Earth system model sea-ice loss experiments are wrong. Are they useful?

Our poor understanding of how the Arctic’s atmosphere, sea ice, and ocean are coupled limits what we can say about Arctic change from greenhouse warming, and what Arctic change means for global weather and climate. Earth system models that simulate Arctic and global change, while complicated and imperfect, are useful to understand drivers of Arctic change and its global influence. In the virtual world of models, you can remove Arctic sea ice and analyze its local and remote response, without greenhouse warming. Or, you can keep sea ice unchanged and investigate a virtual world of greenhouse warming without sea ice loss. But this virtual exploration can fool us: recent work by Mark England and colleagues has shown that this kind of sea ice removal, when carried out in the setting of coupled ocean-atmosphere models, artificially amplifies Arctic warming, with global implications. The basic problem is that these simulations use Arctic sea ice loss as a stand-in for Arctic warming, but targeted ice loss does not account well for the effect on the Arctic of greenhouse warming.

 

We confirm the England et al. result but argue that sea ice loss experiments can nevertheless provide physically reasonable results, if they are linearly combined with greenhouse warming experiments using scaling suggested by simple energy balance models. This post-processing step, along with refined methods for inducing sea ice loss, allows us to gain value from sea ice loss experiments and avoid some of the difficulties arising from interpreting these experiments at face value.