Ice Margin Boundary Condition Alterations

[alexolinhager]

Makes changes to the boundary conditions at the grounding line to improve stability in ice dynamics-hydrology coupled runs. Most problems were arising because of very weak gradients at the grounding line formed from a thinning glacier. The primary changes in this PR are: 1) zeroing out hydropotential and enforcing a hydrostatic waterPressure in the ocean, 2) instating a small outflowing condition at grounding line edges (instead of zeroing out channelArea and channelDischarge to prevent inflow). An additional change ensures no channels form when there is no water in the upstream cell (done to avoid instabilities).

This PR is a compilation of cherry-picked commits originally made on a separate mega-branch ([https://github.com/MALI-Dev/E3SM/tree/alexolinhager/chnlInstabilityDebugging]). The commits in this PR have been tested on the Thwaites coupled domain, as well as within the full integration suite. As expected, the PR passes all integration suite tests, except for the baseline comparison of the Humboldt hydro case (changes only apply to marine termini). The changes here are very minor and only affect waterPressure and hydropotential near the grounding line. A coupled run on Thwaites glacier was successfully ran on this PR for 300 years and the subglacial discharge system behaved as expected.

[alexolinhager]

I pushed another commit making minor changes in regards to yesterday's comments. Hopefully the additional commenting around the hydropotential slope calculations makes it more obvious what's going on. I'll wait until we have a consensus on how to handle hydropotential and waterPressure to make the rest of the changes.

[alexolinhager]

@matthewhoffman, I pushed a few more commits that get the PR to where we want it. The current version passes the hydro_radial validation and baseline comparison; however, it fails the Humboldt baseline comparison because of the changes we made to the marine boundary. The differences between the baseline and this PR are quite minor. This PR also behaves normally for the Thwaites domain. I think we are at a point where this could be merged.

Small change in waterThickness near grounding line:

Small changes in hydropotential near grounding line (black are cells at or below sea level): Beginning of run:

End of run:

All of the differences from develop are near the grounding line, and are what I would expect given the changes to the code

[matthewhoffman]

Testing

Ran compass full_integration suite on Perlmutter. The 4 hydro tests failed baseline comparison, as expected, but otherwise everything passes.

Test Runtimes:
00:26 PASS landice_dome_2000m_sia_restart_test
00:14 PASS landice_dome_2000m_sia_decomposition_test
00:21 PASS landice_dome_variable_resolution_sia_restart_test
00:54 PASS landice_dome_variable_resolution_sia_decomposition_test
00:26 PASS landice_enthalpy_benchmark_A
00:31 PASS landice_eismint2_decomposition_test
00:21 PASS landice_eismint2_enthalpy_decomposition_test
00:29 PASS landice_eismint2_restart_test
00:20 PASS landice_eismint2_enthalpy_restart_test
00:09 PASS landice_greenland_sia_restart_test
00:16 PASS landice_greenland_sia_decomposition_test
00:18 FAIL landice_hydro_radial_restart_test
00:10 FAIL landice_hydro_radial_decomposition_test
00:30 FAIL landice_humboldt_mesh-3km_decomposition_test_velo-none_calving-none_subglacialhydro
00:30 FAIL landice_humboldt_mesh-3km_restart_test_velo-none_calving-none_subglacialhydro
00:26 PASS landice_dome_2000m_fo_decomposition_test
00:29 PASS landice_dome_2000m_fo_restart_test
00:26 PASS landice_dome_variable_resolution_fo_decomposition_test
00:16 PASS landice_dome_variable_resolution_fo_restart_test
00:31 PASS landice_circular_shelf_decomposition_test
00:45 PASS landice_greenland_fo_decomposition_test
00:46 PASS landice_greenland_fo_restart_test
00:26 PASS landice_thwaites_fo_decomposition_test
00:34 PASS landice_thwaites_fo_restart_test
00:18 PASS landice_thwaites_fo-depthInt_decomposition_test
00:39 PASS landice_thwaites_fo-depthInt_restart_test
00:45 PASS landice_humboldt_mesh-3km_restart_test_velo-fo_calving-von_mises_stress_damage-threshold_faceMelting
00:29 PASS landice_humboldt_mesh-3km_restart_test_velo-fo-depthInt_calving-von_mises_stress_damage-threshold_faceMelting
Total runtime 13:00
FAIL: 4 tests failed, see above.