Regression kriging of nitrate levels in upper groundwater in Dutch sandy soils. An analysis at national and regional extents.

Abstract

Sample data concerning nitrate concentrations in groundwater, were collected on farmland and in nature reserves on sandy soils in The Netherlands. Using Regression Kriging modelling, a geostatistical approach that exploits both the spatial variation in the sampled variable itself, and environmental information collected from covariate maps for the target predictor, it is possible to predict groundwater quality maps for the sandy soil regions in The Netherlands, and quantify the uncertainty in accompanying maps. Maps were produced for four different sandy soil regions and three different years in 2007, 2008 and 2009. For a combination of the regions into a nationwide model for the three years maps were made as well. The most successful covariate to be found in the regression part was the groundwater table map. The differences between a regional approach and a combined nationwide approach were explored. The nationwide approach seemed to generate slightly better predictions in a more stable manner, although differences are not very pronounced.