The definition of the distance between sampling grid points directly impacts the development of fertility maps because it affects the spatial dependence of geostatistics and the estimates for locations not sampled in the interpolation. Based on geostatistical concepts, it is common to recommend one or more soil samples per hectare. However, there is a need to understand how a cost-effective grid-sampling scheme can be developed to produce accurate digital maps. This study aimed to assess how + 5% and + 10% additional random points in the original sample grids affect the development of soil fertility maps under Brazilian Cerrado conditions. Four agricultural areas located in different states of Brazil were analyzed by applying the various sampling techniques. In total, 625 points were sampled, and ten sub-samples within 5 m of the central point were collected. Additional sampling points (+ 5% and + 10%) were randomly placed in the four quadrants of the field boundary following the original grid scheme. The soil attributes evaluated were pH, cation exchange capacity, base saturation, and Ca, Mg, K, and P contents. Geostatistical variograms were modeled for each field, attribute, and statistical sampling treatment. The best variogram model was selected based on the minimal difference between the root mean square error and average standard error of the cross-validation procedure, along with an evaluation of the root mean square standardized error value. The maps were compared using the relative deviation coefficients. The inclusion of additional 5% and 10% sampling points in the original grid was found to be effective in generating soil fertility maps, resulting in improved root mean square and average standard error values.

采样网格点之间距离的定义直接影响生育率地图的开发，因为它影响地质统计学的空间依赖性和插值中未采样位置的估计。根据地统计学概念，通常建议每公顷采集一个或多个土壤样本。但是，需要了解如何开发具有成本效益的网格采样方案来生成准确的数字地图。本研究旨在评估原始样本网格中 + 5% 和 + 10% 的额外随机点如何影响巴西塞拉多条件下土壤肥力图的开发。通过应用各种采样技术，对位于巴西不同州的四个农业区进行了分析。总共采样了625个点，并采集中心点 5 m 以内的 10 个子样本。额外的采样点（+ 5% 和 + 10%）按照原始网格方案随机放置在场地边界的四个象限中。评估的土壤属性包括 pH 值、阳离子交换容量、碱饱和度以及 Ca、Mg、K 和 P 含量。针对每个字段、属性和统计抽样处理对地统计变异函数进行建模。根据交叉验证程序的均方根误差和平均标准误差之间的最小差异，以及对均方根标准化误差值的评估，选择最佳变差函数模型。使用相对偏差系数比较地图。