The paper introduces a method of digital mapping of spatial distribution of soil typological units. It implements fuzzy k-means to classify the soil profile data (study area from the Považský Inovec Mountains, Slovakia) and regression-kriging with the selected digital terrain and remote sensing data to draw membership maps of soil typological units. Totally three soil typological units were identified: Haplic Cambisols (Skeletic, Dystric), Albic Stagnic Luvisols, and Haplic Stagnosols (Albic, Dystric). We analysed the membership values to these units with respect to terrain and remote sensing data. The membership values appeared as spatially smoothly dependant on the terrain gradients (linearly or exponentially) whereas the residua showed spatial autocorrelation. Based on regression and kriging analyses, the regression-kriging model was successfully deployed to draw raster membership maps. These maps yield coefficients of determination between R<sup>2</sup> = 56% (Albic Stagnic Luvisols) to R<sup>2</sup>= 79% (Haplic Cambisols (Skeletic, Dystric)) when evaluated by cross validation. The grid-based continuous soil map represents an alternative to the classical polygon soil maps and can offer a wide range of interpretations for landscape studies.
Geographically Weighted Area-to-Point Regression Kriging for Spatial Downscaling in Remote Sensing
