We investigate air entrainment and bubble statistics in three-dimensional breaking waves through novel direct numerical simulations of the two-phase air–water flow, resolving the length scales relevant for the bubble formation problem, the capillary length and the Hinze scale. The dissipation due to breaking is found to be in good agreement with previous experimental observations and inertial scaling arguments. The air entrainment properties and bubble size statistics are investigated for various initial characteristic wave slopes. For radii larger than the Hinze scale, the bubble size distribution, can be described by $N(r,t)=B(V_{0}/2{\rm\pi})({\it\varepsilon}(t-{\rm\Delta}{\it\tau})/Wg)r^{-10/3}r_{m}^{-2/3}$ during the active breaking stages, where ${\it\varepsilon}(t-{\rm\Delta}{\it\tau})$ is the time-dependent turbulent dissipation rate, with ${\rm\Delta}{\it\tau}$ the collapse time of the initial air pocket entrained by the breaking wave, $W$ a weighted vertical velocity of the bubble plume, $r_{m}$ the maximum bubble radius, $g$ gravity, $V_{0}$ the initial volume of air entrained, $r$ the bubble radius and $B$ a dimensionless constant. The active breaking time-averaged bubble size distribution is described by $\bar{N}(r)=B(1/2{\rm\pi})({\it\epsilon}_{l}L_{c}/Wg{\it\rho})r^{-10/3}r_{m}^{-2/3}$, where ${\it\epsilon}_{l}$ is the wave dissipation rate per unit length of breaking crest, ${\it\rho}$ the water density and $L_{c}$ the length of breaking crest. Finally, the averaged total volume of entrained air, $\bar{V}$, per breaking event can be simply related to ${\it\epsilon}_{l}$ by $\bar{V}=B({\it\epsilon}_{l}L_{c}/Wg{\it\rho})$, which leads to a relationship for a characteristic slope, $S$, of $\bar{V}\propto S^{5/2}$. We propose a phenomenological turbulent bubble break-up model based on earlier models and the balance between mechanical dissipation and work done against buoyancy forces. The model is consistent with the numerical results and existing experimental results.

While ocean current and winds certainly play a major role in guiding the trajectories of free-floating icebergs, the direct effect of atmospheric surface pressure gradients can also have an important influence on the trajectories of large icebergs whose horizontal dimensions are sufficiently great to span synoptic systems. This effect is examined as a way of understanding why icebergs B15A, B15J, B15K, and C16 became “trapped” in a limited region immediately north of Ross Island for a period of several years, without being grounded. This limited region is otherwise flushed annually by summer surface winds and currents; thus the delay of the northward drift of the large icebergs (particularly B15A and B15J) defied expectation. The best explanation for this unexpected iceberg behaviour is that the large volcanic massifs on Ross Island create a quasi-permanent surface pressure anomaly patterned as a dipole, with high pressure in the area upwind of the island (an area appropriately called Windless Bight), and low pressure in the downwind area of the iceberg parking lot. The surface pressure regime experienced by two icebergs B15A and B15K is estimated using Automatic Weather Station observations and Global Positioning System receivers deployed on their surfaces to explain why they remained trapped. Breakdown of the atmospheric pressure gradients allowed them to eventually escape from the region to the north-west.