Atmospheric impacts of Arctic sea-ice loss, 1979–2009: separating forced change from atmospheric internal variability

The ongoing loss of Arctic sea-ice cover has implications for the wider climate system. The detection and importance of the atmospheric impacts of sea-ice loss depends, in part, on the relative magnitudes of the sea-ice forced change compared to natural atmospheric internal variability (AIV). This study analyses large ensembles of two independent atmospheric general circulation models in order to separate the forced response to historical Arctic sea-ice loss (1979–2009) from AIV, and to quantify signalto-noise ratios. We also present results from a simulation with the sea-ice forcing roughly doubled in magnitude. In proximity to regions of sea-ice loss, we identify statistically significant near-surface atmospheric warming and precipitation increases, in autumn and winter in both models. In winter, both models exhibit a significant lowering of sea level pressure and geopotential height over the Arctic. All of these responses are broadly similar, but strengthened and/or more geographically extensive, when the sea-ice forcing is doubled in magnitude. Signal-to-noise ratios differ considerably between variables and locations. The temperature and precipitation responses are significantly easier to detect (higher signal-to-noise ratio) than the sea level pressure or geopotential height responses. Equally, the local response (i.e., in the vicinity of sea-ice loss) is easier to detect than the mid-latitude or upper-level responses. Based on our estimates of signal-to-noise, we conjecture that the local near-surface temperature and precipitation responses to past Arctic sea-ice loss exceed AIV and are detectable in observed records, but that the potential atmospheric circulation, upper-level and remote responses may be partially or wholly masked by AIV.