Identify impact of global_cycle time shift

[ClaraDraper-NOAA]

This PR fixed a bug in the time given to global_cycle with IAU turned on.

We need to understand the impact of this PR, by running test cases (for each of the 4 daily cycles) and checking which variables are changed by this bug fix, and whether any of these changes are significant.

[jiaruidong2017]

• (See Help debug the global_cycle crash #37) The ufs-community/UFS_UTILS and ufs-community/ccpp-physics repos have not been updated yet in the global-workflow, which cause the sfcanl job to fail at 06z cycle. Updating both repositories allows the sfcanl job to complete successfully at 06z.
• gfs_marineanlvar and gdas_marineanlvar jobs also failed at 06z and require further debugging. (possible reason is that all marine obs files are missing or empty at 06z)

logs/2021122106/gfs_marineanlvar.log:

 1: Assertion failed: spaces_.size() >0 in ObsSpaces, line 88 of /scratch3/NCEPDEV/global/Jiarui.Dong/noscrub/GlobalWorkflow/global-workflow.cory/sorc/gdas.cd/bundle/oops/src/oops/base/ObsSpaces.h
 2: Assertion failed: spaces_.size() >0 in ObsSpaces, line 88 of /scratch3/NCEPDEV/global/Jiarui.Dong/noscrub/GlobalWorkflow/global-workflow.cory/sorc/gdas.cd/bundle/oops/src/oops/base/ObsSpaces.h

sorc/gdas.cd/bundle/oops/src/oops/base/ObsSpaces.h:

  ASSERT(spaces_.size() >0);
  Log::trace() << "ObsSpaces<MODEL, OBS>::ObsSpaces done" << std::endl;

[ClaraDraper-NOAA]

@jiaruidong2017 While we wait for the marine DA to be sorted out, you can start with this test for the uncoupled model:
global-workflow/dev/ci/cases/pr/C96_atm3DVar.yaml

[ClaraDraper-NOAA]

Copied from chat with David:

here is David's PR with the update submodules for testing the recent ccpp-physics changes: NOAA-EMC/global-workflow#4389 (comment)

I just finished running a full suite of tests on Ursa yesterday and did not run into issues with the marine DA. However, the model ufs_configure does need an update in order to run coupled forecasts. The needed update can be found here in ush/parsing_ufs_configure.sh.

[jiaruidong2017]

Both the C96_atm3DVar and C96C48mx500_S2SW_cyc_gfs runs have completed successfully. I will proceed with the analysis to examine the updated variables from gcycle.

[jiaruidong2017]

Please ignore the latest updates above, because I need to re-run the second tests for GCYCLE2/C96_atm3DVar without a time shift.

[ClaraDraper-NOAA]

Please ignore the latest updates above, because I need to re-run the second tests for GCYCLE2/C96_atm3DVar without a time shift.

Thanks for the update. I was surprised that there was no impact!

[jiaruidong2017]

I have conducted two g-w runs 1) GCYCLE3/C96_atm3DVar with a time shift and 2) GCYCLE2/C96_atm3DVar without a time shift.

GCYCLE3:

commit d095c83d678381e8b293700aef84f214326e7d3d (HEAD)
Author: Clara Draper <33430543+ClaraDraper-NOAA@users.noreply.github.com>
Date:   Thu Dec 18 06:47:26 2025 -0500

GCYCLE2:

commit d4a11f8e5704f5c1f85a7441c71349ceb9208811
Author: Walter Kolczynski <Walter.Kolczynski@noaa.gov>
Date:   Wed Dec 17 15:06:27 2025 -0500

[jiaruidong2017]

https://docs.google.com/presentation/d/1TBUD3SWq9af7wq4DdYSasVyX0_7R-JIx/edit?slide=id.p5#slide=id.p5

I will proceed to launch the C96C48mx500_S2SW_cyc_gfs experiment runs configured with/without a time shift.

[jiaruidong2017]

The C96C48mx500_S2SW_cyc_gfs experiment runs configured with a time shift (left) and without a time shift (right) are compared for four gdas cycles as below:

By examining the variables affected by the global_cycle in the C96C48mx500_S2SW_cyc_gfs experiment, we can draw a conclusion that the changes introduced by a bug fix were not significant.