Reconciled model projections for the Antarctic contribution to sea level rise

Recent numerical model projections of the Antarctic contribution to global mean sea level rise by 2100 have differed by an order of magnitude: from around 10 cm (Ritz et al, 2015, Nature) to over one metre (DeConto and Pollard, 2016, Nature). The latter would double total sea level rise projections from the last IPCC report, with substantial consequences for adaptation. DeConto and Pollard's projections incorporate a new hypothesis of rapid sea level rise by widespread ‘ice cliff instability’, in which tall ice cliffs created by the loss of ice shelves along the coast due to warming are physically unstable and collapse into the ocean. Here we quantify uncertainties for their projections under cliff instability using statistical modelling of the simulator (‘emulation’) to estimate a full probability distribution, and show the probability of substantial sea level rise by 2100 is lower than previously interpreted. Projections are fully reconciled with previous probabilistic numerical model projections when the new cliff instability mechanism is switched off, establishing a new multi-model consensus of 30-40 cm for the upper end (95% probability) of sea level rise by 2100 due to changes in glacier flow alone. Projections with and without cliff instability diverge at the 95% probability level from around 2035, indicating that widespread cliff instability would begin to increase sea level rise projections from around two decades’ time.