The objective of this work was to evaluate the influence of the sampling density (three grids) and of the type of interpolation (kriging, inverse of the distance, inverse of the square distance, and polynomial) in the elaboration of thematic maps of soybean yield in an area of Dystroferric Red Nitosol under precision agriculture from 1998 to 2002. The selected grids had 128, 64 and 32 units using stratified systematic aligned sampling (SSAS). The thematic maps was generated from each sampling grid and using each of the interpolators. The relative deviation coefficient (CDR) was proposed to express the average difference in module of the values interpolated in each grid in relation to the considered standard SSAS grid with 128 units and kriging interpolator. The CDR proved efficient in the comparison of maps and made it possible to conclude that to effectively use the entire potential of the kriging, it is necessary to have several points, so that a good semivariogram is constructed. Otherwise, any of the other analyzed interpolators can be used. The SSAS 128 inverse distance method was most similar to the reference, SSAS 128 Kriging.
precision agriculture; geostatistics; sampling
