twice of temporal dimension numbers are stored for variables with "tpt-al-hxy-u"

[karakita-MRI]

Thank you for your valuable time.

When cmor_version “3.15.2” operates e.g., atmos.clis.tpt-al-hxy-u.3hr.glb or atmos.hus.tpt-al-hxy-u.6hr.glb––i.e., the valuables with “alevel” and “time1” in the dimensions (i.e., tpt-al-hxy-u)––, twice of temporal dimension numbers (“time”) more than expected are stored in cmorized files: then, spurious numbers, such as 0 or 9.8813129168…e-324, are stored in “time” and etc.

Are there any bugs?

Thank you.
Tatsuro KARAKI(MRI)

[mauzey1]

@karakita-MRI

I've identified a potential bug that might be causing your issue. Attached is a script that reproduces that bug.

reproduce_issue_989_standalone.py

It seems to be caused by first writing data to a variable that already has all of its time values filled (no time_vals passed), and then writing the zfactor data with the variable using store_with.

    cmor.write(hus_var, hus_data)
    cmor.write(ps_zfactor, ps_data, store_with=hus_var)

Is this what your code is doing?

[karakita-MRI]

Thank you for your quick response.

cmor.write(hus_var, hus_data)
cmor.write(ps_zfactor, ps_data, store_with=hus_var)

Is this what your code is doing?

Exactly!
Our cmor.py follows similar logical flows to "reproduce_issue_989_standalone.py".