How to interpret expert judgment assessments of 21st century sea-level rise

In a recent paper Bamber and Aspinall (2013) (BA13) investigated the sea-level rise that may result from the Greenland and Antarctic ice sheets during the 21st century. Using data from an expert judgment elicitation, they obtained a final high-end (P95) value of +84 cm integrated sea-level change from the ice sheets for the 2010–2100 period. However, one key message was left largely undiscussed: The experts had strongly diverging opinions about the ice-sheet contributions to sea-level rise. We argue that such (lack of) consensus should form an essential and integral part of the subsequent analysis of the data. By em-
ploying a method that keeps the level of consensus included, and that is also more robust to outliers and less dependent on the choice of the underlying distributions, we obtain on the basis of the same data a considerably lower high-end estimate for the ice-sheet contribution, +53 cm (+38-77 cm interquartile range of “expert consensus”). The method compares favourably with another recent study on expert judgement derived sea-level rise by Horton et al. (2014). Furthermore we show that the BA13 results are sensitive to a number of assumptions, such as the shape and minimum of the underlying distribution that were not part of the expert elicitation itself. Our analysis demonstrates that one should be careful in considering high-end sea-level rise estimates as being well-determined and fixed numbers.