Investigation of the robustness of different contour interpolation models for the generation of contour map and digital elevation models

Abstract

This research seeks to investigate the robustness of ten different methods for contour interpolation in Surfer10.0 software namely: Inverse distance to weighting power (IDP), Kriging (KRG), Minimum Curvature (MC), Nearest Neighbor (NENE), Polynomial Regression (PR), Radial Basis Function (RF), Modified Shepard’s Methods (MSM), Natural Neighbor (NANE), Triangulation with Linear Interpolation (TWLI) and Local Polynomial (LP). Hi Target Differential Global Positioning System (GPS) receiver was used for the spatial data (x, y, z) acquisition within the study area which was used for gridding and generation of contour maps and DEMs using each of the investigated interpolation method. From the surface models generated, twenty sample points were selected and their heights verified on the ground using a DGPS receiver. Comparative evaluation reveals that minimum curvature (236.4269m) and kriging (236.4674m) had the closest value to the observed value while MSM gave a value that is far higher than the observed height (236.8585m). Statistical investigations were carried out and the results also revealed that minimum curvature and kriging do have the best topographic representation of the study area, which is also supported by the low values of their RMSE of 0.2806 and 0.3109 respectively, while the modified shepard’s method produced the highest RMSE value of 0.9100.