The Long-Term Evolution of the Dynamo and Implications for Paleointensity Variations

The geodynamo has persisted for at least 3.5 Gyr because of core convection driven by cooling and inner core solidification [1]. Recent measurements of core thermal conductivity [2-4] imply that driving core convection prior to inner core solidification requires heat flows in excess of 15 TW [1-3]. This high heat flow implies rapid cooling and an inner core inescapably less than 1 Gyr old (i.e. Neoproterozoic). The onset of inner core growth is expected to increase the magnetic field intensity by a factor of ~3, and also to alter the field geometry and rate of reversals [5]. No sudden increase in paleointensity has yet been reported during the Neoproterozoic interval, although observations are limited. Some anomalous paleomagnetic directions have been reported, and inferred to represent an episode of true polar wander [6,7].

Intrusive rocks suggest a more strongly dipolar field at 2-3 Ga compared to 1-2 Ga [8]. Theory suggests that the ancient lower core was stably stratified [1-3]; the inferred change in dipolarity could provide a constraint on the evolution of core stratification with time. The sparse paleointensity data spanning the period from 1 Ga to 3.5 Ga suggest a field intensity constant to within better than a factor of 2 [9]. That in turn implies that the rate of core cooling stayed roughly constant, implying a pervasively molten lower mantle prior to about 3 Ga. When combined with petrological constraints on the rate of upper mantle cooling, these results imply that the surface heat flux changed by only a factor of ~2 over this interval. This change is less than expected for standard convection but consistent with some models of plate tectonics [10], and is also consistent with the apparent ~Gyr mixing timescales of ancient mantle isotopic heterogeneities [11].

[1] Nimmo, Treatise Geophys. [2] Pozzo et al. 2012 [3] Gomi et al. 2013 [4] De Koker et al. 2012 [5] Roberts and Glatzmaier 2001 [6] Li et al. 2004 [7] Maloof et al. 2006 [8] Smirnov et al. 2011 [9] Aubert et al. 2010 [10] Conrad and Hager 1999 [11] O'Neill et al. 2013