资源三号卫星多光谱数据空间分辨率达到5.8 m, 能够很好的应用于地物分类和识别, 由于其缺乏短波红外波段, 无法采用暗目标法进行大气校正, 因此, 提出通过6S辐射传输模型构建大气校正系数查找表, 结合MODIS数据反演的气溶胶光学厚度参数的大气校正方法, 对ZY-3卫星多光谱（MUX）数据进行大气校正。 采用敦煌地区星地同步测量的石膏矿、 戈壁两种地物光谱对大气校正结果进行了验证, 并比较了大气校正前后归一化植被指数（NDVI）。 结果表明: 大气校正后的地面反射率与石膏矿、 戈壁两种地物实测光谱数据相对误差不超过6%; 大气校正增大了植被的NDVI与其他地物的NDVI的差值, 突出在植被监测方面的应用能力。

The resolution of the ZY-3 MUX data is 5.8 meter, which is used to the subject classification. It is very difficult to use the dark target method for the atmosphere correction due to the lack of near infrared band in ZY-3 MUX data. The present paper uses the atmospheric correction coefficient look-up table（LUT）constructed by the radiation transmission model 6S based on the aerosol optical depth retrieved from the MODIS data for the atmospheric correction of ZY-3 image. To validate the results, the paper compares the surface spectral curves of the gypsum mine and Gobi and the NDVI values from the corrected and TOA reflectance, the relative error of the atmosphere corrected and the ground-based surface reflectance is less than 6%; the atmospheric correction increases the difference between vegetation NDVI and other features NDVI, highlights vegetation monitoring application ability.