Isostatic rebound
A Socratic walk-through of isostatic rebound — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why is land in Scandinavia still rising thousands of years after the ice that pressed it down melted away?
Harbours around the Gulf of Bothnia keep going dry. Jetties built by a grandfather sit above the waterline. The usual explanation is that the crust was pressed down by ice and is springing back — but hold that word springing up to the light. A spring released returns at once. The ice left about ten thousand years ago. What kind of spring takes ten millennia, and if it is that slow, is it a spring at all?
Reasoning it through
REASONING #Start with the load. The Fennoscandian ice sheet at its maximum was of order three kilometres thick over the northern Baltic. How far down should that push the crust? Not by its own thickness — the crust floats on denser mantle, so it sinks until the weight it displaces balances the weight added. Ice has a density near 917 kg per cubic metre and the mantle beneath is around 3,300, so the depression is the ice thickness scaled by that ratio: 3,000 metres times 917 over 3,300, about 830 metres. That is a derivation from two densities, not a measurement, but it lands where the field's estimates for the Baltic depression sit.
Now the timing, which is where the intuition fails. The crust is elastic, and elastic really does mean instant — when a load comes off, the elastic part of the recovery happens as fast as a seismic wave can carry it. So if rebound were elasticity, it finished ten thousand years ago and Scandinavian harbours would be stable. They are not: GPS networks measure the northern Gulf of Bothnia rising by about ten millimetres a year, right now.
So the elastic answer is not wrong, it is finished. Something else is still running, and only one part of the process could be slow: the part that has to move material. To let the crust down, mantle rock had to flow sideways out from under the ice sheet; to let it back up, that rock has to flow back in. Mantle rock is solid — it transmits shear waves — yet over long times it creeps, with an upper-mantle viscosity around 10^21 pascal-seconds. The rate is not set by how strong the crust is at all. It is set by how fast rock can ooze.
That reframing makes a prediction worth testing. Squeeze a viscous fluid out from under one place and it has to go somewhere — so a ring of land around the ice sheet should have bulged up while the centre went down, and should now be sinking as the material drains back. It is. Around the former North American ice sheet, the mid-Atlantic coast of the United States is subsiding at on the order of a millimetre a year for exactly this reason. A spring model predicts no such thing. A flow model requires it.
Sweden's highest ancient shoreline, the högsta kustlinjen, sits about 285 metres above present sea level in Ångermanland. Against a derived depression near 830 metres that looks like a shortfall — until you account for two things. Global sea level itself rose roughly 120 metres as the world's ice melted, so emergence records land uplift minus that rise; and much of the rebound happened while the ice was still going, before that coastline existed to be marked.
Where the account is genuinely unfinished: the observed decay does not fit a single exponential. Fitting it requires trading off the thickness of the rigid lithosphere against upper- and lower-mantle viscosity, and the lower-mantle value is contested across more than an order of magnitude. That is not an academic quibble — the correction for ongoing rebound is one of the largest adjustments applied to satellite measurements of sea-level rise and ice mass loss, so the groups producing those numbers have a direct stake in which viscosity model wins.
Where this sits next to its neighbours: the collection's piece on glacier flow turns on the same fact — a solid that creeps under sustained stress — at a different fixed point. Ice deforms fast enough to see in a decade; mantle rock is so much stiffer that the same physics reads as geology rather than weather.
The analogy
THE ANALOGY #Think of a heavy filing cabinet standing for years on a thick carpet with dense foam beneath. Lift it away and the carpet pile springs up at once — the elastic part, done in a moment. But the foam has been squashed and its material pushed outward, and it seeps back over hours, the dent shrinking long after the cabinet is gone, while the slight ridge around the dent slowly settles.
Foam recovers because it stores elastic energy and pushes itself back; the mantle stores nothing and is driven purely by the pressure difference between the depressed column and its surroundings, which is why the motion decays smoothly toward zero rather than stopping at a remembered shape.
Clarifying the model
THE MODEL #Three refinements hold the account together. "Isostasy" is a balance condition, not a mechanism: it says where the crust ends up and nothing about when, and the when belongs entirely to viscosity. Elastic and viscous are not competing explanations but sequential ones — both happen, one instantly and one over millennia, and confusing them is the whole of the puzzle. And the process is not local: near-field uplift and far-field forebulge subsidence are one flow field, which is why an explanation of Scandinavia has consequences for Virginia.
One misconception worth naming as simply wrong: that the land is rising because the ice is melting now. Present-day ice loss does add a small, genuinely elastic uplift near Greenland and Alaska, but Scandinavia's ten millimetres a year is the delayed response to a load that vanished before writing was invented. The land is not reacting to the world it is in. It is still finishing its reply to the last one.
A picture of it
THE PICTURE #How to readRead left to right as real elapsed time, watching which of the two recoveries acts in each era. The 12,000-to-10,000 entry is the whole argument in one line: the elastic part finishes the moment the load goes, so everything to its right is viscous flow alone. The last two entries are the live test — the centre rising while the ring around it sinks is what flow predicts and elasticity does not. The 830-metre depression is derived from the two densities; the 10 mm a year and the 285-metre shoreline are measured.
What became clearer
WHAT CLEARED #The land is not springing back. The spring already sprang, ten thousand years ago, in an instant. What remains is solid rock flowing at the speed rock flows, and the ten-millennium delay is not a puzzle to be explained away but the direct readout of mantle viscosity. Scandinavia's rising harbours are a viscometer the size of a continent — which is precisely how the mantle's stiffness came to be known at all.
Where to go next
ONWARD #- How the same measurements are inverted to estimate mantle viscosity, and why the lower-mantle value stays contested.
- Why present-day ice loss produces an elastic uplift that is fast, local, and easy to confuse with this slow one.
Key terms
TERMS #| Term | What it means |
|---|---|
| Isostasy | the buoyant balance in which the crust floats at a depth set by the weight it carries. |
| Glacial isostatic adjustment (GIA) | the ongoing deformation of the solid Earth in response to past ice loading and unloading. |
| Forebulge | the ring of uplifted ground around a loaded region, formed by displaced mantle and subsiding as that material returns. |
| Viscosity | resistance to flow; for the upper mantle, around 10^21 pascal-seconds, which sets the thousands-of-years timescale of rebound. |
Every term the collection defines is gathered in the glossary.