Mapping soil cation exchange capacity in a semiarid region through predictive models and covariates from remote sensing data. Rev Bras Cienc Solo. 2018;42:e0170183.

Brazilian semi-arid region is a recent frontier in the country for agribusiness. The objective of this study is to apply pedometric tools to zoning areas with distinct potential and limitations to agricultural purposes. The research was set in three main steps: (i) to compile a database with all complete profile data collection; (ii) to analyze the vertical variability of soil properties and select a set of soil key properties useful to define the land potential and limitations; and (iii) to classify the area according to potential for agriculture, considering a medium technological level of the farmers. The quantitative methods applied are supported by geographic information systems (GIS) and spatial statistics. The soil data compilation was based on legacy data, with corresponding topographical data and information from remote sensing images of the area. Tree-based and geostatistical algorithms were applied to predict the spatial variability of the soil key properties. The definition of management zones was based on Iso Cluster and Maximum Likelihood Classification tools. The results pointed three different management zones according to risks of salinization and requirements for irrigation control. The approach showed to be a simple and useful way to select and recommend primarily potential areas for agriculture based on soil properties.