Fix dimensions in wave structure

[raphaeldussin]

routine was failing dimensional testing. Proposed fixes makes it dimensionally consistent.
I also add several dimension checks, available with DEBUG=true

I think the normalization of the vertical structure was wrong and I corrected accordingly.

In spite of the various fixes, tridiagonal solver becomes unstable at step=2 in benchmark :(

[raphaeldussin]

@marshallward how do I figure out where the style problem comes from?

[codecov-io]

Codecov Report

Merging #1252 (42758d1) into dev/gfdl (f3483be) will decrease coverage by 0.01%.
The diff coverage is 0.00%.

@@             Coverage Diff              @@
##           dev/gfdl    #1252      +/-   ##
============================================
- Coverage     45.90%   45.88%   -0.02%     
============================================
  Files           225      225              
  Lines         71395    71422      +27     
============================================
  Hits          32774    32774              
- Misses        38621    38648      +27     

Impacted Files	Coverage Δ
src/ALE/regrid_edge_values.F90	34.57% <ø> (ø)
src/diagnostics/MOM_wave_structure.F90	0.00% <0.00%> (ø)
...c/parameterizations/lateral/MOM_internal_tides.F90	0.00% <0.00%> (ø)
...meterizations/vertical/MOM_internal_tide_input.F90	0.00% <0.00%> (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 f3483be...42758d1. Read the comment docs.

[adcroft]

@marshallward how do I figure out where the style problem comes from?

Evidently I messed up the error reporting when merging the two workflows. It was easier when they were separate. I can fix the reporting once the workshop is over...

[raphaeldussin]

@adcroft it was a dumb whitespace error, I fixed it. It would be helpful though to have github actions print the styling errors found

[adcroft]

@adcroft it was a dumb whitespace error, I fixed it. It would be helpful though to have github actions print the styling errors found

Yes, and it used to. While developing the workflows I initially had separate workflows for style and docs but we decided we needed fewer workflows. In order to avoid a white space problem not block the testing of docs I had to do a workaround. I got it wrong.

[Hallberg-NOAA]

Overall, I think that there is progress with these changes, and suspect that they will all be found to be correct when used with an appropriate test case. However, it appears that there are new definitions for the rescaled units of a number of the variables in this routine, but there have not been any changes to the comments describing these variables to reflect these changes.

Please add comments describing the (now correct) units of the variables when the variables are defined. I am aware that some of the variables are not described in comments at all, but this should be the right opportunity to start doing so.

[raphaeldussin]

I don't know how to fix style if I can't see the errors in the checks

[marshallward]

Looks like one of your lines is exceeding the 120-character limit due to the comment:

$ ./.testing/trailer.py -e TEOS10 -l 120 src config_src | tee style_errors
src/diagnostics/MOM_wave_structure.F90, line 166: Line length exceeded "  real, dimension(SZK_(G)+1) :: Uavg_profile !< Vertical profile of the magnitude of horizontal velocity [L T-1 ~> m s-1]."

We need to clean this up to give better feedback, I think.

[raphaeldussin]

@marshallward this would be very helpful thanks!

[raphaeldussin]

I think this is ready but I'd like to give @sonyalegg a chance to chime in

[adcroft]

https://gitlab.gfdl.noaa.gov/ogrp/MOM6/-/pipelines/11699