Is sea-ice-driven Eurasian cooling too weak in models?
Is sea-ice-driven Eurasian cooling too weak in models?
复制标题
DOI：
10.1038/s41558-019-0635-1
复制
发表时间：
2019-11
影响因子：
30.7
通讯作者：
J. Screen;R. Blackport
中科院分区：
文献类型：
--
作者：
J. Screen;R. Blackport
气候变化和物理上是合理的，因为上升的地表热通量异常可能导致或加强WACE模式，往往滞后于巴伦支-卡拉海冰的减少5,11。相反，wace引起的（更普遍的是，环流引起的）向下的热通量异常往往先于巴伦支-喀拉海冰的减少5,11。Blackport等人5提供的证据表明，这种先行一个月或滞后一个月的方法可以有效地区分“冰驱动大气”和“大气驱动冰”的制度。为了验证我们的假设，我们重复了分析，但在WACE模式前一个月计算了巴伦支-卡拉海冰的r2σ2（见方法）。在使用改进方法时，AOGCM和AGCM对海冰驱动的WACE方差给出了几乎相同的估计（图1d），这支持了我们的假设，并表明海洋耦合对海冰损失的WACE响应几乎没有影响。将改进的方法应用于观测和agcm（图1c），可以得出与Mori等人不同的模型性能结论。现在，所有七个agcm的估计值都接近基于观测的估计值。只有两个agcm的集合成员范围没有跨越观测到的估计值，而且只有非常小的边际（可能在观测不确定性范围内，这里没有考虑到）。我们得出的结论是，海冰驱动的WACE方差的模拟和观测估计值彼此一致。这表明AGCM能够真实地模拟WACE对海冰损失的响应，有效地排除了AGCM误差或缺乏海洋耦合是观测结果中欧亚变冷趋势强于AGCM的主要原因。接下来，我们重复分析并计算了巴伦支-卡拉海冰比WACE模式晚一个月的r2σ2。r2σ2
NATURE CLIMATE CHANGE and is physically justified here because the upward surface heat flux anomalies that might cause or reinforce the WACE pattern tend to lag reductions in Barents–Kara sea ice 5, 11. Conversely, WACE-induced (and more generally, circulation-induced) downward heat flux anomalies tend to precede reductions in Barents–Kara sea ice 5, 11. Blackport et al. 5 presented evidence that this one-month lead or one-month lag approach can effectively distinguish between regimes of ‘ice driving atmosphere’and ‘atmosphere driving ice’. To test our hypothesis, we repeated the analysis, but calculated r2σ2 with Barents–Kara sea ice one month ahead of the WACE pattern (see Methods). The AOGCM and AGCM gave almost identical estimates of the WACE variance driven by sea ice when using the refined method (Fig. 1d), lending support to our hypothesis and suggesting that ocean coupling has little effect on the WACE response to sea-ice loss.Applying the refined approach to the observations and AGCMs (Fig. 1c) leads to a different conclusion on model performance to that in Mori et al. 1. Now, the estimates from all seven AGCMs lie close to observation-based estimates. Only two AGCMs have ensemble member ranges that do not span the observed estimate, and only by very small margins (probably within observational uncertainty, which has not been accounted for here). We conclude that the modelled and observed estimates of the WACE variance driven by sea ice are consistent with each other. This suggests that AGCMs are able to realistically simulate the WACE response to sea-ice loss, effectively ruling out either AGCM error or a lack of ocean coupling as the main reason for the stronger Eurasian cooling trends in observations than in AGCMs. We next repeated the analysis and calculated r2σ2 with Barents–Kara sea ice lagging one month behind the WACE pattern. The r2σ2