Effects Of External Forcing And Internal Variability On Interdecadal Variability In Arctic Sea Ice

The Arctic is a sensitive region in response to climate change.With the global warming forced by the greenhouse gas emissions from human activities,the Arctic sea ice is undergoing rapid shrinking.The internal variability and external forcing are also superimposed on each other to jointly affect the inter-decadal variability of Arctic sea ice.The climate system models participating in the Coupled Model Comparison Program(CMIP)have significant uncertainties in their simulations and future projections of Arctic sea ice,which could challenge policy makers to develop effective climate response policies.In this study,a combination of observations,CMIP5/6 models,and large ensemble simulations are used to systematically evaluate and study model performance for sea ice simulations,and to quantify the relative contribution of internal variability and external forcing to historical and near-term sea ice change on inter-decadal timescales.The main conclusions are summarized as follows:1.The performance of CMIP5 and CMIP6 models in arctic sea ice simulation is compared and evaluated.It is found that although the underestimation of sea ice declining trend by CMIP5models still exists in CMIP6 models,the sea ice extent trend simulated by CMIP6 ensemble mean is more consistent with the observations.In terms of the internal variability,the bias in CMIP6 is larger than that of CMIP5.2.The relative contributions of internal variability and external forcing on inter-decadal variability of Arctic sea ice are quantified.Results show that changes in Arctic sea ice extent in September are dominated by external forcing over a 40-yr timescale.With the intensification of global warming,the external signal on the sea ice extent change on a 20-yr timescale is emerging.On the timescale of at least 50 years,the influence of external forcing on March sea ice extent could exceed the internal variability.Below the 20-yr timescale,even by the end of the 20^th century,the signal of external forcing had not started to emerge in the March sea ice extent.The internal variability that contribute to sea ice change in different regions of the Arctic in September are different,and there are different forms of internal variability that influence sea ice change on different timescales.Unlike September,the modes of internal variability in March are similar over different timescales.This indicates that the internal variability affecting Arctic sea ice changes at different timescales in March might come from the same origins.3.The combined influence of AO and PDO on the inter-decadal variability of Arctic sea ice is discussed.PDO and AO can influence the anticyclonic circulation in Arctic in summertime,and then affect the melting of sea ice through adiabatic subsidence.Moreover,we find that the influence of the two factors on the decadal variability of Arctic sea ice are not only linear but also nonlinear.Arcitc sea ice responds more strongly to PDO when AO is in its negative phase than AO is in is positive phase.From 2012 to 2020,both PDO and AO begin to shift from negative to positive phase,which might temporarily slow down the trend of Arctic sea ice loss over the next few decades.4.Quantifying the contribution of internal atmospheric variability to September Arctic sea ice projection uncertainty.The internal variability of summer atmospheric circulation contributes 12.3%?16.7%to the near-term September Arctic sea ice extent projection uncertainty.It is found that this internal atmospheric variability can increase or decrease the rate of sea ice loss and affect the probability of an ice-free Arctic over the next 30 years.Further analysis shows that the internal variability of atmospheric circulation responsible for the sea ice variability might be mainly due to local internal process on interannual timescale,but might be related to tropical Pacific signals on inter-decadal timescales.5.The model uncertainty of future Arctic sea ice projection is reduced by using the emergent constraint method.The results show that under the SSP5-8.5 emission scenario,the optimal constraint year for ice-free time is 2042,and its 66%confidence interval(equivalent to the IPCC‘possible'interval)is 2031-2053,which is 10 years earlier than the original CMIP6model and reduced the uncertainty interval by 54%.