Characterizing Aerosol Modes and Uncertainty in VIIRS Aerosol Optical Depth Retrievals Over Africa Using AERONET

Abstract

Accurate characterization of aerosol optical depth (AOD) uncertainties is critical for air quality assessment, data assimilation (DA), and environmental studies. In this study, we evaluate two sets of AOD products retrieved from Suomi-NPP VIIRS over Africa. Specifically, we compare the products of NASA's Dark Target (DT) and Deep Blue (DB) algorithms with co-located AERONET observations from 2020 to 2024 over Africa. AERONET visible–near-IR Angstrom Exponents (AE) shows a bimodal distribution and strong monthly variability, with fine-mode dominance in August–September and coarse-mode dominance in March–April in this region. When the VIIRS retrievals are collocated with AERONET over Africa, DB shows a slight overestimation with DT showing a slight underestimation. When examined by AOD value range, DB shows a low bias under heavier aerosol loading, whereas DT exhibits a wider data spread and a less pronounced low bias at higher AOD values. Overall, DB demonstrates a higher correlation and a smaller expected error (EE) envelope compared to DT. Analysis of monthly uncertainty indicates that fine-mode-dominated months, particularly August, September, and October, which also contain the largest number of moderate to heavy aerosol loading cases, exhibit the lowest uncertainty in the DB retrievals, highlighting the improved performance of the updated algorithm. Our analysis shows that, for both DT and DB, AOD retrieval uncertainties are related to the observed AE, suggesting mismatches between algorithm assumptions and the actual dominant aerosol mode, particularly for coarse and mixed-mode aerosols.