Conclusions

• Different mapping techniques can give vastly different results: it is essential to consider which method to use when making maps.
• The best method (from cross-validation results) for us was cokriging log(nitrates) against one additional variable, then retransforming. More variables led to ill-conditioning.
• We would recommend kriging over cokriging, in general, unless the user is sufficiently adept at cross-variogram modelling and cross-validation.
• We have found the Public Domain GIS GRASS to be a useful tool in our effort to link land use to nitrate concentration.