Affected by high anthropogenic emissions and adverse weather conditions, heavy haze events over large areas and long durations occur frequently in the autumn and winter in China. To overcome the disadvantage of discrete ground-based observations on large scale monitoring, this study uses satellites to realize rapid and automatic haze detection based on combination of multi-channel information that shows different scatter and emission characteristics of cloud, ice, snow, haze, dark and bright surface under different wavelengths. Through the correction of air molecular scattering in haze detection, the misjudgment under large-angle conditions in the previous work is effectively eliminated. It also greatly reduces the variation range of threshold, which is very helpful for automatic haze detection. The haze detection method proposed in this study has been applied to MEdium-Resolution Spectral Imager-Ⅱ (MERSI-Ⅱ) of Fengyun-3D (FY-3D) from October 2020 to February 2021, and the spatial distribution of haze identification region is consistent with that of gray areas on true color images. The validation has been done using ground-based meteorological observations of haze weather and environmental observations of PM2.5
concentrations over 75 μg/m3
. The validation results show that the consistency between the haze identification results of FY-3D and observations of ground meteorological stations is 91.1%, and the consistency between the haze identification results of FY-3D and observations of ground environmental stations is 85.6%. As haze observations by meteorological stations depend on the extinction of the atmosphere while satellites observations also rely on this physical characteristics, it is reasonable that the two are more consistent.