Persistent westward drift of the geomagnetic field at the core–mantle boundary linked to recurrent high-latitude weak/reverse flux patches

Observations of changes in the geomagnetic field provide unique information about processes in the outer core where the field is generated. Recent geomagnetic field reconstructions based on palaeomagnetic data show persistent westward drift at high northern latitudes at the core–mantle boundary (CMB) over the past 4000 yr, as well as intermittent occurrence of high-latitude weak or reverse flux patches. To further investigate these features, we analysed time-longitude plots of a processed version of the geomagnetic field model pfm9k.1a, filtered to remove quasi-stationary features of the field. Our results suggest that westward drift at both high northern and southern latitudes of the CMB have been a persistent feature of the field over the past 9000 yr. In the Northern Hemisphere we detect two distinct signals with drift rates of 0.09° and 0.25° yr ⁻¹ and dominant zonal wavenumbers of m = 2 and 1, respectively. Comparisons with other geomagnetic field models support these observations but also highlight the importance of sedimentary data that provide crucial information on high-latitude geomagnetic field variations. The two distinct drift signals detected in the Northern Hemisphere can largely be decomposed into two westward propagating waveforms. We show that constructive interference between these two waveforms accurately predicts both the location and timing of previously observed high-latitude weak/reverse flux patches over the past 3–4 millennia. In addition, we also show that the 1125-yr periodicity signal inferred from the waveform interference correlates positively with variations in the dipole tilt over the same time period. The two identified drift signals may partially be explained by the westward motion of high-latitude convection rolls. However, the dispersion relation might also imply that part of the drift signal could be caused by magnetic Rossby waves riding on the mean background flow.