Papers

Summary

In precision agriculture it is necessary to assess the quality of the maps of soil attributes. This paper aimed to evaluate the performance of geostatistical methods: ordinary kriging and sequential Gaussian simulation on spatial prediction of mean diameter of the goethite crystal in a regular grid with 121 points collected in a grid of 1 ha with regular spacing of 10 to 10 m. After the particle size and concentration analysis of iron oxides the values of the mean diameter of the goethite crystal were calculated. These were analysed by descriptive statistics and geostatistics, ordinary kriging and sequential Gaussian simulation was used for production of maps. With the results which evaluated the method is more faithful to make statistics, the cumulative probability density function conditional epsilon statistics ε_y. E-Type estimates were similar to ordinary kriging, due to the minimization of the variance. However, the ordinary kriging did not perform well in specific locations for crystallinity of goethite, while the E-Type map by sequential Gaussian simulation should be used instead of kriging maps. The E-type maps should be prefered as they present better performance in modeling.