Proposta e comparação de medidas de dependência espacial em geoestatística

Abstract

Geostatistics is a widely used technique to describe the spatial dependence of soil attributes. The semivariogram is the main tool used to describe such dependence and allows the interpretation of the spatial dependence regarding the distance between the samples using the nugget effect (C0), contribution (C1), threshold (C0 + C1), and range (a) parameters. Because the semivariogram is a visual tool for interpreting spatial dependence and indicating parameters, a measure that represents the information contained in the semivariogram in a brief and objective manner is required. In the literature, there are indices that are calculated from the semivariogram parameters and that allow the classification of spatial dependence into strong, moderate, and weak. Despite the existence of such indices, in some cases, it is impossible to find coherence between what is visually verified in the semivariogram and the classification of spatial dependence given by the indices. Thus, an alternative measure was developed to describe spatial dependence, considering the semivariogram parameters and half of the largest distance between the sampled points. A classification was also developed to measure spatial dependence. The data derived from the soil attributes were applied considering this new measure and its respective classification into weak, moderate, and strong spatial dependence. Finally, comparisons were made between the two indices found in the literature.