Optimization of interpolation parameters based on statistical experiment

Abstract

Abstract Interpolation parameters are the basic elements of an interpolation algorithm in terrain environmental modeling. Different interpolation parameters produce different interpolation precisions, and the differences can be substantial. They are divided into deterministic parameters and uncertainty parameters. The optimization of uncertainty interpolation parameters is studied in this article. First, the methods of cross-validation, correlation analysis, and trend surface analysis are systematically used to identify and determine the optimal interval of interpolation parameters, which is helpful to solve the “black box” problem of parameter selection. Second, the significance degree of different parameters on the interpolation accuracy is given by using analysis of variance, which indicated that the high influence parameters are more helpful to improve the accuracy. This is an innovative conclusion.