Is Earth’s magnetic field about to flip?

The intervals between reversals are irregular, with no periodicity anyone has established. Over the past ten million years they have averaged about one reversal every 200,000 years, and the present interval has already run longer than that. The switch itself is quick by geological standards: the directional change appears to take under a thousand years, with a precursor before it and a rebound after, neither longer than a few thousand. Those durations come from recent reversals, and not every reversal was quick: a deep-sea record of two transitions about 40 million years ago, published in 2026, puts them at 18,000 and 70,000 years.

During a transition the field does not simply vanish. It becomes weaker and more complicated, no longer a simple two-pole field, and in simulations several poles appear at once before it reorganises.

The dipole has been weakening since direct measurement began, and a large region of low field strength over the South Atlantic has grown in both directions along its edges, which matters for satellites passing through it.

The present dipole is still strong, far above the values that preceded the last five reversals, and long declines in the record have ended without one.

A 2006 paper asked this question in its title. A 2016 review of the reversal record cites its statistical treatment of the past two million years for the conclusion that the present decline falls within the normal range of variation, and a 2022 reconstruction of the past 9,000 years found anomalies like the South Atlantic one that came and went.

One 2016 study of the anomaly leans the other way. It reads the patches of reversed field beneath it, and the rate of decay, as like those before past reversals and excursions, while finding the dipole itself not especially low.

A 2018 paper in the Proceedings of the National Academy of Sciences is titled, in full, that the Earth's magnetic field is probably not reversing.

Its argument is from the past rather than from a simulation of the core: a model of the field spanning 30,000 to 50,000 years ago finds two earlier episodes, about 49,000 and 46,000 years ago, when the field was comparable to today and carried an anomaly like the South Atlantic one, and neither developed into an excursion or a reversal.

Neither of the two most recent excursions, the Laschamp at 41,000 years and the Mono Lake at 34,000, resembles what the field is doing now either. That is the state of the evidence, and it is not a guarantee.

The field is generated by convecting liquid iron in the outer core, and understanding how a system like that works is not the same as being able to predict it.

Satellite measurements now map the field and its rate of change in detail, and let changes at the surface be traced to the top of the core. Error growth in dynamo simulations still puts the forecast horizon at a few decades to a century, at least an order of magnitude short of the time a reversal takes, so no one can say when the next one will begin.

A weaker, more complex field would raise the radiation dose at satellite altitudes. The South Atlantic Anomaly is the small version of it: satellites crossing that region record measurably more particle hits on their electronics.

Navigation follows the field, and already has to keep correcting for the drift of the magnetic north pole. A reversal would ask the same of it on a far larger scale.

Those are infrastructure problems on a timescale of centuries or longer, which is an unusual combination and makes them easy to either dismiss or exaggerate.