With continual advancements in data assimilation systems, new observing systems, and improvements in model parameterizations, several new atmospheric reanalysis datasets have recently become available. Before using these new reanalyses it is important to assess the strengths and underlying biases contained in each dataset. A study has been performed to evaluate and compare cloud fractions (CFs) and surface radiative fluxes in several of these latest reanalyses over the Arctic using 15 years (1994–2008) of high-quality Baseline Surface Radiation Network (BSRN) observations from Barrow (BAR) and Ny-Alesund (NYA) surface stations. The five reanalyses being evaluated in this study are (i) NASA's Modern-Era Retrospective analysis for Research and Applications (MERRA), (ii) NCEP's Climate Forecast System Reanalysis (CFSR), (iii) NOAA's Twentieth Century Reanalysis Project (20CR), (iv) ECMWF's Interim Reanalysis (ERA-I), and (v) NCEP–Department of Energy (DOE)'s Reanalysis II (R2). All of the reanalyses show considerable bias in reanalyzed CF during the year, especially in winter. The large CF biases have been reflected in the surface radiation fields, as monthly biases in shortwave (SW) and longwave (LW) fluxes are more than 90 (June) and 60 W m−2 (March), respectively, in some reanalyses. ERA-I and CFSR performed the best in reanalyzing surface downwelling fluxes with annual mean biases less than 4.7 (SW) and 3.4 W m−2 (LW) over both Arctic sites. Even when producing the observed CF, radiation flux errors were found to exist in the reanalyses suggesting that they may not always be dependent on CF errors but rather on variations of more complex cloud properties, water vapor content, or aerosol loading within the reanalyses.

Along with significant changes in the Arctic climate system, the largest year-to-year variation in sea-ice extent (SIE) has occurred in the Laptev, East Siberian, and Chukchi seas (defined here as the area of focus, AOF), among which the two highly contrasting extreme events were observed in the summers of 2007 and 1996 during the period 1979-2012. Although most efforts have been devoted to understanding the 2007 low, a contrasting high September SIE in 1996 might share some related but opposing forcing mechanisms. In this study, we investigate the mechanisms for the formation of these two extremes and quantitatively estimate the cloud-radiation-water vapor feedback to the sea-ice-concentration (SIC) variation utilizing satellite-observed sea-ice products and the NASA MERRA reanalysis. The low SIE in 2007 was associated with a persistent anticyclone over the Beaufort Sea coupled with low pressure over Eurasia, which induced anomalous southerly winds. Ample warm and moist air from the North Pacific was transported to the AOF and resulted in positive anomalies of cloud fraction (CF), precipitable water vapor (PWV), surface LWnet (down-up), total surface energy and temperature. In contrast, the high SIE event in 1996 was associated with a persistent low pressure over the central Arctic coupled with high pressure along the Eastern Arctic coasts, which generated anomalous northerly winds and resulted in negative anomalies of above mentioned atmospheric parameters. In addition to their immediate impacts on sea ice reduction, CF, PWV and radiation can interplay to lead to a positive feedback loop among them, which plays a critical role in reinforcing sea ice to a great low value in 2007. During the summer of 2007, the minimum SIC is 31 % below the climatic mean, while the maximum CF, LWnet and PWV can be up to 15 %, 20 Wm -2 , and 4 kg m -3 above. The high anti-correlations (-0.79, -0.61, -0.61) between the SIC and CF, PWV, and LWnet indicate that CF, PWV and LW radiation are indeed having significant impacts on the SIC variation. A new record low occurred in the summer of 2012 was mainly triggered by a super storm over the central Arctic Ocean in early August that caused substantial mechanical ice deformation on top of the long-term thinning of an Arctic ice pack that had become more dominated by seasonal ice.Keywords Trigger and cause of Arctic sea ice retreat Á Cloud-radiation-water vapor feedback to the sea-ice-concentration variation Á Extreme Arctic seaice extent formation mechanisms Á Triggered by atmospheric forcings Á Enhanced clouds-radiation-PWV feedback