资源环境科技发展态势分析平台

The latest version of JAXA aerosol optical depth (AOD) product over China in 2018 from Himawari-8 Advanced Himawari Imager (AHI) was evaluated, including validation with global AErosol RObotic NETwork (AERONET) and China Aerosol Research Network (CARSNET) ground-based observations, and comparison against the Moderate Resolution Imaging Spectroradiometer (MODIS) AOD product. Considering both the quality and quantity of the AHI AOD product, the Quality Assurance Flags (QA_Flags) with “good” and “very good” are chosen for evaluation and analysis. AHI and MODIS AOD products are well correlated and the spatial distributions are similar, even though underestimation from AHI AOD in the morning and overestimation in the afternoon are found when compared with MODIS AOD. Validation with ground-based observations also indicates a good correlation (R > 0.75) when sparsely vegetated surfaces are excluded. It is found that 52.74% fall below the expected error (EE) at 10:30 BJT when compared with CARSNET which has more vegetation sites and 56.91% fall above the EE at 13:30 BJT when compared with AERONET which has more urban sites. The slope of the fit line is around 1.0 for all matched sites, while the scatter plots have divided into two parts as the performance of AHI AOD depends on the observation times and surface types. The temporal variations of AHI and AERONET AODs show that the overestimation from AHI AOD occurs in the afternoon for barren surface while the underestimation occurs in the morning for well-vegetation surface. For monthly variation, the overestimation of AHI AOD is found over Northern China in winter and those seasons when the plants are growing or disappearing rapidly.

The assumption that the surface reflectance remains unchanged and the solar geometries are nearly identical within a month causes the underestimation of surface reflectance and thus contributes to the AHI AOD overestimation for low vegetation and bare surfaces in autumn and winter. While over the vegetated surfaces, neglecting the background aerosols causes overestimation of the surface reflectance, and thus leads to underestimation of AHI AOD. The difference between backward and forward scattering contributes to the larger JAXA AHI AOD error in the afternoon, which aggravates positive deviation for barren land sites and alleviates the underestimation for vegetation surface types in the afternoon. In addition, aerosol model selection and cloud contamination also affect the accuracy of aerosol products, especially when AOD is large.

Overall, the JAXA AHI AOD provides a high temporal resolution and relative reliable aerosol product for environmental research. The analysis on characteristics and uncertainty sources in this study are to further enhance the understanding of the geostationary satellite aerosol products for applications, and provide useful information for improving the aerosol retrieval algorithm in future.