随着我国城市的发展， 霾已经成为大气污染的重要形式。 从霾的物理性质出发， 利用米理论和RT3计算了霾的光学性质和反射特性， 表明环境一号(HJ-1)卫星CCD相机的第一和第二波段最适宜霾的反演。 然后， 利用深蓝算法， 基于MODIS地表反射率产品建立地表反射率库， 实现霾的反演。 最后， 以北京为试验区进行了2009年全年的反演试验， 结合地面观测结果的验证表明， HJ-1监测结果与地面结果有着较好的相关性（相关系数大于0.9）， 但整体大于地面监测结果。 讨论表明， 地表反射率库的误差对区分霾影响较小（带来的霾光学厚度误差小于0.1）， HJ-1的CCD传感器的辐射分辨率尚不能完全满足霾监测需求。

With the urbanizing in China, haze days occur frequently, which have largely impacted air quality. In the present paper, based on haze physical properties, haze optical properties were calculated by Mie theory and apparent reflectance of haze in Environment Satellite 1 (HJ-1) channels was simulated by Radiative Transfer (RT) 3. Simulated results show that it is reasonable to extract the haze from apparent reflectance in the first and second channels. By Deep Blue algorithm, Haze Optical Depth (HOD) was retrieved from HJ-1 data supported by reflectance database from MODIS product. From HJ-1 data in 2009 over Beijing area, the haze days were monitored and validated by AERONET/PHOTONS Beijing site. The validation shows that the correlation coefficient of HOD is greater than 0.9, but HOD from HJ-1 is greater than that from ground-based measurements. The discussions show that the error from reflectance database is less than 0.1 and radiance resolution of HJ-1 needs to advance for haze monitoring.