Estimating the future sea level rise contribution of Thwaites glacier, Antarctica, using an uncertainty quantification approach

Thwaites Glacier in West Antarctica may be the single largest contributor to sea level rise in the coming centuries, but existing projections over such timescales are highly uncertain. A number of factors contribute to this uncertainty and robust predictions involve many complex processes through the interaction between ice, ocean and atmosphere. Here, we use the Úa ice-flow model in conjunction with an uncertainty quantification approach to provide uncertainty estimates for the future (100 years’ time scale) mass loss from Thwaites, and the relative contribution of individual model parameters to that uncertainty. In a first step, we simulate Thwaites glacier from 1997 to present day for a wide variety of uncertain model parameters and compare key outputs from each simulation to observations.  Using a Bayesian probability framework we sample the model parameter space, using informed priors, to build up a model emulator, allowing us to provide uncertainty estimates for a range of future emission scenarios. We show how this framework can be used to quantify the relative contribution of each model parameter to the total variance in our estimation of the future mass loss from the area. This, furthermore, allows us to make clear quantitative statements about different sources of uncertainty, for example, those related to external forcing parameterizations (e.g. surface mass balance) as compared to uncertainties in ice-flow parameters (e.g. basal sliding).