North pole bug

[aekiss]

Issue description

Something's going wrong at the north pole in all OM3 and CM3 configurations. Thanks to @ofa001 and @NoahDay for spotting and reporting this issue.

Here's an animation (click to run) of ice velocity uvel from /g/data/ol01/outputs/access-om3-25km/MC_25km_jra_iaf+wombatlite-test3v2-00532b88/output065/access-om3.cice.1day.mean.2023-12*.nc. Note the persistently odd patterns 25% and 75% along the top boundary - these are at the north pole.

Both u and v components of ice velocity are affected, together with all other ice fields.

[aekiss]

Given this is happening at the north pole itself (not all along the tripole seam as in COSIMA/access-om2#86), I think it must be something to do with regridding (probably wind) from the lat-lon to triplar grid, which means it's a problem in DATM, because every other part of the model uses the tripolar grid, for which nothing special happens at the north pole, ie no singularity (see https://access-om3-configs.access-hive.org.au/inputs/Forcing-data-models/). But if it's also happening in CM3 there must be a similar problem in the UM cap.

[aekiss]

Does it depend on the regridding method used for wind? See #243

[access-hive-bot]

This issue has been mentioned on ACCESS Hive Community Forum. There might be relevant details there:

https://forum.access-hive.org.au/t/cosima-twg-meeting-minutes-2026/5906/11

[anton-seaice]

I never understand the reference direction for these diagnostics. Is uvel/vvel relative to true north/sorth? Or positive x/y on the tripolar grid ?

There isn't a singularity in the tripolar grid, but there is presumably one in the JRA grid (as it's regular lat-lon).

If it's remapping related, it should also show in uatm/vatm

[aekiss]

I never understand the reference direction for these diagnostics. Is uvel/vvel relative to true north/sorth? Or positive x/y on the tripolar grid ?

u/v is in the grid i/j direction in MOM and CICE - see COSIMA/access-om2#86

There isn't a singularity in the tripolar grid, but there is presumably one in the JRA grid (as it's regular lat-lon).

Yes, that's what I was thinking

If it's remapping related, it should also show in uatm/vatm

Yes, we should look at uatm/vatm being received by the mediator. We are rotating the vectors in CICE and MOM (#1045 (comment)), so the mediator will be receiving and providing u/v relative to east/north.

[anton-seaice]

I never understand the reference direction for these diagnostics. Is uvel/vvel relative to true north/sorth? Or positive x/y on the tripolar grid ?

u/v is in the grid i/j direction in MOM and CICE - see COSIMA/access-om2#86

grid direction and true north/east are aligned at the geographic north pole anyway, so this won't have an impact I think ? (the rotation angle is most significant over land at the tripoles )

If it's remapping related, it should also show in uatm/vatm

Yes, we should look at uatm/vatm being received by the mediator. We are rotating the vectors in CICE and MOM (#1045 (comment)), so the mediator will be receiving and providing u/v relative to east/north.

The cice fields uatm vatm would be relative to grid i/j direction

The mediator fields would be medImp_Sa_u and medImp_Sa_v relative to east/north.

[ofa001]

The Uatm vatm seen on the CICE side of the mediator have the issue, but yes @anton-seaice it would be interesting to see what is on the other side of the mediator. In relation to whether this is related to "patch" or bi-linear" etc. I checked all the Mom5 based runs OM2/CM2 ESM1.5 ESM1.6 none of them had the issue.

We only recently switched to patch for CM3 and there have been problems with our wind stresess and the way they have been driving the ice, @kieranricardo has just put in the fix to make sure that the model is seeing the correct UM stresses, but its only been a minor impact on the ice thickness more on velocities.

[aekiss]

grid direction and true north/east are aligned at the geographic north pole anyway, so this won't have an impact I think ? (the rotation angle is most significant over land at the tripoles )

j points northward at the north pole, but i points southward (towards Canada or Siberia, depending on what side of the seam you approach from), not eastward - see diagram.

[anton-seaice]

j points northward at the north pole, but i points southward (towards Canada or Siberia, depending on what side of the seam you approach from), not eastward - see diagram.

Oh yes - doesn't that mean we should rotate the winds back from grid angle to true north/east when making these plots ?

[aekiss]

That depends on what we're trying to show and whether we're bearing these rotations in mind when eyeballing the plots...