Testing with Gaussian hump with amplitude of 4m:

Initial at t=0 hrs:

At 4 hrs:

Note, bathymetry and vertical coordinate system is shifted to avoid instability with bottomDepth < 0, as discussed earlier @mark-petersen.

@sbrus89, can you please test and verify that this works for you? I'm also running your Sandy test case to ensure that it works as expected following these changes.

@mark-petersen, we will verify that Sandy case is intact and then merge. Thanks!

@sbrus89, here is the result from the sandy case, which is stable, but doesn't appear to be as good as some of the recent runs we've done based on plots we've made. All plots are out of QuickViz on the pointStats.

Are the parameters in the namelist.ocean file in the repo the best ones we have? I don't think they are because config_use_const_visc=.false., for instance.

Thus, because this is stable I'm inclined to think it is working and we should have a separate PR to deal with tuning the Sandy case.

Please let me know what you think but I'm inclined to think this is ready to merge and you should do a separate PR to submit the tuned results that we anticipate to be better for Sandy.

Using `ocean/coastal' yield the following result, which is qualitatively equivalent to the above (noting I've removed the superflous modification of Bahama bathymetry so results aren't expected to be BFB.

Given these results I'll merge this PR and we can fine tune the existing Sandy simulation in a follow-up PR (e.g., using the depth-averaging approach that produces a better match against the data).