Now, new insight into the nature of this process is helping make sense of the aberrant behaviour of Earth's magnetic field over thousands and millions of years. Writing in the April 1999 issue of the Geophysical Journal International (GJI), Professor David Gubbins of the University of Leeds describes a new scenario for physical processes in Earth's core that explains in broad terms the pattern of past episodes in which Earth's magnetic field has changed dramatically in strength, direction or both.

Certain rocks preserve a record of the state of Earth's magnetism at the time they formed. From studying and dating them, geophysicists have known for some time that Earth's magnetic poles have often flipped completely in the remote past. The last such reversal took place 700 thousand years ago. But there is also evidence for more frequent episodes when the magnetic poles have moved a large distance - 45 degrees or more away from the geographical pole - then returned. These events, known as 'excursions', are rather like failed attempts at reversal. When they occur, the strength of the magnetic field falls dramatically as well, by a factor of 5 or 10.

Professor Gubbins has drawn on recent experimental results, particularly those from a research group in Utrecht headed by Dr Cor Langereis, which clearly identify six relatively recent magnetic excursions as true global phenomena. All the excursions lasted roughly the same length of time - about 5,000 years. Furthermore, preliminary results from the recent Ocean Drilling Program Leg 172 have revealed more than twenty excursions recorded in sediments of the North Atlantic in the same time period. While these events have not been correlated world-wide, the sediments indicate very clearly that excursions are quite frequent events. Professor Gubbins noted that there are about ten excursions between each full reversal. Every 20 - 50 thousand years, the Earth's magnetic field collapses in a failed attempt to reverse, but then re-establishes itself quickly over a timescale of just a couple of thousand years.

This pattern of behaviour can be explained, says Professor Gubbins, by the much longer period of time it takes for magnetic change to take effect in the solid inner core than in the fluid outer core. In the outer core, the magnetic field changes in response to the flow of the liquid iron, which typically moves 10 or 20 kilometres per year. At that rate a 'parcel' of iron moves from the top of the inner core to the bottom in about 500 years or so, and the magnetic field in the outer core could also change that quickly.

However, the magnetic field also permeates the solid metal inner core. Here change is much slower, governed by the response to electrical resistance. The timescale is more like 5,000 years than 500. The difference of a factor of ten between these two timescales explains why there are about ten excursions for every reversal.

In the scenario painted by Professor Gubbins, excursions represent a reversal of the magnetic field in the liquid outer core, which takes only a few centuries. The much slower process of change in the solid inner core begins, but more often than not, the dynamo in the outer core re-establishes the magnetic field there in its original direction before total reversal in the whole core is achieved. The 5,000-year timescale for magnetic change in the inner core also explains why no excursions lasting longer than about 5,000 years are identified in the best data; an episode lasting more than 5,000 years will end up with total reversal.

So far these ideas are largely qualitative. Computers currently available are not powerful enough for the kind of simulations needed to reproduce the properties of Earth's liquid core in sufficient detail, but Professor Gubbins is hopeful that the next generation of parallel supercomputers will finally help theorists explain the mysteries of Earth's magnetic field reversals.

Note: