Studies on soil fertility mapping based on geostatistics should consider some spatial variability characteristics such as spatial continuity and absence of anisotropy. The present study reports an anisotropic analysis of the spatial variability of the following soil chemical variables: iron (Fe), potential acidity (H + Al), organic matter (OM) and manganese (Mn) of a data set of 128 plots without local chemical management, spatially referenced, studied between 1998 and 2002 in a dystrophic Red Latosol (Oxisol), in Cascavel, Paraná, Brazil. Anisotropy was identified through the construction of directional semivariograms with adjusted models and corrected through linear transformations and by combined models. Thereafter a model was used adjusted to the omnidirectional semivariogram for the construction of thematic variability maps of the studied variables. The existence of geometric anisotropy was observed for the variable H + Al. The variables MO, Mn and Fe evidenced the presence of combined anisotropy. The geometric anisotropy was first corrected and then the zonal anisotropy was adjusted. Based on anisotropy it was possible to eliminate the biased directions, enhancing the semivariogram fitting and precision of the thematic maps of chemical variables.
geostatistics; spatial dependence; precision agriculture
