Evaluation of MODIS and two reanalysis aerosol optical depth products over AERONET sites

Abstract

This study selects 400 sites from Aerosol Robotic Network (AERONET) over 2002–2015 to develop an aerosol optical depth (AOD) evaluation dataset, which is used to evaluate two reanalysis AOD datasets (MERRAero and MERRA-2) from Modern-Era Retrospective analysis for Research and Applications (MERRA) projects, and satellite AOD datasets from Terra and Aqua Moderate Resolution Imaging Spectroradiometer (MODIS) products. Each AOD dataset is first evaluated individually, and then, a comparison of the MODIS, MERRAero and MERRA-2 datasets is conducted. The results show that the reanalysis MERRA-2 AOD dataset has a comparable accuracy (RMSE = 0.119, MBE = −0.008) with the satellite MODIS AOD datasets (RMSE = 0.110, MBE = 0.011). The MERRA-2 AOD dataset has similar spatial distributions with MODIS data and can capture the dynamic changes of AOD temporally. Compared to MERRAero, the MERRA-2 product improves, but it still slightly underestimates AERONET AOD in general. The MODIS AOD datasets perform differently compared to AERONET AOD due to different algorithms and spatial resolutions. Compared to AERONET AOD, the MODIS AOD datasets generated by the deep blue algorithm are negative biased, while the MODIS AOD datasets from dark target algorithm and the merged dataset are positively biased. The precisions of the Aqua MODIS AOD datasets are higher than the corresponding Terra MODIS AOD datasets.