Earth's Magnetic Field

The needle that lies#

Hold a compass flat and let it settle. The needle swings, hunts, and finally points — and almost everywhere on Earth, it is not pointing at the North Pole. In London it points about a degree east of true north; a century ago it pointed a good many degrees west. In parts of Alaska the error is over 20 degrees. Sailors have known this for five hundred years and every serious chart still prints a correction for it. The angle between where the needle points and true geographic north has a name — magnetic declination — and pilots subtract it from every heading.

Worse, the thing the needle is pointing at will not hold still. The north magnetic dip pole, the place where the field points straight down, sat in the Canadian Arctic for the whole of the twentieth century. Then it began to move, and it has been accelerating: from around 15 km a year in the 1990s to something like 50–60 km a year now, striking out across the Arctic Ocean toward Siberia fast enough that the international reference model used by every phone and aircraft has had to be updated off-schedule to keep up.

And on the long view it does more than wander. Dig up a stack of ancient lava flows and read the fossil magnetism frozen into them, and you find that the entire field has completely reversed — north and south magnetic poles trading places — hundreds of times over Earth's history, most recently about 780,000 years ago.

None of this makes sense for a bar magnet buried in the rock. It makes complete sense for what the field actually is: the churning, self-sustaining output of a planet's molten iron heart.

Not a magnet — a dynamo#

The first thing to give up is the picture of a giant permanent magnet inside the Earth. It fails immediately on temperature. Iron loses its permanent magnetism above its Curie point, about 770 °C, and the Earth's interior is far hotter than that everywhere below the shallow crust — thousands of degrees in the core. There is no frozen magnet down there. There cannot be.

What there is, we know from a completely independent line of evidence: seismic waves. Shear waves cannot cross a liquid, and they vanish from a wide shadow on the far side of the planet from every earthquake — the fingerprint of a large sphere of liquid at the centre. That liquid is the outer core: an ocean of molten iron and nickel between about 2,890 and 5,150 km depth, wrapped around a solid inner core, roughly the size of Mars.

That iron ocean is the key, because iron is a metal and a moving metal is a moving conductor. Three ingredients turn it into a geodynamo:

• A conducting fluid. Liquid iron carries electric current beautifully.
• Convection. The core is heated from below and cooled from above, and as the inner core slowly freezes it releases both heat and buoyant light material. That drives the same kind of overturning you get in a pan of soup — hot fluid rising, cool fluid sinking — churning the whole outer core.
• Rotation. The Earth spins, and the Coriolis effect twists the rising and sinking columns into corkscrews, organising an otherwise messy flow into coherent, mostly north–south rolls.

Here is the loop that makes it self-sustaining, and it is exactly the electromagnetism of Maxwell's equations. A moving conductor sweeping through a magnetic field has a voltage induced across it, which drives electric currents; those currents, by Ampère's law, generate magnetic field of their own; and if the flow is arranged right, the field they generate reinforces the field they started from. A seed field is stretched, twisted, and amplified by the fluid's own motion until it settles at a strength where generation balances loss. The energy comes not from magnetism but from heat and gravity — the slow cooling and freezing of the core. Switch off the convection and the field would decay away in a few tens of thousands of years as its currents dissipated into heat. It persists only because the motion never stops feeding it.

That is the sentence to hold onto: the field is generated and maintained by motion, not stored in a permanent magnet. Everything strange about it follows from that.

Driving the core#

The widget is a cutaway of the Earth. The bright churning region is the convecting liquid outer core; the looping curves loft out of the core, thread through the mantle and out into space, and return — that is the dipole field, the same shape as the field around a bar magnet, but generated rather than fixed. What to try:

• Push the convection/rotation slider up. As you drive the fluid harder, the flow organises, the induction loop wins, and the field lines strengthen and comb themselves into a clean dipole aligned close to the spin axis. This is Earth today: a vigorous dynamo running a mostly steady field.
• Now drag the vigour right down. The regeneration falters, the field weakens, and the neat dipole degenerates into a tangle of stubby, disordered loops. There is no floor holding it up — take away the motion and there is nothing left. That is the whole point: no permanent magnet would care how fast the fluid moves.
• Press Trigger reversal. Watch the dipole collapse toward zero, thrash through a complicated multi-lobed state, and then re-form — pointing the other way. The field did not flip like a switch. It sagged, went complicated, and rebuilt itself in the opposite sense, which is exactly what the rock record shows a real reversal does.

The takeaway the widget is built to deliver: turn the motion up and the field is strong and orderly; turn it down and the field falls apart. A magnet cannot do that. A dynamo does nothing else.

The mathematics that decides whether it runs#

Three pieces of quantitative reasoning pin down the behaviour you just watched.

The dipole, and why the field is a surface phenomenon at a distance. Outside its source, a magnetic dipole of moment $m$ produces a field whose strength falls off as the inverse cube of distance:

$$B(r,\theta) = \frac{\mu_0 m}{4\pi r^3}\sqrt{1 + 3\cos^2\theta}$$

where $r$ is distance from the centre, $\theta$ is the magnetic colatitude, and $\mu_0$ is the permeability of free space. Two things fall straight out of that $1/r^3$. First, the field is twice as strong at the magnetic poles ($\theta = 0$) as at the magnetic equator ($\theta = 90^\circ$) — that is the factor $\sqrt{1+3\cos^2\theta}$ going from 2 down to 1. Second, the field drops off fast: it is around 25,000–65,000 nanotesla at the ground but only a hundredth of that a few Earth-radii out, which is why the magnetosphere ends where it does. Earth's dipole moment is about $8\times10^{22}\ \mathrm{A\,m^2}$, and the dipole accounts for roughly 90% of the field at the surface; the remaining ~10% is the messier, faster-changing non-dipole field that makes the poles wander.

The magnetic Reynolds number, which decides whether there is a dynamo at all. A magnetic field embedded in a moving conductor is subject to two competing processes: the flow drags and stretches it (advection, which can amplify it), and electrical resistance lets it diffuse and decay (ohmic diffusion). Their ratio is the dimensionless magnetic Reynolds number:

$$R_m = \frac{U L}{\eta}$$

with $U$ a characteristic flow speed, $L$ the size of the region, and $\eta = 1/(\mu_0\sigma)$ the magnetic diffusivity set by the fluid's conductivity $\sigma$. When $R_m$ is small, diffusion wins and any field bleeds away. Only when $R_m$ exceeds a critical value of order tens does stretching outpace decay and a seed field regenerate itself — a self-sustaining dynamo. Plug in core values — $U \sim 5\times10^{-4}\ \mathrm{m/s}$ (fluid motions of a fraction of a millimetre per second, inferred from how the field pattern drifts), $L \sim 2\times10^{6}\ \mathrm{m}$, and $\eta \sim 1\text{–}2\ \mathrm{m^2/s}$ for liquid iron — and $R_m$ comes out in the hundreds to about a thousand. Comfortably supercritical. The Earth's core is not marginally a dynamo; it is a strong one, which is why the field has run more or less continuously for billions of years.

Declination, the angle the compass makes. What a navigator actually needs is a single number: the angle in the horizontal plane between the direction the needle points (magnetic north) and true geographic north. That is the declination $D$, and if $B_x$ is the local horizontal field component pointing to geographic north and $B_y$ the component pointing east, then

$$D = \arctan\!\left(\frac{B_y}{B_x}\right)$$

$D$ is not a global constant — it is a full map that varies from place to place and drifts year to year, which is exactly why nautical charts print local declination with an annual rate of change, and why the compass is a liar you can trust only once you know the correction.

Why the poles are not the poles#

A persistent misconception deserves killing precisely, because there are three different "north poles" and they are genuinely different places.

The geographic North Pole is where the rotation axis pierces the surface — the top of the globe, fixed by definition. The Earth's dipole axis is tilted about 11° away from that rotation axis, so the geomagnetic pole (where the best-fit dipole axis meets the surface) already sits far from the geographic one, near northwest Greenland. And the magnetic dip pole — the place a compass needle actually points straight down — is different again, because the real field is not a perfect dipole: the ~10% non-dipole part shoves it hundreds of kilometres off the geomagnetic pole, into the Arctic Ocean.

Because the non-dipole field is produced by the shallowest, fastest-changing eddies in the core flow, the dip pole is the most mobile of the three. Its recent sprint from Arctic Canada toward Siberia is not the planet tipping over; the rotation axis has not moved at all. It is the core's fluid reorganising underneath a fixed crust, and the compass faithfully reporting the change. A permanent magnet would give you one fixed pole. A dynamo gives you a pole that goes for a walk.

When the whole field turns over#

Now the deepest strangeness. The field does not only wander — it reverses, north and south swapping entirely, and it has done so hundreds of times.

We know this from the same archive that clinched plate tectonics. New ocean crust erupting at a mid-ocean ridge cools through the Curie temperature and freezes in the direction of the field at that instant, then splits and spreads outward in both directions. The result is a barcode of magnetic stripes on the seafloor, symmetric about every ridge, alternating normal and reversed polarity — a direct tape recording of the field's history, ragged and irregular, the same sequence appearing in every ocean. Reversals are not periodic: sometimes the field holds one polarity for tens of millions of years (a superchron), sometimes it flips every few hundred thousand. The last full reversal, the Brunhes–Matuyama, was about 780,000 years ago.

The other misconception to correct is the disaster-movie version, in which the field switches overnight and chaos ensues. The record says otherwise. A reversal is a process, not an event, taking somewhere between roughly a thousand and ten thousand years. What happens is exactly what the geodynamo widget shows: the dipole weakens to a small fraction of its normal strength, the field becomes multipolar — several weak poles scattered over the globe rather than one clean dipole — and then the dynamo re-establishes a strong dipole that happens to point the opposite way. During the low-field interval the surface is less well shielded, but there is no evidence in the fossil record of mass extinction tied to reversals. It is slow, it is messy, and life has sat through hundreds of them.

The bubble that keeps the weather out#

The field's reach does not stop at the ground. It fills the space around the Earth and inflates a protective cavity called the magnetosphere, and that is where the field earns its keep for life.

The Sun blows a continuous solar wind — a thin plasma of protons and electrons streaming outward at roughly 400 km/s, faster during storms. It carries its own magnetic field and it would, unopposed, strip gas from the top of the atmosphere the way it appears to have done on Mars, which lost its own dynamo and much of its air long ago. Earth's field stands in the way. The solar wind cannot easily cross magnetic field lines, so it piles up against the field and is deflected around the planet, carving out a cavity: compressed on the dayside into a blunt nose about ten Earth-radii out (squashed to six or seven during a big storm), and drawn out into a long magnetotail streaming hundreds of Earth-radii downwind on the nightside. The same shield turns away most of the high-energy cosmic rays that would otherwise reach the surface.

This second widget is a schematic — a cartoon of the geometry, not a plasma simulation — but the shapes are right. The stream from the left is the solar wind; the teardrop cavity is the magnetosphere. What to try:

• Watch the deflection. Most particles hit the dayside boundary and are swept around the flanks and down the tail. They do not get through. That deflected flow is the shielding, drawn.
• Follow the few that leak in. Near the poles the field lines are open funnels, and some particles spiral down them into the upper atmosphere. Where they hit, the sky lights up — the aurora.
• Turn up the solar-wind strength to simulate a solar storm. The dayside boundary is pushed inward as the cavity compresses, the tail loads up, and the auroras brighten and spread toward the equator. This is a real effect: during large geomagnetic storms auroras have been seen from the tropics, and the same storms induce currents that can trip power grids.

The auroras are worth dwelling on, because they are the field made visible. Solar-wind particles funnelled down the polar field lines slam into oxygen and nitrogen atoms high in the atmosphere and kick their electrons up; as those electrons fall back they emit light — green and red from oxygen, blue and violet from nitrogen. The rings of aurora sit around the magnetic poles, not the geographic ones, which is the whole reason they are called the aurora borealis and australis and why they brighten when the field is disturbed. They are the exhaust glow of the shield doing its job.

Reading, using, and losing the field#

The magnetic field threads through far more than navigation. It is a scientific instrument in its own right: because new rock records the field as it cools, paleomagnetism lets geologists reconstruct where continents sat hundreds of millions of years ago, and the reversal barcode is a global clock used to date seafloor and sediment alike. Migratory animals — birds, sea turtles, some fish — carry a magnetic sense and navigate by the very field lines the compass reads, using both direction and, it appears, the local field's inclination and intensity as a coordinate.

It is also, increasingly, an operational concern. The reference field models baked into every smartphone compass and aircraft navigation system have to be revised every few years precisely because the dip pole keeps moving; the sprint toward Siberia forced an unscheduled update. Space weather — the storms that compress the magnetosphere — degrades GPS accuracy, endangers satellites, exposes astronauts and polar-route air crews to radiation, and drives ground currents that have blacked out power grids, most famously across Quebec in 1989.

And the long story is one of a field that will not last forever in its present form. The dynamo runs on the Earth's internal heat, and that heat is slowly leaking away as the planet cools and the inner core freezes. On billion-year timescales the core will eventually solidify and the dynamo will die, as it seems to have died on Mars. When it does, the shield will go with it. For now, though, the churning iron under our feet keeps the compass pointing — not quite true, always drifting, occasionally flipping — and keeps the sky, most of the time, from catching fire.