Department of Soil Science, School of Agricultural Engineering and Technology, University of Tehran, Karaj, Iran.
Environ Monit Assess. 2022 Sep 10;194(10):760. doi: 10.1007/s10661-022-10434-9.
Accuracy and uncertainty of models used for digital soil mapping are important for assessing confidence of predictions and reliable land use planning and management. In this study, two approaches of geostatistical (spatial) and machine learning (ML) models were evaluated for predictive mapping of soil calcium (Ca) and potassium (K). Two spatial models including empirical Bayesian kriging (EBK) and sequential Gaussian simulation (SGS) were compared with machine learning models: Cubist, random forest (RF) and support vector machine (SVM) in terms of their accuracy and uncertainty for mapping soil Ca and K. The study area is in Nowley, New South Wales, Australia, with an area of 2083 ha and a variety of soil types and farming systems. For the models training process, 240 soil samples data and for validation 102 independent samples data were used. For accuracy assessment R, root mean square error (RMSE), concordance and bias and for uncertainty assessment confidence limits were investigated. Also, in order to compare the outcomes for the two soil properties with different measurement units, mean absolute percentage error (MAPE) and relative uncertainty (RU) as accuracy and uncertainty measures, respectively, were evaluated. Results showed that for K map SGS had the highest R (0.74) and lowest RMSE (1.96), followed by EBK with R = 0.72 and RMSE = 2.02. For Ca map, EBK model showed the highest accuracy (R = 0.46; RMSE = 3.21), followed by SVM and SGS with comparable accuracies. Comparing the two soil properties, Ca map showed higher MAPE and RU, compared to K map. The lowest MAPE was obtained for EBK model (for K = 39) and SGS model (for K = 44). Also, the lowest RU values were found for EBK and SGS models. Among the ML models, SVM showed lower sensitivity to higher variance in data input. In general, the spatial models outperformed the ML models with regard to both accuracy and uncertainty. An additional conclusion is that considering the data variance in the two soil properties, geostatistical models with lower RU and MAPE were relatively less susceptible to data variance, compared to ML models.
用于数字土壤制图的模型的准确性和不确定性对于评估预测的置信度和可靠的土地利用规划和管理至关重要。在本研究中，评估了两种地质统计学（空间）和机器学习（ML）模型方法，用于土壤钙（Ca）和钾（K）的预测制图。将两种空间模型（经验贝叶斯克里金（EBK）和序贯高斯模拟（SGS））与机器学习模型（Cubist、随机森林（RF）和支持向量机（SVM））进行比较，比较了它们在土壤 Ca 和 K 制图中的准确性和不确定性。研究区域位于澳大利亚新南威尔士州的诺利（Nowley），面积为 2083 公顷，拥有多种土壤类型和耕作系统。对于模型训练过程，使用了 240 个土壤样本数据，对于验证使用了 102 个独立样本数据。为了评估准确性，研究了 R、均方根误差（RMSE）、一致性和偏差，为了评估不确定性，研究了置信限。此外，为了比较两种具有不同测量单位的土壤属性的结果，分别评估了平均绝对百分比误差（MAPE）和相对不确定性（RU）作为准确性和不确定性的度量。结果表明，对于 K 图，SGS 具有最高的 R（0.74）和最低的 RMSE（1.96），其次是 EBK，R 为 0.72，RMSE 为 2.02。对于 Ca 图，EBK 模型表现出最高的准确性（R=0.46；RMSE=3.21），其次是 SVM 和 SGS，准确性相当。比较两种土壤属性，Ca 图的 MAPE 和 RU 高于 K 图。EBK 模型（对于 K=39）和 SGS 模型（对于 K=44）获得了最低的 MAPE。此外，EBK 和 SGS 模型的 RU 值最低。在 ML 模型中，SVM 对数据输入的较高方差表现出较低的敏感性。总体而言，空间模型在准确性和不确定性方面均优于 ML 模型。另一个结论是，考虑到两种土壤属性的数据方差，与 ML 模型相比，具有较低 RU 和 MAPE 的地质统计模型相对不易受到数据方差的影响。
