Dynamical Sea Ice Model #69

LenkaNovak · 2022-06-23T18:10:24Z

Purpose and Content

the dynamical sea ice from PR#50 needs to be tested and finalized as a stand-alone driver (addressing SDI Implement prescribed sea ice #42 ), separate from the moist_mpi_earth directory, which will contain the prescribed sea ice and bucket model (PR#51 that addresses SDI Land implement bucket model in Land #41 ).

To do:

separate example from the bucket / prescribed (moist_mpi_earth) directory
energy conservation test - , neglecting the initial equilibration over the first 2 timesteps, up to 0.1% error after 27d
- first 100 timesteps (dt =200s)
- 27 days
physical test - h_ice (m)

h_ice.mp4

Future PRs:

update for the latest interface (after Earth bucket #51 is ready)
add area fraction (apply to all examples)
prescribe albedo
make an abstraction for the sea ice model + thermal slab
replace heavisides with smooth functions
add regression test
add buildkite test
investigate if the small conservation error grows for longer integrations

Benefits and Risks

benefit: finalizes the sea ice model development, which will be useful when testing all available ODE surface-model setups, when appraising the AMIP configuration.
risk: the dynamical sea ice is strictly speaking not needed for the first AMIP simulation (however, this is already implemented and only requires some time for testing. Completing it now will save time later. )

Linked Issues

(Provide references to any link issues. Use closes #issuenum to automatically close an open issue)

Fixes #
Closes #

PR Checklist

LenkaNovak · 2022-06-27T21:20:44Z

@kmdeck this is ready to be merged as a checkpoint for the dynamic sea ice. It is now in a separate directory, so it shouldn't conflict with the Bucket PR, and we can pick it up again after AMIP. If you're ok with it, I'll wait for you to merge #57 and then rebase on that. As for documentation, the sea ice model is documented in the experiments/ClimaCore directory, which we can copy over when we have an abstraction for the sea ice model next week.

kmdeck · 2022-06-28T16:19:14Z

Manifest.toml

@@ -0,0 +1,929 @@
+# This file is machine-generated - editing it directly is not advised
+
+julia_version = "1.7.2"


Why did the Manifest for ClimaCoupler change with this PR?

Ooh! Good catch! I think the formatting did this.

kmdeck · 2022-06-28T16:26:26Z

experiments/AMIP/moist_mpi_earth_dynamical_sea_ice/mpi/mpi_postprocess.jl

+            global_sol_u_slab[i] = Fields.Field(global_Y_slab, global_h_space)
+        end
+    end
+    if ClimaComms.iamroot(comms_ctx)


This seems hardcoded to a slab+atmos. Will this work for this example? If not, should it be removed from the PR?

Good point - btw, if we wanted to use it (say in the Bucket example), we'd need to combine all of the surface fields into one, like we do in the coupling loop in the coupler_driver.jl. But since we don't need MPI for this example imminently, and since we'll be changing the interface for this in the bucket example (and then broadcast it to the other examples), I think deleting this here makes sense. :)

kmdeck · 2022-06-28T16:27:32Z

experiments/AMIP/moist_mpi_earth_dynamical_sea_ice/slab_ice/slab_init.jl

+end
+
+
+function semster_zero_layer_model(T_ml, T_sfc, h_ice, F_atm, ∂F_atm∂T_sfc, p, Δt)


kmdeck · 2022-06-28T16:30:05Z

experiments/AMIP/moist_mpi_earth_dynamical_sea_ice/slab_ice/slab_init.jl

+    Y = u
+    FT = eltype(dY)
+
+    if p.Ya.prescribed_sic_data !== nothing # Prognostic eqn for T_sfc


I wonder if only the Dynamic Sea Ice model should be added - i.e. remove code for prescribed_sic = true from this PR. Moving forwards, the prescribed SIC+dynamic T and the dynamic sea ice (h, T_sfc prognostic) should be different types of SeaIceModels, and we would use dispatch - we wouldnt want to have these if/else statements about prescribed_sic, right?

I agree with this plan. I'm tempted to leave it in for now though until we revamp the interface, as it is useful to have the prescribed_sic option as a check in case the Dynamic one is doing something funky.

kmdeck · 2022-06-28T16:30:36Z

experiments/AMIP/moist_mpi_earth_dynamical_sea_ice/slab_ice/slab_init.jl

+    space = nothing,
+    mask = nothing,
+    prescribed_sic_data = nothing,
+    ocean_params = (; ρ = FT(1e3), c = FT(4e3), h = FT(1)),


should these be enforced to be consistent with the slab_ocean params being used?

True, I made it a non-keyword argument, so there is no accidental imposition of the default ocean parameters.

kmdeck

Hey Lenka - not sure of the best way to proceed, it's up to you. We can merge this as is (except for small changes noted in comments), as a copy of the sea ice model, but this code will have to be changed a lot to be used in practice (globals, unused files, etc). This is also not being tested, so it may break at some point (due to other packages changing). It might be better to bring it up to date with moist_mpi_earth changes (but still in it's own directory, so it can have its own versions of conservation checker, etc). Then we can add it to buildkite. and it will be easier to refactor, etc later for interface changes.

kmdeck · 2022-06-28T16:40:46Z

Hey Lenka - not sure of the best way to proceed, it's up to you. We can merge this as is (except for small changes noted in comments), as a copy of the sea ice model, but this code will have to be changed a lot to be used in practice (globals, unused files, etc). This is also not being tested, so it may break at some point (due to other packages changing). It might be better to bring it up to date with moist_mpi_earth changes (but still in it's own directory, so it can have its own versions of conservation checker, etc). Then we can add it to buildkite. and it will be easier to refactor, etc later for interface changes.

I almost wonder if it better to keep this as a branch. Then we could start a new branch off of ClimaCoupler next week (or when slab models are abstracted), and jsut add in the dynamic ice model as a model option with its own methods then. And that should work if we have done the abstractions well.

LenkaNovak · 2022-06-28T21:44:12Z

As discussed on Slack, we will go ahead with the merge, because we be won't be working on the dynamical sea ice for a while (except for perhaps making it an abstract type), so it would be good to checkpoint it now (as a working model integration PR), and then iterate with smaller PRs (with a new interface or enhancements to the physical parameterization) when we have time.

fix coupler_loop_rough spatial BCs, LinearSolver deps update, runs, rough spatial BCs working dry conservation moist conservation, SF.jl, idealised radiation(t) update for new ClimaAtmos inteface changes depend on ClimaAtmos checkpoint wip - mpi race condition fix mpi working + conserving fixes for single processor surface flux clean up readme add deps deps deps for cluster scaling scaling tests finished format rm residual output init Co-authored-by: Akshay Sridhar <akshaysridhar@users.noreply.github.com> Ben Mackay <jbmackay@caltech.edu> jiahe23 <jiahe23@users.noreply.github.com> Zhaoyi Shen <szy21@users.noreply.github.com> runs sst + land-sea - conservation pending Co-authored-by: Akshay Sridhar <akshaysridhar@users.noreply.github.com> stable conservative 60d sst option write vtk from Float32 prescribed sea ice + ssts running - pending tests ice runs for >15d, qualitatively behaves space fix format space fix old bug fixes copy over rad utils conserves with radiation TOA balance maintained (but still needs time log) TOA fluxes conservation check callback sea ice conservation check passes format dynamic sea ice clean up - wip note init from PR#50 mv dir mv dir dynamic sea ice conservation - up to 27d, 1% err clean dynamical sea ice runs format rev fix remove unused mpi hooks

LenkaNovak · 2022-06-29T03:49:13Z

bors r+

bors · 2022-06-29T04:25:21Z

Build succeeded:

LenkaNovak mentioned this pull request Jun 24, 2022

Implement prescribed sea ice #42

Closed

15 tasks

LenkaNovak force-pushed the earth-moist-mpi-real-dynamical branch from 01a3dce to 6681193 Compare June 27, 2022 21:09

LenkaNovak requested a review from kmdeck June 27, 2022 21:13

LenkaNovak mentioned this pull request Jun 28, 2022

Smooth MPI Simple Earth #50

Closed

10 tasks

LenkaNovak force-pushed the earth-moist-mpi-real-dynamical branch from 6681193 to a97869f Compare June 28, 2022 00:45

LenkaNovak marked this pull request as ready for review June 28, 2022 16:03

kmdeck reviewed Jun 28, 2022

View reviewed changes

LenkaNovak force-pushed the earth-moist-mpi-real-dynamical branch from a97869f to 4dc9942 Compare June 28, 2022 21:53

bors bot merged commit 2cc97ab into main Jun 29, 2022

bors bot deleted the earth-moist-mpi-real-dynamical branch June 29, 2022 04:25

LenkaNovak mentioned this pull request Sep 14, 2023

Fix land snow #421

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Dynamical Sea Ice Model #69

Dynamical Sea Ice Model #69

LenkaNovak commented Jun 23, 2022 •

edited

Loading

LenkaNovak commented Jun 27, 2022

kmdeck Jun 28, 2022

LenkaNovak Jun 28, 2022

kmdeck Jun 28, 2022

LenkaNovak Jun 28, 2022

kmdeck Jun 28, 2022

kmdeck Jun 28, 2022

LenkaNovak Jun 28, 2022

kmdeck Jun 28, 2022

LenkaNovak Jun 28, 2022

kmdeck left a comment

kmdeck commented Jun 28, 2022

LenkaNovak commented Jun 28, 2022

LenkaNovak commented Jun 29, 2022

bors bot commented Jun 29, 2022

		@@ -0,0 +1,929 @@
		# This file is machine-generated - editing it directly is not advised

		julia_version = "1.7.2"

		end


		function semster_zero_layer_model(T_ml, T_sfc, h_ice, F_atm, ∂F_atm∂T_sfc, p, Δt)

Dynamical Sea Ice Model #69

Dynamical Sea Ice Model #69

Conversation

LenkaNovak commented Jun 23, 2022 • edited Loading

Purpose and Content

To do:

Future PRs:

Benefits and Risks

Linked Issues

PR Checklist

LenkaNovak commented Jun 27, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kmdeck left a comment

Choose a reason for hiding this comment

kmdeck commented Jun 28, 2022

LenkaNovak commented Jun 28, 2022

LenkaNovak commented Jun 29, 2022

bors bot commented Jun 29, 2022

LenkaNovak commented Jun 23, 2022 •

edited

Loading