Skip to content

Support CISM's SMB budget diagnostics and fractional ice_covered option - #158

Merged
Katetc merged 6 commits into
masterfrom
lipscomb/smb_fixes
Nov 12, 2026
Merged

Katetc merged 6 commits into
masterfrom
lipscomb/smb_fixes

Conversation

@whlipscomb

@whlipscomb whlipscomb commented Nov 11, 2026 •

Copy link
Copy Markdown
Contributor

Branch: lipscomb/smb_fixes
Base: master (d6497d2, tag cismwrap_3_00_002, the CISM-wrapper version in CESM3 alpha11d)
Requires the matching CISM PR, ESCOMP/CISM lipscomb/smb_fixes. Fractional ice_covered additionally requires ESCOMP/CMEPS#720 (merged).
4 commits

Summary

This PR goes with the CISM PR on lipscomb/smb_fixes, which adds SMB budget diagnostics for coupled runs, an option for a fractional ice_covered field, and a fix to GLAD's ocean mask. Together with ESCOMP/CMEPS#720, they remove a 40-60 Gt/yr water conservation error in CESM runs with an evolving Greenland ice sheet (details in the CISM PR). It replaces CISM-wrapper PR #156, which was stacked on #155; the first three commits are cherry-picked from #156 without changes.

It also adds a testmod with the Greenland settings of the current CESM3 coupled runs, so that aux_glc tests the configuration used in those runs.

Changes

  • New namelist variable ice_covered_thck_scale (group glad_climate, default 0.0) (1d12802). If greater than 0, CISM sends ice_covered = min(1, thck/ice_covered_thck_scale). Fractional values require a CMEPS version with Allow fractional ice_covered values in the lnd2glc downscaling CMEPS#720.
  • Default history fields (6623b14): for evolving ice sheets, add the SMB budget fields acab_cpl_tavg, acab_applied_tavg, acab_unapplied_glad_tavg, acab_unapplied_tavg, total_smb_flux_tavg, total_smb_cpl_flux_tavg, total_smb_unapplied_glad_flux_tavg and total_smb_unapplied_flux_tavg to esm_history_vars. Since these are time averages, evolving-ice-sheet runs now write annual h0a history files by default, in addition to the h0i files.
  • New testmod cism/fractional_ice_covered (0d9ab52), setting ice_covered_thck_scale = 4.0. No test in the test list uses it yet.
  • New testmod and aux_glc test cism/gris_powerlaw (d4719af), with the CISM settings of the CESM3 alpha11d runs (b.e30_alpha11d.B1850_MTso.ne30_t233_wgx3.419-422): power-law sliding with a friction coefficient from an inversion, read (with the initial ice sheet state) from a CISM restart file in inputdata (cesm2_init), plus the other settings in those runs' user_nl_cism. Test: ERS_Ly3.f09_g17_gris4.T1850Gg.derecho_intel.cism-gris_powerlaw.
  • source_cism will point to the CISM tag made from the matching CISM PR.

Answer changes

  • Default settings: small changes from the GLAD ocean-mask fix (cold starts only) and from rofi, which now includes positive SMB not applied by CISM (2-5 Gt/yr for Greenland). New h0a history files and new fields in the namelists, so baseline comparisons will show namelist and history differences.

Planned follow-up

Once ESCOMP/CMEPS#720 is in the CESM tag that CISM-wrapper uses:

  • set the default ice_covered_thck_scale = 4.0 for evolving ice sheets (in 20-year Greenland runs, 4 m gave the smallest imbalance, < 0.1 Gt/yr, and the fewest cells switching between ice-covered and ice-free, 14-32 per year);
  • add a testmod with ice_covered_thck_scale = 0.0, so that aux_glc keeps testing the binary path, and add an aux_glc test with the current default.

Testing

All on derecho, with CISM lipscomb/smb_fixes:

🤖 Generated with Claude Code

whlipscomb and others added 4 commits November 11, 2026 11:34
CISM's [GLAD climate] section has a new option, ice_covered_thck_scale.
If it is greater than 0, GLAD sends the coupler a fractional
ice_covered field, min(1, thck/ice_covered_thck_scale), so that cells
with only a trace of ice are treated as partly ice-covered. This removes
most of the water conservation error from the SMB that the land model
assigns to such cells but CISM cannot apply (about 40-60 Gt/yr in CESM
runs with an evolving Greenland ice sheet).

The default is 0.0, which keeps the binary ice_covered field.
Fractional values require a CMEPS version that supports them.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
For evolving ice sheets, add these annual-mean fields to the default
esm_history_vars:
  acab_cpl_tavg, acab_applied_tavg, acab_unapplied_glad_tavg,
  acab_unapplied_tavg, total_smb_flux_tavg, total_smb_cpl_flux_tavg,
  total_smb_unapplied_glad_flux_tavg, total_smb_unapplied_flux_tavg
These are CISM's SMB budget diagnostics: the SMB received from the
coupler, the SMB applied to the ice, and the parts that are not applied
(zeroed by GLAD, or not applied by the dycore). In each cell and
globally, received = applied + unapplied (GLAD) + unapplied (dycore).

Since these are time-average fields, evolving-ice-sheet runs now write
annual h0a history files by default, in addition to the h0i files.

Requires a CISM version with the SMB budget diagnostics.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The testmod cism/fractional_ice_covered sets ice_covered_thck_scale = 4.0,
so that CISM sends the coupler a fractional ice_covered field. It can be
used to test the coupling of fractional ice cover, including exact
restart. It requires a CMEPS version that supports fractional
ice_covered values, so no test in the test list uses it yet.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
New testmod cism/gris_powerlaw, with the CISM settings of the CESM3 alpha11d
coupled runs (b.e30_alpha11d.B1850_MTso.ne30_t233_wgx3.419-422): power-law
basal sliding (which_ho_babc = 9) with a spatially varying friction
coefficient (which_ho_powerlaw_c = 2) obtained by inversion in a standalone
spin-up, plus the other settings in those runs' user_nl_cism. The friction
coefficient and the initial ice sheet state are read from a CISM restart
file of a coupled run, already in inputdata (cesm2_init).

New aux_glc test using it:
ERS_Ly3.f09_g17_gris4.T1850Gg.derecho_intel.cism-gris_powerlaw

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

@Katetc Katetc left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'd like to update the externals so that we can match CESM3 better in CISM-Wrapper. If that's ok with you Bill, then I'll go ahead and do it.

</machine>
</machines>
</test>

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If we include the CMEPS and other externals updates in CISM-Wrapper with these changes (which I'm find with doing) then we could add the fractional_ice_covered test at the same time. I think that's a good idea. Do you want me to make this change or do you want to do it?

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@Katetc, I'll defer to Claude. Claude writes:

I'd suggest Kate updates the externals and adds the test, for two reasons:

  • She knows the compatible set. Moving CMEPS to a version with Allow fractional ice_covered values in the lnd2glc downscaling CMEPS#720 may mean moving cime, cdeps or the share code together with it, and choosing a consistent set of tags is her area. Getting that wrong would cost more time than the change itself.
  • The test entry is small and goes with the externals. It's one testlist_cism.xml entry using the existing cism/fractional_ice_covered testmod (for example, ERS_Ly5.f09_g17_gris4.T1850Gg.derecho_intel.cism-fractional_ice_covered). It belongs in the same commit, because it only works once CMEPS has #720. If she'd rather not write it, I can draft the exact entry for her.

Points worth mentioning to her:

  • Keep the default at ice_covered_thck_scale = 0 for now. CESM uses its own top-level externals, not CISM-wrapper's. If the wrapper default became 4.0 while CESM's CMEPS lacks #720, CESM runs would stop with the range-check error. The planned follow-up (default 4.0, plus a testmod for 0) waits until CESM's own CMEPS includes #720.
  • The CESM baseline will change. With new externals, the Stage 1 results no longer serve as the baseline, since other components may change answers. After her commit, I'd rerun the full aux_glc suite with her externals to make the new reference baseline, then repeat the Stage 2 CESM checks against it.

Effect on Stage 2: small. The six subsets running now use the current externals and still show whether each Stage 2 PR is BFB for the CISM code. After Kate's update, I'll:

  1. rebase the Stage 2 wrapper branches onto her updated lipscomb/smb_fixes;
  2. run the full suite with her externals (new baseline);
  3. rerun the per-PR subsets against it (each about an hour, all in parallel).

By Stage 2, Claude is referring to the next step, where we divide ESCOMP/CISM#94 into six smaller PRs and implement those PRs on top of ESCOMP/CISM#97.

@Katetc
Katetc merged commit a406fa6 into master Nov 12, 2026
@whlipscomb

Copy link
Copy Markdown
Contributor Author

Thanks, @Katetc!

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.

2 participants