1State Key Laboratory of Remote Sensing Science, jointly sponsored by the Institute of Remote Sensing Applications of Chinese Academy of Sciences and Beijing Normal University, Institute of Remote Sensing Applications, Chinese Academy of Sciences, Beijing
2Faculty of Computing, London Metropolitan University, 166-220 Holloway Road, London N7 8DB, UK
3Department of Physics, University of Helsinki, Helsinki, Finland
4Finnish Meteorological Institute, Climate Change Unit, Helsinki, Finland
5Netherlands Organisation for Applied Scientific Research TNO, Utrecht, The Netherlands
6German Remote Sensing Data Center, German Aerospace Center, Oberpfaffenhofen, 82234 Wessling, Germany
7School of Geography, Beijing Normal University, Beijing, China
8Center for Earth Observation and Digital Earth, Chinese Academy of Sciences, No. 9 Dengzhuang South Road, Haidian District, Beijing 100094, China
9Graduate University of the Chinese Academy of Sciences, Beijing 100049, China
Abstract. A novel approach for the joint retrieval of aerosol optical depth (AOD) and surface reflectance, using Meteosat Second Generation – Spinning Enhanced Visible and Infrared Imagers (MSG/SEVIRI) observations in two solar channels, is presented. The retrieval is based on a time series (TS) technique, which makes use of the two visible bands at 0.6 μm and 0.8 μm in three orderly scan times (15 min interval between two scans) to retrieve the AOD over land. Using the radiative transfer equation for plane-parallel atmospheres two coupled differential equations for the upward and downward fluxes are derived. The boundary conditions for the upward and downward fluxes at the top and at the bottom of the atmosphere are used in these equations to provide an analytic solution for the surface reflectance. To derive these fluxes, the aerosol single scattering albedo (SSA) and asymmetry factor are required to provide a solution. These are provided from a set of six pre-defined aerosol types with the SSA and asymmetry factor (g). We assume one aerosol type for a grid of 1° × 1° and the surface reflectance changes little between two consequent scans. A k approximation was used in the inversion to find the best solution of atmospheric properties and surface reflectance. The algorithm makes use of numerical minimisation routines to obtain the optimal solution of atmospheric properties and surface reflectance by selection of the most suitable aerosol type from pre-defined sets. Also, it is assumed that the surface reflectance is little influenced by aerosol scattering at 1.6 μm and therefore the ratio of surface reflectances in the solar band for two consequent scans can be well-approximated by the ratio of the reflectances at 1.6 μm. A further assumption is that the surface reflectance varies only slightly over a period of 30 min.
A detailed analysis of the retrieval results show that it is suitable for AOD retrieval over land. Six Aerosol Robotic Network (AERONET) sites with different surface types were used for detailed analysis and 42 other AERONET sites were used for validation. From 445 collocations representing stable and homogeneous aerosol type, we found that >75% of MSG-retrieved AOD values compared to AERONET observed values with an error envelope of ±0.05 ± 0.15τ and a high correlation (R > 0.86). The AOD datasets derived using the TS method with SEVIRI data was also compared with collocated AOD products derived from the NASA TERRA and AQUA MODIS data using the dark dense vegetation (DDV) method and the Deep Blue algorithms. Using the TS method, AOD could be retrieved for more pixels than with the NASA Deep Blue algorithm. The AOD values derived compare favourably.