Seasonality biases arise from the interplay of retrieval quality and solar zenith angle effects in passive sensor AOD products

Abstract

Station observations of surface Arctic aerosol have long shown a pronounced seasonal cycle, with burdens characteristically peaking in the late winter/early spring. Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) aerosol optical depth (AOD) products replicate this seasonality, but passive sensor and reanalysis data products do not. We find that the sub- and low-Arctic seasonality of gridded AOD products from six passive sensors diverges from that of CALIOP data products during the months of September-April, even when controlling for sampling biases. Using colocated CALIOP and Moderate Resolution Imaging Spectroradiometer (MODIS) (Aqua) retrievals, we find that for colocations characterized by low-quality MODIS retrievals, the bias between MODIS and CALIOP strongly depends on the solar zenith angle (SZA), with MODIS AODs showing a 132 % reduction relative to the instrument-mean over a theoretical 0-90° SZA domain. As the fraction of MODIS retrievals flagged as "low-quality"increases with higher SZAs, retrieval quality mediates the relationship between the SZA and dataset biases in gridded products. The dependency may be the result of cloud adjacency effects, and likely also affects midlatitude AOD seasonality. Though additional sources of uncertainty in high latitude retrievals remain, the observed dependency impacts passive sensor data products' representations of (sub-)Arctic aerosol burdens in boreal spring and autumn, which are important for understanding aerosol processes in a highly sensitive yet understudied region. This work also contributes to improved understanding and quantification of the effects of viewing geometry on satellite AOD retrievals, which can help constrain aerosol observations and associated forcings, globally.