Name and subject of the action

Last edition: 06/08/21 16:13:31 by jamesharle

The PI is responsible to closely follow the progress of the action, and especially to contact NEMO project manager if the delay on preview (or review) are longer than the 2 weeks expected.

  1. Summary
  2. Preview
  3. Tests
  4. Review
  5. SETTE tests for merge:

Summary

Action AGRIF-03_jchanut_vert_coord_interp
PI(S) Jérôme Chanut and James Harle
Digest Online vertical interpolation with AGRIF (2019 cont.)
Dependencies Nesting tools update, i.e. #2129 + new step flowchart
Branch source:/NEMO/branches/{YEAR}/dev_r{REV}_{ACTION_NAME}
Previewer(s) S. Masson
Reviewer(s) Mattia Almansi and James Harle
Ticket #2222

Description

This is a continuation from the IMMERSE task started in 2019. Changes to allow for vertical interpolation in NEMO have been implemented in December 2019 in the trunk. This task will eventually provide additional corrections and an update of the Nesting Tools. Since the idea is to use the DOMAINcfg tool for AGRIF, the task tackled in AGRIF-05_rblod_CMEMS (#2129) should ideally be completed first.

Implementation

Documentation updates

Preview

Tests

VORTEX test case

  • Functionality: Detect nesting errors. Evaluate new algorithms. Check volume/tracer conservation with 2 way nesting.
  • Setup: Self advected vortex over a beta plan. No forcing, flat bottom. Analytical definition of geostrophically balanced vortex.
  • Verification value: No analytical solution of the solution exists. The parent domain at high resolution is usually used to have a quantitative measure of the nesting errors.

One can nevertheless implement the following unit tests:

Global volume should be conserved at machine accuracy

Perfect symmetry of the results if switching hemisphere or reversing earth rotation

1:1 nesting (i.e. no refinement in the zoom) and fully explicit time stepping mode should not change the results

  • Status: available as part of NEMO test cases suite on svn.

2DV OVERFLOW test case

  • Functionality: Detect nesting errors with 1d varying topography. Evaluate vertical grid change within AGRIF zooms.
  • Setup: based on existing OVERFLOW test case with an AGRIF zoom over the slope. Since offline bathymetry connection is important here, input domain file have to be provided. That's the main difference with existing test case.
  • Verification value: No analytical solution of the solution exists. The parent domain at high resolution can be used to have a quantitative measure of the nesting errors. This test can be used to easily detect errors in the vertical coordinate change algorithm if vertical grids are actually identical.
  • Status: Test case already set up but not transferred on any repository yet.

BOWL OVERFLOW test case (TBC)

  • Functionality: Detect nesting errors with 2d varying topography. Evaluate vertical grid change within AGRIF zooms.
  • Setup: Similar to the 2d OVERFLOW test case above but with 2d varying topography.
  • Verification value: No analytical solution of the solution exists. The parent domain at high resolution can be used to have a quantitative measure of the nesting errors.
  • Status: Not available yet.

2DV UPWELLING test case (TBC)

  • Functionality: Detect nesting errors with 1d varying topography. Evaluate vertical grid change within AGRIF zooms and potential artefacts in the connection of surface/bottom boundary layers.
  • Setup: issued form COMODO test cases, that's a stationary solution of a wind forced shelf overflow.
  • Verification value: Analytical solution.
  • Status: Not available yet.

DOME test case (TBC)

  • Functionality: That's the classical experiment used in the literature to evaluate the numerical impact on dense overflows. It illustrates here the potential benefit of using a local vertical coordinate change and/or horizontal refinement on a longer time scale.
  • Setup: Forced dense water flow in a channel over a linear slope (hence need bdy forcing).
  • Verification value: No analytical solution of the solution exists. The parent domain at high resolution can be used to have a quantitative measure of the nesting errors.
  • Status: Not available yet.

Review

To allow for vertical interpolation in NEMO/AGRIF a vertical remapping the conservative piecewise polynomial reconstruction library of Engwirda and Kelley (2016) has be successfully implemented. All proposed tests (bar the bowl overflow case) have been successfully undertaken. This development extends the existing code, which previously only supported z-coordinate to z-coordinate nesting. Due to the nature of the development task there are differences in the solution when activating the new feature. These differences are acceptable small. The development is ready for merging.

As the DOMAINcfg tool for AGRIF had not been implemented a collection of python scripts were employed to produce configuration files for a series of test cases: Flat Bottom, 2DV Overflow, VORTEX and DOME experiments. While the domain_cfg files could not be re-created using the existing NEMO NESTING or DOMAINcfg Tools, existing domain_cfg.nc files where used to verify the test cases. All of which performed as described in the IMMERSE WP3.2 final report.

Testing of ORCA2_PISCES (with multiple nests) was also undertaken. This currently fails when turning key_vert_remap = .true., as nests 2 and 3 have not been refactored to match the volume along the parent-child boundary. No explicit error message is issued to say this cannot be done (something we believe will be addressed in ticket #2638)

In the namelist, the namagrif block includes both ln_vert_remap and ln_chk_bathy. It is suggested that ln_chk_bathy may well be a redundant: if ln_vert_remap = .true., ln_chk_bathy should be .true. by default, in order to remove any ambiguity.

This development has been documented with sufficient explanation and examples in the IMMERSE WP3.2 final report. Documentation of project literature will most likely come as an output from ticket #2638 (development of the DOMAINcfg Tool to enable vertical remapping of NEMO/AGRIF configurations employing different vertical grids).

SETTE tests for merge:

SETTE validation report generated for :

NEMO/branches/2020/dev_r13312_AGRIF-03-04_jchanut_vinterp_tstep @ r13961+ (last changed revision)

on openmpi_KARA_MERCATOR_XIOS arch file

!!———————-1st pass—————————!!

!——restart——!

WGYRE_PISCES_ST run.stat restartability passed : 13961+

WGYRE_PISCES_ST tracer.stat restartability passed : 13961+

WORCA2_ICE_PISCES_ST run.stat restartability passed : 13961+

WORCA2_ICE_PISCES_ST tracer.stat restartability passed : 13961+

WORCA2_OFF_PISCES_ST tracer.stat restartability passed : 13961+

WAMM12_ST run.stat restartability passed : 13961+

WORCA2_SAS_ICE_ST run.stat restartability passed : 13961+

WAGRIF_DEMO_ST run.stat restartability passed : 13961+

WWED025_ST run.stat restartability passed : 13961+

WISOMIP+_ST run.stat restartability passed : 13961+

WOVERFLOW_ST run.stat restartability passed : 13961+

WLOCK_EXCHANGE_ST run.stat restartability passed : 13961+

WVORTEX_ST run.stat restartability passed : 13961+

WICE_AGRIF_ST run.stat restartability passed : 13961+

!——repro——!

WGYRE_PISCES_ST run.stat reproducibility passed : 13961+

WGYRE_PISCES_ST tracer.stat reproducibility passed : 13961+

WORCA2_ICE_PISCES_ST run.stat reproducibility passed : 13961+

WORCA2_ICE_PISCES_ST tracer.stat reproducibility passed : 13961+

WORCA2_OFF_PISCES_ST tracer.stat reproducibility passed : 13961+

WAMM12_ST run.stat reproducibility passed : 13961+

WORCA2_SAS_ICE_ST run.stat reproducibility passed : 13961+

WORCA2_ICE_OBS_ST run.stat reproducibility passed : 13961+

WAGRIF_DEMO_ST run.stat reproducibility passed : 13961+

WWED025_ST run.stat reproducibility passed : 13961+

WISOMIP+_ST run.stat reproducibility passed : 13961+

WVORTEX_ST run.stat reproducibility passed : 13961+

WICE_AGRIF_ST run.stat reproducibility passed : 13961+

!——agrif check——!

ORCA2 AGRIF vs ORCA2 NOAGRIF run.stat unchanged - passed : 13961+ 13961+

Last modified 4 months ago Last modified on 2021-06-08T16:13:31+02:00