Space-time statistics for decision support to smart farming.

Abstract

This paper summarizes statistical procedures which are useful for precision farming at different scales. Three topics are addressed: spatial comparison of scenarios for land use, analysis of data in the space-time domain, and sampling in space and time. The first study compares six scenarios for nitrate leaching to ground water. Disjunctive cokriging reduces the computing time by 80% without loss of accuracy. The second study analyses wind erosion during four storms in a field in Niger measured with 21 devices. We investigated the use of temporal replicates to overcome the lack of spatial data. The third study analyses the effects of sampling in space and time for soil nutrient data in a Southwest African field. We concluded that statistical procedures are indispensable for decision support to smart farming.