If all the ice in Greenland and Antarctica melted, it would cause a rise in sea level of around 70 metres. Such a nightmare scenario is extremely unlikely under any circumstances. However, even the melting of a small proportion of these ice caps could have devastating effects, flooding low-lying coastal regions and reducing the safety margin of coastal defences. Now, calculations by Philippe Huybrechts of Vrije Universiteit Brussel, Brussels, Belgium, and Jan De Wolde of Universiteit Utrecht, Utrecht, the Netherlands (a country with a vested interest in sea-level rise), put limits on how much sea level could rise due to melting ice over the next millennium.

Ice caps respond to climate warming in various ways. If ice becomes warm enough, then it melts. However, if the ice remains below freezing point, then increased snowfall may cause the ice cap to grow. Also, warmer ice is softer than cold, and so flows faster. Previous calculations indicate that over the next hundred years sea level will rise by about 10 cm, with melting in Greenland roughly balanced by increased snowfall in Antarctica. However, these calculations assume that the ice caps are in equilibrium with the present climate or that they react like blocks of ice.

Neither assumption is realistic, so these predictions are only valid for about a hundred years. Writing in the August issue of the Journal of Climate, Huybrechts and De Wolde describe a computer model of the Greenland and Antarctic ice caps. The model takes into account snowfall, temperature, the flow of the ice and the response of the rock beneath to the weight of the ice, and includes a model of global climate.

Their computer program started with a representation of the ice cap as it was two glacial cycles ago, and ran until it arrived at the ice cap as it is today, and then carried on for the next 1,000 years. This long 'running-in' period allowed the model to 'forget' its starting conditions, ensuring that it correctly represented the present-day size and shape of the ice caps, with all their imbalances. It also allowed the model to be checked against present conditions, which it reproduced with an accuracy of a few per cent.

The first prediction from the model is that without any warming, imbalances in the present ice caps are causing sea level to rise by about 4 cm per century. Huybrechts and De Wolde observe that this background rise accounts for a significant part of a previously unexplained increase in sea level observed during the past century.

Huybrechts and De Wolde also calculate the effect of various levels of global warming. They assumed that CO₂ levels would stabilise by 2130 AD, and used maximum CO₂ concentrations of two, four and eight times the present value, corresponding to low, middle and high forecasts by the Intergovernmental Panel on Climate Change, an international advisory body. They calculate a range of values for sea level rise by the year 2130 AD between 6 and 30 cm, with a value corresponding to the middle concentration of CO₂ of 14 cm. The range for the year 3000 AD is from 55 to 650 cm, with a middle value of 250 cm.

The expected rise in 2130 AD would reduce the safety margin of existing coastal defences so that they would be more frequently overwhelmed by storm surges which global warming would make more likely. Even the mid-range value predicts considerable coastal flooding by the year 3000 AD. All of which suggests that many coastal cities that have existed for millennia will need radical and expensive defences if they are to survive the next 1,000 years.