Skip to content

[production/RRFS.v1]Fixes for REFS restart reproducibility and crash under DEBUG#90

Merged
jkbk2004 merged 2 commits into
NOAA-PSL:production/RRFS.v1from
JiliDong-NOAA:spp_restart_fix
Jan 20, 2026
Merged

[production/RRFS.v1]Fixes for REFS restart reproducibility and crash under DEBUG#90
jkbk2004 merged 2 commits into
NOAA-PSL:production/RRFS.v1from
JiliDong-NOAA:spp_restart_fix

Conversation

@JiliDong-NOAA
Copy link
Copy Markdown
Contributor

@JiliDong-NOAA JiliDong-NOAA commented Sep 16, 2025

This PR addresses:

  1. REFS ensemble restart reproducibility issues when running with 32 bit physics by
  • changing the SPP random seeds and spectral coefficient variable names for consistency when reading these variables for restart (spppert_seed->spp_seed; ppcpert_spec->spp_spec)
  • modifying data type from single to double precision when reading and writing the above two variables
  1. crash when running REFS under DEBUG mode

The issues are related to LSM-SPP. It appears that LSM-SPP perturbations were added to the whole domain without masking out the water/ice points. This caused:

  • 0 index error under DEBUG mode for smcmin/smcmax(stype) when stype=0 over water
  • floating overflow error under DEBUG mode when applying LSM-SPP to zorll where zorll would have missing values over water/ice (9x10e30)

Temporary patches were added to fix these crashes when running REFS under DEBUG. Could we have @ClaraDraper-NOAA and @barlage review these changes in lndp_apply_perts.F90? A more general question: should water/ice points be masked out when applying LSM-SPP when calling lndp_apply_perts or am I missing something here?

Fixes issue #92

@JiliDong-NOAA JiliDong-NOAA changed the title Fixes for RRFS restart reproducibility and running under DEBUG Fixes for REFS restart reproducibility and crash under DEBUG Sep 16, 2025
@JiliDong-NOAA JiliDong-NOAA changed the title Fixes for REFS restart reproducibility and crash under DEBUG [production/RRFS.v1]Fixes for REFS restart reproducibility and crash under DEBUG Sep 19, 2025
@MatthewPyle-NOAA
Copy link
Copy Markdown

@JiliDong-NOAA Have you heard anything about this one? Wondering if it might be a good idea to create the associated UFS WM and UFSATM PRs so it has more visibility.

@JiliDong-NOAA
Copy link
Copy Markdown
Contributor Author

@JiliDong-NOAA Have you heard anything about this one? Wondering if it might be a good idea to create the associated UFS WM and UFSATM PRs so it has more visibility.

Not yet. I was hoping to create single UFSATM and WM PRs by including all changes necessary after submodule components getting merged first. @pjpegion @ClaraDraper-NOAA @barlage could you review this PR when you have time? RRFSv1 implementation is approaching and we have a code handoff date as Oct. 31.

@MatthewPyle-NOAA
Copy link
Copy Markdown

@pjpegion I'm hoping at long last to get this change merged into production/RRFS.v1. Can you review/approve? Thanks!

@MatthewPyle-NOAA
Copy link
Copy Markdown

Thanks @pjpegion Can you merge it? I don't seem to have that permission for this repository.

@pjpegion
Copy link
Copy Markdown
Collaborator

@MatthewPyle-NOAA @jkbk2004 is the code manager.

@MatthewPyle-NOAA
Copy link
Copy Markdown

@jkbk2004 Could we get your help to merge this one? Thanks!

@jkbk2004 jkbk2004 merged commit e8d56dd into NOAA-PSL:production/RRFS.v1 Jan 20, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants