The Earth's global magnetic field is generated and maintained through the dynamo process, whereby thermal convection of liquid iron in the outer core drives magnetohydrodynamic currents that self-sustain a large-scale field dominated by its dipole component.
Evidence from the remanent magnetization of ocean-floor rocks formed at mid-ocean ridges shows that this dipole has reversed its polarity many times over geological history, at irregular intervals spanning several hundred thousand to tens of millions of years, with individual reversals completing over timescales of roughly one thousand years. Despite a large body of simulation work including the NIFS group's own landmark Science paper in 2002 reproducing aperiodic reversals, the physical trigger mechanism has remained completely unknown.
In the current study, the team - led by Assistant Professor Hiroki Hasegawa, Associate Professor Hiroaki Ohtani, and Professor Emeritus Tetsuya Sato - simplified the problem deliberately, setting aside full reversal reproduction in favor of a detailed examination of what determines polarity in the first place.
They applied a three-dimensional MHD code using the Yin-Yang grid, a dual-component spherical computational grid developed at Kobe University that avoids the extreme grid compression and associated numerical stiffness that arise near the poles in conventional spherical polar coordinates. Calculations were performed on the NIFS Plasma Simulator supercomputers "Raijin" and its successor "Sosei," the latter entering service in July 2025 with a total theoretical peak performance of 40.4 petaflops across three heterogeneous subsystems.
The team ran 50 simulations, each starting from the same nearly steady helical convection field but using a different random weak magnetic perturbation as the initial condition. All 50 runs converged to a dipole-dominated magnetic field, and northward and southward polarities emerged with approximately equal probability across the ensemble.
When the researchers repeated the experiment with the convection direction reversed, the roughly equal probability of the two outcomes was unchanged, demonstrating that the convective flow structure does not determine polarity. Instead, polarity is set by the initial weak magnetic perturbations - an analogue of symmetry breaking driven by microscopic fluctuations.
The time evolution observed in the simulations fell into two clearly separated stages. In the first stage, the magnetic field grows and the polarity flips periodically on the magnetic-diffusion timescale, the characteristic time over which finite electrical resistance in the conducting fluid allows the field to diffuse and decay.
In the second stage, one polarity asserts dominance, the field saturates at a stable amplitude, and periodic reversals cease. The team further confirmed the robustness of this stable state by re-applying weak perturbations after saturation: the pre-existing dipole field was unaffected, showing that small disturbances alone cannot break the equilibrium once it is established.
The bi-stability result has direct implications for interpreting the geological reversal record. Previous simulation studies have largely framed reversals as events tied to the magnetic-diffusion timescale, but real reversals are aperiodic and occur on far longer timescales - a discrepancy the standard MHD framework has not fully explained.
The NIFS team argues that the extreme robustness of the stable states identified here implies that real reversals must be triggered by physics beyond conventional MHD theory. The leading candidates they propose are anomalous magnetic diffusivity or anomalous viscosity arising from microscopic plasma instabilities that occur when particle distributions deviate from equilibrium - effects that are absent from continuum MHD but present in the actual particle-based plasma of the outer core.
The team also raises a methodological caution: magnetohydrodynamic simulations discretize continuous space onto a finite grid, and the finite-difference errors this introduces can become amplified at stagnation points - regions where flow energy concentrates. Some prior simulation results reporting irregular magnetic reversals may, they suggest, have been caused by this numerical artifact rather than by genuine physical mechanisms. Future work will use high-accuracy simulations to carefully delineate what conventional models can and cannot reproduce before new physically motivated models are constructed.
Research Report:Bi-stable dipole polarity in spherical shell dynamo with quadruple convection
Related Links
National Institute for Fusion Science
Earth Observation News - Suppiliers, Technology and Application
| Subscribe Free To Our Daily Newsletters |
| Subscribe Free To Our Daily Newsletters |
