不同地貌类型区DEM空间内插算法精度评价

Accuracy Assessment of DEM Interpolation Algorithms in Different Landform Regions

数字高程模型(DEM)是数字地貌形态表达最为常用的基础数据,在工程建设规划设计、地貌定量分析、流域水文和土壤侵蚀模拟分析、土地利用、数字地质、遥感图像辅助分类等方面具有重要的用途.通过数字化等高线空间内插生成DEM是当前的主要技术手段之一,不同的空间内插算法和不同的地貌区域应用对DEM生成的质量和精度有重要的影响.本文选择了丘陵、山地和高山地3种地貌类型区,利用国际上流行的DEM专业化插值ANUDEM方法及另外5种典型常用的空间插值方法生成DEM,然后通过任意点法、等高线回放法及三维可视化叠加分析方法对生成的DEM进行了精度对比分析.结果表明,3种地貌区ANUDEM算法生成的DEM精度最高,IDW算法其次,THIESSEN算法、Kriging算法和自然邻域法精度相似,样条函数法精度最低.同一种空间插值方法,在地形起伏越大的地区,生成的DEM精度越低.

As one of the most important basic data for digital topographic expression, digital elevation model （DEM）has wide range of applications. DEM spatial interpolation methods based on contour digitization are commonly utilized. Different space interpolation algorithms and landform regions play an important influence on the accuracy of DEM generation. In the paper, six spatial interpolation algorithms, including an internationally popular ANUDEM method and five other commonly used interpolation methods, were applied in three different landform regions, that was, hills, mountains and alpine area. Comparative analysis and quality evaluation were carried out using random point check, comparasion between derived contours with original visualization overlay analysis etc. Results showed that the accuracies of DEMs ANUDEM were the highest. IDW method ranked second. Thiessen, Kriging ones, and 3D generated by and natural neibourhood methods had similar accuracy, and the spline-function method was the worest. For a specific interpolation method, the greater the terrain undulates, the lower the accuracy of the generated DEM is.