Antarctic Bottom Water (AABW) is the densest watermass present in the Southern, Atlantic, Indian, and Pacific oceans. It is formed in seveal distinct regions of the Antarctic Continental Shelf, and as it is exported to the oceans to its north its derivatives (after dilution with other watermasses) fill about 1/3 of their total volume! Hence the thickness, stratification, and distribution of AABW has large and global impacts on ocean circulation and watermass properties.  Yet AABW export pathways from the Southern Ocean are poorly observed due to the depths, remoteness, and extent of the area that needs to be sampled. In particular, the literature is divided on if AABW formed in different Antarctic regions is exported separetly northwards (the "conduits" paradigm) or is blended together in the Southern Ocean before export together (the "blender paradigm). Since AABW has experienced regional property changes over the last decades (generally warming, freshening, and decrease in thickness), the regional export pathways determine which basins to the north would be imacted by these changes.

To address these questions, we analyzed a state of the art ocean and sea ice numerical model, ACCESS-OM2-01. It is a global model in almost unprecedented high resolution (0.1 degree nominally) ran for 61 years (1958-2019). It is particularly suitable for investigating AABW export pathways due to its eddy-resolving resolution, global extent, and high fidelity in capturing AABW formation rates and locations (e.g., on the continental shelves rather than in the open ocean). We also included in the model online passive tracers which track the distribution of each source of AABW (i.e., AABW sourced from the Weddell Sea, Prydz Bay, Adelie Land, and the Ross Sea).

Our findings show that over the model integration time, AABW export is accomplished in a way which is halfway between the "conduits" and "blender" paradigms. In particular, AABW sourced from the Weddell Sea and Prydz Bay are blended and exported together, mainly to the Atlantic and Indian Oceans. In contrast, AABW sourced from Adelie Land and the Ross Sea are blended together with each other, but minimally so with the former pair, and exported mainly to the Pacific Ocean. These export pathways apear to be dictated by topographic steering, as described on our paper, where we also explore the locations of blending between AABW sourced from differernt regions and the degree of vertical homogenization. The results are deomnstarted by the figure below, showing seabottom distributions of the tracers representing AABW sourced from the four different formation regions, 61 years after model initialization. As described above, the results show where recent changes in each AABW source properties are likely to impact in the coming decades. It is still an open question how these property changes may feedback into AABW export pathways over longer (e.g., centennial) time scales.