中文摘要：

针对我国＂三大地形阶梯＂划分偏于定性描述,边界不够精确的问题,提出了面向对象的定量划分方法。以GTOPO30 DEM数据为基础,提取地形起伏度、光照晕渲及平均高程3因子组合成特征影像,结合海拔频率分布,利用遥感面向对象分类法进行我国地形三大阶梯的自动划分。实验所得分界线与传统的描述性边界吻合;与以单个栅格为基本单位进行定量划分的方法相比,该方法所得界线既兼顾了细节上的精细、合理,又具有地貌边界的光滑性,边界地貌实体的完整性,且后期编辑处理工作量小,减少了人为因素的干扰,提高了三大阶梯划分的自动化水平。实验结果为揭示我国地势总体的空间分异格局提供了数据基础。

英文摘要：

In view of the qualitative description and the lack of accurate boundary for China＇s ＂three terrain grades＂, the quantitative classification method is proposed. In this paper, based on the GTOPO30 DEM data, three topographical factors, which are local relief,hill shading and mean elevation, combined with altitude frequency distribution were utilized for automatic segmentation for China＇s ＂ three terrain grades＂ using object-oriented classification. The results show that the experimental boundary is mainly consistent with the traditional defined boundary. The new ＂three terrain grades＂ boundary based on objectoriented classification has more complete terrain body and finer boundary detail than results of traditional quantitative classification based on grids. Moreover,this method needs less human interference. Results of this paper provide basic data for revealing the spatial pattern of the whole terrain in China.