複数の衛星間で一貫したエアロゾル推定を可能にする手法を開発（New Method Enables Consistent Aerosol Retrieval Across Satellites）

Polarimetric observations provide complementary constraints on aerosol size and absorption, offering enhanced sensitivity for characterizing aerosol variability and its impacts on climate, environment, and human health. The Particulate Observing Scanning Polarimeter (POSP), deployed aboard the GaoFen-5(02) and DaQi-1 satellites, is a multispectral (from ultraviolet to shortwave infrared) cross-track scanning polarimeter to deliver twice-daily global surface-atmosphere observations. Despite the growing use of artificial intelligence for aerosol retrieval from high-dimensional satellite measurements, most existing approaches retrieve aerosol parameters independently, without explicitly enforcing the internal relationships among aerosol optical quantities. In this study, we present a physics-guided learning approach for optically consistent multi-parameter aerosol retrieval from dual-satellite polarimetric observations. The method jointly retrieves primary aerosol optical quantities, including multi-wavelength aerosol optical depth (AOD), fine- and coarse-mode AOD, and absorbing and scattering AOD, and derives secondary parameters such as Ångström exponent (AE), fine-mode fraction (FMF), and single scattering albedo (SSA) from these primary quantities. Importantly, the proposed approach supports direct cross-satellite transfer between POSP-equipped platforms (GF-5(02) and DQ-1), allowing consistent aerosol retrievals to be applied across different satellites without retraining. Validation against ground-based AERONET measurements demonstrates high accuracy for POSP multi-wavelength AOD, with correlation coefficients (R) ranging from 0.93 to 0.96 and root mean square errors (RMSE) between 0.046 and 0.085. Independent cross-satellite evaluations further confirm the robustness and transferability of the retrieval framework under completely independent observational conditions, yielding R values of 0.89–0.95 and RMSE values of 0.054–0.107 across different wavelengths. Retrievals of aerosol size- and absorption-related parameters also show good agreement with AERONET products. The resulting dual-satellite POSP aerosol products are used to characterize global and seasonal variations in aerosol size and absorption during 2023, and to construct multi-year absorption aerosol optical depth (AAOD) time series from November 2021 to December 2023. Applications to extreme aerosol events, including the Canadian wildfires, further highlight the unique value of polarimetric observations for capturing aerosol size and absorption evolution, demonstrating the potential of the proposed approach for advancing aerosol monitoring and environmental applications.