AMM15 CO9p2 SSTs become unreastically high (initially 300°C, then 70°C).

[hadjt]

In James' recent runs, the SSTs have become unrealistically high in shallow, inshore waters.

Ocean only runs had temperuatures >45°C, coupled ocean wave runs had >300°C. Similar temperature blow ups, near rivers, have been occuring in the REP system (without WAD). the Operational System doesn't appear to have an issue.

Several hypotheses for the cause of the high temperatures have been proposed including:
- An offset in the oasis mapping between the wave grid and ocean grid, meaning zero stress was passed to the NE of the coastal points.
- Wind on the u grid was averaged with zeros over land when interpolating on the T grid, leading to a ring of low wind stresses around the coast.
- Seapoints with 6 or 7 (out of a possible 8) surrounding land points
- Sea points with only a few depth levels (2 points).
- SBCs being interpolated from the global fields to the regional fields without accounting for land contamination.

Fixing the wave remapping offset reduced hte maximum temperatuers from 300°C to 70°C.

The problem appears to happen with or without WAD, and with or without waves.

We met on the 24/7/2025 to discuss this in depth
The actions are:

1. Juan wrote another branch to check whether the T to U grid interpolation of momentum was problematic and involved 0s. James tested it, the runs are identical, so we don't need the branch, this is NOT a problem - but Alex is double-checking this with CO8.
2. James will test masking out land fluxes with the UM mask before the NEMO interpolation takes place, to make sure no land flux is passed onto ocean points - James will re-run a long run, to see if >40°C is solved
3. It was discovered that some points in CO9.2 operate on only 2 or 3 vertical levels close to coastlines (so only ~34cm of valid ocean point) - this should not be the case - if 2) doesn't resolve the issue, then a "quick fix" could be to mask out the points which have less than 10? vertical levels in the ocean. - we may need to go back to NOC/Enda/Oliver if this is still not working. This may also involve having to redefine the wave grid, remapping weights and DA mask, so is quite involved.

Background

Here is a short summary of the CO9.2 journey towards PS48:

• initial ocean-only tests done by Enda & NOC
• then initial testing from OCC
  • then coupling with waves:
  • Jian Guo created a new wave grid based on the new land sea mask and allowing for wetting & drying, wave team worked on waves side of wetting/drying, ensured fluxes passed to the ocean were decent, and wave model not crashing.
  • tracers were not conserved with wave coupling in the ocean, this caused crashes very early on in the runs - this was fixed by Mike and Juan, this is now in a NEMO branch, but Juan, Daley, Oliver working on getting this into NEMO central trunk.
  • longer runs coupled with runs & with/without DA were started by James While, and they crashed 6-months in. In parallel, Alex had lots of crashes with the fully coupled system with CO8 - the problem was found as being a miscalculation of the wave lat/lon point values, which meant the remapping weights were wrong and the waves were sending 0 momentum on NorthEast coast. This is now fixed for both CO8 and CO9.2 (though still three points sending 0 values to ocean in CO9.2 - minor issue).
  • James re-ran tests, and they now crash 10 months in, with large current values. ~100 grid points at the coast have SSTs >40°C.