Spatial interpolation methods of determination of digital elevation models for forestry planning

Abstract

Topographic data acquisition methods are generally subject to measurement errors and subsequent digital terrain models (DTM) interpolation models can propagate these errors. For the forestry sector, the use of the DTM facilitates the planning stage in the road construction and mechanization phase in mountainous areas, especially in subsoiling, pesticide application and logging operations. To evaluate the results of interpolation methods, it is very common to use indicators such as the multiple determination coefficient and the residual error. This study aimed to compare and choose the best interpolation method in an elevation dataset to construct a DTM by applying the Taylor diagram to graphically analyze the results. Seventeen different methods of spatial interpolation were tested. The Spline method was selected as the best model tested over geostatistical models, most commonly adopted in spatial variability assays. The statistics of all methods were very similar, with slight variations, with the square mean square root error and the Spline method correlation being closer to the observed data, as easily shown by the Taylor diagram.