The Greenland Ice Sheet Large Ensemble (GrISLENS): simulating the future of Greenland under climate variability

The Greenland ice sheet has lost ice at an increasing pace over recent decades, driven by a combination of human-caused climate change and internal variability in the climate system. In projections of future ice sheet evolution, internal variability in climate results in uncertainty that cannot be reduced through model improvements due to the intrinsically chaotic nature of the climate system. This study describes the Greenland Ice Sheet Large Ensemble (GrISLENS), the first large-ensemble study of ice sheet evolution under climate variability, which resolves individual outlet glaciers and climate variability calibrated to observations. GrISLENS combines multiple advanced modeling methods, including a stochastic ice sheet model, a coupled atmosphere–ocean model, dynamical surface mass balance downscaling, and statistical techniques for constraining stochastic parameterizations of climate forcing. We quantify the role of internal climate variability in 185-year projections of the Greenland ice sheet under both a high-emission scenario and pre-2000 climate conditions. We find that spread between ensemble members due to internal climate variability represents a substantial fraction of the mean ice sheet change in the first 20–30 years of simulations, which may be important for coastal planning efforts on decadal timescales. This spread between ensemble members decreases to a small fraction of the total ice sheet change past 2050. At the ice sheet scale, uncertainty in ice loss is dominated by the response to surface mass balance variability, while the response to ocean variability is relatively small, though its influence is more important within individual catchments. The GrISLENS ensemble spread is relatively small compared to that of previous studies estimating uncertainty from climate variability in coarse models, which indicates that resolving small-scale features in climate forcing and ice sheet dynamics substantially affects the quantification of internal variability in ice sheet mass change. On longer timescales, human emissions of greenhouse gases and structural and parametric uncertainties in climate and ice sheet models are larger contributors to projection uncertainties. Through our analysis, we identify the need for more robust initialization methods and extension of these large-ensemble methods to the Antarctic ice sheet.