Long-term validation and error analysis of MODIS and VIIRS dark target aerosol products in Africa

Abstract

This study reports the accuracy and uncertainty of aerosol optical depth (AOD) retrievals for the dark target (DT) algorithm over Africa. Three AOD products from the Aqua Moderate Resolution Imaging Spectroradiometer (MODIS), the NOAA-20 Visible Infrared Imaging Radiometer Suite (VIIRS), and the Suomi-NPP (SNPP) VIIRS were evaluated against ground-based Aerosol Robotic Network (AERONET) observations during the period from 2018 to 2023. The validation results indicate that all three products show good validation accuracy in northern Africa, with a slight underestimation in southern Africa. However, it is worth noting that a severe systematic underestimation was observed for all three products in western Africa. Although the correlation coefficients (R) between these three products and the AERONET observations in western Africa are all greater than 0.9, they exhibit significant negative mean biases (MB) of less than −0.18 and root mean square errors (RMSE) exceeding 0.25. Subsequently, an in-depth analysis of error dependence was conducted to explore the underlying causes of this underestimation in this region. The analysis reveals that the aerosol model assumptions and surface reflectance estimation in the DT algorithm jointly lead to this underestimation. Specifically, the underestimation is more pronounced under conditions of strongly absorbing aerosols and sparse vegetation surfaces with high reflectance. Additionally, it was found that as the scattering angle (SCA) increases, the degree of AOD underestimation also intensifies. These results provide critical insights to guide the ongoing refinement of the DT algorithm in western Africa.