Conclusions

• Small rotation and FREE-SLIP
• One corner and FREE-SLIP
• Remarks on ADCROFT and MARSHALL
• General conclusion

Small rotation and FREE-SLIP

• One step is enough to modify the circulation.
• High viscosity kills the sensitivity.
• But low viscosity with high resolution doesn't. For low value of the viscosity, higher resolution-Reynolds ( ) number dependence.

One corner and FREE-SLIP

• FREE-SLIP does not imply vorticity=0 at corners
• Changing the way vorticity is specified at corners dramatically alters the circulation
• Drawback: Value somewhat arbitrary, yields spurious oscillations in the vicinity of the corner.
• we don't recover the NO-SLIP circulation.

Remarks on ADCROFT and MARSHALL

• dependence of the advocated stress formulation on the artificial vorticity=0 at the corner.

General conclusion

• FREE-SLIP not robust. Too sensitive. No chance of convergence for higher resolution and lower viscosity => too dangerous.
• NO-SLIP robust. although remains some difficulties like a very thick boundary layer. The steady eddy in the WOCE model (Treguier 92)
• Solution: NO-SLIP with increased resolution at the bdy (unstructured grid)?