The Geoid: Earth's True Lumpy Shape

Why 'sea level' is not level

Ask what shape the Earth is and most people picture a sphere; a careful answer upgrades that to an oblate ellipsoid — a ball squashed at the poles because it spins. That is genuinely true and genuinely important: Earth's equatorial radius is 6378.137 km while its polar radius is 6356.752 km, so the equator bulges out about 21 km farther than the poles. The centrifugal effect of one rotation every 24 hours flings the equator outward, and the planet's own gravity can't quite pull it back into a perfect sphere. If you stand on the equator you are, in a real sense, 21 km 'taller' — farther from the center — than someone at the North Pole.

But even this smooth, dimple-free ellipsoid is a fiction. The real Earth has continents, ocean trenches, mountain roots, and — most importantly — a lumpy interior. Some regions of the mantle are dense and cold; others are hot and buoyant. Gravity is stronger over the dense patches and weaker over the light ones. Because the ocean surface is free to flow, it piles up slightly over strong-gravity regions and slumps over weak ones. Mean sea level is not a level surface at all — it is a gently rippling one, and the geoid is the name for the exact shape of those ripples extended across the entire globe, including under the continents.

The key word is equipotential. The geoid is the surface on which Earth's gravitational potential is everywhere constant — the surface a still, global ocean with no winds, tides, or currents would adopt. A ball placed anywhere on the geoid has no tendency to roll, because there is no 'downhill.' That is exactly what we mean by 'level,' and it is why every surveyor's spirit bubble and every plumb line ultimately references the geoid, not the tidy ellipsoid on your GPS.

The mechanism: mass, potential, and the plumb line

Gravity does not care about the visible surface of the Earth — it responds to mass, wherever that mass sits. A buried slab of dense rock, a plume of hot, buoyant low-density mantle material 2000 km down, an excess of ice, a deficit of water: all of it tugs on a test mass and warps the potential field. The geoid is the fingerprint of the planet's entire three-dimensional density distribution, projected onto a single surface.

Here is the crucial and counterintuitive part. You might think the geoid bulges upward over a mountain, since there's extra mass there. Over broad scales the opposite tendency often wins for deep sources, but at the surface the intuition holds: an excess of mass warps the equipotential surface so that it bows toward the mass. A local concentration of dense material makes the geoid rise (a 'geoid high'); a local mass deficit makes it fall (a 'geoid low'). The undulations you actually observe are the integrated effect of density anomalies at every depth, which is why the biggest geoid features come from the deep mantle, not from surface topography — the mantle's density contrasts, though modest per cubic meter, are spread over enormous volumes.

Because the geoid is an equipotential surface, a plumb line always hangs perpendicular to it, and the local vertical points straight into it. Where the geoid tilts relative to the ellipsoid, the plumb line is deflected — near a big mountain range the deflection of the vertical can reach tens of arc-seconds, enough to throw off precise surveys and, famously, to have confused early attempts to weigh the Earth. The angle between the true vertical (perpendicular to the geoid) and the ellipsoid normal is called the deflection of the vertical, and mapping it was one of the first practical windows into the planet's hidden mass.

• Geoid high: excess mass, stronger gravity, sea level piles up (e.g. +85 m near New Guinea, in the western Pacific; a smaller ~+60–70 m high sits over the North Atlantic near Iceland).
• Geoid low: mass deficit, weaker gravity, sea level sags (e.g. −106 m south of India).
• Undulation (N): the signed height of the geoid above or below the reference ellipsoid at each point.

The numbers: how lumpy is lumpy?

The vertical distance between the geoid and the reference ellipsoid at any location is the geoid undulation, written N. Globally, N ranges from roughly −106 m at its deepest to about +85 m at its highest — a total spread near 191 m. Compared with Earth's ≈6371 km radius, that is a relative deviation of about 3 parts in 100,000. Counting only these geoid undulations — and setting aside the ±km of real topography — by some out-of-roundness tolerance metrics the Earth's departure from a sphere is comparable to a billiard ball's, which is where the popular claim that 'the Earth is smoother than a billiard ball' comes from; by strict surface-roughness measures a polished ball is far smoother. The word 'lumpy' is honest, but the geoid's lumps are exquisitely subtle.

The single most dramatic feature is the Indian Ocean Geoid Low (IOGL), a roughly circular depression centered south of the Indian peninsula where the geoid plunges about 106 m below the ellipsoid. It is the deepest geoid low on the planet, covering an area larger than the Indian subcontinent. On the other side of the ledger, the strongest geoid high — about +85 m — sits over the western Pacific east of New Guinea, while the North Atlantic near Iceland hosts a secondary high of roughly +60–70 m; in both places sea level bulges up by tens of meters.

To keep the scales straight, contrast the geoid's undulations with the shapes it lives inside:

• Sphere → ellipsoid: the rotational flattening is about 21 km — a hundred times larger than the geoid ripples.
• Ellipsoid → geoid: the undulations are ±100 m — the true 'lumpiness.'
• Geoid → topography: Everest rises 8.8 km above the geoid and the Challenger Deep sinks ~11 km below it — the visible relief, which is what we actually walk on.

The everyday payoff is height. The elevation on a trail sign — say 2000 m — is an orthometric height, measured straight up from the geoid. GPS, by contrast, natively reports ellipsoidal height above WGS84. The two differ by the local undulation N, which can be tens of meters, so every GPS receiver silently applies a geoid model (such as EGM2008 or EGM96) to convert its raw fix into the 'height above sea level' you expect.

Weighing the planet from orbit: GRACE

For most of history the geoid was inferred laboriously from ground surveys, plumb-line deflections, and tide-gauge networks. The modern maps come from space. The landmark mission is GRACE — the Gravity Recovery and Climate Experiment — a pair of identical satellites launched in 2002 that flew in the same orbit about 220 km apart, chasing each other around the poles until 2017. Its successor, GRACE Follow-On (GRACE-FO), launched in 2018 and continues the record today.

The trick is beautifully simple in concept. As the lead satellite approaches a region of stronger gravity — a mountain range, a dense mantle blob, an accumulation of groundwater — it gets tugged forward and the gap between the two satellites widens by a hair. Once it passes the mass and the trailing satellite arrives, the gap closes again. By measuring the changing distance between them with a microwave (and, on GRACE-FO, a laser) ranging system, the mission tracks their separation to about one micron — roughly the width of a blood cell — over a baseline of 220 km. Those minute speed-ups and slow-downs are literally the satellites feeling the planet's gravity field, and inverting them yields the geoid.

What makes GRACE revolutionary is that it repeats the map every month, so it sees the geoid change over time as mass moves around. It watches Greenland and Antarctica lose hundreds of gigatons of ice a year, tracks the depletion of aquifers under India and California, follows the seasonal breathing of the Amazon, and even registered the mass redistribution from great earthquakes. In this sense the geoid stopped being a static portrait of Earth's interior and became a moving ledger of where the planet's water and ice are going — one of the most important climate-monitoring tools ever flown.

The Indian Ocean's ghost: a gravity hole 20 million years in the making

The IOGL has puzzled geophysicists since the Dutch geodesist Felix Andries Vening Meinesz detected gravity anomalies over the Indian Ocean from a submarine in the 1920s and 1930s. Why should the ocean south of India sag more than 100 m? A mass deficit in the deep interior must be responsible, but the details resisted explanation for the better part of a century.

The most compelling account came from a 2023 study in Geophysical Research Letters by Debanjan Pal and Attreyee Ghosh at the Indian Institute of Science. Using global mantle-convection models tied to plate reconstructions, they traced the low to the slow-motion aftermath of a vanished ocean. As the Indian plate rifted from Gondwana and rammed northward into Eurasia, the intervening Tethys seafloor was subducted and sank as a cold, dense slab deep into the mantle. Sinking through the lower mantle, it disturbed the great African Large Low-Shear-Velocity Province — a continent-sized pile of hot, buoyant material near the core-mantle boundary — and stirred up plumes of low-density mantle. Those light plumes rose and spread beneath the Indian Ocean, creating exactly the mass deficit needed to pull the geoid down. Their simulations suggest the low took roughly its present form about 20 million years ago.

Not everyone is fully convinced — the geoid is an integrated signal and multiple density models can fit it, so the plume interpretation is a leading hypothesis rather than a closed case. But the story captures why the geoid matters scientifically: its bumps are a CT scan of a planet's insides. Read carefully, a single warped surface encodes the ghosts of oceans that closed tens of millions of years ago and the churning of rock a thousand kilometers beneath our feet.

Common misconceptions and where the geoid actually bites

The geoid attracts a surprising amount of muddled thinking, so it's worth clearing up a few things directly.

• 'The geoid is the shape of the solid Earth.' No. The geoid is an imaginary equipotential surface — where a global, undisturbed ocean would sit. Under the continents it runs through the rock, defined by continuing the sea-level potential inland. It is not the physical surface; it is the reference the physical surface is measured against.
• 'It's the same as mean sea level.' Close, but not exact. The real ocean surface departs from the geoid by up to a meter or two because of currents, winds, temperature, salinity, and atmospheric pressure — a difference called dynamic ocean topography. The Gulf Stream, for instance, stands about a meter tall relative to the geoid. The geoid is the ocean's equilibrium shape, which the moving ocean only approximates.
• 'Earth is basically a sphere, so the geoid barely matters.' The undulations are only ±100 m, but in a GPS age that is enormous. Confuse ellipsoidal and orthometric height by 40 m and you drain the wrong reservoir, mis-grade a canal so water runs backward, or botch a flood map. Every precise elevation on Earth is anchored to a geoid model.
• 'The geoid never changes.' It shifts continuously as ice melts, aquifers drain, and great quakes rearrange mass — exactly the signal GRACE was built to watch.

There is also a long, fruitful history behind all this. Eratosthenes measured the round Earth's circumference around 240 BC; Isaac Newton predicted the equatorial bulge from rotation in 1687; 18th-century French expeditions to Lapland and Peru confirmed the flattening; and the German mathematician Carl Friedrich Gauss conceived of the 'mathematical figure of the Earth' in the 1820s, a notion his student Johann Benedict Listing named the geoid in 1872. Each step traded a simpler ideal for a truer, lumpier reality — a fitting arc for a surface whose whole point is that the Earth is not as tidy as it looks.

Three ways to describe the shape of the Earth, from simple to real

Model	What it captures	Typical departure from the geoid
Sphere	First approximation; one radius (≈6371 km)	Up to ≈21 km at the poles — ignores the rotational bulge
Reference ellipsoid (WGS84)	The rotational flattening: equator 21 km wider than poles	±100 m — a smooth mathematical surface, no density bumps
Geoid	True equipotential (mean sea-level) surface, all mass included	0 by definition — this is the reference; topography sits on top of it
Physical topography	Actual land and seafloor: Everest +8.8 km, Mariana −11 km	Measured as height above the geoid (orthometric height)