High-Resolution Aerosol Optical Depth Retrieval Over Land

Atmospheric aerosol, which directly or indirectly affects the solar radiation, cloud and atmospheric chemical processes, has become an important factor affecting global or local climate change, urban air quality and ecological environment. Wtih the rapid development of satellite remote sensing technology, remote sensing of aerosol retrieval approaches can obtain wide-scale spatial distributions of aerosol for long time periods. The estimation of surface reflectance and assumption of aerosol types are two of the most important factors affecting the AOD retrieval accuracy. However, due to complex surface types and various aerosol types,aerosol retrieval over land has faced great challenges. Moreover, due to low and coarse spatial resolutions of operational AOD products, they are seriously limited in the air pollution or eco-environmental quality monitoring in the urban or local small-scall areas. Therefore, aimed at the these problems, a new and improved aerosol retrieval algorithm with priori database of land surface parameters support is purposed in this paper, and the AOD dataset at 1 km resolution over global land is firt generated and published on line. The main contents of this paper mainly include as followings:1) Dynamic threshold cloud detection algorithm research. Cloud detection is the key and indispensable data preprocessing for AOD retrieval. It can improve overall accuracy of aerosol retrieval by accurately identifying and masking the cloud in the images. Due to the effects of mixed pixels and atmospheric factors, traditional cloud detection algorithms are limited and show overall low accuracy, especially for thin and broken clouds. To solve this problem, a new dynamic threshold cloud detection method is proposed based on a priori surface reflectance database, which is construdtced using the MODIS surface reflectance products through the minimum synthesis technology. Relationships of top-of-atmosphere reflectance (TOA) and surface reflectance in different conditions are simulated using the Second Simulation Satellite Signal Solar Spectrum (6S) model. Then the dynamic threshold cloud detection models for different channels are built. The MODerate Resolution Imaging Spectroradiometer (MODIS)data are selected to perform cloud detection experiments, and the remote sensing interpretation cloud results, Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO),MOD35 and MOD04 cloud mask products are obtained for validation and comparison purposes.Results showed that the new algorithm can realize good and effective cloud detection resoluts with an average high accuracy great than 80%, especially for the thin and broken clouds over bright surfaces.2) Retrieval of aerosol optical depth over bright urban areas. Due to the complexity of the underlying surfaces in urban areas, the surface reflectance is high and varies obviously, leading to large difficultities of surface reflectance estimation. To solve this problem, a new approach for surface reflectance estimation based on a pre-constructed surface reflectance database is presented using the MODIS 8-day composite surface reflectance (MOD09A1) product. The Beijing-Tianj in-Hebei region, which features complex underlying surfaces, is chosen and the MODIS data is collected for aerosol retrieval experiments. The continental aerosol is selected as the dominant type in this paper. Finally, the Aerosol Robotic Network (AERONET) ground-based AOD measurements in four local stations including the Bejing, XiangHe, Bejing CAMS and Beijing RADI, and the MOD04 aerosol products are selected to validate and compare the retrievals. Results show that the proposed algorithm can achieve a high accuracy of AOD retrieval in urban areas, and compared with MOD04 AOD products, the accuracy and spatial continuity are improved obviously.3) High-resolution aerosol retrieval over global land. Based on previous studies, the scale is extended to global, and the key problems for aerosol retrievals are the estimation of the surface reflectance and asuumptions of the aerosol types. Aimed at these problems, an improved high-resolution aerosol retrieval algorithm supported by prior global land surface parameters <I-HARLS>, and a global monthly land surface reflectance (LSR) and a seasonal land aerosol type(LAT) are created based on the MODIS surface reflectance products and aerosol products with the minimum and majority synthesis technologies. The MODIS data of the entire 2013 are selected to perform the aerosol experiments. Then the 1km high-resolution AOD products,including daily (Level 2) and monthly (Level 3) products. Meanwhile, the AOD measurements from a total of 172 AERONET ground-based stations are obtained for validation purposes. In addition, four typical local regions are also selected to verify the accuracy of AOD datasets.Results show that the I-HARLS algorithm can realize aerosol retrievals over both bright and dark surfaces with approximately 73% of the collections falling with the accuracy requirements,and the AOD products can provide high-accuracy and continues spatial distributions, especially for monitoring the air quality from small or local scales. Moreover, the AOD products are uploaded and opened freely for all users on the Intenet.