How Do We Know Where the Sea Used to Be? Marks Left Above and Below

Several kinds of evidence form at or very near the surface of the sea and therefore mark where it stood. Corals of certain species grow only within a few metres of the surface, so a fossil reef gives both a level and, through radiometric dating, an age. Salt marsh sediments accumulate within a narrow band related to the tide, and the microscopic organisms preserved in them are strongly zoned by how often they are submerged, which allows the level to be reconstructed to within a few centimetres. Beach deposits, notches cut into cliffs and marine terraces record former shorelines directly. Drowned forests, peat layers and archaeological structures built in relation to the water supply further evidence.

A mark showing where the sea was does not by itself show how much the sea has changed:

• •Land rises and falls, so a raised beach may mean the land went up rather than the sea down
• •Ice sheets depress the crust beneath them and the surrounding region bulges upwards
• •When the ice melts, that adjustment reverses over thousands of years and is still running
• •Northern Britain and Scandinavia are still rising while southern Britain sinks
• •Tectonic uplift and subsidence add further local movement
• •Sediment compaction lowers a site without any change in the sea

Distinguishing global change in ocean volume from local movement of the land is the central technical problem, and the approach is to model the land movement physically. The adjustment of the crust to vanished ice sheets can be calculated from the known ice history and the properties of the earth's interior, and the resulting predictions are tested against records from many sites, since a site far from any former ice sheet behaves differently from one at its centre. Sites at intermediate distances are particularly informative because the predicted patterns differ most there. Combining many records with such models produces a global curve, and the modelling itself is refined by the fit, which makes the whole exercise iterative rather than a simple measurement.

Modern measurement uses two instruments with different strengths and a persistent problem reconciling them. Tide gauges are fixed to land and record the level relative to that land, which means they measure the combination of ocean change and local movement, and the longest series run back to the eighteenth century at a few European ports. Satellite altimeters bounce radar off the sea surface and measure height relative to the earth's centre, which removes the land movement problem entirely and covers nearly the whole ocean rather than the coasts where gauges sit. Precise positioning of the gauges themselves, using satellite navigation, now separates their two components. Combining all three gives both a global figure and a map showing that the rise is far from uniform, with some regions rising several times the average.

The resulting picture is reasonably clear over the past several hundred thousand years. Sea level fell by roughly one hundred and twenty metres during glacial maxima, with that water locked into ice sheets, and rose rapidly during deglaciations, at rates reaching more than a metre per century for sustained periods and considerably faster during brief pulses. The last such rise ended around six thousand years ago, after which level was relatively stable until the late nineteenth century. Instrumental records from tide gauges and, since 1993, from satellite altimetry show a rise that has accelerated, and reconciling those instruments with the geological record is what allows the recent change to be placed in context. Past interglacial periods with temperatures modestly above the present had sea levels several metres higher.

Corals, salt marsh sediments and cliff notches all form within a narrow band of the surface and record where it stood. The difficulty is that land rises and falls too, including the continuing adjustment to vanished ice sheets, which is still lifting northern Britain and lowering the south. Separating the two requires modelling that adjustment physically and testing it against many sites.