[A-CG41] Satellite Earth Environment Observation

[ACG41-09] Improved algorithm for the acquisition of atmospheric aerosol optical properties from multi-wavelength, multi-pixel satellite observations

*Miho Sekiguchi¹, Teruyuki Nakajima¹, Makiko Hashimoto², Makiko Nakata³, Chong Shi⁴, Masahiro Momoi⁵ (1.Tokyo University of Marine Science and Technology, 2.Japan Aerospace Exploration Agency, 3.Kinki University, 4.Chinese Academy of Science, 5.GRASP SAS)

Keywords:satellite observation, aerosol, algorithm

The Multi-Wavelength and Multi-Pixel Method (MWPM) is the standard aerosol retrieval algorithm for land surfaces in the GOSAT-2/CAI-2 mission. This method estimates aerosol and surface parameters simultaneously using multi-wavelength and multi-pixel information and is particularly effective for complex terrain, including urban areas and megacities. An updated version of the algorithm, MWPM2, has been developed to improve accuracy. In the present study, the accelerated radiative transfer algorithm was replaced by a state-of-the-art radiative transfer model, RPstar-3, and the aerosol optical properties were revised to better reflect realistic conditions. The aerosol microphysical model consisted of an external mixture of three aerosol types: fine particles, soil dust and sea salt. For fine particles, a mixture of sulphate and black carbon was considered; however, brown carbon was also included in the updated model as black carbon can have unrealistic absorption characteristics, especially at low concentrations. In addition, dust particles were represented as non-spherical Voronoi particles.
In this study, the MWPM2 algorithm, an updated version of the MWPM, was applied to GOSAT-2/CAI-2 and GCOM-C/SGLI observations to estimate aerosol optical thickness (AOT), single scattering albedo (SSA) and Ångström exponent (AE). The comparison with GOSAT-2/CAI-2 data showed a significant improvement in accuracy compared to results derived from the original MWPM. In particular, the estimated accuracy of the aerosol optical thickness in this case was less than 0.09. In contrast, the analysis using SGLI data showed a lower accuracy compared to the GOSAT-2/CAI-2 results. Possible contributing factors include differences in the wavelengths of the bands used and variations in surface reflectance. Further investigation of these discrepancies is required to better understand the underlying causes and to improve accuracy in future applications.