Methodology to determine the soil sampling grid for precision agriculture in a coffee field

Abstract

The objective of this study was to develop and propose a methodology to evaluate the quality of different sampling grids. In addition, it allows us to choose the sampling grid that better suits one or a set of variables. The structure and magnitude of the spatial dependence were characterized by semivariogram. It allowed us to apply validation techniques as a base to create an index for evaluating the grid quality, and to develop an indicator that points out the best sampling grid. To test the proposed methodology, an experiment was performed at the Brejão farm, in Brazil. We have developed and compared twenty sampling grids, which were applied to four soil variables sampled in georeferenced locations. An accuracy index (AI), a precision index (PI) and the optimum grid indicator (OGI) were developed and proposed, which allowed us to choose the best grid (grid 5) among the sampling grids.