Fuzzy logic and geostatistics in studying the fertility of soil cultivated with the rubber tree

Abstract

Knowledge of the spatial variability of soil attributes is an aid to soil management. The aim of this study was to apply fuzzy logic and geostatistics in defining the spatial variability of soil fertility, in soil cultivated with the rubber tree. Soil samples from the 0-0.20 m and 0.20-0.40 m layers were collected in a stratified random sampling grid, with a shortest distance of 6 m, for a total of 60 points. The chemical attributes were P, K, Ca, Mg, BS, CEC and V. The spatial dependence of the attributes was determined, and the maps were constructed using ordinary kriging interpolation. The interpolated values were transformed into degrees of pertinence (FI), with the lower and upper limits previously defined, using an increasing model. The nutrient P displayed values in both layers below the recommended lower limit (< 20 mg dm-3). The rules of inference indicated that the total area shown in the fuzzy maps of soil fertility requires the application of correctives and fertiliser, as it has an FI < 0.50 and a percentage area > 50%. This methodology reduced the number of maps for interpreting soil fertility in the area, enabling visualisation of the spatial and gradual variability of the needs of the region.