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.摘要：了解土壤属性的空间变异特征，可为土壤管理提供科学依据。本研究旨在结合模糊逻辑与地统计学方法，明确橡胶园种植土壤的肥力空间变异特征。研究采用分层随机采样网格法采集土壤样品，最短采样间距为6米，分别采集0~0.20米与0.20~0.40米土层的样品，共计设置60个采样点。本次测定的化学属性包括磷（P，Phosphorus）、钾（K，Potassium）、钙（Ca，Calcium）、镁（Mg，Magnesium）、盐基饱和度（BS，Base Saturation）、阳离子交换量（CEC，Cation Exchange Capacity）及V。通过分析各属性的空间依赖性，采用普通克里金（Ordinary Kriging）插值法构建空间分布图。基于预先设定的上下限值，利用递增型隶属度函数模型将插值结果转换为隶属度（FI，Fuzzy Index）。两个土层中的磷（P）含量均低于推荐下限（<20 mg·dm⁻³）。推理规则表明，该区域土壤肥力模糊图覆盖的总面积中，FI<0.50的区域占比超过50%，因此需对该区域施加土壤改良剂与肥料。该方法减少了该区域土壤肥力解译所需的图件数量，可直观展现该区域肥力需求的空间渐变变异特征。