CAM7 test failure - zi (geopotential at interfaces) not read from input

[jimmielin]

What happened?

CAM7 analytic IC test fails to run

 ERROR:
 Required variables missing from registered list of input variables: geopotentia
 l_height_wrt_surface_at_interface

What are the steps to reproduce the bug?

Use the branch in #316 (if it was not yet merged to SIMA)

Run test suite and observe test case SMS_Ln9.ne5pg3_ne5pg3_mg37.FCAM7.derecho_intel.cam-outfrq_se_cslam_analy_ic

What CAM-SIMA hash were you using?

4bbb920

What machine were you running CAM-SIMA on?

CISL machine (e.g. cheyenne)

What compiler were you using?

Intel

Path to a case directory, if applicable

/glade/derecho/scratch/hplin/aux_sima_intel_20241212144921/SMS_Ln9.ne5pg3_ne5pg3_mg37.FCAM7.derecho_intel.cam-outfrq_se_cslam_analy_ic.GC.aux_sima_intel_20241212144921

Will you be addressing this bug yourself?

Any Software Engineer can do this

Extra info

It appears that if a variable is only intent(out) in tphysbc, write_init_files.py will not write input for this variable into phys_init/physics_inputs.F90.

This only happens in CAM7 because sima_state_diagnostics was moved to tphysac in CAM7; other suites have sima_state_diagnostics (which have zi as intent(in)) in tphysbc.

[nusbaume]

Just to provide an update, @peverwhee and I were able to find that the issue is likely due to each variable only being included once in the list of variables in the CCPP database, which means if the first group has a variable that is intent(out) then that variable will only be listed as an output variable when CAM-SIMA is generating the initialization routines, even if all other groups have the same variable listed as intent(in). This then prevents CAM-SIMA from properly locating the variable at runtime, hence the error.

@peverwhee is going to try out some possible solutions (thanks Courtney!), but it will likely require an update to the CCPP-framework before this bug can be fixed in CAM-SIMA.