High spatial–temporal image fusion model for retrieving aerosol optical depth based on top-of-atmosphere reflectance

With the growth in industrialization and urban development, air pollution has become an increasing serious health concern. Although ground stations can effectively monitor air quality, they generally observe only located phenomena and limited in the spatial distribution. Remote-sensing approaches have thus been employed by many scholars for air quality monitoring in an entire region. However, no single satellite equips with sufficient spatial and temporal resolutions for detecting rapidly changing local phenomena, such as air quality variations. A top-of-atmosphere reflectance–based spatial–temporal image fusion model (TOA-STFM) is proposed in this paper to solve this problem. The proposed TOA-STFM is modified based on the spatial–temporal adaptive reflectance fusion model (STARFM) and yields fused images in which atmospheric properties are retained. A key process in the TOA-STFM is blurring effect adjustment (BEA), which is performed to match the atmospheric effects caused by aerosols in images with different spatial resolutions. The feasibility of fusing Himawari-8 images with SPOT-6 images was evaluated in this study. We used the proposed model to extract aerosol optical depths (AODs) from images produced by fusing Himawari-8 and SPOT-6 images and compared the extracted AODs with corresponding in-situ observations made by the AErosol RObotic NETwork (AERONET). The AOD relative errors of the proposed TOA-STFM were 2.3%–7.6%, which is a significant improvement comparing to a relative error of 8.4%–13.5% from Himawari-8 images and existing AOD products.