Implements CCPP-ized tropopause_find

[jimmielin]

Companion PR: ESCOMP/atmospheric_physics#112

Note: I think I brought in more changes than necessary in ncar_ccpp due to the my atmospheric_physics branch being based off development rather than the released used in CAM-SIMA. I can do a rebase if necessary...

Changes made to support CCPP-ized tropopause_find

• Copied utility module interpolate_data.F90 from CAM
• Added utility module tropopause_read_file.F90 to provide climatological tropopause data to tropopause_find from CAM-SIMA
• &tropopause_nl tropopause_climo_file is read from tropopause_read_file.F90 inside SIMA and not in the parameterization, to avoid passing this file name back and forth between the parameterization to CAM-SIMA
• Adds get_curr_calday() result to new standard name variable fractional_calendar_days_on_end_of_current_timestep
• Submodule update (possibly need a rebase?)

I recommend manually merging in the history PR for testing. It might be easier to merge my branch (jimmielin:hplin/tropopause_find) into peverwhee:history_court.

[jimmielin]

Verified that CAM and CAM-SIMA should give the same results now (via good old write(6,*)...), although not in the history tape. The same numbers written to the history tape in CAM and CAM-SIMA differ slightly.

My bad, namelist error when setting the output precision, user error as I suspected 😄 False alarm!

[jimmielin]

Updated generate_registry_data.py to check for valid numeric values before assuming it's a variable.

[jimmielin]

After the fillvalue change I'm getting very small differences in the cold-point-tropopause history output and bit-for-bit in others, still figuring out why...

Seems like the fillvalues are being written as 9.999...E+35

 TROPF_P   (ncol,time)  t_index =      1     1
         23     1352  (  1017,     1) (  1260,     1) (  1017,     1) (  1017,     1)
                1352   9.999999616903162E+35   3.451314683962001E+03 3.8E+28  9.999999616903162E+35 6.5E-10  9.999999616903162E+35
                1352   1.000000000000000E+36   3.451314683962001E+03          1.000000000000000E+36          1.000000000000000E+36
                1352  (  1017,     1) (  1260,     1)
          avg abs field values:    1.701183366780863E+34    rms diff: 5.0E+27   avg rel diff(npos):  6.5E-10
                                   1.701183431952662E+34                        avg decimal digits(ndif):  7.4 worst:  7.4
 RMS TROPF_P                          4.9967E+27            NORMALIZED  2.9372E-07

 TROPF_T   (ncol,time)  t_index =      1     1
         23     1352  (  1017,     1) (   862,     1) (  1017,     1) (  1017,     1)
                1352   9.999999616903162E+35   1.786520363629979E+02 3.8E+28  9.999999616903162E+35 6.5E-10  9.999999616903162E+35
                1352   1.000000000000000E+36   1.786520363629979E+02          1.000000000000000E+36          1.000000000000000E+36
                1352  (  1017,     1) (   862,     1)
          avg abs field values:    1.701183366780863E+34    rms diff: 5.0E+27   avg rel diff(npos):  6.5E-10
                                   1.701183431952662E+34                        avg decimal digits(ndif):  7.4 worst:  7.4
 RMS TROPF_T                          4.9967E+27            NORMALIZED  2.9372E-07

 TROPF_Z   (ncol,time)  t_index =      1     1
         23     1352  (  1017,     1) (  1045,     1) (  1017,     1) (  1017,     1)
                1352   9.999999616903162E+35   9.015423193848925E+03 3.8E+28  9.999999616903162E+35 6.5E-10  9.999999616903162E+35
                1352   1.000000000000000E+36   9.015423193848925E+03          1.000000000000000E+36          1.000000000000000E+36
                1352  (  1017,     1) (  1045,     1)
          avg abs field values:    1.701183366780863E+34    rms diff: 5.0E+27   avg rel diff(npos):  6.5E-10
                                   1.701183431952662E+34                        avg decimal digits(ndif):  7.4 worst:  7.4
 RMS TROPF_Z                          4.9967E+27            NORMALIZED  2.9372E-07

[jimmielin]

I had to specify _kind_phys in the <initial_value>1.e36_kind_phys</initial_value> to get the correct definition in physics_state.F90 and bit-for-bit results with CAM/previous version of CAM-SIMA. The regex in generate_registry_data.py has also been updated. Thanks!

[nusbaume]

Thanks @jimmielin for ensuring the tropopause code works in CAM-SIMA! I do have some change requests, but hopefully none that are too difficult to implement. Of course if you find that there is a problem or issue with anything I requested just let me know. Thanks!

[jimmielin]

Thanks @nusbaume for the review. I've addressed most of the concerns. There are a few questions that I wanted to confirm for some suggested changes, could you please take a look? I appreciate your help!

[nusbaume]

Super-close! Just one last code change and a standard-name change.

[nusbaume]

Looks great now, thanks @jimmielin!

[peverwhee]

One tiny comment to take or leave! Thanks @jimmielin

[jimmielin]

Thanks @nusbaume and @peverwhee for the reviews, merging it in! I'll address the standard names and the check_allocate issues in separate PRs following this.