We urgently need to know how far and how fast the sea will rise, but the latest attempt to put figures on it is dangerously misleading

IMAGINE your job is to protect London from surging seas. In one way it is easy: unlike most coastal cities, London has a formidable flood defence system in the form of the Thames Barrier, capable of protecting it from all but the highest storm surges.

But as the seas rise, the risk of the barrier being breached will increase steadily. With a 1-metre rise in local sea level, London will get flooded every 10 years. So when do you start building new flood defences, and how high do you make them?

The stakes are enormous. Building new defences will cost tens of billions and involve decades of planning and controversy before construction even begins. Get it wrong, and storm surges could kill thousands and displace millions. So all around the world, planners are clamouring to know how fast the seas will rise as the planet warms.

Until recently, scientists could not give them any reliable numbers. There were no computer models capable of simulating the melting of the world’s ice sheets and glaciers.

The 2007 report of the Intergovernmental Panel on Climate Change (IPCC) handled this uncertainty really badly. It acknowledged that we don’t know how fast all the ice will melt, but then gave some numbers anyway – between 18 and 59 centimetres of sea level rise by 2100 – based on highly dubious assumptions such as glaciers continuing to flow at the same rate and the Antarctic ice sheet growing larger. The numbers also assumed a maximum warming of 5.4 °C, even though the report’s highest projection was 6.4 °C. Unsurprisingly, many people wrongly took 59 cm of sea level rise to be the worst case.

Now we have some more numbers. A European-funded project called ice2sea has developed computer models of glaciers and ice sheets. Earlier this month it announced that melting ice would contribute between 4 and 37 cm to global sea level by 2100. Adding this to the other causes of sea level rise – the main one being the expansion of the oceans as they warm – gives figures of between 16 and 69 cm by 2100.

Some media reports focused on the fact that this is less than some other recent estimates of at least a metre. “Seas will rise no more than 69 centimetres by 2100,” proclaimed this magazine.

Others focused on the fact that even this relatively small rise could have devastating consequences. “Floods could overwhelm Thames Barrier by end of century,” declared The Guardian in London.

How much trust can we put in these numbers, though? The whole point of the ice2sea programme was to “reduce the uncertainty”, but its numbers come with some rather large caveats.

For starters, the modellers didn’t have the computing power to look at a range of scenarios for how much carbon dioxide we will pump into the atmosphere. Instead, they looked at just one – a “mid-range” scenario predicted by the 2007 report to lead to warming of around 3 °C.

Yet actual emissions today are much closer to the worst-case scenario, which some recent studies predict could lead to warming of 6 °C or more. And far from falling, annual global emissions are rising ever faster. With hundreds more coal-fired power stations being built and new sources of fossil fuels like tar sands being exploited, there is good reason to think emissions will continue to soar for many decades to come.

What’s more, to account for the fact that warming will not be uniform across the globe, the modellers had to produce regional projections of warming, snowfall and so on to feed into the ice models. But regional projections are highly unreliable, with different models often producing wildly varying results. The prime example is the Arctic, where the sea ice is disappearing much faster than anyone expected.

To understand why regional climate predictions are so much less reliable than global ones, think of the heat entering the atmosphere and oceans as water pouring into a bath. Predicting the average level of the bath is much easier that predicting the height of the waves sloshing around.

So the climate information being fed into these latest ice models could be way off the mark.

And even if it isn’t, how do we know the models are right? Well, say the researchers, they can reproduce some of the observed responses to the actual 0.5 °C warming of the past few decades, such as the retreat of glaciers. But that doesn’t prove they can predict the response to future warming of 3 or 6 °C. There are similar issues with global climate models.

This kind of research is vital. But when such a limited study is presented as the “best estimate” available, the danger is that it will be misinterpreted in the same way as the 2007 IPCC report. Its numbers do not encompass the worst-case scenario – far from it. They don’t even represent the most likely scenario. The narrow range implies a degree of certainty that simply doesn’t exist. Nobody should be basing life-and-death decisions such as how to protect Londoners on these numbers.