How does the atmospheric response to Arctic sea-ice decline compare to the full effect of the Arctic Amplification?
Most modeling studies evaluating linkages between the Arctic weather/climate and midlatitudes have prescribed changes in Arctic sea ice into the 21st century. However, sea-ice loss is only one contributing factor to Arctic amplification (AA), and this warming is mostly confined to the surface. Future projections indicate that warming will extend into the mid-troposphere as well. Thus, experiments that prescribe only changes in sea-ice cover will not capture the total warming of the atmosphere due to Arctic climate change.
We show results from numerous climate model perturbation experiments (coupled and uncoupled) where we assess the role of Arctic sea-ice loss compared to the total effect of AA on the large-scale circulation. In the first set of experiments, we conduct a series of sea-ice loss simulations following the PAMIP protocol using a forcing at 2°C of global warming relative to the pre-industrial era. In our second set of experiments, we impose warming from AA as found in the CESM-Large Ensemble (RCP8.5) near years around 2030, 2060, and 2090.
Despite the differences in model set-up and forcing, we find a strikingly similar pattern of sea level pressure response – increase in sea level pressure over Eurasia and decrease in sea level pressure over the Arctic Ocean and North Pacific. However, the strength of this zonal wavenumber 1-like sea level pressure response is substantially greater when there is more warming in the middle to upper Arctic troposphere. The reinforcement of the Siberian/Ural high is connected to a cooling downstream over eastern Asia, which is due to cold air advection. Thus, without including the total effect of Arctic amplification (compared to only sea-ice loss), studies may not observe the complete role of Arctic-mid-latitude climate teleconnections.
The figure shows the response in Z500 thickness averaged over the polar cap (>65N) vs the response in Siberian High index for the above experiments, AMIP simulations over 1979-2016 comparing the last decade to the first, two reanalysis products and others.
