Attribution of late summer early autumn Arctic sea ice decline in recent decades

Lejiang Yu,Bo Sun,Timo Vihma,Shiyuan Zhong

doi:10.1038/s41612-020-00157-4

Abstract

The underlying mechanisms for Arctic sea ice decline can be categories as those directly related to changes in atmospheric circulations (often referred to as dynamic mechanisms) and the rest (broadly characterized as thermodynamic processes). An attribution analysis based on the self-organizing maps (SOM) method is performed to determine the relative contributions from these two types of mechanisms to the Arctic sea ice decline in August–October during 1979–2016. The daily atmospheric circulations represented by daily 500-hPa geopotential height anomalies are classified into 12 SOM patterns, which portray the spatial structures of the Arctic Oscillation and Arctic Dipole, and their transitions. Due to the counterbalance between the opposite trends among the circulation patterns, the net effect of circulation changes is small, explaining only 1.6% of the declining trend in the number of August–October sea ice days in the Arctic during 1979–2016. The majority of the trend (95.8%) is accounted for by changes in thermodynamic processes not directly related to changes in circulations, whereas for the remaining trend (2.6%) the contributions of circulation and non-circulation changes cannot be distinguished. The sea ice decline is closely associated with surface air temperature increase, which is related to increasing trends in atmospheric water vapor content, downward longwave radiation, and sea surface temperatures over the open ocean, as well as to decreasing trends in surface albedo. An analogous SOM analysis extending seasonal coverage to spring (April–October) for the same period supports the dominating role of thermodynamic forcing in decadal-scale Arctic sea ice loss.

Highlights

Observations in the Arctic Ocean have shown a rapid decline in sea ice extent especially in recent decades[1]
To understand how much the sea ice trends can be explained by changes in atmospheric circulations, we examine the corresponding large-scale circulation patterns, represented by the 500-hPa geopotential height fields, for the same period
The self-organizing maps (SOM) method is utilized to statistically estimate the contributions to the declining trends in the August–October Arctic sea ice cover over the 1979–2016 period from changes in large-scale atmospheric circulations known as dynamic forcing and in other mechanisms not directly related to circulations that are broadly categorized as thermodynamic forcing

Summary

Introduction

Observations in the Arctic Ocean have shown a rapid decline in sea ice extent especially in recent decades[1]. The predicted rates of sea ice retreat by general circulation models (GCM) under various greenhouse gas emissions scenarios have been, in general, smaller than the observed rates[6], indicating that other mechanisms may have played a role. One such mechanism is changes in aerosol emissions that, through aerosol-radiation feedback, can contribute to Arctic sea ice loss[7,8]. Estimates of the internal variability from GCMs vary significantly, between 20 and 50% (refs. 19–21), and large discrepancies exist between the observed and GCMsimulated sea ice concentrations[22,23,24] due possibly to a combination of factors, including low sea ice sensitivity to greenhouse gas emissions[25], errors, and uncertainties in sea ice and atmospheric models, and inadequate treatment of sea ice–atmosphere interactions[26,27]

Objectives

Methods

Results

Conclusion