A new, high resolution depth anomaly map covering the anomalously deep and rough Australian-Antarctic Discordance (AAD) has been constructed using crustal ages derived from a detailed aeromagnetic survey. The map shows a large, arcuate-shaped, negative depth anomaly that is centered on the Southeast Indian Ridge and trends NNE across the Australian plate (SSE across the Antarctic plate). Within this broad scale feature, two prominent depth anomaly lows are observed at 45 ¿S, 128 ¿E (the northern flank) and 54 ¿S, 125 ¿E (the southern flank). Both lows are associated with 15 Ma oceanic crust. The observed depth anomaly patterns are compared with the distinctive patterns predicted by coldspot, downwelling limbs of convection cells, and thin crust models of the discordance source. The observed depth anomaly does not result from absolute plate motions over a fixed coldspot source because the predicted ENE trend on the Antarctic plate is not in agreement with the SSE trend observed. The symmetric arrangement of the large-scale depth anomaly and prominent lows about the ridge axis suggests instead a source that has varied in strength but remained located at the ridge axis as the ridge migrated northeastward in the absolute reference frame. The organized pattern of elongated depth anomaly highs and lows predicted for upper mantle convection (cells) is not evident in the observed depth anomaly map. Thus a convergence of downwelling limbs of convection cells beneath the discordance is not indicated. If the source of cooler upwelling that produces less magma and hence thin crust has not varied over time, nor migrated along the ridge, then the predicted depth anomaly would persist unchanged with distance from the ridge axis, and trend in the direction of relative plate motion (parallel to fracture zones). The observed depth anomaly trends obliquely across fracture zones and changes in both amplitude and location relative to the ridge axis, and is therefore not consistent with cool upwelling producing thin crust.

To explain the features of the depth anomaly map, we propose that asthenospheric material flowing from the Amsterdam hotspot in the west, and the Balleny and Tasmantid hotspots in the east, collides within the discordance. Propagating rifts converging on the AAD provide evidence for such asthenospheric flow. Attenuated shear velocities beneath the George V fracture zone complex, and lavas geochemically identical to those from propagating rifts associated with hotspots, suggest a thermal anomaly producing additional asthenospheric flow east of the AAD. The increased flow and greater proximity of the discordance to the thermal anomaly and hotspots to the east produce a higher pressure gradient, and hence greater driving force, which results in a westward migration of the collision zone with time. Seafloor spreading over the westward moving collision zone has produced the observed arcuate-shaped anomaly with the accompanying oblique depth anomaly trends. We cannot decipher from depth anomalies alone whether the converging flows downwell within the AAD or simply mix with upwelling materials. ¿1990 American Geophysical Union