A METHOD FOR OPTIMAL SITE-SPECIFIC NITROGEN FERTILISATION V. HAAHR, R. N. JØRGENSEN, A. JENSEN National Laboratory Risø, 4000 Roskilde, Denmark J. OVERGAARD Thoustrup & Overgaard a/s, 8900 Randers, Denmark E-mail address: vagner.haahr@risoe.dk ABSTRACT Yield mapping of crops obtained by harvest with a combine equipped with yield metering device and GPS positioning system, has in recent years, become an important tool in crop management. With respect to nitrogen fertilisation, it is now technically possible to change practices from uniform application to differentiated, site-specific application. The presentation introduces and discusses a method for determining the spatial variation in yield response to nitrogen fertilisation, based on large scale field experiments in 1997-98. The spatial variation in yield response, based on two amounts of uniformly applied nitrogen fertiliser is suggested for generating application maps for optimal site- specific nitrogen fertilisation. The Optimal method for predicting fertilisation rates seems to give a better background for generating application maps than predictions based on yield maps and the principle of replacement of removed nitrogen. The presented results show that the effect of spatially varied nitrogen fertilisation is reduced at increasing nitrogen input levels, and that this effect increases with decreasing nitrogen input levels. INTRODUCTION Yield maps are only to some extent correlated with the predicted yield in the following year (Lamb et al. 1997; Moore 1998), but at present yield mapping contributes as the most important basic information for the prediction of spatial variation in yield. For spatial nitrogen fertiliser application, yield maps for one or, optimally, several years, are important as background information for the generation of a fertiliser application map. From an environmental point of view, it seems correct to vary the amount of nitrogen fertiliser in relation to the predicted yield (Verhagen et al. 1995) but this will not appeal to the farmer unless he has economic gain from spatial application of nitrogen fertiliser. During resent years uniform application of nitrogen has been compared with varied application using predicted yield. The spatial application of nitrogen was based on information about yield and the principle of replacement. Thus high yield predictions