Papers

Summary

The spatial characterization of Fe oxides (hematite and goethite content) has usually been made by ordinary kriging (OK) considering the variogram parameters. However, OK softens local details of the spatial variation, overestimating small values and underestimating high ones. Thus, Trans-Gaussian Kriging (TGK) becomes an alternative to have a robust estimation of the variogram, reducing outlier effects. The objective of this study was to evaluate OK and TGK algorithm performances in estimating and mapping goethite and hematite iron oxides in two hillslope curvatures on an Alfisol in Catanduva, São Paulo State, Brazil. Two sampling areas were selected, one concave landscape and another convex landscape. Then, over each area, a 1-ha sample grid with regular spacing of 10 × 10 m, totaling 121 sample points of soil per area, was selected. The mineralogical analysis was performed in each sample to determine hematite and goethite contents. Moreover, to meet TGK criteria, data were previously converted to standard normal transformation, whereas OK data were not transformed. The TGK estimates presented improved accuracy mapping from 0.84 to 11.1% for the Gt and from 8.23 to 0.76% for the Hm content in concave and convex hillslope curvature, respectively. In general, the TGK estimates reproduced the best results. Moreover, the conditional cumulative distribution function and experimental variogram were better reproduced by TGK estimates than OK. The TGK is recommended for estimation of a more stable robust variogram in Fe oxide mapping with strong variability, when higher efficiency and accuracy are required. Hillslope curvatures influenced the interpolation efficiency and accuracy of interpolation. Relief classification is as much important as the variogram modeling for a greater efficiency and it would improve digital modeling of Fe oxides. The OK maps for Fe oxides should be cautiously used due to its uncertainty, especially in different hillslope curvatures mappings.