On core surface flows inferred from satellite magnetic data

Satellite-data allows the magnetic field produced by the dynamo within the Earth’s core to be imaged with much more accuracy than previously possible with only ground-based data. Changes in this magnetic field can in turn be used to make some inferences about the core surface flow responsible for them. In this paper, we investigate the improvement brought to core flow computation by new satellite-data based core magnetic field models. It is shown that the main limitation now encountered is no longer the (now high) accuracy of those models, but the “non-modelled secular variation” produced by interaction of the non-resolvable small scales of the core flow with the core field, and by interaction of the (partly) resolvable large scales of the core flow with the small scales of the core field unfortunately masked by the crustal field. We show how this non-modelled secular variation can be taken into account to recover the largest scales of the core flow in a consistent way. We also investigate the uncertainties this introduces in core flows computed with the help of the frozen-flux and tangentially geostrophic assumptions. It turns out that flows with much more medium and small scales than previously thought are needed to explain the satellite-data-based core magnetic field models. It also turns out that a significant fraction of this flow unfortunately happens to be non-recoverable (being either “non-resolvable” because too small-scale, or “invisible”, because in the kernel of the inverse method) even though it produces the detectable “non-modelled secular variation”. Applying this to the Magsat (1980) to Ørsted (2000) field changes leads us to conclude that a flow involving at least strong retrograde vortices below the Atlantic Hemisphere, some less-resolved prograde vortices below the Pacific Hemisphere, and some poorly resolved (and partly non-resolvable) polar vortices, is needed to explain the 1980–2000 satellite-era average secular variation. The characteristics of the fraction of the secular variation left unexplained by this flow are also discussed.