监测自然衰减作为一种经济高效的地下水污染修复技术,其精准实施高度依赖于对污染羽演化的可靠预测。本研究针对传统数值模拟方法的效率瓶颈,构建了"正交实验-Flopy自动化建模-机器学习"三位一体的通用预测模型框架。通过情景设计与数值模拟批量生成200-31200组不同规模、多维时空演化数据集,利用这些数据集,结合不同数据集容量及模型训练参数,训练模型参数及数据集容量对静态预测（Multilayer Perceptron,MLP）与动态预测（Spatiotemporal-MLP,ST-MLP）模型性能的差异化影响,从而揭示了数据规模与模型架构的耦合作用机制：静态MLP模型在1000组数据量级即可实现0.97的污染范围预测精度,而动态ST-MLP模型需23400组数据达到0.991的时空解析精度。此外,研究发现时间节点采样频率优化（180天/次）可使ST-MLP的数据生成成本降低40%,并且维持精度大于0.98的技术指标。本模型体系具有潜在的应用价值,一方面可为监测自然衰减技术的可行性评估提供量化依据;另一方面,其污染羽时空演化快速预测能够突破传统风险评估的静态耗时局限,可有效建立地下水污染扩散风险的快捷动态预警机制。
As a cost-effective groundwater remediation technology, the precise implementation of Monitored Natural Attenuation critically relies on reliable predictions of contamination plume evolution. To address the efficiency limitations of traditional numerical simulation methods, this study developed a general-purpose predictive modeling framework integrating three key components: orthogonal experimental design, Flopy-based automated modeling, and machine learning. Through scenario design and batch numerical simulations, we generated 200-31,200 multidimensional spatiotemporal datasets representing groundwater contamination plumes of varying scales. Leveraging these datasets, we systematically investigated the differential impacts of dataset sizes and the model training parameters on the performance of the static prediction (Multilayer Perceptron,MLP) and the dynamic spatiotemporal prediction (Spatiotemporal-MLP, ST-MLP) models. Our findings elucidate the coupling mechanism between the data scale and the model architecture: the static MLP achieved a contamination extent prediction accuracy of 0.97 with 1,000 training samples, whereas the dynamic ST-MLP required 23,400 samples to attain a spatiotemporal resolution accuracy of 0.991. Furthermore, optimizing temporal node sampling frequency (every 180 days) reduced the data generation costs by 40% for ST-MLP while maintaining accuracy above 0.98. This framework demonstrates the dual application values: (1) providing quantitative benchmarks for MNA feasibility assessments, and (2) enabling rapid spatiotemporal plume predictions to overcome the static and time-consuming limitations of conventional risk assessments, thereby establishing an efficient dynamic early-warning mechanism for groundwater contamination risks.
[1] Wang W R, Jia J L, Zhang B, et al.A review of Sustained release materials for remediation of organically contaminated groundwater: Material preparation, applications and prospects for practical application[J]. Journal of Hazardous Materials Advances, 2024, 13:100393. DOI:10.1016/j.hazadv.2023.100393.
[2] Cundy A B, Bardos R P, Church A, et al.Developing principles of sustainability and stakeholder engagement for “gentle” remediation approaches: The European context[J]. Journal of Environmental Management, 2013, 129: 283-291. DOI: 10.1016/j.jenvman.2013.07.032
[3] Ravindiran G, Rajamanickam S, Sivarethinamohan S, et al.A review of the status, effects, prevention, and remediation of groundwater contamination for sustainable environment[J]. Water, 2023, 15(20): 3662. DOI: 10.3390/w15203662.
[4] 李元杰, 王森杰, 张敏, 等. 土壤和地下水污染的监控自然衰减修复技术研究进展[J]. 中国环境科学, 2018, 38(3): 1185-1193. DOI: 10.19674/j.cnki.issn1000-6923.2018.0141
[5] 沈晓芳. 大数据产业集聚及其对经济增长的影响研究[D]. 贵阳: 贵州大学, 2021. DOI: 10.27047/d.cnki.ggudu.2021.000987.
[6] McConnell L, Karimi Askarani K, Cognac K E, et al. Forecasting groundwater contaminant plume development using statistical and machine learning methods[J]. Groundwater Monitoring & Remediation, 2022, 42(3): 34-43. DOI: 10.1111/gwmr.12523.
[7] Deng H, Gharasoo M, Zhang L W, et al.A perspective on applied geochemistry in porous media: Reactive transport modeling of geochemical dynamics and the interplay with flow phenomena and physical alteration[J]. Applied Geochemistry, 2022, 146: 105445. DOI: 10.1016/j.apgeochem.2022.105445.
[8] 杨小芳, 王明玉, 王丽亚, 等. 永定河生态修复地下水位主控因素与数值模拟预测不确定性[J]. 中国科学院大学学报, 2015, 32(2): 192-199. DOI: 10.7523/j.issn.2095-6134.2015.02.007.
[9] 闫龑, 王明玉, 陈建平, 等. 场地地下水1, 2-二氯乙烷污染的修复实验与数值模拟研究[J]. 地球与环境, 2021, 49(3): 250-259. DOI: 10.14050/j.cnki.1672-9250.2020.48.115.
[10] Alshahri A H, Elbisy M S.Assessment of using artificial neural network and support vector machine techniques for predicting wave-overtopping discharges at coastal structures[J]. Journal of Marine Science and Engineering, 2023, 11(3): 539. DOI: 10.3390/jmse11030539.
[11] Davis S E, Cremaschi S, Eden M R.Efficient surrogate model development: Impact of sample size and underlying model dimensions[J]. Computer Aided Chemical Engineering, 2018, 44: 979-984. DOI: 10.1016/B978-0-444-64241-7.50158-0.
[12] Zou Y H, Yousaf M S, Yang F Q, et al.Surrogate-based uncertainty analysis for groundwater contaminant transport in a chromium residue site located in Southern China[J]. Water, 2024, 16(5): 638. DOI: 10.3390/w16050638.
[13] 王燕, 钟建, 张志远. 支持向量回归的机器学习方法在海浪预测中的应用[J]. 海洋预报, 2020,37(03):29-34. doi:10.11737/j.issn.1003-0239.2020.03.004.
[14] Gad M, Gaagai A, Eid M H, et al.Groundwater quality and health risk assessment using indexing approaches, multivariate statistical analysis, artificial neural networks, and GIS techniques in el kharga oasis, Egypt[J]. Water, 2023, 15(6): 1216. DOI: 10.3390/w15061216.
[15] 李子乐, 安永凯, 闫雪嫚. 耦合敏感性分析与两阶段马尔科夫链蒙特卡洛算法的地下水污染溯源辨识[J]. 地球科学与环境学报, 2024, 46(05):702-710.DOI:10.19814/j.jese.2024.05006.
[16] Papernot N, Abadi M,Erlingsson, et al. Semi-supervised knowledge transfer for deep learning from private training data[J]. ArXiv e-Prints, 2016: arXiv: 1610.05755. DOI: 10.48550/arXiv.1610.05755.
[17] Papernot N, McDaniel P, Goodfellow I, et al. Practical black-box attacks against deep learning systems using adversarial examples[EB/OL]. arXiv:1602.02697(2016-02-08)[2025-05-20]. https://doi.org/10.48550/arXiv.1602.02697.
[18] Mohammadi M, Jamshidi S, Rezvanian A, et al.Advanced fusion of MTM-LSTM and MLP models for time series forecasting: An application for forecasting the solar radiation[J]. Measurement: Sensors, 2024, 33: 101179. DOI: 10.1016/j.measen.2024.101179.
[19] Verma G, Kumar B.Multi-layer perceptron (MLP) neural network for predicting the modified compaction parameters of coarse-grained and fine-grained soils[J]. Innovative Infrastructure Solutions, 2021, 7(1): 78. DOI: 10.1007/s41062-021-00679-7.
[20] Fang Z, Liu Z G, Wang G W, et al.FloPy for optimizing the structure of hydraulic-driven groundwater circulation wells[J]. Earth Science Informatics, 2024, 18(1): 94. DOI: 10.1007/s12145-024-01568-0.
[21] Liu Y T, Wang W, Li J H, et al.A novel simulation-optimization model built by FloPy: Pollutant traceability in a chemical park in China[J]. Applied Sciences, 2023, 13(19): 10707. DOI: 10.3390/app131910707.
[22] 魏亚强, 陈坚, 张铎, 等. 基于Python的地下水模拟研究进展与应用[J]. 计算机技术与发展, 2021, 31(5): 150-156. DOI: 10.3969/j.issn.1673-629X.2021.05.026.
[23] 张鹏伟. 滹沱河对超采区地下水增补与水位控制优化模拟[D]. 北京: 中国地质科学院, 2022. DOI: 10.27744/d.cnki.gzgdk.2022.000028.
[24] 康燕楠, 降亚楠, 苏振辉. 基于NSGA-Ⅲ和FloPy的灌区水资源多目标模拟优化模型[J]. 水利与建筑工程学报, 2021, 19(3): 17-23. DOI: 10.3969/j.issn.1672-1144.2021.03.003.
[25] 王凯航. 基于AquaCrop-FloPy和NSGA-Ⅲ耦合模型的TK601玉米新品种灌溉制度多目标优化[D]. 杨凌: 西北农林科技大学, 2023. DOI: 10.27409/d.cnki.gxbnu.2023.002290.
[26] 吕婧妤, 徐超, 刘昱君, 等. 基于模拟优化模型的干旱风沙草原区水-粮食-能源关系[J]. 排灌机械工程学报, 2023, 41(3): 296-304. DOI: 10.3969/j.issn.1674-8530.21.0206.
[27] China T.M.o.E.a.E.o., Quality Standard for Groundwater (GBT 14848-2017). 2017.
[28] Vapnik V.The Nature of Statistical Learning Theory[M]. Berlin: Springer, 1999.
[29] Samandi V, Mukhopadhyay D.Workflow scheduling in cloud computing environment with classification ordinal optimization using SVM[J]. International Journal of Computational Science and Engineering, 2021, 24(6): 563. DOI: 10.1504/ijcse.2021.119970.
