Dissertations and Theses

Geostatistical methods in the spatial characterization of iron oxides in different pedoforms.

Author: João Fernandes da Silva Junior

Keywords: goethite, hematite, ordinary kriging, stochastic simulation, accuracy.

Link

Summary

Iron oxides are important pedoenvironmental indicators and have great influence on the physical and chemical attributes of the soil. However, for the better understanding of its formation and occurrence, studies of spatial variability are necessary. Geostatistics is the best tool for this study, and the Ordinary Kriging (KO) is a least squares interpolator, most commonly used in soil science. However, it softens the local details of spatial variation, overestimating small values ​​and underestimating high values. Therefore, it proves to be inadequate in evaluating the true spatial variability of iron oxides. Although kriging shows the best estimate of a Z attribute, it does not represent the variation. In this way, the alternative is geostatistical simulation that evaluates and quantifies the uncertainty of an attribute, in a more realistic way. The most used stochastic simulation technique is the Gaussian Sequence Simulation (SSG), because it is fast and direct in the construction of the conditional cumulative probability density function (f.d.p.a.c), and is therefore preferable to the KO estimation. Thus, it is important to test other methods for spatial characterization of iron oxides. In order to evaluate the performance of (KO), standard kriging of standardized data (KOP) and Gaussian sequential simulation (SSG) of the spatial variability of iron oxides in an area with concave and concave pedoforms, 121 samples of two meshes were collected in both reliefs classified as: concave and convex with spacing of 10 meters. The mean error, the mean square error, and the standardized mean square error were the criteria for the accuracy of the prediction. The high values ​​of the coefficients of variation of the iron oxides affected the performance of the methods. Iron oxides presented differentiated variability by SSG in the pedoforms. The KOP and E-type maps presented greater accuracy and greater detail regarding KO in the pedoforms. E-type maps are recommended instead of KO maps. SSG is preferable to KO for iron oxides. Besides the use of KOP, it can be another interpolation alternative in the evaluation of the spatial variability of iron oxides.