Quantifying uncertainties in the land ice contribution to sea level from ISMIP6 and GlacierMIP

The land ice contribution to global mean sea level has not yet been predicted for the latest generation of socio-economic scenarios, nor with coordinated assessment of uncertainties from the various computer models involved (climate, Greenland and Antarctic ice sheets, and global glaciers). Two recent projects generated a large suite of projections but used previous generation scenarios and climate models and could not fully explore uncertainties. Here we estimate probability distributions for their projections, using statistical emulation, and find uncertainty does not diminish if greenhouse gas concentrations are reduced: the sea level contribution of land ice is 28 [5, 57] cm from 2015 to 2100 under no mitigation (median and 90% range), and 16 [-5, 46] cm under very stringent mitigation. Greenland is projected to contribute around 2.5 cm/ºC of global warming, and Alaskan and Arctic glaciers a total of around 2 cm/ºC, but Antarctic uncertainties are too large to determine temperature-dependence. Knowing future global mean temperature exactly for a given socio-economic scenario would reduce the uncertainty for glaciers by up to two thirds (6 cm) but have little effect for ice sheets. Quantifying how ice sheet margins respond to ocean warming would reduce uncertainty by up to one third (Antarctica 15 cm; Greenland 7 cm). The remaining uncertainty for a given scenario is dominated by the climate and glaciological models themselves. Improved modelling and observations of polar regions, rather than global warming and glaciers, would therefore have the greatest effect in reducing uncertainty in future sea level rise.