Atmospheric River Response to Arctic Sea Ice Loss in the Polar Amplification Model Intercomparison Project

The atmospheric river (AR) response to Arctic sea ice loss in the Northern hemisphere winter is investigated using simulations from the Polar Amplification Model Intercomparison Project. Results have shown that the midlatitude responses are dominated by dynamic effects. Poleward of around 60°N, the dynamic and thermodynamic effects cancel each other, resulting in relatively small responses. The response uncertainty can be characterized by leading uncertainty modes, with the responses over the Pacific and Atlantic projecting onto the northeastward extension and equatorward shift mode, respectively. In addition, the responses seem to be mean state‐dependent: under the same forcing, models with more poleward‐located climatological ARs tend to show stronger equatorward shifts over the Atlantic; over the Pacific, models with more westward‐located climatological AR core tend to show stronger northeastward extensions. These relationships highlight the importance of improving the AR climatology representation on reducing the response uncertainty to Arctic sea ice loss. Plain Language Summary It has been hypothesized that Arctic sea ice loss can influence midlatitude weather extremes. Atmospheric rivers (ARs), a type of extreme weather, are characterized by intense water vapor transport in the atmosphere. When these “rivers” in the sky make landfall and encounter mountain barriers, they can produce torrential rainfall and even cause floods. However, how Arctic sea ice loss may influence ARs remains largely unknown. Using data from climate model simulations driven by Arctic sea ice loss, we found that sea ice loss can cause ARs to occur at lower latitude regions over the Atlantic. Over the Pacific, ARs will occur more often at regions closer to the North American west coast. Although the above‐mentioned responses are quite consistent across models, model‐to‐model differences in the responses exist. We further found that such inter‐model differences in the responses are related to how well models simulate the AR climatology. Therefore, improving the model's ability to capture the AR characteristics in the present climate can potentially help us make better predictions on how ARs will change under future Arctic sea ice loss. Key Points Midlatitude AR response to Arctic sea ice loss dominated by the effects driven by the changes in the westerly jet AR response to sea ice loss characterized by northeastward extension over the Pacific and equatorward shift over the Atlan