Soil organic carbon (SOC) is the regulatory soil property for soil fertility. The SOC status and its variability pattern can be studied through spatial interpolation techniques. In the current study we compared deterministic (Inverse Distance Weightage, IDW), geostatistical (spherical and exponential kriging (OK) and Empirical Bayesian Kriging, EBK) and Machine Learning (Random Forest, RF, Support Vector Machine, SVM) method for samples collected at four grid spacings (20, 40, 60 and 80 m) to find out the combination of best interpolation method and sample spacing to produce a variability map for SOC. Geostatistical models with corresponding highest R² (31.9%), Lin CCC (0.49) and Pearson Correlation Coefficient (0.57) performed better than deterministic and machine learning models. A more precise prediction technique of geostatistical methods than deterministic interpolation may be due to differences in weightage calculation technique and a higher biasness in terms of Mean Error associated with IDW as compared to EBK. The presence of higher local uncertainty and lack of auxiliary supporting parameters e.g., crop and nutrient management) made ML (R² 19.4%, 21.7% in case of RF and SVM respectively) to be a poor predictor than EBK. Among the six interpolation methods, EBK was found to be the best interpolation method in each sample grid spacing [31.9% (p < 0.001), 10.5% (p < 0.01), 13.5% (p < 0.01), 10.5% (p < 0.05)]. Better simulation technique of variogram generation through EBK makes it the best fit model among the three geostatistical techniques. A 20 m grid spacing was found to be the best minimum spacing for studying SOC at a small scale as a higher density sampling could better capture the variability pattern of a heterogenous field having moderate autocorrelation. High local variability and lack of covariates have resulted in poor performance of ML as compared to kriging. However, the prediction performance of models can be improved by using covariate data which has a high correlation with soil properties combined with intense sampling in regions of high variability.

土壤有机碳 (SOC) 是调节土壤肥力的土壤特性。可以通过空间插值技术研究 SOC 状态及其变化模式。在当前的研究中，我们比较了确定性（反距离权重，IDW）、地统计（球面和指数克里金法 (OK) 和经验贝叶斯克里金法，EBK）和机器学习（随机森林、RF、支持向量机、SVM）方法收集的样本在四个网格间距（20、40、60 和 80 m）处，找出最佳插值方法和样本间距的组合，以生成 SOC 的可变性图。具有相应最高 R ^{2 的}地统计模型(31.9%)、Lin CCC (0.49) 和 Pearson 相关系数 (0.57) 的表现优于确定性和机器学习模型。地统计方法比确定性插值更精确的预测技术可能是由于权重计算技术的差异以及与 EBK 相比与 IDW 相关的平均误差方面的更高偏差。较高的局部不确定性和缺乏辅助支持参数（例如作物和养分管理）的存在使 ML（在 RF 和 SVM 的情况下分别为R ² 19.4%、21.7%）成为比 EBK 更差的预测指标。在六种插值方法中，EBK被发现是每个样本网格间距中最好的插值方法[31.9% ( p < 0.001 ), 10.5% ( p < 0.01 ), 13.5% (p < 0.01 ), 10.5% ( p < 0.05 )]。通过 EBK 生成变异函数的更好模拟技术使其成为三种地质统计技术中最适合的模型。发现 20 m 网格间距是小规模研究 SOC 的最佳最小间距，因为更高密度的采样可以更好地捕获具有中等自相关的异质场的变异模式。与克里金法相比，高局部变异性和缺乏协变量导致 ML 的性能较差。然而，模型的预测性能可以通过使用与土壤特性具有高度相关性的协变量数据结合高变异区域的密集采样来提高。