Floodplain storage

Why does a river need land that it does not touch except once a decade?

Walk along a river in an ordinary month and the flat ground on either side looks like wasted space. The water is comfortably inside its banks; the field beside it is dry, and has been dry for years. It is very hard to look at that field and see part of the river.

So here is the question worth sitting with: what is that land doing on the ninety-nine per cent of days when the river never reaches it? If the answer were "nothing", building on it would be free. It is not free, and the reason has less to do with getting flooded than with what the field does to the flood.

Start with a channel and a volume of water arriving faster than the channel can pass it. Two things can happen. The water can rise — deeper, faster, more forceful — or it can spread sideways over flat ground. Which of those two does more to reduce what arrives downstream an hour later?

Spreading, decisively, and for a reason that is mostly geometry. A channel is narrow; adding depth to it adds relatively little cross-section. A floodplain is wide and almost flat, so a few centimetres of depth across it holds an enormous volume. That volume is briefly taken out of transit. It is not destroyed — it drains back over the following hours and days — but it arrives later, and later is the whole game. A flood peak is not a quantity of water; it is a quantity of water per unit time.

There is a second effect on top of storage. Water moving over a rough surface — grass, hedgerows, standing trees, ploughed furrows — travels far slower than water in a smooth channel. So the floodplain both holds and delays. The result, measured on a hydrograph, is a peak that is lower and later than it otherwise would be.

Now ask how often this land is genuinely needed. Rivers in temperate settings typically fill their channels to the brim — bankfull discharge — at intervals of roughly one to two years, an observation associated with Leopold and Wolman's work in the 1950s. Anything larger than that goes over the top by design. So the floodplain is not an occasional accident; it is the part of the channel used by the flows that build the valley. The river, over centuries, made that flat land by depositing sediment during exactly these events.

Here is where the reasoning becomes uncomfortable. Suppose you build a levee to keep water in the channel. You have not removed the water; you have removed its storage. That volume now stays in the channel, moving faster, and reaches the next town sooner and higher. Local protection can export risk downstream, and the effect is well documented on heavily engineered reaches. The honest qualifier is that the magnitude depends strongly on the reach, on what fraction of the plain was disconnected, and on how tributary peaks happen to time against each other — so "levees always worsen downstream flooding" is too strong a statement.

And there is a behavioural consequence that may be larger than the hydraulic one. Gilbert White observed in 1945 that flood protection encourages development on the protected land, so the value at risk rises. The levee reduces the frequency of flooding and increases the consequence of the flood that eventually overtops or breaches it. That is the levee effect, and it is why the Dutch Room for the River programme, after the 1993 and 1995 near-disasters, deliberately gave land back to the rivers rather than raising defences further.

Three refinements.

First, storage is not the same as absorption. Very little of a flood soaks away in the hours that matter; saturated ground infiltrates poorly, which is often precisely why the flood happened. The floodplain's service is temporary detention, not disposal.

Second, "once a decade" understates the connection. Small overbank flows happen every year or two and are the events that deposit sediment, replenish soil, recharge shallow aquifers and sustain wetland ecology. A floodplain cut off from its river is a different ecosystem within a few decades, not merely a drier one.

Third, this is genuinely a redundancy argument, and redundancy is only visible when it is missing. A system with spare capacity behaves identically to one without it, right up to the moment the primary path saturates. That is why the floodplain looks like idle land — looking idle is exactly what functioning spare capacity looks like.

• How natural flood management upstream compares with floodplain reconnection downstream.
• Why flood recurrence intervals are widely misread as "once in a hundred years".
• What managed realignment costs, and who bears it, when the land already has houses on it.

Every term the collection defines is gathered in the glossary.