An examination of simulated geomagnetic reversals from a palaeomagnetic perspective

Abstract

Four magnetic polarity reversals that occurred during two numerical simulations of the Glatzmaier–Roberts geodynamo display a range of behaviour that resembles records of real reversals of the Earth‘s magnetic field in some ways, and suggests additional insights in others. Two reversals happened during the homogeneous simulation, which prescribes spatially uniform heat flux at the core–mantle boundary (CMB); and two occurred during the tomographic simulation, which specifies variable CMB heat flux patterned after a low–order seismic velocity model from tomographic investigation of the lower mantle. All but one were accomplished within 2000–7000 (model) years, whereas the second tomographic reversal took 22 000 years. The two homogeneous transitions display low intensities typical of real reversals, with longer–term variation resembling what has been called ‘sawtooth’ behaviour. During the first tomographic reversal extremely high non–dipole fields occur in some regions, the result of strong patches of vertical flux that appear in less than 100 years and grow rapidly for several hundred more. The intensity during the second tomographic reversal is unusually low for a long time, and large–amplitude oscillations in direction are common. The fields in the middle of the polarity transitions are dominantly non–dipolar for all but the first tomographic reversal. One consists of spherical harmonics that are mainly antisymmetric about the equator, two by symmetric harmonics, and one by a mixture of symmetric and antisymmetric harmonics. Despite this wide variety of characteristics, all reversals occur when the non–dipole energy trend is upward. Finally, after running 300 kyr and reversing twice, the density of transitional virtual geomagnetic poles in the tomographic simulations exhibits a crude statistical correlation with areas of higher–than–average CMB heat flux, offering some support for hypotheses of preferred bands and patches.