How many soil samples are neccessary to obtain a reliable estimate of mean nitrate concentrations in an agricultural field?

Abstract

In Germany, field-average soil NO3- measurements are used to identify agricultural risks of groundwater pollution and to evaluate the compliance of land users with environmental regulations. In the present study, it was tested at three typical agricultural sites if common practices of soil NO3- sampling were appropriate to obtain reliable estimates of the mean field NO3- content in fall. Three sites of 1 ha area were extensively sampled for NO3-. Ordinary statisticial analyses were conducted on the NO3- data in order to calculate how many samples per ha were required to predict the mean field NO3- with a given accuracy. Variograms were derived to assess applicability and relevance of ordinary statistical methods for heterogeneous fields. Results from statistical analysis suggest that at two sites common practices of soil NO3- sampling would have been adequate to obtain estimates of the actual mean field NO3- content with a sampling error less than 10kg NO3--N ha-1 at a 95% probability level. At the other site, common practices obviously would have failed because NO3- contents varied much more spatially. It remains a problem of soil sampling for NO3- analysis that information on field heterogeneity is frequently not available a priori.