Data from the Orsted and CHAMP satellites have allowed modelling of the geomagnetic secular variation (SV) with unprecedented accuracy. The spectrum of the SV is consistent with generation by advectively dominated processes. Based on the SV model, it is not possible to reject the frozen-flux hypothesis, but the spectrum of the SV implies that a conclusive test of frozen-flux is not possible. We parametrize the effects of diffusion as an expected misfit in the flow prediction due to departure from the frozen-flux hypothesis; at low spherical harmonic degrees, this contribution dominates the expected departure of the SV predictions from flow to the observed SV, while at high degrees the SV model uncertainty is dominant. We construct fine-scale core surface flows to model the SV. Flow non-uniqueness is a serious problem because the flows are sufficiently small scale to allow flow around non-uniqueness contours. Nevertheless, we find evidence to support previously suggested polar vortices. For this model of field and SV, predicted variations in length of day from modelled core angular momentum vary over a large range, although there is evidence that this effect is reduced with longer time-series of magnetic data and better parametrization of the external magnetic field.
Geophysical Journal International, 2006, Vol 166, Issue 2, p. 518-528