A negative (or positive) value of the North Atlantic Oscillation (NAO) index, which measures the variability in sea-level atmospheric pressure, would imply an increase (or decrease) in intense cold air outbreaks and the number of storms in the eastern parts of North America and Northern Europe. The NAO may be influenced by several climate factors. Using a data science approach, here we aim to study the complex dynamics that NAO has with the sea surface temperature (SST) and sea ice extent (SIE), and show that there exists a critical instability (through positive feedback loops) in the complex dynamics of the climate variables of melting Arctic SIE, rising SST, and NAO index. Our statistical machine learning approach shows that the melting SIE and increasing SST significantly affect the NAO, resulting in the changing weather pattern of the North Atlantic region. We also develop a Bayesian Granger-causal dynamic linear model to establish the relationship between the predictor and dependent variable. Our study indicates that there would be a critical instability with more frequent bouts of very cold climate in eastern North America and northern Europe than previously seen, marking a significant climate change.

翻译：北大西洋涛动（NAO）指数通过测量海平面大气压的变率，其负值（或正值）意味着北美东部和北欧地区强冷空气爆发及风暴次数增加（或减少）。NAO可能受多个气候因素影响。本研究采用数据科学方法，旨在探究NAO与海表温度（SST）及海冰范围（SIE）的复杂动态关系，并揭示北极海冰融化、海表温度升高与NAO指数这三个气候变量之间存在通过正反馈循环导致的临界不稳定性。我们的统计机器学习方法表明，融化的海冰范围与升高的海表温度显著影响NAO，进而改变北大西洋地区的天气格局。此外，我们构建了贝叶斯格兰杰因果动态线性模型以确立预测变量与因变量之间的关系。研究指出，未来北大西洋地区将出现比以往更频繁的极寒天气事件，标志着显著的气候变化。