地形校正可以减小地形起伏对地物光谱的影响，提高计算机分类在山区的精度。设计了针对全球土地覆盖分类的Landsat TM/ETM+数据地形校正方法SCOS（Smoothed COS余弦），首先对地形的坡度角进行抹平处理，很大程度上削弱了地表非朗伯性对地形校正的影响，然后利用简单有效的余弦校正去除地形效应。该方法与其他常用地形校正算法的对比分析是通过对全球不同区域、不同地表覆盖的有代表性的6景Landsat TM/ETM+数据的试验，采用统计分析与目视判读的方式，从过度校正和类内均一性两个方面进行的。结果表明，该方法在目视效果和统计结果上优于常规方法，并且更加简单有效，无需复杂的大气参数及传感器参数，满足全球地表覆盖分类对地形校正的需求。

Topographic correction reduces terrain effects and improves radiation quality, such enhancements are beneficial to the improvement of computer classification accuracy in mountains. This paper designed a topographic correction method specifically for the global land cover classification of Landsat TM/ETM+ data. This method first smoothens the terrain slope angle and then performs a cosine correction based on the smoothed terrain. This method effectively removes radiation abnormalities in the results of Cosine correction (COS) and other Lambertian-based methods. This method and several other common methods, such as C, SCS, and Minnaert corrections, were tested on six Landsat TM/ETM+ data from different regions of the world with different land cover types. Visual and statistical analyses were conducted to assess the performance of these methods in terms of two aspects: overcorrection and homogeneity within a land cover class. Comparison results indicate that the smoothing step significantly reduces overcorrection. This simple and effective algorithm avoid complex atmospheric parameter estimations. Engineering requirements of terrain correction for global land cover classification are met.