Investigating Uncertainty in the Mid-latitude Response to Sea-Ice Loss with Idealised General Circulation Model Experiments

The Arctic is undergoing rapid climate change, manifested by substantial sea-ice loss and Arctic amplification (AA) of global warming. In turn, sea-ice loss (and associated AA) can drive changes in mid-latitude weather and climate, for example through the effect of a reduced equator-to-pole temperature gradient causing a weakening of the mid-latitude westerlies. This and other mid-latitude responses to Arctic climate change have been extensively investigated using climate model simulations in which sea-ice loss (or a local Arctic heating) is prescribed instead of increasing greenhouse gas concentrations. However, there is uncertainty in the magnitude of the 'true' climate response to Arctic sea-ice loss. This due to uncertainty regarding the methodology used to induce sea-ice loss, as well as inter-model spread in the strength of atmospheric eddy feedbacks, which can amplify the circulation response to sea-ice loss and are often too weak.

In this work, we investigate sources of uncertainty in the mid-latitude response to sea-ice loss using an idealised general circulation model with thermodynamic sea-ice. Simulations where sea-ice loss is imposed using a range of methods, and with the sea-ice module 'switched off', are compared against a control simulation with ice. This process is repeated for a range of control climatologies with different eddy feedback strengths. For each experiment, we quantify the magnitude of AA, the mid-latitude jet strength, location, and waviness, and the persistence of surface weather anomalies. By comparing our experiments, we demonstrate the sensitivity of each diagnostic to eddy feedback strength and the method used to impose sea-ice loss. These results are placed in context through discussion with existing work on this topic.