An improved aerosol retrieval algorithm based on non-linear surface model from FY-3D/MERSI-II remote sensing data

This study explores a new scheme to retrieve the global aerosol optical depth over land for the advanced Medium Resolution Spectral Imager (MERSI-II) aboard on Fengyun-3D (FY-3D) satellite, on the basis of dark target (DT) algorithm. The main improvement is that the global surface reflectance model non-linearly varying with solar zenith angle and NDVIswir is made, which is more complex relative to that of MODIS operational algorithm. Our AOD retrievals are compared with AERONET AOD and cross-evaluated with Aqua/MODIS respectively. Overall, the MERSI-II retrieved results over global scale have good consistency with the AERONET observations, on the same condition, the percentage of matchups within the expected error (EE: ±0.05±0.15AOD) is 67.06%, which is slightly lower than the percentage of MODIS (79.84%). On a spatial scale, the coverage of MERSI-II retrievals at one granule is significantly higher than that of MODIS, which is related to successful inversion of haze pixels and has retrieval ability in urban, grassland and other surface types. The monthly mean AOD values retrieved by MERSI-II are close to those of MODIS, indicating that MERS-II has similar quantitative capability and application potential as international counterparts.