Dev master candidate from NCAR 2019/10/02

[gustavo-marques]

Summary:

• Corrects time-related arguments in ocean_model_init in the MCT and NUOPC caps. This fixed the issue of overwriting ocean.stats every time the model was restarted;
• Obsoletes ADD_KV_SLOW and removes visc%Kv_slow from restart files since it is no longer needed;
• Checks for consistency of lat, lon and mask between mesh and mom6 grid in the NUOPC cap;
• Fixes the coding issues with GME described in Coding issues with GME code NCAR/MOM6#109;
• Adds NCAR version of MOM6 Doxygen (https://ncar.github.io/MOM6/APIs/), and improves NUOPC documentation.

Changes in answers and MOM parameters

The following parameter will be added in MOM_parameter_doc.all when USE_MEKE = True:

MEKE_EQUILIBRIUM_ALT = False    !   [Boolean] default = False
                                ! If true, use an alternative formula for computing the (equilibrium)initial
                                ! value of MEKE.

The following will be removed:

ADD_KV_SLOW = False             !   [Boolean] default = False
                                ! If true, the background vertical viscosity in the interior (i.e., tidal +
                                ! background + shear + convection) is added when computing the coupling
                                ! coefficient. The purpose of this flag is to be able to recover previous
                                ! answers and it will likely be removed in the future since this option should
                                ! always be true.

I could not figure out why the answers change for the following configuration (perhaps compiler version? I've used intel/17.0.1):

modified: regressions/coupled_AM2_LM3_SIS/CM2G63L/ocean.stats.intel

All other configurations in MOM6-examples are bit for bit.

[codecov-io]

Codecov Report

❗ No coverage uploaded for pull request base (dev/master@5b994f3). Click here to learn what that means.
The diff coverage is 29%.

@@              Coverage Diff              @@
##             dev/master    #1013   +/-   ##
=============================================
  Coverage              ?   43.17%           
=============================================
  Files                 ?      213           
  Lines                 ?    62163           
  Branches              ?        0           
=============================================
  Hits                  ?    26836           
  Misses                ?    35327           
  Partials              ?        0

Impacted Files	Coverage Δ
src/core/MOM_variables.F90	77.27% <ø> (ø)
src/diagnostics/MOM_obsolete_params.F90	87.66% <100%> (ø)
src/parameterizations/lateral/MOM_hor_visc.F90	62.14% <19.23%> (ø)
...c/parameterizations/vertical/MOM_set_viscosity.F90	65.22% <50%> (ø)
...c/parameterizations/vertical/MOM_vert_friction.F90	72.03% <50%> (ø)
...arameterizations/lateral/MOM_thickness_diffuse.F90	63.18% <61.53%> (ø)
src/parameterizations/lateral/MOM_MEKE.F90	69.52% <75%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 5b994f3...bccdf19. Read the comment docs.

[Hallberg-NOAA]

Thank you for these suggested changes. They are very helpful for cleaning up aspects of the code.

However, a number of the comments documenting the dimensional rescaling of variables in MOM_hor_visc.F90 have been removed or perhaps omitted by basing the changes on an older version of the code. Please consider restoring these comments (i.g., [T-1 ~> s -1] instead of [s-1] on line 244 of MOM_horvisc.F90, and similarly on numerous following lines.

[gustavo-marques]

@Hallberg-NOAA, sorry for missing the comments documenting the dimensional rescaling of variables. These have been restored now. Thank you for catching this.

[kshedstrom]

I approve this PR.

[marshallward]

The FrictWorkMax and FrictWork_diss diagnostics appear to have been removed, along with blocks of code dedicated to computing them. Was that intentional?

[marshallward]

Ignore my last comment, I see from ba2f186 that this was intentional. (This may require an update of the OpenMP directive, btw).

[marshallward]

Github is not reporting any conflicts here, but if I try to do a local merge then I get a conflict in src/parameterizations/lateral/MOM_hor_visc.F90 at L1379.

@gustavo-marques I'm not really sure the best way to resolve this on our end, but if you wanted to do a dev/gfdl merge into your PR, resolve the conflict, and then update the PR, maybe that would work?

[adcroft]

If there is no conflict with dev/master then all is good for this PR. Merging dev/master back into dev/gfdl is our problem. This PR was OK'd already (before we lost the machine) so complete the merge on dev/master and respectively update MOM6-examples and the stats repo. We'll look at merging onto dev/gfdl after that.

[marshallward]

Right, sorry, wrong branch. Will deal with this as a dev/master merge.