We present a new parameterisation for the Greenland ice sheet (GIS) that relates ice sheet surface mass balance (SMB: surface accumulation minus ablation) to changes in surface elevation (Edwards et al., 2014a). The motivation is to dynamically adjust SMB in ice sheet models run “offline” from a climate model, in order to account for known feedbacks between ice sheet elevation and SMB. Based on results from the MAR regional climate model, we estimate four geographically distinct “SMB lapse rates”, gradients that relate SMB and elevation changes over different regions of the GIS. We assess uncertainties within a Bayesian framework, estimating probability distributions for each gradient, from which we present best estimates and confidence intervals. Below the equilibrium line, our gradient estimates are mostly positive, because SMB usually increases with elevation. Above the equilibrium line, gradients are much because SMB can either increase or decrease in response to increased ice sheet elevation.
In Edwards et al. (2014b), we apply this statistically founded approach in order to make probabilistic assessments for the effect of the SMB - elevation feedback on future sea level projections from the GIS. We apply the parameterisation to projections of future climate change using five next-generation ice sheet models, which are forced offline by HadCM3 and ECHAM5 under the A1B emissions scenario. At 2100 (2200), the additional sea level contribution due to the feedback is 4.3% (9.6%). In all results the elevation feedback is significantly positive, amplifying the sea level contribution from the GIR relative to projections in which the ice sheet topography is assumed fixed.
Our methods are novel in sea level projections because we propagate three types of modeling uncertainty - climate and ice sheet model structural uncertainties, and elevation feedback parameterisation uncertainty - along the causal chain, from emissions scenario to sea level, within a coherent experimental design and statistical framework. The relative contributions to uncertainty depend on the timescale of interest. At 2100, the climate model uncertainty is largest, but by 2200 both the ice sheet model and parameterisation uncertainties are larger.