C. Eymin and G. Hulot
Institut de Physique du Globe, Paris, France.
Computations of core surface flows from observations of the main magnetic field of the Earth and its secular variation provide an important insight into outer core dynamics. During the last years, our knowledge of the Earth's magnetic field improved thanks to satellite measurements. It can be hoped that this better spatial resolution of the magnetic field would benefit to core surface flow models. Indeed, the spatial resolution of these flows was mainly limited by the resolution of the secular variation models prior to satellites. In our first attempt to derive core surface flows from satellite data, we use two main field models derived from Magsat and {\O}rsted data for 1980 and 2000 by Langlais \textit{et al.} (PEPI, 2002), and the difference between them as secular variation model. It is possible to account for this secular variation with flows having a relatively high energy level for their small scale components. However, the truncation error, produced by interactions of the non-modeled and non-observed small scale components of the flow and the magnetic field, is too large for these flows. For the truncation error to be acceptable, we have to select a ``consistent'' flow, with less energy in the small scale components, for which the misfit between the observed and predicted secular variation is not larger than the truncation error but may possibly be larger than the observational error. Because of this truncation error, it is not possible to fully exploit the high resolution secular variation model. The hiding of the small scale components of the main magnetic field by the crustal field now seems to be the main limiting factor. To check the coherence of the obtained flow patterns, we also compute ``semi-synthetic'' core flow models derived with synthetic random prolongations of the main magnetic field for spherical harmonic degrees 13 to 30. The observation of 50 different semi-synthetic core surface flows shows that the main features of the consistent flow are visible in all the computed semi-synthetic flows. Finally, we compare the core surface flows derived from the 20 year average secular variation with others computed with an ``instantaneous'' secular variation model derived from 18 months of satellite data by Olsen \textit{et al.} (AGU Spring Meeting, 2002).