Sharp corners problems at the ice-land-open water in compression about mitgcm

[dringeis]

Velocity at the corner:

The fact that the velocity goes inward at the border is kinda weird.

Shear at the corner:

Nothing too weird here

Numerical residual

The LSR residual diagnostic does not work at the moment, problem in the INM branch

Parameters

• Folder: /storage/dringeisen/MITgcm_runs/MITgcm_CornerTest/run00
• Branch: master (same as upstream/master)
• Default parameters from MITgcm except:
  • Replacement pressure off
  • kt = 0.05

[dringeis]

The bug in the INM branch is fixed.

The residual

Nothing special appears here.

I do not a big problem with numerical convergence here.

One thing not to forget is that U and V and H/A are not at the same place on the grid. The quiver representation is then not very accurate. To change.

[dringeis]

New velocity plot

We now plot the velocities UV at the correct place in the ArakawaC grid

We see that the velocity in U is small but can be inverted.

Here is the velocity in time. It becomes negative suddenly, then decreases slowly to zero.

The velocity becomes negative as the stress SIsigI suddenly jumps from 0 to almost -1.

Still no problem in the residuals (checked but not shown)

[dringeis]

• The velocity inversion does not appear in the free drift setup. (folder run01)

• The velocity inversion does appear in the more converged setup (folder run02)
  

  • But it is not the main problem here: The velocities are much much higher (2 orders of mag) in run02 than in run00. The fracture is also nicer, more precise and defined.

With the residuals:

[dringeis]

There is often an episode of inversed velocities in the cells newly covered by sea ice

, often when the stress goes suddenly from 0 to ca. -1.