Een landsdekkend beeld van veranderingen in de Nederlandse grondwaterkwaliteit op 5 tot 17 meter diepte

Abstract

For 12 groundwater quality variables statistical maps of temporal changes were made, focusing on the long-term components of temporal changes. For each monitoring well screen, short-term predictions were calculated from time series of yearly measurements over 5-7 years, by using a simple regression model. Within each soil-land use category these predictions were interpolated spatially, taking location specific prediction accuracies, local level, and the size and structure of spatial variation into account. Spatial estimates concern 4 km x 4 km block median concentrations. For two extrapolation times (1980 and 2000) estimates of block median values are shown in maps as 95% confidence intervals, related to four levels that were derived from critical levels for human consumption. The resulting maps show a wealth of trend indications. Primarily, we can conclude that a trend approach with such a high spatial differentiation is attainable and advisable. Advisable because cells with a pronounced trend can be distinguished from a neighbourhood with less pronounced trends, and because trends in adjacent or nearby cells (also within the same soil-land use category) can have opposite directions. From a policy point of view the maps can be used directly for the assessment of diffuse groundwater contamination (the measurements that were used for the maps represent only non-location specific contamination situations). More indirectly they can be used for the validation of regional and national groundwater quality models, that are used for policy development. More generally, the presented methods are relevant for other compartments and for environmental outlooks.