Rarely do major geological events happen in a 'flash'. But the bulk of a huge flood that provided enough water to fill the Mediterranean Sea more than 5 million years ago may have occurred within less than two years. “That makes it an instantaneous event on geological timescales,” says Daniel Garcia-Castellanos, a geophysicist at the Institute for Earth Science Jaume Almera in Barcelona, Spain.

About 5.6 million years ago, the Mediterranean Sea became isolated from the Atlantic Ocean and almost dried up in an event known as the Messinian salinity crisis. Some 270,000 years later, another change — a shift in tectonic plates, alterations in global sea levels or erosion of land (or possibly all three) — reopened the connection between the Mediterranean and the Atlantic at the Gibraltar Strait. The resulting flood, named after its geological timescale the Zanclean, was the largest in Earth's history and has long fascinated geologists. But because little is known about the flood's dynamics, estimates of its duration have been highly variable.

Garcia-Castellanos and his colleagues have successfully modelled the dynamics of the flood using new formulations based on how rivers change landscapes through incision — the water-driven erosion that cuts a path through rock (see page 778). This idea to incorporate incision-based processes stems back to research Garcia-Castellanos did several years ago as a postdoc at the Free University in Amsterdam, where he studied tectonic lakes.

Unless an underlying tectonic process keeps them in place, Garcia-Castellanos explains, most lakes 'quickly' disappear — they fill up with sediment, overspill their banks and the water finds a way out, incising along an outlet. “In the models we were running in Amsterdam, this transition was very fast,” he says. At around the same time that he was doing his postdoctoral work he started to learn about the Messinian salinity crisis. It struck him that the feedback between water flow and incision should be similar in the Zanclean flood.

At peak discharge water poured in at a rate of 100 million cubic metres per second.

By combining a variation of the river-incision model with the proper hydrodynamic equations for the Zanclean flood, Garcia-Castellanos and colleagues predicted that although the initial stages of the flood may have taken thousands of years — with water trickling down a ramp of rock at the strait — 90% of the flood water was transferred to the Mediterranean basin in a period lasting between only a few months and two years. According to the authors' model, at peak discharge water poured in at a rate of 100 million cubic metres per second, cutting down into the bedrock by almost half a metre per day and possibly raising the level of the Mediterranean Sea by more than 10 metres per day.

Garcia-Castellanos was shocked by the numbers that the model produced. “When we got the first predictions, I was very surprised and thought maybe something was wrong with the formulations,” he says. To test the model's accuracy, Garcia-Castellanos looked for physical evidence. “If the model was correct, we would expect to find traces of the flood erosion preserved under the sedimentary layers in the strait,” he says. In other words, such a tremendous rush of water would have carved a gorge, and its remnants should still be present under the sea floor.

To look for such traces, the team turned to two available data sets gathered by other groups. The first was seismic line data, which can be used to generate maps of a cross-section of the layers of rock under the sea floor. The second was rock cores drilled from the strait area during exploratory work for the Africa–Europe tunnel project, which aims to build a train tunnel connecting Spain and Morocco. Both data sets showed a clear channel running from the Eastern Atlantic to the Western Mediterranean through the strait. Researchers who first detected the channel thought it was the result of erosion caused by a river running through the dried up strait, but there is no evidence for such a river ever existing there. The conclusion that flood waters had shaped the channel makes more sense and is consistent with the team's model.

With the flood's intensity resolved, there is still much to learn about the ramifications of such a drastic event. The flood makes for a “natural laboratory”, Garcia-Castellanos says, for discovering how it might have changed oceanic or atmospheric circulations, and so climate, at the time. He also conjectures that the quick disappearance of a land bridge between Europe and Africa affected the migration of mammals, noting that had the connection not disappeared, “hominins might have arrived in Europe much earlier than 1.2 million or so years ago.”