Sogacheva, L; Popp, T; Sayer, AM; Dubovik, O; Garay, MJ; Heckel, A; Hsu, NC; Jethva, H; Kahn, RA; Kolmonen, P; Kosmale, M; de Leeuw, G; Levy, RC; Litvinov, P; Lyapustin, A; North, P; Torres, O; Arola, A (2020). Merging regional and global aerosol optical depth records from major available satellite products. ATMOSPHERIC CHEMISTRY AND PHYSICS, 20(4), 2031-2056.

Satellite instruments provide a vantage point for studying aerosol loading consistently over different regions of the world. However, the typical lifetime of a single satellite platform is on the order of 5-15 years; thus, for climate studies, the use of multiple satellite sensors should be considered. Discrepancies exist between aerosol optical depth (AOD) products due to differences in their information content, spatial and temporal sampling, calibration, cloud masking, and algorithmic assumptions. Users of satellite-based AOD time-series are confronted with the challenge of choosing an appropriate dataset for the intended application. In this study, 16 monthly AOD products obtained from different satellite sensors and with different algorithms were inter-compared and evaluated against Aerosol Robotic Network (AERONET) monthly AOD. Global and regional analyses indicate that products tend to agree qualitatively on the annual, seasonal and monthly timescales but may be offset in magnitude. Several approaches were then investigated to merge the AOD records from different satellites and create an optimised AOD dataset. With few exceptions, all merging approaches lead to similar results, indicating the robustness and stability of the merged AOD products. We introduce a gridded monthly AOD merged product for the period 1995-2017. We show that the quality of the merged product is as least as good as that of individual products. Optimal agreement of the AOD merged product with AERONET further demonstrates the advantage of merging multiple products. This merged dataset provides a long-term perspective on AOD changes over different regions of the world, and users are encouraged to use this dataset.