Bayesian calibration of an ice sheet model for the Amundsen Sea Embayment region.

Abstract

Mass loss from the Amundsen Sea Embayment of the West Antarctic Ice Sheet has been increasing over recent decades and is a major contributor to global sea level rise. Predictions of future sea level rise are increasingly modelled using ensembles of simulations within which model parameters and external forcings are varied widely then scored according to observations. Accurately reporting the uncertainty associated with these predictions is vital to enable effective planning for, and maybe construction of defences against, rising sea levels. Here we constrain, or calibrate, an ensemble of simulations of ice loss from the Amundsen Sea Embayment using the BISICLES ice sheet model with remotely sensed observations of surface elevation change and ice speed. The calibrations make it possible to reduce the 90% credibility bounds of predicted contributions to sea-level rise by 40%.