Abstract:
The darkening Arctic that results from global warming was hypothesized 50 years ago and confirmed only recently by satellite observations. Compared to the notable surface darkening as reported in extensive publications, the darkening at the top-of-atmosphere (TOA) has been studied insufficiently and shows a significant divergence in magnitude among the limited references. Hence, it raises an urgent need to quantify the Arctic TOA darkening from an observational perspective. To assess the operational TOA albedo observations, both instrument calibration and retrieval algorithm have been checked between the Clouds and the Earth’s Radiant Energy System (CERES) and Multiangle Imaging SpectroRadiometer (MISR). While the intercalibration experiment confirmed the stability of CERES long-term measurements and accounted for most of the discrepancies in their TOA albedo trends, the remaining differential trends were likely due to the biased MISR albedo retrievals that mainly resulted from the inaccurate scene identification. These results strongly enhanced the confidence in the temporal trends of TOA albedo measured by the CERES. The magnitude, seasonality, and contributing sources of the Arctic TOA albedo changes were then examined by CERES and compared to four state-of-the-art reanalysis data sets. A significant declining TOA albedo of −0.012±0.003 per decade was observed within the Arctic Ocean between 1982 and 2015. The darkening occurred in every sunlit month and experienced an accelerated (damped) darkening in the early (late) melt season. Furthermore, changes in surface albedo not only accounted for most of the TOA darkening trends but also largely explained the variations in the TOA albedo anomalies. Nevertheless, no reanalysis product was able to fully reproduce this change, particularly in the summer months. Being closely related to the underlying sea ice condition, June TOA reflected solar radiation showed a robust 3-month lag correlation with September sea ice extent (SIE). The small hindcast prediction errors, high forecast skill, and the capability of predicting September SIEs with large negative anomalies are similar to or better than other complex models. The results emphasized the particular importance of the early summer sea ice state for the subsequent ice evolution and served as an application of the satellite TOA albedo product.