Support for single process single cell standalone test cases by akturner · Pull Request #5117 · E3SM-Project/E3SM

akturner · 2022-08-09T14:47:51Z

Added several options for controlling which thermodynamic processes are active
Added option for conservation AM to include ocean heat, mass, energy content in analysis with mixed layer model
Added another mixed layer ocean model that mimics ocean model in E3SM for standalone testing
Added another forcing option to support idealized single cell test case

Fixes #5118
Fixes #5119

[NML]
[BFB]

eclare108213 · 2022-08-15T15:13:53Z

    case ("ncar")
       call init_oceanic_forcing_ncar(domain)
+    case ("ncar_free_surface")
+       call init_oceanic_forcing_ncar_free_surface(domain)


I think the nomenclature here is incorrect. POP is a free-surface ocean model, although it is constant volume. So the original ncar forcing data is usable for a "free surface" ocean model.

Good point, I have renamed the option to som_varying_thickness

eclare108213 · 2022-08-15T15:16:42Z

+!> \details Representation of ocean processes and coupling in e3sm. This
+!>  routine mimics the ocean/sea ice coupling in e3sm using MPAS-Ocn,
+!>  including updates to ocean thickness, temperature and salinity from
+!>  interactions with sea ice and frazil formation.


Is there a (design) document somewhere that explains these differences in true detail? We will need that for the Icepack documentation.

No design document. This is an additional subroutine in MPAS rather than the column package so doesn't necessarily need to be in icepack. Making this proxy was a good way to understand how frazil is currently done and I plan to try a better formulation so this may change soon anyway.

eclare108213 · 2022-08-15T15:19:27Z

+               config_use_prescribed_ice, &
+               config_use_congelation_basal_melt, &
+               config_use_lateral_melt, &
+               config_use_latent_processes)


What is the computational burden of adding 'if' conditionals for all of these individual processes? They have to be parsed for each category, on each grid cell with ice, perhaps multiple times per timestep.

@eclare108213: Good point. I have recoded this so that there are no conditionals now, just multiplication by a real flag taking values (0,1). I will test performance of these changes as well

rljacob · 2022-08-25T17:30:12Z

@akturner please respond to review comments.

akturner · 2022-09-15T14:33:49Z

Single cell test case for idealized processes added here: MPAS-Dev/MPAS-Seaice_standalone_framework#13

eclare108213

Thanks for getting rid of the conditionals - this is a more elegant approach in my opinion and perhaps not as computationally impactful. The title of this pull request does not fully represent what is in it -- the new conservation checks are particularly important. I'm curious to know whether these tests have turned up anything surprising or needing to be fixed, but answering that question is not necessary for this PR. Approving.

rljacob · 2022-10-06T17:35:59Z

telecon: waiting on review from @alicebarthel who is out of town.

alicebarthel · 2022-10-21T17:39:13Z

I ran a G-case 1-year test with this code. It ran successfully and the sea ice fluxes look consistent visually (see attached plots). My automated check technically failed but I think it may be an issue with the test itself. Looking into this now.

alicebarthel · 2022-10-21T17:40:17Z

My only suggestion at this stage would be to change the streams.seaice file to replace:
clobber_mode="append" with clobber_mode="truncate"
to match other streams (and avoid errors popping up when restarting/overwriting output files)

eclare108213 · 2022-10-21T17:51:02Z

What are the units on the y-axes? E.g. I'd expect the radiative fluxes to be ~50x larger in magnitude.

alicebarthel · 2022-10-21T22:25:27Z

@eclare108213 These fluxes are all in W/m2. Note that these are only the sea-ice ones. The atm/ocn radiative fluxes are much bigger (~+/-150 for net SW and ~+/-200-300 for LW).
@akturner Can you remind us if there is area scaling: e.g. are we summing over the sea ice area then dividing by the whole hemisphere area? That could explain the order of magnitude difference. I have vague memories of discussing this with you a while back.

akturner · 2022-10-24T18:46:39Z

@alicebarthel: Which streams file are you referring to?

akturner · 2022-10-24T18:47:28Z

@eclare108213, @alicebarthel : I believe these are fluxes over whole earth surface?

eclare108213 · 2022-10-24T18:52:38Z

But they are for a single-cell column configuration. Can they be made consistent with that view of the "world"?

alicebarthel · 2022-10-25T14:34:17Z

@akturner I meant the conservationCheck AM in streams.seaice. The code I checked out had them in "append" mode instead of "truncate" mode which means errors popped up when trying to overwrite the AM output files. Looking at the files, it does not seem to be something introduced by your changes, so I am happy to approve the PR.
@eclare108213 The test case I ran here is a "normal" (global) G-case, not the single-column mode. Adrian had asked me to check introducing these changes did not modify the standard e3sm flux check I have worked on.
@eclare108213 does that address your concern or do we need to double-check what a "hemispheric scaling" would look like in single-column mode? Is the idea to have the same sea-ice vs. coupler check for flux consistency in a single-column test? In that case, I don't know how the coupler would handle writing out the SH and NH fluxes...

alicebarthel

I was asked to check that these changes did not impact the standard e3sm fluxes. I ran a G-case and checked the consistency of sea-ice heat fluxes between the Seaice AM and the coupler log. These match to <e-8.

eclare108213 · 2022-10-25T15:04:55Z

I see, thanks for the clarification @alicebarthel. Eventually we would like to look at MPAS-SI output in realistic magnitudes for at least one sea ice grid point in each hemisphere, but others are working on that problem now.

rljacob · 2022-11-03T17:29:15Z

telecon notes: needs a rebase and build system fixes.

Added several options for controlling which thermodynamic processes are active Added option for conservation AM to include ocean heat, mass, energy content in analysis with mixed layer model Added another mixed layer ocean model that mimics ocean model in E3SM for standalone testing Added another forcing option to support idealized single cell test case

Active column processes now toggled with multiplication by real rather than conditional for performance Renamed ncar_free_surface forcing type to som_varying_thickness

Removed trailing whitespace from Registry namelist options

jonbob · 2022-11-21T18:36:34Z

@akturner - the rebase and bld updates look good. I'll test today and hopefully merge soon

…to next (PR #5117) Merge branch 'akturner/mpas-seaice/idealized-single-cell-testcase' into next (PR #5117) Support for single process single cell standalone test cases Changes to the mpas-seaice standalone testing capabilities: * Added several options for controlling which thermodynamic processes are active * Added option for conservation AM to include ocean heat, mass, energy content in analysis with mixed layer model * Added another mixed layer ocean model that mimics ocean model in E3SM for standalone testing * Added another forcing option to support idealized single cell test case Fixes #5118 Fixes #5119 [NML] [BFB]

jonbob · 2022-11-21T19:47:48Z

passes:

SMS_D_Ld1.T62_oEC60to30v3.DTESTM.chrysalis_intel
SMS_D_Ld1.ne30pg2_EC30to60E2r2.WCYCL1850.chrysalis_intel.allactive-wcprod
ERS.ne11_oQU240.WCYCL1850NS.chrysalis_intel

with expected NML DIFFs

merged to next

jonbob · 2022-11-22T17:11:14Z

merged to master and expected NML DIFFs blessed

akturner requested review from alicebarthel and eclare108213 August 9, 2022 14:48

akturner added MPAS-seaice Concerning the MPAS-seaice model BFB PR leaves answers BFB labels Aug 9, 2022

rljacob assigned jonbob Aug 10, 2022

eclare108213 reviewed Aug 15, 2022

View reviewed changes

akturner requested a review from eclare108213 September 26, 2022 15:16

eclare108213 approved these changes Sep 26, 2022

View reviewed changes

alicebarthel approved these changes Oct 25, 2022

View reviewed changes

jonbob added the bug fix PR label Oct 31, 2022

akturner added 3 commits November 17, 2022 16:30

Changes to how active process flags implemented

f395679

Active column processes now toggled with multiplication by real rather than conditional for performance Renamed ncar_free_surface forcing type to som_varying_thickness

Added new namelist options to build-namelist

c28aeff

Removed trailing whitespace from Registry namelist options

akturner force-pushed the akturner/mpas-seaice/idealized-single-cell-testcase branch from 3083180 to c28aeff Compare November 18, 2022 17:43

jonbob added the NML label Nov 21, 2022

jonbob merged commit 9495ddd into E3SM-Project:master Nov 22, 2022

Uh oh!

Conversation

akturner commented Aug 9, 2022 • edited by jonbob Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

eclare108213 Aug 15, 2022

Choose a reason for hiding this comment

Uh oh!

akturner Sep 15, 2022

Choose a reason for hiding this comment

Uh oh!

eclare108213 Aug 15, 2022

Choose a reason for hiding this comment

Uh oh!

akturner Sep 15, 2022

Choose a reason for hiding this comment

Uh oh!

eclare108213 Aug 15, 2022

Choose a reason for hiding this comment

Uh oh!

akturner Sep 15, 2022

Choose a reason for hiding this comment

Uh oh!

rljacob commented Aug 25, 2022

Uh oh!

akturner commented Sep 15, 2022

Uh oh!

eclare108213 left a comment

Choose a reason for hiding this comment

Uh oh!

rljacob commented Oct 6, 2022

Uh oh!

alicebarthel commented Oct 21, 2022

Uh oh!

alicebarthel commented Oct 21, 2022

Uh oh!

eclare108213 commented Oct 21, 2022

Uh oh!

alicebarthel commented Oct 21, 2022

Uh oh!

akturner commented Oct 24, 2022

Uh oh!

akturner commented Oct 24, 2022

Uh oh!

eclare108213 commented Oct 24, 2022

Uh oh!

alicebarthel commented Oct 25, 2022

Uh oh!

alicebarthel left a comment

Choose a reason for hiding this comment

Uh oh!

eclare108213 commented Oct 25, 2022

Uh oh!

rljacob commented Nov 3, 2022

Uh oh!

jonbob commented Nov 21, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jonbob commented Nov 21, 2022

Uh oh!

jonbob commented Nov 22, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

akturner commented Aug 9, 2022 •

edited by jonbob

Loading

jonbob commented Nov 21, 2022 •

edited

Loading