Jonathan Bamber is Professor of Physical Geography at University of Bristol (U.K.); Michael Oppenheimer is Professor of Geosciences and International Affairs at Princeton University.

Antarctica is further from civilization than any other place on Earth. The Greenland ice sheet is closer to home but around one tenth the size of its southern sibling. Together, these two ice masses hold enough frozen water to raise global mean sea level by 65 meters if they were to suddenly melt. But how likely is this to happen?The Antarctic ice sheet is around one and half times larger than Australia. What’s happening in one part of Antarctica may not be the same as what’s happening in another — just like the east and west coasts of the US can experience very different responses to, for example, a change in the El Niño weather pattern. These are periodic climate events that result in wetter conditions across the southern US, warmer conditions in the north and drier weather on the north-eastern seaboard.
The ice in Antarctica is nearly 5 kilometers thick in places and we have very little idea what the conditions are like at the base, even though those conditions play a key role in determining the speed with which the ice can respond to climate change, including how fast it can flow toward and into the ocean. A warm, wet base lubricates the bedrock of land beneath the ice and allows it to slide over it.These issues have made it particularly difficult to produce model simulations of how ice sheets will respond to climate change in future. Models have to capture all the processes and uncertainties that we know about and those that we don’t — the “known unknowns” and the “unknown unknowns” as Donald Rumsfeld once put it. As a result, several recent studies suggest that previous Intergovernmental Panel on Climate Change reports may have underestimated how much melting ice sheets will contribute to sea level in future.What the experts sayFortunately, models are not the only tools for predicting the future. Structured Expert Judgement is a method from a study one of us published in 2013. Experts give their judgement on a hard-to-model problem and their judgements are combined in a way that takes into account how good they are at assessing their own uncertainty. This provides a rational consensus.The approach has been used when the consequences of an event are potentially catastrophic, but our ability to model the system is poor. These include volcanic eruptions, earthquakes, the spread of vector-borne diseases such as malaria and even airplane crashes. Since the study in 2013, scientists modeling ice sheets have improved their models by trying to incorporate processes that cause positive and negative feedback. Impurities on the surface of the Greenland ice sheet cause positive feedback as they enhance melting by absorbing more of the sun’s heat. The stabilizing effect of bedrock rising as the overlying ice thins, lessening the weight on the bed, is an example of negative feedback, as it slows the rate that the ice melts. The record of observations of ice sheet change, primarily from satellite data, has also grown in length and quality, helping to improve knowledge of the recent behavior of the ice sheets.With colleagues from the U.K. and U.S., we undertook a new Structured Expert Judgement exercise. With all the new research, data and knowledge, you might expect the uncertainties around how much ice sheet melting will contribute to sea level rise to have got smaller. Unfortunately, that’s not what we found. What we did find was a range of future outcomes that go from bad to worse.Rising uncertaintyWe gathered together 22 experts in the U.S. and U.K. in 2018 and combined their judgements. The results are sobering. Rather than a shrinking in the uncertainty of future ice sheet behavior over the last six years, it has grown.