Brooklyn College of The City University of New York, Department of Earth and Environmental Sciences, 2900 Bedford Avenue, Brooklyn, NY 11210, USA; Graduate Center of The City University of New York, PhD Program in Earth and Environmental Sciences, 365 5(th) Avenue, New York, NY 10016, USA; RUDN University, Agrarian-Technological Institute, Miklukho-Maklaya Street, 6, Moscow 117198, Russian Federation; School of Geosciences, University of Louisiana, Lafayette, LA 70504, USA.
RUDN University, Agrarian-Technological Institute, Miklukho-Maklaya Street, 6, Moscow 117198, Russian Federation.
Sci Total Environ. 2022 Feb 25;809:151107. doi: 10.1016/j.scitotenv.2021.151107. Epub 2021 Oct 21.
The successful use of visible and near-infrared (Vis-NIR) reflectance spectroscopy analysis requires selecting an optimal procedure of data acquisition and an accurate modeling approach. In this study, Vis-NIR with 350-2500 nm wavelengths were applied to detect different forms of lead (Pb) through the spectrally active soil constituents combining principal component regression (PCR) and Partial least-square regression (PLSR) for the Vis-NIR model calibration. Three clouds with different soil spectral properties were divided by the Linear discriminant analysis (LDA) in categories of Pb contamination risks: "low," "health," "ecological," ranging from 200 to 750 mg kg. Farm soils were used for calibration (n = 26), and more polluted garden soils (n = 36) from New York City were used for validation. Total and bioaccessible Pb concentrations were examined with PLSR models and compared with Support Vector Machine (SVM) Regression and Boosting Regression Tree (BRT) models. Performances of all models' predictions were qualitatively evaluated by the Root Mean Square Error (RMSE), Residual Prediction Deviation (RPD), and coefficient of determination (R). For total Pb, the best predictive models were obtained with BRT (R = 0.82 and RMSE 341.80 mg kg) followed by SVM (validation, R = 0.77 and RMSE 337.96 mg kg), and lastly by PLSR (validation, R = 0.74 and RMSE 499.04 mg kg). The PLSR technique is the most accurate calibration model for bioaccessible Pb with an R value of 0.91 and RMSE of 68.27 mg kg. The regression analysis indicated that bioaccessible Pb is strongly influenced by organic content, and to a lesser extent, by Fe concentrations. Although PLSR obtained lower accuracy, the model selected many characteristic bands and, thus, provided accurate approach for Pb pollution monitoring.
成功应用可见近红外（Vis-NIR）反射光谱分析需要选择最佳的数据采集程序和准确的建模方法。在这项研究中，采用 350-2500nm 波长的 Vis-NIR，通过与主成分回归（PCR）和偏最小二乘回归（PLSR）相结合的光谱活性土壤成分来检测不同形态的铅（Pb），用于 Vis-NIR 模型校准。通过线性判别分析（LDA）将三个具有不同土壤光谱特性的云分类为 Pb 污染风险类别：“低”、“健康”、“生态”，范围从 200 到 750mgkg。使用校准农场土壤（n=26），并使用来自纽约市的污染更严重的花园土壤（n=36）进行验证。使用 PLSR 模型检查总 Pb 和生物可利用 Pb 浓度，并与支持向量机（SVM）回归和提升回归树（BRT）模型进行比较。通过均方根误差（RMSE）、剩余预测偏差（RPD）和决定系数（R）对所有模型预测的性能进行定性评估。对于总 Pb，最好的预测模型是使用 BRT（R=0.82，RMSE 为 341.80mgkg），其次是 SVM（验证，R=0.77，RMSE 为 337.96mgkg），最后是 PLSR（验证，R=0.74，RMSE 为 499.04mgkg）。PLSR 技术是生物可利用 Pb 最准确的校准模型，R 值为 0.91，RMSE 为 68.27mgkg。回归分析表明，生物可利用 Pb 强烈受有机含量的影响，其次受 Fe 浓度的影响。尽管 PLSR 获得的准确性较低，但该模型选择了许多特征波段，因此为 Pb 污染监测提供了准确的方法。
