Bug in DatasetRestoring

[simone-silvestri]

With @francispoulin we were trying to revamp PR #142, but we realized that most likely some bug has been introduced in the refactor. A simulation with restoring blows up pretty drastically after little time starting from the boundary.

acc_surface_e.mp4

It looks like the restoring is trying to restore the temperature to zero. See:
https://clima.github.io/ClimaOceanDocumentation/previews/PR142/literated/acc_regional_simulation/

This possibly hints at the fact that the fields are not correctly set! in case of a DatasetRestoring

[francispoulin]

I would add that it seems like all the fields, temperature, salinity and the velocities, are not being set correctly.

[simone-silvestri]

However, in the one degree example https://github.com/CliMA/ClimaOcean.jl/blob/main/examples/one_degree_simulation.jl we have restoring and it does not affect the simulation so drastically.

[francispoulin]

In that case the restoring is only at the poles I presume? There can't be a lot happening there, I would guess. In the ACC case we have restoring happening closer to 25 degrees south. This is a much bigger region and small errors could make a bigger impact on the simulations.