Abstract The Arctic is heating far more rapidly than the global mean, and clarifying the influence of aerosols in this intensification demands accurate and reliable observational records. The Arctic exhibits a distinct seasonal aerosol cycle, springtime ”Arctic Haze” with elevated AOD and summertime “Clean Air” with low AOD. Thus, it is critical to evaluate how well various datasets capture this seasonality relative to ground-based observations. This study analyzes spring and summer AOD variability using CAMSRA and MERRA-2 reanalyses, MODIS Terra and Aqua satellite observations, AERONET measurements, AEROSNOW retrievals, and GEOS-Chem model simulations. Results show that satellite-derived and satellite-assimilated reanalyses are far from capturing the expected seasonal Arctic Haze and Clean Air pattern, except at Bonanza Creek and Yakutsk, where anthropogenic pollution alters it. The inability of reanalyses to capture Arctic aerosol seasonality likely stems from the assimilation of satellite retrievals influenced by cloud contamination and surface reflection from snow and ice, as well as inherent biases in the underlying models used to generate these datasets. In contrast, AERONET observations and GEOS-Chem simulations consistently capture Arctic Haze in spring, driven by long-range transport, and Clean Air in summer, associated with efficient wet removal of aerosols. CAMSRA further underestimates emissions from Arctic forest fires and inadequately represents long-range pollution transport. These findings suggest that independent model simulations align more closely with ground-based observations than satellite products or reanalyses, and that adjusting wet-scavenging parameters to fit such reanalyses may misrepresent aerosol processes and their contribution to Arctic warming. Incorporating advanced retrieval algorithms like AEROSNOW into reanalyses offers a pathway to reduce these biases and improve representation of Arctic aerosol seasonality.
Vountas et al. (Mon,) studied this question.