion: handling optional components within JEDI, presentation by Steve H:

 

Q&A:

What is the main goal of this?

Steve H: simplify handling of optional components; start testing different configurations.

Dom: obscure finding of packages can be a problem; with this proposal the configuration will be at the bundle level.

  • No labels