Study On The Retrieval Of Aerosol Optical Thickness Over Land Based On HJ-1 Data

Aerosol has great influence on human life and global climate. AOT (Aerosol Optical Thickness) is one of the most important parameters of aerosol. It is not only an important physical quantity to attribute the atmospheric turbidity but also a key parameter to qualify the climatic effect. Satellite remote sensing as an effective method to detect aerosol has been widely used. With the launch of HJ-1, China has the satellite exclusively for monitoring and forecasting the environment and disaster. Next, how to use HJ-1 to retrieve AOT is an important thing. The method that will be selected to retrieve AOT should be utility and accurate.This paper is based on the wide coverage and multi-spectral camera data (CCD) from HJ-1 to research the inversion method of AOT. The scanning width of HJ-1 CCD data is 360km (combining of CCD1 and CCD2> 700km), and the nadir ground pixel resolution is 30 meter. The CCD camera has three visible bands (0.43-0.52 um, 0.520-0.60μm,0.63-0.69pm) and a near infrared band (0.76-0.90μm). Then for the purpose of using high-resolution HJ-1 CCD camera to monitoring atmospheric environment, especially the concentration of particles on ground layer (ground particle concentration generally related with aerosol optical thickness), we would firstly need to fix the AOT retrieval algorithm.The core of aerosol optical thickness inversion is how to remove the ground contribution from the energy of satellite received. For the MODIS AOT inversion algorithm, it uses the ground reflectance relationship between 2.1μm and visible light channels to retrieve AOT. But this method is not fit for HJ-1 because of HJ-1 CCD lack of 2.1μm channel. For other method, such as the inversion method using Multi-angle Imaging (MISR) and multi-band iteration method (BAER, Bremen Aerosol Retrieval), they can also not be used on HJ-1. Therefore, we need find other ways to retrieve AOT based on HJ-1 CCD data.After the summary and comparing of different inversion methods of AOT, this paper studied on the retrieval method based on building the surface reflectance database. The method of using surface reflectance database uses other satellites' ground surface reflectance and BRDF products to retrieve AOT, through effectively selection of ground reflectance and BRDF, conversion between reflectance from different sensors to build the surface reflectance of HJ-1 CCD. The study area in this paper contains southern Jiangsu, northern Zhejiang, and Shanghai, all located in the Yangtze River Delta. The data source is from the MODIS ground reflectance and bidirectional reflectance products. By reference of the operational MODIS AOT retrieval process, we built look-up table and used the MODIS approach to process look-up table to achieve the inversion of AOT. Relative to other method, this technique can be applied to a variety of surface and different areas. Because of its surface reflectance database considering the directional reflectance characteristics of the surface, this method avoids the difficulty of the selection of reference images that has the same observation geometry. Key issues of this algorithm include:transition between HJ-1CCD and MODIS surface reflectance, construction rules of surface reflectance database, the building of look-up table. This paper discussed and analyzed these questions in detail, and gave the solution of them.The AOT inversion results from method based on building the surface reflectance database were also compared and validated. For comparison and validation, two methods were used. One is using other AOT retrieval results to compare with this method AOT in space, and the other is to validate the AOT results by sun photometer data in point. Because the MODIS aerosol product has greatly different spatial scales with HJ-1 AOT, we used MODIS LIB data to get lkm resolution AOT to compare with HJ-1 inversion results in space. HJ-1 retrieval results have good correlation with MODIS results, and the correlation coefficient is 0.909.The HJ-1 inversion results are slightly smaller. Besides, this paper used the CE318 data from aerosol Automatic observation network (AERONET) to validate the AOT inversed from HJ-1. The result shows AOT retrieval values have high consistency with sun photometer observations. The correlation coefficient is 0.896 and the linear fitting formula is y=0.927x+0.038. Most points are within the MODIS AOT inversion error.