| 摘要: |
| 【目的】以地形类型辨识和区分为目的，对比分析不同粗糙度算法的优缺点，为定量描述地形形态、划分地形类型提供新思路。【方法】基于SRTM3高程数据，使用常见的11种地表粗糙度算法（坡度、局地高差、高程均方差、高程均方根偏差、均方根坡度、绝对坡度、坡度标准差、面积比、矢量离差、余弦特征值、二维变异性）提取研究区的地表粗糙度，从空间格局特征和统计分布特征两方面对提取结果进行对比，并利用典型断面和频率曲线分析方法探讨各算法区分研究区地形类型的能力。【结果】坡度、局地高差、高程均方差、高程均方根偏差、均方根坡度和绝对坡度6种算法所得地表粗糙度的空间格局特征与统计分布特征极其相似，且能有效区分研究区的地形类型。坡度标准差、面积比、矢量离差、余弦特征值和二维变异性5种算法难以表达研究区地表粗糙度的宏观变异，不能有效区分研究区的地形类型。【结论】基于高程空间变异的坡度、局地高差、高程均方差、高程均方根偏差、均方根坡度、绝对坡度6种算法是表达地表粗糙程度和区分地形类型的较佳算法。 |
| 关键词: 地表粗糙度 提取算法 地形类型区分 空间格局 |
| DOI： |
| 分类号: |
| 基金项目:国家自然科学基金项目（41371274）；国家重点研究计划项目（2017YFD0800502） |
|
| Comparison of surface roughness algorithms for terrain type separation |
|
LI Yuru,YANG Qinke,WANG Chunmei,et al
|
| Abstract: |
| 【Objective】Aiming at distinguishing terrain types,this study quantitatively evaluated different algorithms to provide new ideas for terrain type separation.【Method】A total of 11 surface roughness algorithms (slope,local elevation range,root-mean-square of height,root-mean-square of height deviation,root-mean-square of slope,absolute slope,standard deviation of slope,area ratio,direction cosine eigenvalues,vector dispersion,and two-dimensional variability) were applied to the SRTM3 digital elevation data to extract the surface roughness of the study area.The results were analyzed from the spatial pattern and statistical distribution.Then,the ability of different methods to distinguish terrain types was analyzed by typical section analysis and distribution curve analysis.【Result】The slope,local elevation range,root-mean-square of height,root-mean-square of height deviation,root-mean-square of slope and absolute slope showed similar spatial pattern and statistical distribution characteristics,and they identified the terrain types effectively.The standard deviation of slope,area ratio,direction cosine eigenvalues,vector dispersion and two-dimensional variability methods cannot express the macroscopic variation of surface roughness and distinguish the terrain types effectively.【Conclusion】Elevation variation based algorithms,including slope,local elevation range,root-mean-square of height,root-mean-square of height deviation,root-mean-square of slope and absolute slope are better algorithms to represent surface roughness and distinguish terrain types. |
| Key words: surface roughness extraction algorithms terrain types separation spatial pattern |
