Reconstruction Of Greenland Ice Sheet Surface Mass Balance Over Past 200 Years And Causes Analysis

Under the condition of global warming,mass loss over the Greenland ice sheet（Gr IS）has become a major source of the sea level rising（SLR）.Known as the“Ice Silk Road”,the navigation condition of the Arctic route will also be impacted.As an important component of mass loss,the proportion of surface mass balance（SMB）has increased and become the biggest contributor to the development of Arctic amplification.Therefore,estimating the SMB over the Gr IS and its effect on the SLR has become a hotspot.Lacking the knowledge of SMB for long time series,regional climate models（RCMs）have difficulties in simulating SMB,which limits the understanding of SMB over the Gr IS and its response to climate change.This study reconstructs the SMB over the Gr IS for the past 200 years based on the Kriging-like interpolation,Degree-day Model and Simplified meltwater refreezing scheme by global reanalyses,RCMs,the snow accumulation rate of ice cores,ground precipitation measurements.First,this study analyses spatiotemporal patterns of SMB and explores the relationship between climate,and atmospheric circulation.The main results of this paper are as follows:（1）As the only input of SMB,the cycle of precipitation determined the spatiotemporal patterns of snow accumulation.To confirm the force datasets of snow accumulation,this study assesses the quality of eight global precipitation reanalyses（including 20CR,CERA20C,CFSR,ERA20C,ERA5,ERAI,JRA55 and MERRA2）over the Gr IS.In comparison,the wet deviation in space of CERA20C is the smallest,and the mean biases and mean RMSE compared with the satellite remote sensing datasets Cloud Sat are-1.73 and 14.8 mm/month,respectively.ERA5 has the highest correlation in simulating annual precipitation.This indicates that CERA20C and ERA5 have the greatest precision in spatial distribution and temporal cycle.Therefore,CERA20C and ERA5 can be used for covariance variable background in the reconstruction of snow accumulation.（2）Global precipitation reanalyses CERA20C and ERA5,coupled with the Regional Climate Models MAR and RACMO were selected as covariance variable backgrounds.Based on the datasets above,four snow accumulation datasets using the kriging-like interpolation by snow accumulation rate of ice cores and ground precipitation measurements of Gr IS for 1800-2011were reconstructed.The four snow accumulation reconstruction datasets are named SA_CERA20C,SA_ERA5,SA_MARand SA_RACMO,respectively.The uncertainty and spatiotemporal characteristics are analyzed.The independent verification and cross-verification show that SA_RACMOhas the best quality.By carrying out 399 cross-validations,the results show that the removal of individual years and ground measurements will not affect the robustness of reconstructions.The SA_MARshows the biggest uncertainty with±27.29 Gt/a.The snow accumulation over the Gr IS has not changed significantly since 1800,the trend of SA_RACMOis only 0.02 Gt/10a2,which means that the snow accumulation is in stable status.After 2000,the trend of SA_RACMOis up to-57.85 Gt/10a².But similar decreasing trends can be found in history.Therefore,the rapid decline of snow accumulation in recent years may be a cyclical change over a long time series.（3）To gain the variables of runoff for 1851-2011,this study inputs daily temperature of20CR into Degree-day Model and Simplified meltwater refreezing scheme.The trends of runoff were 54.20 and 53.53 Gt/10a²,respectively（p<0.05）during the 1920s-1930s and 1990s-2011.Combined with the snow accumulation reconstruction results above,four SMB datasets(SMB_CERA20C,SMB_ERA5,SMB_MARand SMB_RACMO)were reconstructed.By method of Empirical Orthogonal Function（EOF）,the results indicate that the SMB increases in the western edge and eastern ice sheet in the cold period before the 1920s.For the1920s-1930s and 1990s-2011,the EOF results show that SMB decreases,especially in the western ice sheet.The trends of SMB over the Gr IS mainly depend on the variations in the west and southwest,where high runoff occurs,and variations in the southeast,where high snow accumulation occurs.（4）The relationships between SMB and 2m temperature,precipitation,terrain and atmospheric circulation were analyzed,especially after the 1990s.The mechanisms of temperature,precipitation,terrain and atmospheric circulation on SMB in different periods are complicated.These elements cannot individually explain the variations of SMB.The SMB for the west ice sheet is very sensitive to the changing of temperature with the fastest warming trends.Before 1968-1992,temperature caused the increasing of SMB by increases in precipitation.Terrain contributes to the spatial differentiation of the surface mass balance,mainly by influencing precipitation.After 1993,the increase in runoff caused by temperature will lead to a decrease in SMB.Greenland Blocking Index（GBI）and Northern Atlantic Oscillation（NAO）have opposite correlation distributions.The positive phase of GBI（negative phase of NAO）will cause increasing temperature and runoff.In all,the causes of SMB are complex after the 1990s.In the early 1990s,SMB showed an increasing trend due to the increase in snow accumulation and a decrease in the runoff.After1996,increased runoff led to a rapid decline in SMB.After 2007,SMB decreases over the Gr IS were determined by the reduction of snow accumulation and increasing runoff.The correlation between SMB and NAO（r=0.66,p<0.05）,GBI（r=-0.78,p<0.05）,indicates the important impact of GBI on SMB.In conclusion,temperature,precipitation and atmospheric circulation cannot individually explain the changes in SMB.