The Evaluation of the Accuracy of Interpolation Methods in Crafting Maps of Physical and Hydro-Physical Soil Properties

Abstract

The goal of this study was the spatial processing and showcasing selected soil properties (available water capacity, total organic carbon content and the content of clay fraction <0.001 mm) in the Nitra River Basin (Slovakia) via the usage and the subsequent evaluation of the quality of applied interpolation methods (Spline, inverse distance weighting (IDW), Topo to Raster). The results showed the possibilities of “conversion” of point information obtained by field research as well as research in the laboratory into a spatial expression, thus providing at least relevant estimation of the soil properties even in localities not directly covered by soil sampling. Based on the evaluation and mutual comparison of the accuracy of the used interpolation methods (by using the so-called cross-validation and trust criteria), the most favorable results were achieved by the Spline method in the GRASS GIS environment, and in the ArcGIS environment. When comparing the measured and estimated values of given soil properties at control points, the interpolated values classified as very accurate up to accurate prevailed in the verification dataset. Qualitatively less favorable (but still acceptable) were the results obtained with Topo to Raster (ArcGIS) interpolation method. On the contrary, the Spline method in the ArcGIS environment turned out to be the least accurate. We assume that this is most likely not only a consequence of insufficient density of points (resources), but also an inappropriate implementation of the method into the ArcGIS environment.