Tc4 ice shelf #1210

MJHarrison-GFDL · 2020-09-18T01:10:53Z

@marshallward . tc4.rotate is failing due to intermediate timestats diverging although the final stats are identical.

- .testing/tc4 includes a static ice shelf which thermodynamic coupling - tc4.rotate fails due to divergence of very small energetic values at timesteps intermediate within the 2 day run.

codecov-commenter · 2020-09-18T01:30:27Z

Codecov Report

Merging #1210 into dev/gfdl will increase coverage by 1.80%.
The diff coverage is 41.22%.

@@             Coverage Diff              @@
##           dev/gfdl    #1210      +/-   ##
============================================
+ Coverage     46.08%   47.89%   +1.80%     
============================================
  Files           214      225      +11     
  Lines         69399    71065    +1666     
============================================
+ Hits          31984    34036    +2052     
+ Misses        37415    37029     -386

Impacted Files	Coverage Δ
...g_src/external/GFDL_ocean_BGC/FMS_coupler_util.F90	`0.00% <0.00%> (ø)`
...fig_src/external/GFDL_ocean_BGC/generic_tracer.F90	`0.00% <0.00%> (ø)`
...c/external/GFDL_ocean_BGC/generic_tracer_utils.F90	`0.00% <0.00%> (ø)`
config_src/external/ODA_hooks/kdtree.f90	`0.00% <0.00%> (ø)`
config_src/external/ODA_hooks/ocean_da_core.F90	`0.00% <0.00%> (ø)`
config_src/external/ODA_hooks/ocean_da_types.F90	`0.00% <0.00%> (ø)`
config_src/external/ODA_hooks/write_ocean_obs.F90	`0.00% <0.00%> (ø)`
config_src/solo_driver/MESO_surface_forcing.F90	`0.00% <0.00%> (ø)`
config_src/solo_driver/user_surface_forcing.F90	`0.00% <0.00%> (ø)`
src/ALE/MOM_remapping.F90	`70.38% <0.00%> (+0.11%)`	⬆️
... and 245 more

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 7cb606e...3e7a73c. Read the comment docs.

…tc4_ice_shelf

adcroft · 2020-09-18T13:28:52Z

It's not just the rotation test that is falling. There's restart and dimensional tests too.

adcroft · 2020-09-18T13:36:57Z

The tc4 test seems broken in many ways. Even layout is failing trying to read the grid. Is everything committed?

FATAL from PE     0: fms_io(read_data_3d_new), field x in file ./ocean_hgrid.nc: field size mismatch 1

adcroft · 2020-09-18T13:44:58Z

That last was because the size has changed (which is good) - cleaning the work dir sorted that out. The actual fails I get interactively are

FAIL: Solutions tc4.restart have changed.
FAIL: Diagnostics tc4.dim.t.diag have changed.
FAIL: Diagnostics tc4.dim.l.diag have changed.
FAIL: Diagnostics tc4.dim.h.diag have changed.
FAIL: Solutions tc4.rotate have changed.

which I think is consistent with what Travis reported.

adcroft · 2020-09-18T14:56:47Z

After a restart fluxes%sens and other member fields are zero where there aren't before the restart?

MJHarrison-GFDL · 2020-09-18T14:59:13Z

From a clean build I get the same. The restart and diagnosric issues I haven't looked at but the rotate fail is weird . The results only differ at timesteps 54 and 63 but are otherwise identical through 144.

MJHarrison-GFDL · 2020-09-18T15:00:27Z

buoy_config="zero".

MJHarrison-GFDL · 2020-09-18T15:31:54Z

Energetics differ temporarily at the last significant digit .Timestats too precise?

marshallward · 2020-09-18T15:37:15Z

I see on my desktop that SB_max is undefined, raising a FPE:

Thread 1 "MOM6" received signal SIGFPE, Arithmetic exception.
0x00005555559356b0 in mom_ice_shelf::shelf_calc_flux (sfc_state_in=..., fluxes_in=0x555558970190, time=..., time_step=1200, cs=0x555558e676d0)
    at ../../../ac/../src/ice_shelf/MOM_ice_shelf.F90:592
592	                if (Sb_max_set .and. (Sbdry(i,j) > Sb_max)) &
(gdb) p Sb_max
$2 = nan(0x4000000000000)

but it is worrying that Travis was unable to detect this.

adcroft · 2020-09-18T15:41:41Z

Should

if (Sb_max_set .and. (Sbdry(i,j) > Sb_max))

be

if (Sb_max_set) then
  if (Sbdry(i,j) > Sb_max)) &

?

marshallward · 2020-09-18T15:42:46Z

Yes, I agree, no short-circuit logic can be assumed in Fortran.

MJHarrison-GFDL · 2020-09-18T16:22:29Z

Registering sfc_state%tau[xy]_shelf restarts fixes that issue.

MJHarrison-GFDL · 2020-09-18T16:43:16Z

tc4 has thermal coupling turned on so buoyancy forcing is non-zero, but not required for restarts.

Two short-circuit logic operations were in the ice shelf code, which were causing failures in newer GCC compilers. This patch replaces them with nested if-blocks.

…tc4_ice_shelf

A few ice shelf diagnostics were tagged with scalar_pair, clarifying their rotational rules. Diagnostics were also computed for the rotated rigidity_ice_[uv], rather than the original field. Some of the handling of frac_shelf_h in the MOM.F90 initialization was restructured, although it should be functionally identical. This reduces some of the diagnostic diffs in the rotated ice shelf tc4, but other issues remain.

- This is stored in the forcing type (unrotated) the MOM_CS (rotated). - Now this field is updated by a query to the ice shelf module prior to calling MOM_state_initialization .

Tc4 ice shelf

BT_rem_[uv], which applys a fractional weight to the barotropic velocity, was incorrectly computed as a vector pair, which computes the incorrect checksum. This patch flags this as a scalar pair, which removes any discrepancy in standard and rotated debug output.

… add_shelf_flux

…tc4_ice_shelf

marshallward · 2020-10-20T13:36:27Z

https://gitlab.gfdl.noaa.gov/ogrp/MOM6/-/pipelines/11417

marshallward · 2020-10-20T14:27:43Z

There appears to be some kind of error in coupled_driver which is causing most of the builds to fail. This would not have been detected by Travis.

This is what popped up but it's buried amongst too many errors in a parallel build to make more sense:

build/gnu/debug/dynamic/mom6/libmom6.a: Error: Actual argument for 'forces_in' must be a pointer at (1)
build/gnu/debug/dynamic/mom6/libmom6.a: ../../../../../..//config_src/coupled_driver/ocean_model_MOM.F90:529:43:

@MJHarrison-GFDL can you have a look?

MJHarrison-GFDL · 2020-10-20T15:17:29Z

PR on the way.

- A bug was introduced in a previous commit on this branch - This version of the code retains previous answers for non ice-shelf cases and properly builds hycom1 and rho grids under ice shelf cavities.

MJHarrison-GFDL · 2020-11-06T16:56:30Z

The current regression failure in tc4 is due to a bugfix for the sponge code

marshallward · 2020-11-06T17:14:13Z

Do you mind cherry-picking that one and submitting separately? Might be better for the commit log as well.

MJHarrison-GFDL and others added 5 commits September 15, 2020 08:42

Initialization of temporary arrays.

e7602ad

*Fixes tracer index bugs in ALE sponge code.

14a81e4

*Fixes mask for target sponge values when SPONGE_ONGRID=True

32b1d2e

Various commits related to testing rotated of ice shelves

5e172cc

- .testing/tc4 includes a static ice shelf which thermodynamic coupling - tc4.rotate fails due to divergence of very small energetic values at timesteps intermediate within the 2 day run.

Merge branch 'dev/gfdl' into tc4_ice_shelf

db291da

MJHarrison-GFDL added 2 commits September 18, 2020 09:04

fix excessive line length

7e67d23

Merge branch 'tc4_ice_shelf' of github.com:MJHarrison-GFDL/MOM6 into …

2fdc914

…tc4_ice_shelf

Fix excessive line length

2504143

doxygen syntax issues.

81f6b4b

fix restart issue

786fdb8

marshallward and others added 7 commits September 18, 2020 15:22

Ice shelf: Removed short-circuit operations

aba74df

Two short-circuit logic operations were in the ice shelf code, which were causing failures in newer GCC compilers. This patch replaces them with nested if-blocks.

Merge branch 'tc4_ice_shelf' of github.com:MJHarrison-GFDL/MOM6 into …

a1f255f

…tc4_ice_shelf

Consolidates initialization of frac_shelf_h

bb83604

- This is stored in the forcing type (unrotated) the MOM_CS (rotated). - Now this field is updated by a query to the ice shelf module prior to calling MOM_state_initialization .

cleanup commented lines.

42f3b3a

Merge pull request #5 from MJHarrison-GFDL/tc4_ice_shelf

d319e07

Tc4 ice shelf

MJHarrison-GFDL and others added 12 commits October 9, 2020 10:43

cleanup

04b7a99

add dimensional rescaling arguments

32f60fe

cleanup

8412658

cleanup

0c5fe32

cleanup

572d605

correct dimensional rescaling

8853b04

turn on T/S sponge in tc4

a25a1a7

Merge branch 'dev/gfdl' into tc4_ice_shelf

9a699a2

Merge branch 'dev/gfdl' into tc4_ice_shelf

e2b1205

Retaining ocean model grid argument in calls to add_shelf_forcing and…

70b23f5

… add_shelf_flux

Merge branch 'tc4_ice_shelf' of github.com:MJHarrison-GFDL/MOM6 into …

56296e1

…tc4_ice_shelf

Merge branch 'dev/gfdl' into tc4_ice_shelf

9fe403c

marshallward self-assigned this Oct 20, 2020

MJHarrison-GFDL added 9 commits October 20, 2020 12:41

Add pointer attribute to forces, fluxes and sfc_state types

b6149e2

ice shelf initialization in coupled mode

22e1e82

Fixes bug in hycom1 grid build

b93b01a

- A bug was introduced in a previous commit on this branch - This version of the code retains previous answers for non ice-shelf cases and properly builds hycom1 and rho grids under ice shelf cavities.

revert tc4 changes

6b3cbaa

more changes to revert to tc4 answers

ecf11e4

Merge branch 'dev/gfdl' of github.com:NOAA-GFDL/MOM6 into tc4_ice_shelf

5b8a7f7

additional flags to recover old tc4 configuration

2e71a6f

further change to recover tc4 answers

ee5dbd2

more parameters to recover tc4

7450ff1

MJHarrison-GFDL closed this Nov 9, 2020

marshallward mentioned this pull request Nov 18, 2020

Ice shelf travis upgrade #1249

Closed

MJHarrison-GFDL deleted the tc4_ice_shelf branch April 10, 2023 16:55

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Tc4 ice shelf #1210

Tc4 ice shelf #1210

MJHarrison-GFDL commented Sep 18, 2020

codecov-commenter commented Sep 18, 2020 •

edited

Loading

adcroft commented Sep 18, 2020

adcroft commented Sep 18, 2020

adcroft commented Sep 18, 2020

adcroft commented Sep 18, 2020

MJHarrison-GFDL commented Sep 18, 2020

MJHarrison-GFDL commented Sep 18, 2020

MJHarrison-GFDL commented Sep 18, 2020

marshallward commented Sep 18, 2020

adcroft commented Sep 18, 2020

marshallward commented Sep 18, 2020 •

edited

Loading

MJHarrison-GFDL commented Sep 18, 2020

MJHarrison-GFDL commented Sep 18, 2020

marshallward commented Oct 20, 2020

marshallward commented Oct 20, 2020

MJHarrison-GFDL commented Oct 20, 2020

MJHarrison-GFDL commented Nov 6, 2020

marshallward commented Nov 6, 2020

Tc4 ice shelf #1210

Tc4 ice shelf #1210

Conversation

MJHarrison-GFDL commented Sep 18, 2020

codecov-commenter commented Sep 18, 2020 • edited Loading

Codecov Report

adcroft commented Sep 18, 2020

adcroft commented Sep 18, 2020

adcroft commented Sep 18, 2020

adcroft commented Sep 18, 2020

MJHarrison-GFDL commented Sep 18, 2020

MJHarrison-GFDL commented Sep 18, 2020

MJHarrison-GFDL commented Sep 18, 2020

marshallward commented Sep 18, 2020

adcroft commented Sep 18, 2020

marshallward commented Sep 18, 2020 • edited Loading

MJHarrison-GFDL commented Sep 18, 2020

MJHarrison-GFDL commented Sep 18, 2020

marshallward commented Oct 20, 2020

marshallward commented Oct 20, 2020

MJHarrison-GFDL commented Oct 20, 2020

MJHarrison-GFDL commented Nov 6, 2020

marshallward commented Nov 6, 2020

codecov-commenter commented Sep 18, 2020 •

edited

Loading

marshallward commented Sep 18, 2020 •

edited

Loading