Processed model data used for plots in Bondzio et al., GRL 2018

Large uncertainties in model parameterizations and input datasets make projections of future sea level rise contributions of outlet glaciers challenging. Here, we introduce a novel technique for weighing large ensemble model simulations that uses information of key observables. The approach is robust to input errors and yields calibrated means and error estimates of a glacier's mass balance. We apply the technique to Jakobshavn Isbr\ae{}, using a model that includes a dynamic calving law, and closely reproduce the observed behavior from 1985 until 2018 by forcing the model with ocean temperatures only. Our calibrated projection suggests that the glacier will continue to retreat and contribute about 5.1 mm to eustatic sea level rise by 2100 under present-day climatic forcing. Our analysis shows that the glacier's future evolution will strongly depend on the ambient oceanic setting.

模型参数化方案与输入数据集存在的巨大不确定性，使得流出冰川（outlet glacier）未来海平面上升贡献的预估工作极具挑战性。为此，本文提出一种全新的权重赋值方法，可基于关键观测信息对大集合模式模拟结果进行加权处理。该方法对输入误差具有良好的鲁棒性，可输出冰川物质平衡的校准均值与误差估计值。我们将该方法应用于雅各布港冰川（Jakobshavn Isbræ），采用包含动态崩解法则的模式，仅通过海洋温度数据驱动模型，便精准复刻了1985年至2018年间的观测演化特征。经校准后的预估结果显示，在当前气候强迫条件下，该冰川将持续退缩，至2100年时将贡献约5.1毫米的全球海平面上升量。本研究分析表明，该冰川的未来演化将极大程度取决于周边海洋环境条件。