Arctic sea ice melting over the past few decades has been one of the most striking features in changing climate system. Previous studies have shown that Arctic sea ice melting has altered the meridional temperature gradient and the mid-latitude jet streams. This alteration has resulted in the cooling temperature trend that is counter to the global warming. However, evidence from Atmospheric General Circulation Models (AGCMs) have not consistently supported this conclusion, with considerable spreads among different AGCMs. This divergence in model outputs may be attributed to the oversimplified representation of sea ice states in AGCMs. For instance, current models simplistically fix the thickness of Arctic sea ice at 2 meters, not correctly reflecting the recent melting and thinning that has occurred. This oversimplification can lead to inaccuracies in the calculation of heat fluxes related to sea ice thickness. Turbulent heat flux (THF), caused by turbulent motion, is the exchange of heat between the boundary layer and the atmosphere, serving as a crucial link between the sea ice and the atmosphere. This exchange encompasses both sensible heat flux and latent heat flux. 

To address these shortcomings in simulation, we employed a nudging method to adjust the model's Arctic sea ice surface THF to align with ERA5 reanalysis data. This adjustment led to a marked improvement in the model's simulation of THF, especially in areas with sea ice concentrations greater than 85%, where the model's previous simulation of heat fluxes deviated significantly from reality, indicating an underestimation of the impact of Arctic sea ice in the models. Further analysis of the atmospheric circulation response revealed that the model results, when forced with THF, were highly consistent with observational analyses, successfully capturing the weakening of the eddy-driven jet, the negative-positive-negative teleconnection wave train over the Eurasian continent, the intensified Siberian High, and the warm Arctic-cold Eurasia temperature pattern. However, the model results, when driven solely by sea ice concentration (SIC), did not exhibit these features. This discrepancy may relate to the vertical structure of Arctic warming, with the introduction of THF leading to a strong warming of the Arctic atmosphere in the model, whereas the warming in the SIC experiment was confined to below 700hPa. Profound Arctic warming, by reducing the eddy-driven jet, weakens the Siberian storm track. Storm track refers to the regions in the atmosphere where cyclonic activity is most concentrated and intense. The weakened Siberian storm track implies a decrease in cyclonic activity, therefore favors the anti-cyclonic anomalies on the winter mean time-scale. To put this another way, the weakened Siberian storm track provides feedback on the time-mean atmospheric circulation through barotropic eddy forcing feedback.

This implies that the inaccurate simulation of THF over the sea ice surface in AGCMs has underestimated the impact of Arctic sea ice melting on mid-latitudes. Therefore, it is imperative to focus on improving parameterization schemes related to sea ice THF to ensure models can accurately depict these processes. On another note, since the THF in ERA5 is an output from the Integrated Forecasting System (IFS) given observationally constrained sea-ice and large-scale circulation states, the reanalysis THF data also contain errors. To better address these issues, it is essential to augment Arctic observational capabilities, focusing particularly on the heat flux over the sea ice surface. Generating more precise datasets of Arctic surface heat flux is crucial for advancing our comprehension of the Arctic's sensitivity and its far-reaching influence on other regions of the globe.