中文摘要：

地表反射率对定量遥感有重要意义,其在复杂地形区会受到地形条件影响。地形校正可以减小或消除山区遥感影像中地形因素影响。以梵净山区为研究区,利用Landsat-5 TM影像、DEM及坡度、坡向数据,采用10种地形校正模型进行比较实验,以发现其在反演地表反射率时存在的问题,并为选择适合地形复杂山区地表反射率反演的地形校正模型及相关研究提供参考。结果显示：1.SCS＋C、Minnaert、b、Teillet-回归、VECA、C模型地形校正效果较好,可用于反演山区的地表反射率,但是当坡度为0°时,Minnaert、Teillet-回归、VECA模型存在问题；2.Cosine-b、Cosine、SCS和Cosine-C模型存在过度校正现象,Cosine和SCS模型甚至可能出现反射率大于1的异常值,不宜选择；3.在通过采样获得经验参数的情况下,Minnaert模型对样点选取较为敏感,而SCS＋C和b模型对样点选取的敏感性较低。在较大范围的地形复杂山区反演地表反射率时,SCS＋C和b模型更合适。此外,从DEM数据、大气校正模型、地表覆被类型方面分别探讨了其对地表反射率的影响并提出了改进建议。

英文摘要：

Important to quantitative remote sensing,surface reflectance is affected by complicated terrain conditions. Topographic correction can reduce or eliminate the terrain effects of image in mountainous area. By taking Fanjing Mountain of Guizhou,China as study area,using Landsat- 5 TM image,DEM,slope and aspect data,and applying 10 kinds of topographic correction models to make comparison in research area,this paper aimed to find problems existing in the application,select suitable model for surface reflectance calculating in rugged terrain area,and provide reference to related research. Results shows that：（ 1） SCS ＋ C,Minnaert,b,C,Teillet-regression,VECA and C model,with good topographic correction effects,can be used in surface reflectance calculating in mountainous area,while Minnaert,Teillet-regression and VECA model may go wrong where slope is 0°.（ 2） Since Cosine- b,Cosine,SCS and Cosine- C model may lead to overcorrection,and results of Cosine and SCS model may even be greater than 1,they are not appropriate to choose.（ 3） If achieving the empirical parameters by sampling,Minnaert model showed higher sensitivity of sample points,while SCS ＋ C and b model lower. The two are more suitable for surface reflectance calculating in large rugged terrain area. Moreover,from DEM data,atmospheric correction model,land cover types,its influence on surface reflectance were discussed respectively and improved suggestions were put forward.