Judging by the growing procession of research grant proposals aimed at studying the inner workings of the Earth’s core through computer modelling, it would be easy to assume that a major breakthrough was just over the horizon. What you need is some kind of supercomputer to handle the massive complexity of core fluid dynamics and then channel that through one of several concepts of a geodynamo, first towards simulating the present field and then to how the geomagnetic field swirls and occasionally flips. The fourth biggest there is belongs to the Japanese geophysical community; the Earth Simulator, which is certainly well ahead, in terms of power and speed, of facilities available to less endowed scientists. Recently, about 10% of its power was let loose for a 9 month modelling run that focussed on complex motion in the liquid outer core that theory should generate (Takahashi, F. et al. 2005. Simulations of a quasi-Taylor state geomagnetic field including polarity reversals on the Earth Simulator. Science, v. 309, p. 459-461). Hitherto, modelling had produced pictures of varying magnetic intensity that bore some resemblance to the real magnetic field at the Earth’s surface, and did indeed come up with reversals. Yet a variety of models all produced similarly plausible patterns in space and time. The snag was the limit to matching the viscosity of liquid iron with spin rate. Geomagnetists suspect that the Ekman number, which represents that relationship, is very low in the Earth’s core, i.e. there is very low drag in core circulation, and that adds to complexity. Until the Earth Simulator was built, no power on Earth could deal with the high spatial resolution needed to simulate properly motions at low Ekman numbers. Takahashi and colleagues were able to drop the Ekman number 10 times below any previous simulation.

Real-looking features did begin to emerge in the time sequence for the field at the core’s surface. The most interesting was the formation of zones of opposed polarity at high latitudes, soon (in about 1000 years of simulated time) to be followed by a reversal. The zones move progressively polewards to coalesce, when the overall magnetic polarity all but disappears, and then a reversed field becomes established. However, this is not real but a model dependant phenomenon, even though it is possible to see patterns akin to those observed today – many geophysicists believe the Earth is on a magnetic cusp before a reversal. Will it ever be real is an obvious question, in the same way that related climate simulations may flatter to deceive. The problem is not a lack of models, nor conceivably computing power, but a lack of real data. The ocean floor contains masses of information on past reversals, and cunning analyses of palaeomagnetism in lavas that cooled slowly through the Curie point at the time of a reversal show astonishing things that happened. Excellent maps of the modern field are available, but reality in a reversal is a time series of that mapped field. Without such data, and the time to collect it (the modelling simulates evolution over 5200 years) before the next order-of-magnitude jump in computing power (perhaps 10 years off), it is very difficult to see a justification for this kind of modelling, as opposed to that for climate, which does have a more rapid response time.

See also: Kerr, R.A. 2005. Threshold crossed on the way to a geodynamo in a computer. Science, v. 309, p. 364-365.