New Mesh: IcoswISC30E3r6

[xylar]

Long name: IcoswISC30L64E3SMv3r6

This nearly uniform 30 km mesh is the dual mesh of a subdivided icosahedron (Icos).

This mesh differs from IcoswISC30E3r2 (#691) in having topography smoothing using a Gaussian filter with a characteristic length scale of 100 km except under ice shelves. It differs IcoswISC30E3r4 (#734) in not smoothing bathymetry under ice shelves and not smoothing land-ice draft or land-ice thickness at all.

The horizontal mesh is identical to IcoswISC30E3r4 and IcoswISC30E3r5 (it is not being recreated).

Mesh, initial condition, dynamic adjustment and files for E3SM will be on Chrysalis at:

/lcrc/group/e3sm/ac.xylar/compass_1.2/chrysalis/e3smv3-meshes/icoswisc30e3r6

Checklist

• Document (in a comment titled Testing in this PR) any testing that was used to verify the changes

[xylar]

This branch requires MPAS-Dev/MPAS-Tools#541 followed by an MPAS-Tools release v0.29.0

[xylar]

Initial Condition

The smoothing seems to work as expected.

Original bathymetry:

Smoothed bathymetry:

The ice draft after smoothing showing that it is not smoothed:

The bottom depth (basically the negative of the smoothed bathymetry):

[proteanplanet]

A review page for this mesh has been created at https://acme-climate.atlassian.net/wiki/spaces/OO/pages/4021223426/Review+IcoswISC30E3r6

[xylar]

@proteanplanet, very much appreciated!

[proteanplanet]

This mesh lacks key oceanographic features required to answer some E3SM Phase 3 science questions for polar science, and should be reviewed also by @njeffery due to its potential impact on ocean BGC:

1. Continental shelves almost completely removed along critical coastlines, including the western Americas, the Eastern Antarctic, Bellingshausen and Amundsen Seas, Africa, and Eastern Australia, with the expectation of a decay in ocean water properties, coastal precipitation, ocean carbon uptake, and ice shelf melt rates.
2. Near-complete removal of the Bering Sea as a separate entity from the main Pacific Basin by removing the Aleutian Island chain, important for BGC, and almost complete removal of Lomonosov Ridge, important for differentiating Arctic water properties for predicting sea ice loss.
3. Complete removal of bathymetry associated with key trenches and banks in the global ocean.

If this mesh were to be adopted as the default standard mesh for E3SM V3, the polar group may need to consider a parallel simulation campaign to address the overarching polar science question in the Phase 3 proposal: What is the likelihood of rapid sea level rise due to tipping points induced by ice shelf instabilities and the likelihood of total Arctic sea ice loss spanning more than a month of each year from the present out to 2060?

Conversely, fully coupled simulations with this aggressive mesh smoothing help address one significant bias cited in the Phase 3 polar proposal: sea ice in the Labrador Sea, only partly removed with the unsmoothed version of this mesh. All other aspects of targeted sea ice extent biases in the Phase 3 proposal for E3SMv3 have been improved without the need for mesh smoothing: Barents, Weddell and Ross seas, but have been shown to be partly decayed by bathymetric smoothing, now being tested again with Antarctic changes in this PR.

A side-by-side graphical comparison is provided here globally, and for Antarctic ice shelves, and for the Arctic, for this mesh versus the un-smoothed version:

Acceptance of this mesh from the sea ice perspective depends on removal of this odd artifact detected in fully coupled simulations with the predecessor of this mesh with ice shelf cavity smoothing, shown for years 101-150 of the B-case spinup:

[xylar]

I rebased so the mpas_tools update can happen in #742 instead of being hidden here.

[njeffery]

@proteanplanet : Although it's likely that smoothing will have consequences for marine bgc (possibly both negative and positive), HES is not planning on using a mesh with ice shelves for other reasons - concerns with MarBL behaving properly under iceshelves and the Lagrangian tracer scheme.

[xylar]

Will not use.