Assessing the possibility of a runaway ice-albedo feedback

If the ice-albedo feedback becomes dominant in high latitudes at some point as the climate warms, a runaway feedback will ensue in which the climate transitions to a new state. Such a transition occurs at a bifurcation point and is characterized by a discontinuity in the stable states available to the climate system. The possibility of such a runaway feedback in the Arctic has been investigated previously in a number of studies using both single-column models and comprehensive global models, and they have produced widely differing results. Here, I will use a toy model to evaluate the varied previously published results. The toy model represents the physics of sea ice growth and melt below a single column of the atmosphere, and the strength of interacting climate feedbacks such as albedo, clouds, and water vapor are controlled by scalable parameters. The toy model results suggest a novel measure, which can be assessed based on the current climate state, of how prone to instability the sea ice cover will be under global warming. I will demonstrate this measure in simulations from two state-of-the-art climate models, one of which has been shown to simulate a runaway ice-albedo feedback in the Arctic climate response to substantial greenhouse forcing and the other of which has been shown not to. This measure, which is based only on the simulated 20th century climate in each model, correctly identifies the model more prone to Arctic sea ice instability under substantial global warming.