Towards a Soil Information System with quantified accuracy : a prototype for mapping continuous soil properties

This report describes the potential and functionality of software for spatial analysis, prediction and stochastic simulation of continuous soil properties using data from the Dutch Soil Information System (BIS). A geostatistical framework and R codes were developed. The geostatistical model of a soil property has a deterministic component representing the mean value within a soil category, and a stochastic component of standardized residuals. The standardized residuals are interpolated or simulated based on the simple kriging system. The software was tested in four case studies: exchangeable soil pH, clay content, organic matter content and Mean Spring Water table depth (MSW). It is concluded that the geostatistical framework and R codes developed in this study enable to predict values of continuous soil properties spatially, and to quantify the inaccuracy of these predictions. The inaccuracy of a spatial prediction at a certain location is quantified by the kriging variance, which can be interpreted as an indication of the uncertainty about the true value.