search

MPAS-Dev / compass

Configuration Of MPAS Setups
Other
10 stars 37 forks source link

For initialization with ice-shelf cavities, adjust SSH not land ice pressure #813

Closed xylar closed 2 months ago

xylar commented 5 months ago

Checklist

xylar commented 5 months ago

This merge requires https://github.com/E3SM-Ocean-Discussion/E3SM/pull/90

xylar commented 4 months ago

This worked for both the Icos30 and SORRM meshes.

xylar commented 4 months ago

Testing

I have successfully run the mesh, init and performance_test tests on the IcoswISC30 and IcoswISC240 meshes. I also created new cache files and ran the pr suite. All tests were on Chrysalis with Intel and OpenMPI.

xylar commented 4 months ago

Waiting on https://github.com/E3SM-Project/E3SM/pull/6375

xylar commented 2 months ago

I rebased this and successfully ran the pr suite on Chrysalis (Intel, OpenMPI) so I think this is ready to go in once the E3SM-Project submodule has been updated with the changes from https://github.com/E3SM-Project/E3SM/pull/6375.

mark-petersen commented 2 months ago

Testing. See notes at https://github.com/E3SM-Project/E3SM/pull/6375#issuecomment-2181297050

mark-petersen commented 2 months ago

Tested together with https://github.com/E3SM-Project/E3SM/pull/6375. Successfully ran QU240, QUwISC240, and IcoswISC240. I set everything up for RRS6to18 and my new Arctic 3km mesh, and I am in the process of running it now. Thanks!

mark-petersen commented 2 months ago

@xylar, when I use this branch with https://github.com/E3SM-Project/E3SM/pull/6375 for the executable and create the RRS6to18, I can successfully complete the mesh steps (base_mesh, cull_mesh, remap_topography).

In the init/adjust_ssh steps, it is successful for these steps:

01_init_with_draft_from_constant_density
02_ssh_from_surface_density
03_init
04_adjust_ssh
05_init

and then dies in 06_adjust_ssh with a state validation failure after the second time-step of 3 minutes. It is here on Chicoma:

/lustre/scratch5/mpeterse/runs/aerosi_240620/RRS6to18/ocean/global_ocean/RRSwISC6to18/WOA23/init/adjust_ssh/06_adjust_ssh

Did you have similar problems? I can try reducing the time step, but my run should be the same as the one you used to create the RRS6to18 with this branch. Can you post the path here of the RRS6to18 init steps so I can compare the namelist files? Thanks. Perhaps you changed a namelist parameter locally and it didn't get back into this PR.

Also, I see this file:

compass/ocean/tests/global_ocean/init/ssh_adjustment.py

but where are the time steps specified for each mesh and each step for the init/adjust_ssh process? I was expecting to see a file like this one for adjust_ssh:

compass/ocean/tests/global_ocean/mesh/rrs6to18/dynamic_adjustment.yaml

but I don't see it.

mark-petersen commented 2 months ago

Another question: I am generating the initial conditions for the new Arctic refined meshes (3km and another 2.4km Arctic, RRSwISC6to18 background). I am doing this with both the old way (current compass) and this PR. We hope to use these meshes for the Aerosi/RASSM project for several years.

Would you advise me to use the initial conditions from the new process, with this PR? I assume that we catch the important updates for the new HR IC that way.

xylar commented 2 months ago

@mark-petersen, didn't use this branch to create the RRSwISC6to18 mesh and I haven't tried that mesh with these changes.

xylar commented 2 months ago

where are the time steps specified for each mesh and each step for the init/adjust_ssh process?

It's defined in the code for the init test case: https://github.com/MPAS-Dev/compass/blob/f9843fefe341a2569d8ee4dcc2a1d9836d5c99fa/compass/ocean/tests/global_ocean/init/__init__.py#L82-L147

I was expecting to see a file like this one for adjust_ssh: compass/ocean/tests/global_ocean/mesh/rrs6to18/dynamic_adjustment.yaml but I don't see it.

I suppose it could be possible to treat init in a similar way to dynamic adjustment. I didn't implement it that way here.

xylar commented 2 months ago

Would you advise me to use the initial conditions from the new process, with this PR? I assume that we catch the important updates for the new HR IC that way.

No, I would advise you to use the current compass. This process is important for meshes that will be used (eventually) for Antarctic Ice Sheet coupling, which I do not believe you are planning to do.

mark-petersen commented 2 months ago

Thanks, that is all very helpful information.

xylar commented 2 months ago

In the meantime, tough, I will attempt to get RRSwISC6to18 working with these changes.

xylar commented 2 months ago

Testing

I ran the pr suite on Chrysalis with Intel and OpenMPI one more time after the rebase and submodule update. I used the final test from #834 as a baseline. As expected, Icos240, IcoswISC240 and IcoswISC test cases pass pass execution and test comparison but fail baseline comparison.