College of Engineering, Peking University, Beijing 100871, China.
Guangdong Provincial Key Laboratory of Soil and Groundwater Pollution Control, School of Environmental Science and Engineering, Southern University of Science and Technology, Shenzhen 518055, China.
Environ Sci Technol. 2020 Aug 4;54(15):9454-9463. doi: 10.1021/acs.est.0c03617. Epub 2020 Jul 23.
Recent advances in machine learning methods offer the opportunity to improve risk assessment and to decipher factors influencing the spatial variability of groundwater arsenic ([As]). A systematic comparison reveals that boosted regression trees (BRT) and random forest (RF) outperform logistic regression. The probability of [As] exceeding 5 μg/L (approximate median value of Bangladesh [As]), 10 μg/L (WHO provisional guideline value), and 50 μg/L (Bangladesh drinking water standard) is modeled by BRT and RF methods for Bangladesh and its four subregions demarcated by major rivers. Of the 109 geo-environmental and hydrochemical predictor variables, phosphorus and iron emerge as the most important across spatial scales, consistent with known As mobilization mechanisms. Well depth is significant only when hydrochemical parameters are not considered, consistent with prior studies. A peak of probability of [As] exceedance at ∼30 m depth is evident in the partial dependence plots (PDPs) for spatial-parameter-only models but not in the equivalent all-parameter models, suggesting that sediment depositional history explains interdependent spatial patterns of groundwater As-P-Fe in Holocene aquifers. The South region exhibits a decrease of probability of [As] exceedance below 150 m depth in PDPs for spatial-parameter-only and all-parameter models, supporting that the deeper Pleistocene aquifer is a low-As water resource.
近年来，机器学习方法的进步为改进风险评估和解析影响地下水砷 ([As]) 空间变异性的因素提供了机会。系统比较表明，提升回归树 (BRT) 和随机森林 (RF) 优于逻辑回归。通过 BRT 和 RF 方法对孟加拉国及其四条由主要河流划定的子区域进行建模，模拟了 [As] 超过 5μg/L（孟加拉国 [As] 的近似中位数）、10μg/L（世界卫生组织暂定指导值）和 50μg/L（孟加拉国饮用水标准）的概率。在 109 个地质环境和水化学预测变量中，磷和铁在各个空间尺度上都是最重要的，这与已知的砷活化机制一致。只有在不考虑水化学参数时，水井深度才具有重要意义，这与先前的研究一致。在仅考虑空间参数的模型的部分依赖图 (PDP) 中，[As] 超标概率在约 30m 深度处出现峰值，但在等效的所有参数模型中则没有，这表明沉积历史解释了全新世含水层中地下水 As-P-Fe 的相互依存的空间模式。在仅考虑空间参数和所有参数的模型的 PDP 中，南部地区在 150m 以下深度的 [As] 超标概率下降，这支持了更深的更新世含水层是低砷水资源的观点。
