Abstract/Summary

The Northern Hemisphere stratospheric polar vortex (SPV) plays a key role for mid-latitude weather and climate. However, in what way the SPV will respond to global warming is not clear, with climate models disagreeing on the sign and magnitude of projected SPV strength change. Here we address the potential role of Barents and Kara (BK) sea ice loss in this. We provide evidence for a non-linear response of the SPV to global mean temperature change, coincident with the time the BK Seas become ice-free. Using a causal network approach, we demonstrate that climate models show some partial support for the previously proposed link between low BK sea ice in autumn and a weakened winter SPV, but that this effect is plausibly very small relative to internal variability. Yet, given the expected dramatic decrease of sea ice in the future, even a small causal effect can explain all of the projected ensemble-mean SPV weakening, approximately one-half of the ensemble spread at the middle of the 21st century, and one-third of the spread at the end of the century. Finally, we note that most models have unrealistic amounts of BK sea ice, meaning that their SPV response to ice loss is unrealistic. Bias-adjusting for this effect leads to pronounced differences in SPV response of individual models at both ends of the spectrum, but has no strong consequences for the overall ensemble mean and spread. Overall, our results indicate the importance of exploring all plausible implications of a changing Arctic for regional climate risk assessments.