implements the forcing class (wind only) by alicebarthel · Pull Request #426 · E3SM-Project/Omega

alicebarthel · 2026-06-02T15:03:26Z

Adds the new Forcing class (with minimal changes).
Wind Forcing has been moved to the Forcing class and renamed a more general "SrfStress" to account for sea ice stress.
The docs have been updated accordingly.

The longer term vision is that Forcing will own 3 classes of AuxVars for momentum forcing, surface restoring, and surface tracer forcing. This PR is 1/3 to make #418 more digestible.

A key design question I'd like review on is the ownership/dependencies across AuxState vs Forcing.
I have moved all Forcing out of the AuxState (in the code and ctest). Instead, Forcing is initialized at the beginning of the timestep and used by tendencies directly.
Ultimately, the forcing compute will be called at the beginning of a coupling timestep.
The new/modified ctests include:

TendencyTest comparing velocity tendencies with and without SrfStress to ensure that SrfStress can be read it and is utilized by the tendency class.
ForcingTest now tests the computeStressOnEdge capability.

Checklist

alicebarthel · 2026-06-02T18:29:20Z

There has been a call to rename the momentum-related AuxVars from MomForcingAuxVars (or MomentumForcingAuxVars) to the more specific SfcStressForcingAuxVars to make it more understandable to non-ocean modelers. Any objections before I do that? Other naming discussion/suggestions should happen here and will be folded in the full review so that I don't make too many updates.

alicebarthel · 2026-06-02T21:04:06Z

waiting for omega_pr job 53810182 (set up using polaris/rename-wind)

alicebarthel · 2026-06-04T00:58:13Z

Update: omega_pr suite passes on pm-cpu (gnu) using the polaris/rename-wind branch

Polaris `omega_pr` suite

PR build: /global/homes/a/abarthel/pscratch/omega_scratch/develop-cpu
PR workdir: /global/homes/a/abarthel/pscratch/polaris-scratch/add-forcing-class-cpu
Machine: pm-cpu
Compiler: gnu
Build type: Debug
Log: /global/homes/a/abarthel/pscratch/polaris-scratch/add-forcing-class-cpu/polaris_omega_pr.o53863290
Result: All tests passed

alicebarthel · 2026-06-04T01:32:47Z

After the last commit (renaming), I simply did a build and ctest on pm-cpu (gnu). All pass.
Reviewers, it'd be great if you run tests on other machines. Note that you need the new polaris dev branch to run the omega_pr suite.

katsmith133

@alicebarthel I took a quick look at this and just had mostly clarification questions. I guess I am a bit confused that you've pulled the surface stress forcing out of the Aux stuff, but its still kind of an auxiliary variable? Can you explain why the partial move? Perhaps you've hashed this out with others, so sorry for the repetition.

Also, there is almost no commenting. I think it would be helpful to add commenting to the new files you've added and the new testing.

katsmith133 · 2026-06-06T15:33:53Z

@@ -1,24 +1,24 @@
-#ifndef OMEGA_AUX_WIND_H
-#define OMEGA_AUX_WIND_H
+#ifndef OMEGA_AUX_MOM_H


I thought we wanted to stay away from mom? Maybe this was skipped.

Good catch, thanks!

katsmith133 · 2026-06-06T15:38:20Z

+   Err += OmegaConfig->get(SfcStressConfig);
+
+   std::string SfcStressInterpTypeStr;
+   Err += SfcStressConfig.get("InterpType", SfcStressInterpTypeStr);


I assume the InterpType will stay the same for all forcing? Or rather if it doesn't this will be modified when those things are added?

yes, at this stage, this interp is only used for the surface stress, not the other fluxes which are digested straight from the cell-centered values.

katsmith133 · 2026-06-06T15:40:41Z

@@ -0,0 +1,138 @@
+#include "Forcing.h"


A little bit of comments here like in AuxiliaryState.cpp would be good

done, but let me know if you need more detail.

katsmith133 · 2026-06-06T15:44:42Z

I'm assuming that as we add more forcing this will become more generic?

@katsmith133 What other kinds of stress forcing are you thinking we would add?

katsmith133 · 2026-06-06T15:48:17Z

@@ -0,0 +1,277 @@
+//===-- ocn/ForcingTest.cpp - Forcing Unit Test ---------------*- C++ -*-===//


Its not entirely clear to me why some of the testing is in AuxVars and some is here?

Also, there is almost no commenting in this file. I would prefer if there were at least a little bit more. Helps to understand what you are trying to test and what the expected outcome is.

@katsmith133 this was historical. Since the Stress was still in AuxVars, I had left the testing in AuxVars. I have now moved the SfcStress testing to the ForcingTest. That set-up is much cleaner now, thanks for your feedback!

alicebarthel · 2026-06-10T18:41:36Z

Thanks @katsmith133 for your review. I can definitely add comments for clarification.
@sbrus89 @brian-oneill could you give me a sense of your timeline for this review?

cbegeman · 2026-06-12T14:05:25Z

@alicebarthel I noticed you have a polaris branch. Feel free to open the Polaris PR with the Omega PR required tag.

xylar · 2026-06-12T14:16:31Z

@cbegeman, my understanding is that's E3SM-Project/polaris#602.

cbegeman · 2026-06-12T14:47:46Z

@cbegeman, my understanding is that's E3SM-Project/polaris#602.

Thanks for pointing that out @xylar. Sorry, @alicebarthel, I didn't see the PR linked.

cbegeman · 2026-06-12T14:54:01Z

@katsmith133 What other kinds of stress forcing are you thinking we would add?

brian-oneill

I ran the ctests and omega_pr suite successfully on Frontier with craygnu and craygnu-mphipcc. Found a few other instances, mostly comments where might want to change "wind" to "surface stress".

My take on some of the other comments above:

Instead of having the surface stress and other future forcing contributions mixed in with the other aux vars, maybe they should go in their own forcingVars directory. Or if there are just going to be a small number of them, they could just be defined in Forcing.h and Forcing.cpp
A small header in Forcing.h and Forcing.cpp, similar to what's in most other header/source files, along with some comments, go a long way towards making the code look cleaner and more readable.

brian-oneill · 2026-06-15T01:59:02Z

 | TracerDiffOnCell | horizontal diffusion of thickness-weighted tracers
 | TracerHyperDiffOnCell | biharmonic horizontal mixing of thickness-weighted tracers
-| WindForcingOnEdge | forcing by wind stress, defined on edges
+| SfcStressForcingOnEdge | forcing by wind stress, defined on edges


Suggested change

| SfcStressForcingOnEdge | forcing by wind stress, defined on edges

| SfcStressForcingOnEdge | forcing by surface stress, defined on edges

brian-oneill · 2026-06-15T02:00:08Z

Suggested change

"zonal surface stress", // long name/describe

brian-oneill · 2026-06-15T02:00:42Z

-   // Meridional wind stress
   auto MeridStressCellField =
       Field::create(MeridStressCell.label(),  // field name
                     "meridional wind stress", // long Name or description


Suggested change

"meridional wind stress", // long Name or description

"meridional surface stress", // long Name or description

brian-oneill · 2026-06-15T02:03:02Z

@@ -620,23 +622,25 @@ void Tendencies::computeVelocityTendenciesOnly(
   Pacer::stop("Tend:computeVelocityVAdvTend", 2);

   // Compute wind forcing


Suggested change

// Compute wind forcing

// Compute surface stress forcing

brian-oneill · 2026-06-15T02:03:38Z

@@ -301,12 +301,12 @@ class VelocityHyperDiffOnEdge {
 };

 /// Wind forcing


Suggested change

/// Wind forcing

/// Surface stress forcing

brian-oneill · 2026-06-15T02:04:18Z

+   SfcStressForcingOnEdge(const HorzMesh *Mesh, const VertCoord *VCoord);

   /// The functor takes the edge index, vertical chunk index, and arrays for
   /// normal wind stress and edge pseudo-thickness, outputs tendency array


Suggested change

/// normal wind stress and edge pseudo-thickness, outputs tendency array

/// normal surface stress and edge pseudo-thickness, outputs tendency array

brian-oneill · 2026-06-15T02:05:04Z

+   Array2DReal ExactSfcStressForcing("ExactSfcStressForcing", Mesh->NEdgesOwned,
+                                     NVertLayers);

   // Note: this computes wind forcing at every layer


Suggested change

// Note: this computes wind forcing at every layer

// Note: this computes surface stress forcing at every layer

I am less worried about the comments in the test - since in the test, we can just decide that our simple case only has wind forcing. But for clarity/consistency, I can change it here too.

brian-oneill · 2026-06-15T02:05:41Z

+       ExactSfcStressForcing, EdgeComponent::Normal, Geom, Mesh,
+       ExchangeHalos::No);

   // Reset wind forcing to zero below the surface


Suggested change

// Reset wind forcing to zero below the surface

// Reset surface stress forcing to zero below the surface

alicebarthel · 2026-06-17T12:59:19Z

I think I have addressed all comments.
Ctests pass on pm-cpu (gnu).
I have run out of time for other tests on pm due to maintenance.

alicebarthel · 2026-06-17T13:29:47Z

of course I just noticed I forgot a couple of comment changes suggested by brian in the Tests...

sbrus89 · 2026-06-17T17:15:10Z

@alicebarthel, I'm seeing compile errors in the ForcingTest on Frontier:

/autofs/nccs-svm1_proj/cli115/sbrus/polaris_worktrees/testing/e3sm_submodules/Omega/components/omega/test/ocn/ForcingTest.cpp: In function ‘int {anonymous}::testForcingInitAndConfig()’:/autofs/nccs-svm1_proj/cli115/sbrus/polaris_worktrees/testing/e3sm_submodules/Omega/components/omega/test/ocn/ForcingTest.cpp:147:20: error: ‘class OMEGA::Forcing’ has no member named ‘SfcStressForcingAux’; did you mean ‘SfcStressForcing’?
  147 |    if (DefForcing->SfcStressForcingAux.InterpChoice != ExpectedChoice) {
      |                    ^~~~~~~~~~~~~~~~~~~
      |                    SfcStressForcing
/autofs/nccs-svm1_proj/cli115/sbrus/polaris_worktrees/testing/e3sm_submodules/Omega/components/omega/test/ocn/ForcingTest.cpp: In function ‘int {anonymous}::testForcingComputeAll()’:
/autofs/nccs-svm1_proj/cli115/sbrus/polaris_worktrees/testing/e3sm_submodules/Omega/components/omega/test/ocn/ForcingTest.cpp:171:40: error: ‘class OMEGA::Forcing’ has no member named ‘SfcStressForcingAux’; did you mean ‘SfcStressForcing’?
  171 |    auto &ZonalStressCell = DefForcing->SfcStressForcingAux.ZonalStressCell;
      |                                        ^~~~~~~~~~~~~~~~~~~
      |                                        SfcStressForcing
/autofs/nccs-svm1_proj/cli115/sbrus/polaris_worktrees/testing/e3sm_submodules/Omega/components/omega/test/ocn/ForcingTest.cpp:172:40: error: ‘class OMEGA::Forcing’ has no member named ‘SfcStressForcingAux’; did you mean ‘SfcStressForcing’?
  172 |    auto &MeridStressCell = DefForcing->SfcStressForcingAux.MeridStressCell;
      |                                        ^~~~~~~~~~~~~~~~~~~
      |                                        SfcStressForcing
/autofs/nccs-svm1_proj/cli115/sbrus/polaris_worktrees/testing/e3sm_submodules/Omega/components/omega/test/ocn/ForcingTest.cpp:173:40: error: ‘class OMEGA::Forcing’ has no member named ‘SfcStressForcingAux’; did you mean ‘SfcStressForcing’?
  173 |    auto &NormalStress    = DefForcing->SfcStressForcingAux.NormalStressEdge;
      |                                        ^~~~~~~~~~~~~~~~~~~
      |                                        SfcStressForcing

alicebarthel · 2026-06-17T17:24:11Z

@sbrus89 I am sorry! I know I corrected these but somehow must have forgotten to push the changes.
Let me check out that branch on Chrysalis and correct these.

alicebarthel · 2026-06-19T18:48:56Z

I think this new version addresses all the comments. The design is a cleaner separation of ForcingVars outside of AuxiliaryVars. I mirrored the existence of a PLANE and SPHERE test for the forcing test (based on AuxiliaryVarsTest) to maintain the same test coverage of WindForcingAux-->SfcStressForcing.

Currently, ctests:

pm-cpu(gnu): 100% tests passed, 0 tests failed out of 42
pm-gpu: 100% tests passed, 0 tests failed out of 42

alicebarthel · 2026-06-19T18:49:56Z

Polaris `omega_pr` suite

PR build: /global/homes/a/abarthel/pscratch/omega_scratch/add-forcing-class-cpu-gnu2
PR workdir: /global/homes/a/abarthel/pscratch/polaris-scratch/add-forcing-class-cpu3
Machine: pm-cpu
Compiler: gnu
Build type: Debug
Log: not found
Result: All tests passed

alicebarthel · 2026-06-19T18:56:39Z

@mwarusz I was able to pass the testWindForcingAuxVars (moved to ForcingTest and renamed testSfcStressForcingVars) by following the structure in AuxiliaryVarsTest.cpp producing a SPHERE and PLANE test. Hopefully that was an ok plan? Could you review the test section to confirm?

alicebarthel · 2026-06-19T20:07:09Z

omega_pr on pm-gpu
(set up using polaris/rename-wind with the fix from E3SM-Project/polaris#618)

Polaris `omega_pr` suite

PR build: /global/homes/a/abarthel/pscratch/omega_scratch/add-forcing-class-gpu2
PR workdir: /global/homes/a/abarthel/pscratch/polaris-scratch/add-forcing-class-gpu
Machine: pm-gpu
Compiler: gnugpu
Build type: Debug
Log: /global/homes/a/abarthel/pscratch/polaris-scratch/add-forcing-class-gpu/polaris_omega_pr.o54715042
Result: All tests passed

cbegeman · 2026-06-20T17:40:20Z

@alicebarthel Is this PR expected to be BFB with develop for wind-forced tests?

cbegeman · 2026-06-22T14:42:11Z

Polaris `omega_pr` suite

Baseline workdir: /lcrc/group/e3sm/ac.cbegeman/scratch/MPAS-Ocean-output/chrys/polaris-main-omega-20260512
Baseline build: /lcrc/group/e3sm/ac.cbegeman/scratch/MPAS-Ocean-output/chrys/polaris-main-omega-20260512/build
PR build: /lcrc/group/e3sm/ac.cbegeman/scratch/MPAS-Ocean-output/chrys/polaris-wind-omega-wind-20260620/build
PR workdir: /lcrc/group/e3sm/ac.cbegeman/scratch/MPAS-Ocean-output/chrys/polaris-wind-omega-wind-20260620
Machine: chrysalis
Partition: compute
Compiler: oneapi-ifx
Build type: Release
Log: /lcrc/group/e3sm/ac.cbegeman/scratch/MPAS-Ocean-output/chrys/polaris-wind-omega-wind-20260620/polaris_omega_pr.o1238760
Result: All tests passed

alicebarthel · 2026-06-22T16:53:35Z

@alicebarthel Is this PR expected to be BFB with develop for wind-forced tests?

Yes, I would hope so. The functions have been moved but the implementation should be the same. Let me know if it does not come out as BFB.

cbegeman · 2026-06-23T14:09:58Z

@alicebarthel I didn't get a chance to check the velocity field of the barotropic_gyre case to make sure that this branch and develop are still reading in the forcing field. Do you have time to do this?

alicebarthel · 2026-06-23T14:24:48Z

@cbegeman can you share the path to the long forward? I can have a look at the outout.

cbegeman · 2026-06-23T14:27:27Z

@alicebarthel I didn't have time to look at it before chrys went down so it would be best if you just ran it as setup with your latest suite on perlmutter.

…fcStress Compute check

sbrus89

I tested a local merge of this branch (in conjuction with a local merge of E3SM-Project/polaris#602) on Frontier with craygnu-mphipcc. Both the omega_pr suite and CTests pass.

Polaris `omega_pr` suite

Baseline workdir: /ccs/home/brus/run/polaris_omega_pr_baseline_craygnu-mphipcc_mpich
Baseline build: /ccs/home/brus/run/polaris_omega_pr_baseline_craygnu-mphipcc_mpich/build
PR build: /ccs/home/brus/run/polaris_forcing_class_omega_pr_craygnu-mphipcc_mpich/build
PR workdir: /ccs/home/brus/run/polaris_forcing_class_omega_pr_craygnu-mphipcc_mpich
Machine: frontier
Partition: batch
Compiler: craygnu-mphipcc
Build type: Release
Log: not found
Result: All tests passed

sbrus89 · 2026-06-24T20:57:21Z

@alicebarthel, let me know when you're able to verify the state of the long_forward barotropic gyre case and I'll merge.

alicebarthel · 2026-06-24T22:25:07Z

Thanks to @cbegeman for running a barotropic_gyre/long_forward.

To me, this confirms that the wind forcing is adequately read in as forcing for this case.
@cbegeman?
@sbrus89 I am ready for a merge.

cbegeman · 2026-06-25T00:01:03Z

@alicebarthel Thank you so much for taking the time to confirm the behavior! And for this very important development

alicebarthel changed the title ~~Omega/add forcing class~~ implement the forcing class (for wind only) Jun 3, 2026

alicebarthel requested review from brian-oneill, katsmith133 and sbrus89 June 3, 2026 13:41

alicebarthel changed the title ~~implement the forcing class (for wind only)~~ implements the forcing class (wind only) Jun 3, 2026

alicebarthel marked this pull request as ready for review June 4, 2026 01:33

katsmith133 reviewed Jun 6, 2026

View reviewed changes

cbegeman reviewed Jun 12, 2026

View reviewed changes

brian-oneill reviewed Jun 15, 2026

View reviewed changes

sbrus89 reviewed Jun 16, 2026

View reviewed changes

Comment thread components/omega/src/ocn/.Forcing.h.swo Outdated

sbrus89 reviewed Jun 16, 2026

View reviewed changes

Comment thread components/omega/src/ocn/.Forcing.h.swp Outdated

sbrus89 reviewed Jun 16, 2026

View reviewed changes

Comment thread components/omega/src/ocn/forcingVars/SfcStressForcingVars.h

alicebarthel force-pushed the omega/add-forcing-class branch 2 times, most recently from 8049c5b to 94de041 Compare June 19, 2026 17:57

alicebarthel requested a review from sbrus89 June 19, 2026 19:14

alicebarthel added 12 commits June 23, 2026 12:01

implemented Forcing class with wind only

df35b5c

renaming Wind-->SrfStress

ebd75e4

adding a ForcingTest

1d50217

updated docs

a0c10c1

clean-up: removed unnecessary dependencies on Forcing

e705646

increased test coverage in ForcingTest and TendencyTest

41af2de

renaming MomAux and Srf->Sfc

0e092dc

PR review modifications

a41f754

refactored ForcingTest into Planar and Sphere version to enable the S…

6d5733c

…fcStress Compute check

rename windStress to srfStress

3dca4af

more PR clean-up

7e2d0e3

clean up tests

362ae2e

alicebarthel force-pushed the omega/add-forcing-class branch from 523ee23 to 362ae2e Compare June 23, 2026 20:14

katsmith133 mentioned this pull request Jun 23, 2026

Change shear mixing to LMD94 formulation and add ri smoothing #352

Merged

17 tasks

sbrus89 self-assigned this Jun 24, 2026

sbrus89 approved these changes Jun 24, 2026

View reviewed changes

brian-oneill approved these changes Jun 24, 2026

View reviewed changes

sbrus89 merged commit 0c58868 into E3SM-Project:develop Jun 24, 2026
1 of 2 checks passed

cbegeman mentioned this pull request Jun 25, 2026

renamed windForcing to match upcoming omega options E3SM-Project/polaris#602

Merged

cbegeman mentioned this pull request Jul 4, 2026

Add del4 variant of overflow test case E3SM-Project/polaris#642

Draft

5 tasks

		@@ -0,0 +1,277 @@
		//===-- ocn/ForcingTest.cpp - Forcing Unit Test ---------------- C++ --===//

	\| SfcStressForcingOnEdge \| forcing by wind stress, defined on edges
	\| SfcStressForcingOnEdge \| forcing by surface stress, defined on edges

	"meridional wind stress", // long Name or description
	"meridional surface stress", // long Name or description

		@@ -620,23 +622,25 @@ void Tendencies::computeVelocityTendenciesOnly(
		Pacer::stop("Tend:computeVelocityVAdvTend", 2);

		// Compute wind forcing

		@@ -301,12 +301,12 @@ class VelocityHyperDiffOnEdge {
		};

		/// Wind forcing

	/// normal wind stress and edge pseudo-thickness, outputs tendency array
	/// normal surface stress and edge pseudo-thickness, outputs tendency array

	// Note: this computes wind forcing at every layer
	// Note: this computes surface stress forcing at every layer

	// Reset wind forcing to zero below the surface
	// Reset surface stress forcing to zero below the surface

Uh oh!

Conversation

alicebarthel commented Jun 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

alicebarthel commented Jun 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

alicebarthel commented Jun 2, 2026

Uh oh!

alicebarthel commented Jun 4, 2026

Polaris omega_pr suite

Uh oh!

alicebarthel commented Jun 4, 2026

Uh oh!

katsmith133 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

alicebarthel commented Jun 10, 2026

Uh oh!

cbegeman commented Jun 12, 2026

Uh oh!

xylar commented Jun 12, 2026

Uh oh!

cbegeman commented Jun 12, 2026

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

brian-oneill left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

alicebarthel commented Jun 17, 2026

Uh oh!

alicebarthel commented Jun 17, 2026

Uh oh!

sbrus89 commented Jun 17, 2026

alicebarthel commented Jun 2, 2026 •

edited

Loading

alicebarthel commented Jun 2, 2026 •

edited

Loading

Polaris `omega_pr` suite

Polaris `omega_pr` suite

alicebarthel commented Jun 19, 2026 •

edited

Loading

Polaris `omega_pr` suite

Polaris `omega_pr` suite

Polaris `omega_pr` suite

alicebarthel commented Jun 24, 2026 •

edited

Loading