与现有的大气卫星传感器相比,环境一号卫星(HJ-1)CCD相机具有较高的空间分辨率(30 m),使得在城市地区找到光谱纯像元的机率大大增加。本文提出一种基于纯像元提取的城市地区气溶胶光学厚度(Aerosol Optical Depth,AOD)反演算法,利用像元纯净指数来提取CCD影像上的纯像元,并由HJ-1 A星和B星的多时相CCD观测数据结合地表双向反射率模型确定纯像元的地表反射特性,在此基础上反演AOD。与AERONET地基测量数据的对比表明,该算法具有较高精度,相关系数为0.83,线性拟合斜率为1.091,截距为0.053。基于该方法的AOD反演结果作为输入,能较大程度提高HJ-1卫星CCD影像大气校正的精度。

Compared to moderate/low resolution satellite sensors for aerosol detection, the high spatial resolution of HJ-1 CCD camera (30 m) provides an opportunity to easily find spectrally pure pixels (SPP) over urban areas. In this study, we developed an SPP algorithm for urban Aerosol Optical Depth (AOD) retrieval using HJ-1 CCD data. Pixel Purity Index (PPI) is used to identify the pure pixels in the image. The surface reflectances of the pure pixels were estimated from the multi-temporal CCD measurements of HJ-1A and HJ-1B based on the surface Bi-directional Reflectance Distribution Function (BRDF) model. Then the AOD can be retrieved from satellite measurements. The comparison with ground-based AErosol RObotic NETwork AERONET AOD measurements shows good performance of our algorithm. A significant correlation coefficient with R=0.83 was obtained with a linear regression slope close to 1 and a intercept of 0.053. With the retrieved AOD as an input, HJ-1CCD data over urban areas was significantly improved after the atmospheric correction.