The Weddell Polynya, a large expanse of water that originated over Maud Rise (a bathymetric protrusion centred near 64°30′S, 3°E) and remained open during winter in the late 1970s, may have manifested a mode of deep ocean convection where despite large heat loss at the surface, sustained heat transport from below prevents lasting ice formation. In a different dominant mode (the present one), sea ice forms early in the winter and subsequently provides a thermal barrier that quickly quells incipient deep convection, thus preventing wholesale destruction of the ice cover. A possible mechanism for overcoming the thermal barrier is thermobaricity, the pressure dependence of the thermal expansion factor for seawater. An idealized, two-layer version of actual temperature and salinity profiles from the Weddell illustrates that thermobaric mixing can persist for extended periods in an ice-covered ocean, provided realistic melt rates (controlled by salt exchange at the ice/ocean interface) are specified. This furnishes a possible explanation for transient winter polynyas sometime observed in the ice-covered Southern Ocean. Thermobaricity may provide a trigger for widespread convection with possible climate impact.

Byrd Glacier has one of the largest ice catchment areas in Antarctica, delivers more ice to the Ross Ice Shelf than any other ice stream, and is the fastest of these ice streams. A force balance, combined with a mass balance, demonstrates that stream flow in Byrd Glacier is transitional from sheet flow in East Antarctica to shelf flow in the Ross Ice Shelf. The longitudinal pulling stress, calculated along an ice flowband from the force balance, is linked to variations of ice thickness, to the ratio of the basal water pressure to the ice overburden pressure where Byrd Glacier is grounded, and is reduced by an ice-shelf buttressing stress where Byrd Glacier is floating. Longitudinal tension peaks at pressure-ratio maxima in grounded ice and close to minima in the ratio of the pulling stress to the buttressing stress in floating ice. The longitudinal spacing of these tension peaks is rather uniform and, for grounded ice, the peaks occur at maxima in surface slope that have no clear relation to the bed slope. This implies that the maxima in surface slope constitute a “wave train” that is related to regular variations in ice-bed coupling, not primarily to bed topography. It is unclear whether these surface “waves” are “standing waves” or are migrating either upslope or downslope, possibly causing the grounding line to either retreat or advance. Deciding which is the case will require obtaining bed topography in the map plane, a new map of surface topography, and more sophisticated modeling that includes ice flow linked to subglacial hydrology in the map plane.