Publications

Ji, Z; Ma, Y; de Leeuw, G; Shi, Z; Li, ZQ (2025). An Enhanced Aerosol Optical Depth Retrieval Algorithm for Particulate Observing Scanning Polarimeter (POSP) Data Over Land. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 63, 4100418.

Abstract
Single-angle sensors typically use radiative transfer simulations based on Lambertian surface, even though the surface reflectance obtained exhibits directional characteristics. Fully accounting for the contribution of surface directional characteristics to the top of atmosphere (TOA) reflectances can further improve the accuracy of aerosol retrievals. In this study, we propose an enhanced aerosol retrieval algorithm for the Particulate Observing Scanning Polarimeter (POSP), by further considering the surface directional characteristics. Combined with an updated aerosol model, this approach achieves high-accuracy retrievals. We used historical bidirectional reflectance distribution function (BRDF) products to construct stable surface constraints. By exploring the strong empirical statistical relationships between adjacent blue bands, we have realized the joint inversion of multiple blue bands. In addition, we used an optimization algorithm that incorporates boundary constraints, simultaneously accounting for errors in the surface constraint model, and satellite observation errors. The global aerosol optical thickness (AOD) at 550 nm over land was retrieved from November 2021 to April 2022. Validation of POSP AOD versus AErosol RObotic NETwork (AERONET) data shows a high consistency, with correlation coefficient (R) of 0.93, root mean square error (RMSE) of 0.086, bias of 0.004, fraction within expected error (EE) of 80.8%, and fraction within Global Climate Observing System (GCOS) of 52.1%. Comparison with MODIS aerosol products shows that the accuracy of POSP AOD is better than that of MODIS AOD. According to the matching results, for DB, R of 0.936/0.907 and fraction within EE of 82.2%/74.9% (POSP/MODIS DB); for DT, R of 0.937/0.915 and fraction within EE of 83.5%/ 72.0% (POSP/MODIS DT). The spatial distribution difference between POSP AOD and DB AOD is small, indicating good consistency, and POSP AOD captured the intensity of aerosol pollution well. In summary, the enhanced aerosol algorithm achieves reliable high-precision AOD retrieval and because of its generality could also be applied to other sensors.

DOI:
10.1109/TGRS.2024.3514170

ISSN:
1558-0644