关键词: 气溶胶光学厚度, 气溶胶反演, FY-4B;中分辨率成像光谱仪 (MODIS)

Abstract: Objective　Satellite remote sensing technology has become a significant tool for monitoring atmospheric aerosols, offering distinct advantages in spatiotemporal coverage, rapid data acquisition, and high resolution. The Advanced Geostationary Radiation Imager (AGRI) aboard the Fengyun-4B (FY-4B) satellite provides high-frequency, full-disk observations, demonstrating great potential for dynamic aerosol monitoring. Although various aerosol retrieval algorithms have been developed for its predecessor, FY-4A/AGRI, the altered spectral channel settings of FY-4B/AGRI introduce inevitable uncertainties when directly migrating the existing aerosol optical depth (AOD) retrieval algorithms. In addition, the inherent incomplete applicability of traditional Dark Target (DT) algorithms over bright surfaces and the absence of key blue bands required for the Deep Blue (DB) algorithm on FY-4B/AGRI persist. To address these challenges, this study developed a method to retrieve AOD from FY-4B/AGRI by introducing the Moderate Resolution Imaging Spectroradiometer (MODIS) surface reflectance product as auxiliary data. And then, the performance of the method was validated and analyzed over the Beijing-Tianjin-Hebei (BTH) region from June to September 2022. Methods　The aerosol retrieval algorithm developed for FY-4B/AGRI consists of three core stages: data preprocessing, obtaining FY-4B AGRI surface reflectance, and pixel-level AOD retrieval. Initially, FY-4B/AGRI Level 1 (L1) data are subjected to preprocessing, including geometric and radiometric correction. Next, a gaseous absorption correction look-up table (LUT) is constructed using the second simulation of satellite signals in the solar spectrum (6S) radiative transfer model, with the reanalysis data including water vapor and ozone concentrations from the fifth generation atmospheric reanalysis (ERA5) developed by the European Center for Medium-Range Weather Forecasts (ECMWF) as inputs. And the FY-4B/ AGRI surface reflectance is obtained using the MODIS MCD43A4 Nadir BRDF-Adjusted Reflectance (NBAR) product, where a spatiotemporal matching scheme is applied to extract the nearest MCD43A4 pixel value as the optimal estimate for the corresponding FY-4B/AGRI surface reflectance. Finally, an aerosol retrieval LUT is generated using the 6S model, which is further refined through interpolation into a high-density, multi-dimensional grid, and then used for pixel-by-pixel AOD retrieval in the subsequent stages. Results and Discussion　The accuracy of the retrieved FY-4B AOD dataset was validated against the ground-based measurements from the Aerosol Robotic Network (AERONET), with key metrics including the Pearson correlation coefficient (R), mean error (ME), root mean square error (RMSE), and the percentage of retrievals falling within the Expected Error (EE) envelope. Meanwhile, the spatiotemporal distribution consistency of FY-4B AOD products was compared with the MODIS Multi-Angle Implementation of Atmospheric Correction (MAIAC) AOD products. Finally, the spatiotemporal distribution characteristics of AOD over the Beijing-Tianjin-Hebei (BTH) region from June to September 2022 were analyzed. The main results are summarized as follows: 1) FY-4B AOD products exhibit a high level of spatial consistency with MAIAC AOD products. Both datasets display a consistent spatial pattern, characterized by lower aerosol loading in the northwestern mountainous areas and substantially higher aerosol concentrations in the eastern coastal and southern plain regions. 2) A strong agreement is observed between FY-4B AOD retrievals and AERONET ground observations in BTH region across nine distinct time windows from 08:00 to 17:00 Beijing Standard Time (BST) every day during the observation period, which further validates the effectiveness and applicability of the proposed retrieval method. 3) The aerosol loading in BTH region shows distinct seasonal variations, with AOD levels increasing in July and August. 4) Analysis of the diurnal variations in monthly averaged daytime AOD reveals a peak at approximately 11:00 BST and a minimum at 17:00 BST in BTH area. However, it should be noted that this distribution pattern is only derived from a four-month dataset and may deviate from long-term statistical trends. Conclusions　This study presents a method aimed to address the critical challenge of surface reflectance estimation in FY- 4B/AGRI aerosol retrievals. Rather than complex calculations, this approach effectively obtains surface reflectance by leveraging MODIS MCD43A4 products. The methodology, centered on the 470 nm channel, successfully overcomes the limitations of traditional algorithms and demonstrates robust performance even under conditions with limited satellite observations. When applied to the BTH region from June to September 2022, the retrieved AOD exhibits strong agreement with ground-based AERONET measurements, which effectively demonstrates the effectiveness of the method and the accuracy of the retrieval results. This method offers a generalizable method for generating high-frequency AOD products from FY-4B and similar geostationary sensors, particularly those lacking specific aerosol channels or only providing limited and small-scale observations. Future refinements, including the integration of dynamic aerosol models and enhanced cloud detection schemes, are planned to improve the accuracy of the product.

Key words: aerosol optical depth, aerosol retrieval, FY-4B, MODIS (Moderate Resolution Imaging Spectroradiometer) 565