Research on the Robustness of Different Interpolation Methods Based on DEM Data

In this paper, 50% random sampling of DEM data with different error levels are used as input data to evaluate the interpolation and filtering effects of four commonly used interpolation methods. It can be seen from the analysis results that the minimum curvature method (MC) is the best one from the view of errors resistance, while the differences between the standard deviation of filtering and interpolation are unconspicuous among Kriging method, the radial basis function method (RBF) and the triangulation based on linear interpolation (TLI), and the interpolation effect of the minimum curvature method (MC) is slightly better than that of the filtering effect. By calculating the residuals of the measured data and interpolation values which using DEM 50% random sampling data with additional ±15 m error, it is found that the residuals derived from the minimum curvature method are the most concentrated distribution in the range of -15 m~+ 15 m. The comprehensive analysis shows that the minimum curvature method is the best robust data interpolation algorithm in the four methods.