ZY-3是我国首颗民用高空间分辨率光学传输型立体测图卫星, 可为国土资源调查、 生态环境监测等发挥重要作用, 而大气校正是制约其广泛定量应用的关键问题之一。由于实测地面光谱数据和大气参数难以实时获取, 针对这种情况下如何反演得到高空间分辨率卫星精确的地表反射率这一问题, 基于2012年内蒙古野外实验实测数据, 对四种暗像元大气校正方法进行了分析与评价研究。分析了四种暗像元大气校正算法中的关键参数对ZY-3 CCD数据应用效果的影响, 结果表明: (1)四种暗像元大气校正方法在第1, 2和3波段均有明显的校正效果, 其中DOS4方法在第4波段大气校正效果最好, DOS1和DOS3方法在第4波段大气校正效果不明显, DOS2方法在第4波段大气校正效果最差。(2)DOS1方法大气校正结果在4个波段的相对误差均大于10%。DOS2方法在第1波段校正效果最好(AE=0.001 9和RE=4.32%), 而在第4波段校正误差最大(AE=0.0464和RE=19.12%)。DOS3方法大气校正结果在4个波段的相对误差均约10%左右。(3)DOS4方法大气校正结果在4个波段的绝对误差均小于0.02和相对误差均小于10%, 大气校正精度最高。

The present paper performed the evaluation of four dark-object subtraction(DOS) atmospheric correction methods based on 2012 Inner Mongolia experimental data. The authors analyzed the impacts of key parameters of four DOS methods when they were applied to ZY-3 CCD data. The results showed that (1) All four DOS methods have significant atmospheric correction effect at band 1, 2 and 3. But as for band 4, the atmospheric correction effect of DOS4 is the best while DOS2 is the worst; both DOS1 and DOS3 has no obvious atmospheric correction effect. (2) The relative error (RE) of DOS1 atmospheric correction method is larger than 10% at four bands; The atmospheric correction effect of DOS2 works the best at band 1(AE (absolute error)=0.001 9 and RE=4.32%) and the worst error appears at band 4(AE=0.046 4 and RE=19.12%); The RE of DOS3 is about 10% for all bands. (3) The AE of atmospheric correction results for DOS4 method is less than 0.02 and the RE is less than 10% for all bands.Therefore,the DOS4 method provides the best accuracy of atmospheric correction results for ZY-3 image.