Condutividade elétrica aparente do solo como variável adicional para a krigagem por deriva externa no mapeamento de atributos químicos e físicos do solo

Abstract

The measurement of apparent soil electrical conductivity (ECa) has been indicated in the literature as a suitable tool to understand the spatial variability of agricultural soils in a fast and practical way. This ECa information has the capacity to guide soil sampling, generate homogeneous zones, assist in interpolation methods, among other benefits. This tool, combined with multivariate statistics, has great potential for improving predictions of chemical and physical attributes in agricultural soils. This work was divided into two articles, the first aiming to compare the variability maps generated by two different equipment, namely CND1040 - Terram® (Falker Automação Agrícola Ltda.) and Veris® 3100 (Veris Technologies), which calculate ECa by the resistive method at similar depths (0 - 0.3 m). The study was conducted in a 77-hectare grain commercial farm located in Imbituva/PR, Brazil. The comparison between the generated maps was performed using Pearson's method (p < 0.05) and the Bivariate Moran's Index (Moran's II), and the results showed high spatial correlation regardless of the method used, indicating that there is no significant difference between the generated maps and both sensors have the same ability to identify soil variability at the study site. The second article aimed to use the geostatistical method of external drift kriging applied to chemical and physical elements of the soil, using a thematic map of (ECa) as an additional variable with the objective of improving the quality of interpolated maps of such elements through comparative analysis between the results of cross-validation of the elements independently, with cross-validation of the elements by external drift, using ECa as an additional variable. The study was conducted on a 135-hectare commercial farm located in Sidrolândia, Mato Grosso do Sul, Brazil. A resistive ECa sensor, model CND1040 - Terram® (Falker Automação Agrícola Ltda.), was used, and 38 soil samples were collected, from which the attributes clay, calcium, magnesium, organic matter, and CEC were selected for comparative analysis. The low spatial dependence of soil attributes did not allow the execution of the proposed method in this study, except for clay, which, even allowing variographic modeling, did not achieve the proposed result in improving the spatial structure of the generated maps.