Refactoring internal tide reflection

[raphaeldussin]

perform trig operations with the discrete angles instead of radians, which simplify code quite a bit.

[codecov]

Codecov Report

Merging #1468 (7019dd5) into dev/gfdl (bb9cb35) will increase coverage by 0.00%.
The diff coverage is 0.00%.

❗ Current head 7019dd5 differs from pull request most recent head 650fe9a. Consider uploading reports for the commit 650fe9a to get more accurate results

@@            Coverage Diff            @@
##           dev/gfdl    #1468   +/-   ##
=========================================
  Coverage     29.06%   29.06%           
=========================================
  Files           237      237           
  Lines         71645    71642    -3     
=========================================
  Hits          20822    20822           
+ Misses        50823    50820    -3     

Impacted Files	Coverage Δ
...c/parameterizations/lateral/MOM_internal_tides.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 bb9cb35...650fe9a. Read the comment docs.

[Hallberg-NOAA]

From inspection of the code, it looks to me like angle_wall and angle_wall0, and angle_r and angle_r0 all must the same units (probably [nondim], because as integers I don't see how they could have units of [rad]). Please verify that the documented units of these variables are correct.

It might also be worthwhile to document in a comment that the beams are being reflected into a single discrete angle, and are not splitting into multiple angles to approximate an arbitrary new angle.

[raphaeldussin]

fixed

[marshallward]

The tc5 that you sent me showed some partial success. Most of the issues seem to mostly be related to diagnostics, but there's a more serious one related to restarts.

PASS: Solutions tc5.grid agree.
PASS: Solutions tc5.layout agree.
PASS: Solutions tc5.dim.t agree.
PASS: Solutions tc5.dim.l agree.
PASS: Solutions tc5.dim.h agree.
PASS: Solutions tc5.dim.z agree.
PASS: Solutions tc5.dim.q agree.
PASS: Solutions tc5.openmp agree.
PASS: Solutions tc5.rotate agree.
PASS: Solutions tc5.nan agree.
PASS: Solutions tc5.dim.r agree.

FAIL: tc5.restart failed at runtime.
FAIL: Diagnostics tc5.grid.diag have changed.
FAIL: Diagnostics tc5.layout.diag have changed.
FAIL: Diagnostics tc5.dim.t.diag have changed.
FAIL: Diagnostics tc5.dim.l.diag have changed.
FAIL: Diagnostics tc5.dim.h.diag have changed.
FAIL: Diagnostics tc5.dim.z.diag have changed.
FAIL: Diagnostics tc5.dim.q.diag have changed.
FAIL: Diagnostics tc5.openmp.diag have changed.
FAIL: Diagnostics tc5.nan.diag have changed.
FAIL: Diagnostics tc5.dim.r.diag have changed.

We don't need to necessarily sort through all of these now, but the dimensional ones ought to be easy enough to fix.

[Hallberg-NOAA]

I am now satisfied that these changes are correct, and are moving the code in the right direction, even if the internal tide capability is not yet fully operational and passing all of the tests.

[marshallward]

Gaea Regression: https://gitlab.gfdl.noaa.gov/ogrp/MOM6/-/pipelines/13701 ✔️