From the paper[1], it sounds like the same linear motion with the film wrapping around to the top:
> The sphere is dipped with
> its north pole pointing downward. The maximum error on the
> northern hemisphere is within 2mm. However, near its south pole
> the error is much larger (about 5mm). This is because after the
> water surface passes the sphere’s equator, the film gets stretched
> largely, and near the south pole the relative angle between the
> water surface and the object surface approaches to 180◦, leading
> to an ill-posed boundary condition for our simulation (recall
> Equation (1), when θ ≈ 180◦).
You can see the potential for a similar wraparound even on e.g. the mask dips.
[1] http://www.kunzhou.net/2015/hydrographics.pdf