Soil sampling and initial mapping process, when performed inefficiently, can lead to erroneous planning regarding the application of fertilizers. The objectives of this study were to establish a sampling density that provides accurate maps of the chemical attributes generated for different sampling densities in two depths, and to measure their practical implications in fertilization and liming for soybean on an Oxisol in the region of Guarapuava. Maps were compared for the attributes pH, H+Al, P, K, Ca, Mg, base saturation and the lime requirement, in the layers 0-10 and 0-20 cm, in the sampling densities of ¼, ½, 1, 2 and 4 ha, being the first density taken as a reference for comparison. The methods of comparison were the overall accuracy and Kappa index. The decrease in sample density depreciated the quality of the generated maps. All sampling densities differ as to the accuracy, however, showed good levels of accuracy for the sample density of one sample per hectare. Precision agriculture uses in a better way the amount of lime and fertilizer than conventional agriculture.
precision agriculture; Kappa; kriging
