New Mesh: WCAtl12to45E2r4

[xylar]

Full description to come once we iterate a bit...

[xylar]

@vanroekel, here is the very first draft:

The transition in resolution in the Arctic needs to change. My feeling is that the Bering Straight is probably where the transition should happen and the Arctic needs to be part of the RRS30to10 region, but let me know if you want to try something else.

A few more questions:

• Do you want to include ice-shelf cavities around Antarctica?
• Do you want to ease up on the highest resolution around Antarctica?
• What about the Arctic? Is 10 km resolution higher than we really need?
• Do you want to go with 45 km instead of 60 as the coarsest resolution in the southern Indopacific (as we do in SORRM meshes)?

[xylar]

Here's the next draft:

I'm worried the transition at Bering Straight is too abrupt (from ~10 km to ~60 km resolution over ~1600 km).

[xylar]

We likely want a different name. Just WC?

[vanroekel]

I'm thinking WC_Atl would be a reasonable choice. We may want Atl as just having the resolution min/max could be misleading.

[xylar]

Done!

[xylar]

I think WCAtl (without any underscores) would be better. I think I recall there were problems with punctuation in mesh names (possibly including underscores).

[vanroekel]

WCAtl sounds good to me.

[xylar]

Update based on @vanroekel and @lconlon's new strategy.

[xylar]

Not sure we want topography smoothing.

[vanroekel]

My thinking is we probably do, but for consistency, I'd like to stick with whatever we did for the WC14 mesh.

[xylar]

Okay, I'll check on that.

[xylar]

WC14 didn't include any smoothing, but the SORRM mesh did. For now, I have taken out smoothing to be consistent with WC14, but let me know if you'd prefer to be consistent with SORRM.

[vanroekel]

After pondering this a bit more, I'd say we put our best configuration forward and I think that would include smoothing. Let's switch to consistent with SORRM. Sorry for changing again.

[xylar]

No problem, that's why I asked.

[vanroekel]

Thanks for starting this so quickly @xylar! It is looking pretty great already. A couple replies

1. I agree the Bering Strait transition looks pretty rough. What if we do two things, (1) make the max resolution 45 km as you said instead of 60 and (2) come back to 12 km in the arctic instead of 10km.

2. I don't really want to coarsen Antarctic resolution. The transports through the S. Ocean in SORRM look really good, I'd hate to mess with that much.

3. I'd like to go no ice cavities for one version of the mesh for WC.

[xylar]

@vanroekel, that all sounds good. I'll make revisions and start a new build in Anvil soon.

[vanroekel]

No worries, thanks for asking the questions and the review!

[xylar]

@vanroekel, culling this mesh is going to take many, many hours. I have a 4-hour job just for this on Anvil but I think it's going to time out. I'm going to put in a 10-hour job now to make sure it's done by the time I get up tomorrow.

[xylar]

@vanroekel, one more update.

1. Culling seems to be almost done, actually.
2. I'll see if I can get an initial state. But took out a template that needed to be there, so need to fix that.
3. In the process, I realized that the SORRM mesh doesn't have topography smoothing. I'll make completely sure but the template I though was adding topographic smoothing is, in fact, the one that sets a bunch of other parameters (see above).

[darincomeau]

@xylar do you have the number of horizontal cells (or do you need the culled mesh for that)? For reference, the RRS30to10v3 has 1,445,361.

[xylar]

The new version will show up here. I just submitted an 8-hour job to create it. Fingers crossed:

/home/ac.xylar/e3sm/ac.xylar/compass/test_WCAtl12/ocean/global_ocean/WCAtl12

[xylar]

@xylar do you have the number of horizontal cells (or do you need the culled mesh for that)?

Yes, I need the culled mesh for that but I have it:

	nCells = 851664 ;
	nEdges = 2576825 ;
	nVertices = 1724606 ;

[xylar]

For comparison, SORRMr4 has:

nCells = 569915 ;
nEdges = 1723935 ;
nVertices = 1153614 ;

and my proposed SORRMr5 mesh would have:

nCells = 609013 ;
nEdges = 1841499 ;
nVertices = 1232053 ;

[xylar]

If anyone wants to just look at the culled mesh, it's at:

/home/ac.xylar/e3sm/ac.xylar/compass/test_WC_SO_Atl/WC_Atl12/ocean/global_ocean/WC_Atl12/init/culled_mesh/culled_mesh.nc

The same thing should eventually end up in the newer directory I listed above:

/home/ac.xylar/e3sm/ac.xylar/compass/test_WCAtl12/ocean/global_ocean/WCAtl12

[vanroekel]

This is great progress. Thanks @xylar! I'm happy to see this come in around 2x WC14. Hopefully this helps WC issues!

[xylar]

@vanroekel, the mesh and initial condition are now available on Anvil. I had to fix the namelist options related to the topography (because we are using the GEBCO/BedMachine topography, which has different variable names).

The short test run completed successfully. I didn't try to run the spinup because I think you and @lconlon are going to completely change that anyway.

Please push to this branch or make suggestions for modifying spinup, and I'll run the rest to get the files that @darincomeau and @jonbob need to get this ready in E3SM.

[vanroekel]

@xylar I've modified the spin up. WC never had a chance to switch to the new spin up, so I think it is best to do spin up as we have done but set parameters more closely aligned with target settings for E3SM, I've done so here.

[xylar]

Thanks @vanroekel. I tried spinning up but we seem to be getting CFL violations within the first 10 hours. I have a meeting but will try to play with parameters in an hour or so.

[vanroekel]

Thanks @xylar. Please let me know if you'd like any help tinkering to get this stable. Happy to jump in if it would be useful.

[xylar]

It seems like it's just a time step that's too long. Should be easy to fix.

[xylar]

Reducing the time step in the first stage seems to have done it. Probably overkill but I don't feel like playing around with it.

I'm working on the final stage (the files for E3SM and MPAS-Analysis).

[xylar]

@vanroekel, @darincomeau and @jonbob, the files are in place, as I just emailed you:

/lcrc/group/e3sm/ac.xylar/compass/test_WCAtl12/ocean/global_ocean/WCAtl12/files_for_e3sm/files_for_e3sm/assembled_files_for_upload/inputdata/

@vanroekel, let me know when you think this mesh revision is "finalized" and I'll merge this PR.

[vanroekel]

Approving based on visual inspection, figures posted to the PR and testing from @darincomeau

[vanroekel]

sorry for the letting this one slip @xylar I think we can finalize this and integrate to compass. Thanks again for the quick and great work!

[xylar]

Thanks @vanroekel. Definitely no rush but it's also good to get this merged so we don't forget.