Assessment of Non-Linear Interpolators to Construct Digital Elevation Models

摘要

Digital elevation models are numerical data structures that represent spatial elevation distribution over the land surface. Characterization of complex land topographic features traditionally has been performed from the Triangular Irregular Network (TIN) interpolator developed by obeying a linear function while the geometry of nature does not. This researches ought to evaluate the quality of the interpolation of distinct non-linear algorithms through the cross-validation technique without ignoring the results of investigations that have worked profoundly on the structure (TIN) during the last 30 years with good products, like the hybrid structures between raster and break lines. The results revealed that the minimum curvature interpolator (SPL) presented more fidelity of the surface?s topographic features. In the density distribution analysis of interpolation errors, it was noted that these do not fit a gaussian distribution, rather a logistics distribution.