Comparison of thin plate spline, polynomial, CI—function and Shepard’s interpolation techniques with GPS-derived DEM

Abstract

The digital elevation model (DEM) is an important part of mapping and is used for several purposes including orthoimage production, image interpretation, contours derivation, and several Geographic Information System (GIS) applications. However, without a sophisticated digital photogrammetric system, which could provide DEMs “automatically”, and without well-distributed control points, which are usually used in surveying, creating a DEM is not an easy task. Interpolation is often required to create a DEM from a sparse number of points. In this paper, the interpolation accuracy of several techniques, namely thin plate spline, polynomial, local CI-function and weighted-distance (Shepard’s) interpolation are tested for comparison using two Global Positioning System (GPS) derived DEMs. The results of two tests with five cases of control points, which are different in number and distribution, show that the Shepard’s technique is most efficient with respect to accuracy as well as surface representation, followed by the spline technique.