September Arctic sea ice minimum prediction – a skillful new statistical approach

摘要

Sea ice in both polar regions is an important indicator of theexpression of global climate change and its polar amplification.Consequently, broad interest exists on sea ice coverage, variability andlong-term change. However, its predictability is complex and it dependsstrongly on different atmospheric and oceanic parameters. In order toprovide insights into the potential development of a monthly/seasonal signalof sea ice evolution, we applied a robust statistical model based ondifferent oceanic and atmospheric parameters to calculate an estimate of theSeptember sea ice extent (SSIE) on a monthly timescale. Although previousstatistical attempts of monthly/seasonal SSIE forecasts show a relativelyreduced skill, when the trend is removed, we show here that the Septembersea ice extent has a high predictive skill, up to 4 months ahead, based onprevious months' oceanic and atmospheric conditions. Our statistical modelskillfully captures the interannual variability of the SSIE and couldprovide a valuable tool for identifying relevant regions and oceanic andatmospheric parameters that are important for the sea ice development in theArctic and for detecting sensitive/critical regions in global coupledclimate models with a focus on sea ice formation.