Prediction Of Arctic Air Temperature And Sea Ice Based On The High-resolution Prediction System DePreSys3

Under global warming,drastic climate changes of surface air temperature rising rapidly and sea ice decreasing markedly in the Arctic attracts more and more attention than before.Thus,accurate prediction of Arctic climate and deep understanding of its interannual climate variabilities become the focus in climate science.Using hindcasts from various prediction models and multiple observations and reanalysis datasets,the present work mainly examines the seasonal to interannual predictive skills of 2-m air temperature(T2m)and sea ice concentration(SIC)over the Arctic region with a high-resolution model called the Met Office Decadal Prediction System version 3(DePreSys3),in the views of deterministic and probabilistic forecasts.Then the possible physical mechanisms responsible for internal variations of T2 m and SIC anomalies over the Barents-Kara Seas(BKS)during boreal autumn have been investigated.The primary drivers of extreme climate events taking place in the BKS region are also probed into.Here are the main conclusions:(1)The seasonal to interannual deterministic prediction skills of Arctic T2 m and SIC anomalies by DePreSys3 hindcasts are of relatively good performance.The model shows statistically significant skills at lead times up to 16 months,which is mainly due to the contribution of strong temperature rising and sea ice decreasing decadal trends.While the predictive skill appears declining for the detrended variations,the performance of DePreSys3 hindcasts is still better than other low-resolution models'.Moreover,the study found a close relationship between the tropical Pacific El Ni?o–Southern Oscillation(ENSO)and the Arctic detrended T2 m anomalies in boreal winter.This indicates ENSO is an important potential source of predictability.The physical relationship can also be confirmed in observations.(2)The DePreSys3 shows that the probabilistic T2 m and SIC forecasts after decadal trends removing over the Arctic are predictive to a certain extent,although the quality of skills strongly depends on the selection of target seasons and Arctic subregions.In general,this high-resolution model exhibits more stable probability skills for positive and negative categories of the Arctic T2 m and SIC during freezing period,when comparing with forecasts assessment during melt period.Taking advantage of DePreSys3 with relatively high resolution,it is shown that the probabilistic forecasts for the internal climate variations of the BKS are better than that of other Arctic subregions.However,the skills for regional T2m and SIC anomalies in boreal autumn need to be further improved.(3)The Arctic Dipole(AD)and the North Atlantic Tripole(NAT)in the preceding month are both the primary drivers responsible for internal variations of climate anomalies over the BKS during boreal autumn.During a negative AD phase,the positive surface temperature advection anomalies associated with the atmospheric circulations warm up the BKS region and promote sea ice melting,which in turn maintains the below-normal sea ice and warmer-thannormal conditions via local positive feedbacks.During a negative NAT phase,a Rossby wavetype atmospheric response to warm SST leads the anomalous temperature advections from the surface to the upper troposphere located over the Kara-Sea region,which intensifies local SST anomalies and further contribute to sea ice melting.Hence,focusing on the forecasts of AD and NAT modes at several months lead is helpful to improving the BKS regional climate predictions.(4)The discrepancies of spatial patterns of SIC anomalies are observed among all extreme sea ice area(SIA)events in autumn over the BKS region,and the physical mechanisms are also quite different from each other.The results suggest that anomalous condition of sea ice area in the preceding year,changes of wind-driven sea ice fluxes and the thermal forcings of ocean and atmosphere are all the primary factors.The local positive feedbacks also play a vital role via intensifying the anomalous climate signals.Only a third of extreme SIA events over the BKS could be predicted at 9-11 lead months by the DePreSys3 ensemble hindcasts.The difficult predictions for physical processes during the melting seasons are the main cause of missing forecasts.