Delta subsidence

Why do river deltas start sinking once dams upstream stop the mud from reaching them?

A dam is built eight hundred kilometres inland. Within a few decades the delta at the river's mouth is losing land to the sea, salt is climbing into its groundwater, and villages that stood dry for centuries are flooding at ordinary high tides. The obvious reading is that removing the mud started the sinking. But hold that phrasing up to the light: how would the absence of a delivery make solid ground go down? Absence causes nothing. Something else must already have been happening.

Ask what a delta is made of. Not bedrock — a delta is a heap of young sediment, waterlogged, laced with peat and rotting vegetation, laid down over the last few thousand years on top of whatever the river flowed over before. Think what such a heap does under its own weight. Water squeezes from the pore spaces, grains repack, organic layers oxidise and shrink. It consolidates. That is not an event but a condition: every delta on Earth is compacting today and was compacting a thousand years ago.

Add a second, slower term: the crust itself flexes downward under the accumulating load, and in some regions is still adjusting to ice sheets that melted long ago — the process treated at length in isostatic-rebound.md, which reaches this collection from the opposite direction, describing the ground that is going up.

So the delta surface has always been sinking. Why then was it ever above water? Because of the other half of the ledger. During floods the river left its channel, spread across the plain, and laid a film of sediment on top — and periodically it abandoned its course altogether and built somewhere else. Height gained from deposition roughly matched height lost to compaction, and the difference between two large numbers was small. That is the equilibrium: not a static delta, but a fast-sinking one that was being resurfaced just as fast.

Now the intervention makes sense. A reservoir traps sediment — effectively all the coarse fraction and much of the fine, so that a river which once carried a great annual load past a given point carries very little. But notice that the dam is only one of two cuts, and possibly not the larger. Embankments built to stop the river flooding also stop it depositing: the sediment that does get through is conveyed neatly down a walled channel and dropped off the edge of the continental shelf, where it does the delta plain no good whatever. Both interventions attack the deposition term. Neither touches compaction, which continues at its old rate, unopposed.

And that is the whole answer in shape: the delta did not start sinking. It never stopped. What stopped was the resurfacing, and a rate that had been hidden by cancellation became visible in full.

Two consequences follow, worth deriving rather than being told. First, the fastest-sinking deltas should be the ones where humans accelerate compaction as well — and they are. Pumping groundwater or gas from beneath a delta city drains the pore pressure holding the grains apart, and land in the worst-affected districts of cities such as Jakarta has dropped at rates measured in centimetres per year, against a global sea-level rise of a few millimetres per year: locally the land is often the larger term by an order of magnitude. Second, the shoreline should retreat as well as the surface drown, because waves that used to rework a fresh supply of sand now have only the existing coast to work on — which is what happened to the Nile's promontories after the sediment stopped arriving.

"The sea is rising and drowning the delta" gets the outcome right and the accounting wrong. What floods a place is relative sea level — the sea's height minus the land's — and in heavily pumped deltas the land's motion is the larger contributor. That matters practically, because one term is global and slow to influence while the other is local and can be stopped by regulating a well field.

"Dams cause subsidence" is likewise a mislabel worth resisting. Dams cause the loss of accretion; the subsidence was already paid for by geology.

And the deposition term is not simply a volume of mud: it must arrive and be spread across the plain. That is why a delta can be starved even where the river still carries sediment, and why embankment policy can matter as much as dam policy.

Two testable claims hold the account up. If dated cores through a delta showed its surface had been essentially static before the dams rather than continuously accumulating, the "always sinking, always resurfaced" model would be refuted outright. And if deltas whose rivers were never dammed or embanked showed the same acceleration in relative sea-level rise as the starved ones, sediment supply would not be doing the work attributed to it here.

A closer neighbour is silted-harbours.md, which follows the same sediment budget with its sign reversed — there the mud arrives where it is not wanted and buries a port. Both files rest on the same fixed point: a coast is a running balance of supply and removal, and human works rarely change the balance where they are aimed.

• How sediment diversions attempt to rebuild delta land deliberately, and why they trade farmland and fisheries against elevation.
• Why some deltas switch their main channel every few centuries, and what happens when engineering forbids the switch.

Every term the collection defines is gathered in the glossary.