关键词: 插值参数, 搜索方式, 核函数, 交叉验证法, 相关分析, 趋势面分析, 方差分析

Abstract:

Interpolation parameter is one of the basic elements of the DEM interpolation method. Different interpolation parameters produce different interpolation precisions and, the differences may be huge by using various interpolation parameters. But for ordinary users, it is difficult to select an appropriate set of interpolation parameters during the interpolation process. In the related researches on interpolation method, using the appropriate interpolation algorithm or interpolation parameters may lead to potentially serious consequences, even completely opposite experimental conclusions. Therefore, based on the difference of the weight determination method in interpolation algorithm, this paper selected related parameters with inverse distance weighted interpolation method, radial basis functions interpolation method and ordinary kridging interpolation method, and focused on selecting "Optimized" interpolation parameters in interpolation method. First according to the different effects of the interpolation parameters on interpolation accuracy, took the relevant interpolation parameters as the experimental object. Then chosen six different regions of landform types of the sparse distribution and discrete sampling data as data source, and applied a series of methods such as cross validation, correlation analysis, trend surface analysis and variance of analysis to systematic research on choosing optimized DEM interpolation parameters and proposed the uncertainty of the interpolation parameters of the "optimum" value range.

Key words: Interpolation Parameter, Search Type, Kernel Function, Cross Validation, Correlate Analysis, Trend Surface Analysis, Analysis of Variance