The aim of this study was to use the georeferencing and geostatistics techniques to evaluate the spatial variability of coffee fruit detachment force by semivariograms adjustments and kriging interpolation. The study was conducted in Três Pontas, MG, Brazil. The detachment force of green and mature coffee fruit was obtained throughout a prototype dynamometer on georeferenced locations. The spatial dependence data was evaluated by classical and robust semivariogram adjustments, using ordinary least square and weighted least squares methods. Maximum likelihood and restricted maximum likelihood methods were also evaluated, but only for the classical semivariogram. Spherical, exponential and Gaussian models were compared for all the evaluated methods of semivariogram estimation. The isoline maps obtained by kriging were generated based on the best adjustment method and model of the semivariogram function obtained by the statistical validation. The studied variables showed spatial dependence, which were modeled by semivariograms that allowed plot isoline maps of the spatial distribution obtained by kriging. It was possible to identify the best places to start the mechanical and selective coffee fruit harvest.
sun exposure side; precision agriculture; geostatistics; selective and mechanized harvest; coffee
