[mvertens]

This PR creates a new med_enthalpy_mod, plus a multi-instance fix for the branch-run pointer check.

1. Move enthalpy flux calculations into a new med_enthalpy_mod

The enthalpy flux code was split across med_phases_prep_ocn_mod and med_phases_prep_atm_mod, with shared state duplicated between them. This collects it into a single new module, med_enthalpy_mod.

What was wrong

• component_computes_enthalpy_flux existed as two independent module variables, one in each prep module, each populated by its own NUOPC_CompAttributeGet call.
• global_htot_corr and global_hrof_corr lived in med_phases_prep_atm_mod but were computed by routines invoked from med_phases_prep_ocn_mod.
• Consequently med_phases_prep_ocn_mod had to use med_phases_prep_atm_mod from inside two of its subroutines, giving prep_ocn a dependency on prep_atm purely for enthalpy.

What this does

• Adds mediator/med_enthalpy_mod.F90 holding all five enthalpy routines and one copy of each shared variable.
• Drops the prep_ocn -> prep_atm dependency.
• Adds med_enthalpy_init, which performs the attribute read that was previously duplicated in both prep modules. Called from med_phases_prep_ocn_init and med_phases_prep_atm.

The module is not named med_phases_* because it contains no NUOPC run phases; it follows the med_io_mod / med_methods_mod convention instead.

Interface change

med_phases_prep_atm_mod no longer exports:

• med_phases_prep_atm_enthalpy_correction
• med_phases_prep_atm_enthalpy_runoff

These become med_enthalpy_correction and med_enthalpy_runoff in med_enthalpy_mod, alongside med_enthalpy_init, med_enthalpy_med_computation and med_enthalpy_atm_computation. The module
also makes component_computes_enthalpy_flux, global_htot_corr and global_hrof_corr public, as they are now shared rather than private to prep_atm.

---

2. Fix the branch-run DRV_RESTART_POINTER check for multi-instance cases

The check in cime_config/buildnml constructed the expected pointer name as

"rpointer.cpl.{}-{}".format(RUN_REFDATE, RUN_REFTOD)

Testing

This exact code was tested in noresm using the prealpha test suite and all tests passed and were bit-for-bit

<fill in: which test suite, which machine/compiler, bit-for-bit vs. baseline>

🤖 Generated in collaboration with Claude Code

[billsacks]

Approving based on a cursory look at the enthalpy changes and a somewhat more careful look at the buildnml changes for multi-instance branch cases.

[billsacks]

@mvertens - I'm approving this but, given the conservative period we're in with the run-up to the CESM release, I want to get feedback from a few other people about what testing we want to do here before merging.