Frontier advances and challenges of machine learning in groundwater science

doi:10.13745/j.esf.sf.2025.10.19

Abstract

Abstract:

Groundwater is a vital component of the global freshwater supply and plays a critical role in ensuring the water resource security,maintaining ecosystem stability,and supporting the sustainable development of human society. As a deeply buried,dynamically evolving,and highly heterogeneous natural system,groundwater poses persistent challenges in observation,modeling,and management,primarily due to the sparsity and uneven distribution of monitoring points and the strong nonlinearity of subsurface physical processes. In recent years,the rapid advancement of artificial intelligence—particularly machine learning—has demonstrated significant potential in feature extraction,high-dimensional nonlinear modeling,and optimization,thereby driving groundwater science into a new paradigm that integrates data-driven approaches with intelligent modeling. This paper provides a comprehensive review of machine learning applications in addressing three core scientific problems in groundwater research: element characterization,process modeling,and management optimization. Key topics include aquifer structure identification,groundwater level,storage and quality analysis,forward simulation,inverse modeling,physics-informed modeling,and groundwater resource management. Based on these developments,a generalized research framework is proposed,structured around the cycle of “scientific problem-data-driven modeling-model selection-result interpretation.” Given existing challenges such as limited incorporation of physical constraints,weak generalization capability,and poor model interpretability,this study also discusses future directions,including the integration of multi-source heterogeneous data,embedding of physical priors,and the use of causal inference methods. This work aims to provide a conceptual and methodological foundation for the application of machine learning in groundwater science and to outline pathways toward developing intelligent and sustainable groundwater systems.

Key words: groundwater science, machine learning methods, research framework, frontier advances

CLC Number:

KANG Jinzheng, MO Shaoxing, KANG Xueyuan, DANG Jingxuan, CHENG Chijitai, XU Peijie, SHI Xiaoqing. Frontier advances and challenges of machine learning in groundwater science[J]. Earth Science Frontiers, 2026, 33(1): 483-499.

Figures/Tables 3

References 136

[1]	RODELL M, FAMIGLIETTI J S, WIESE D N, et al. Emerging trends in global freshwater availability[J]. Nature, 2018, 557(7707): 651-659. DOI
[2]	胡庆芳, 张根瑞, 方琼, et al. 联合国世界水发展报告述评[J]. 水利水运工程学报, 2025(3): 1-13.
[3]	杜朝阳, 钟华平. 地下水系统风险分析研究进展[J]. 水科学进展, 2011, 22(3): 437-444.
[4]	孙才志, 潘俊. 地下水脆弱性的概念,评价方法与研究前景[J]. 水科学进展, 1999, 10(4): 444-449.
[5]	SHEN C. A transdisciplinary review of deep learning research and its relevance for water resources scientists[J]. Water Resources Research, 2018, 54(11): 8558-8593. DOI URL
[6]	VAN S P, LE H M, THANH D V, et al. Deep learning convolutional neural network in rainfall-runoff modelling[J]. Journal of Hydroinformatics, 2020, 22(3): 541-561. DOI URL
[7]	ZHAO X, WANG H, BAI M, et al. A comprehensive review of methods for hydrological forecasting based on deep learning[J]. Water, 2024, 16(10): 1407. DOI URL
[8]	LI Y. Research and application of deep learning in image recognition[C]// 2022 IEEE 2nd international conference on power, electronics and computer applications (ICPECA). Shenyang: IEEE, 2022: 994-999.
[9]	WANG J, LI X, JIN Y, et al. Research on image recognition technology based on multimodal deep learning[C]//2024 IEEE 2nd International Conference on Image Processing and Computer Applications (ICIPCA). Shenyang: IEEE, 2024: 1363-1367.
[10]	CHEN X, LI C, BERNARDS M T, et al. Sequence-based peptide identification,generation,and property prediction with deep learning: a review[J]. Molecular Systems Design & Engineering, 2021, 6(6): 406-428.
[11]	刘博, 王明烁, 李永, et al. 深度学习在时空序列预测中的应用综述[J]. 北京工业大学学报, 2021, 47(8): 925-941.
[12]	潘志松, 黎维. 基于深度学习的时空序列预测方法综述[J]. Journal of Data Acquisition & Processing/Shu Ju Cai Ji Yu Chu Li, 2021, 36(3):436-438.
[13]	MO S, SCHUMACHER M, VAN DIJK A I, et al. Near‐real‐time monitoring of global terrestrial water storage anomalies and hydrological droughts[J]. Geophysical Research Letters, 2025, 52(7): e2024GL112677.
[14]	WANG P, BAYRAM B, SERTEL E. A comprehensive review on deep learning based remote sensing image super-resolution methods[J]. Earth-Science Reviews, 2022,232: 104110.
[15]	CHATTOPADHYAY A, NABIZADEH E, BACH E, et al. Deep learning-enhanced ensemble-based data assimilation for high-dimensional nonlinear dynamical systems[J]. Journal of Computational Physics, 2023,477: 111918.
[16]	HAMITOUCHE M, MOLINA J L. A review of ai methods for the prediction of high-flow extremal hydrology[J]. Water Resources Management, 2022, 36(10): 3859-3876. DOI
[17]	XU T, LIANG F. Machine learning for hydrologic sciences: an introductory overview[J]. Wiley Interdisciplinary Reviews: Water, 2021, 8(5): e1533. DOI URL
[18]	ZHANG J, CAO C, NAN T, et al. A novel deep learning approach for data assimilation of complex hydrological systems[J]. Water Resources Research, 2024, 60(2): e2023WR035389.
[19]	LATIF S D, AHMED A N. A review of deep learning and machine learning techniques for hydrological inflow forecasting[J]. Environment,Development and Sustainability, 2023, 25(11): 12189-12216. DOI
[20]	REICHSTEIN M, CAMPS-VALLS G, STEVENS B, et al. Deep learning and process understanding for data-driven Earth system science[J]. Nature, 2019, 566(7743): 195-204. DOI
[21]	XU R, ZHANG D. Forward prediction and surrogate modeling for subsurface hydrology: a review of theory-guided machine-learning approaches[J]. Computers & Geosciences, 2024,188: 105611.
[22]	JARDANI A, VU T, FISCHER P. Use of convolutional neural networks with encoder-decoder structure for predicting the inverse operator in hydraulic tomography[J]. Journal of Hydrology, 2022,604: 127233.
[23]	ALAM M J, KAR S, ZAMAN S, et al. Forecasting underground water levels:LSTM based model outperforms GRU and decision tree based models[C]//2022 IEEE International Women in Engineering (WIE) Conference on Electrical and Computer Engineering (WIECON-ECE). Naya Raipur: IEEE, 2022: 280-283.
[24]	SHI J, WANG S, QU P, et al. Time series prediction model using LSTM-Transformer neural network for mine water inflow[J]. Scientific reports, 2024, 14(1): 18284. DOI PMID
[25]	TACCARI M L, WANG H, NUTTALL J, et al. Spatial-temporal graph neural networks for groundwater data[J]. Scientific Reports, 2024, 14(1): 24564. DOI PMID
[26]	舒伟, 孟胤全, 邓芳, 等. 基于 PINNs 算法的一维潜水流方程的渗流参数反演[J]. 南京大学学报 (自然科学版), 2024, 60(2): 317-327.
[27]	TRIPATHY K P, MISHRA A K. Deep learning in hydrology and water resources disciplines: concepts,methods,applications,and research directions[J]. Journal of Hydrology, 2024,628: 130458.
[28]	KOLLET S J, ZLOTNIK V A. Influence of aquifer heterogeneity and return flow on pumping test data interpretation[J]. Journal of hydrology, 2005, 300(1/2/3/4): 267-285. DOI URL
[29]	PARADIS D, TREMBLAY L, LEFEBVRE R, et al. Field characterization and data integration to define the hydraulic heterogeneity of a shallow granular aquifer at a sub-watershed scale[J]. Environmental Earth Sciences, 2014, 72(5): 1325-1348. DOI URL
[30]	PANAHI M, SADHASIVAM N, POURGHASEMI H R, et al. Spatial prediction of groundwater potential mapping based on convolutional neural network (CNN) and support vector regression (SVR)[J]. Journal of Hydrology, 2020,588: 125033.
[31]	XUANYU S, MIN X, SHICHANG K, et al. Modeling of hydrological processes in cryospheric watersheds based on machine learning[J]. Earth Science Frontiers, 2023, 30(4): 451. DOI
[32]	WANG Y, ZHA Y. Comparison of transformer,LSTM and coupled algorithms for soil moisture prediction in shallow-groundwater-level areas with interpretability analysis[J]. Agricultural Water Management, 2024,305: 109120.
[33]	BAI T, TAHMASEBI P. Graph neural network for groundwater level forecasting[J]. Journal of Hydrology, 2023,616: 128792.
[34]	KANG X, KOKKINAKI A, POWER C, et al. Integrating deep learning-based data assimilation and hydrogeophysical data for improved monitoring of DNAPL source zones during remediation[J]. Journal of Hydrology, 2021,601: 126655.
[35]	XU Y, LU W, PAN Z, et al. Groundwater contaminant source identification considering unknown boundary condition based on an automated machine learning surrogate[J]. Geoscience Frontiers, 2024, 15(1): 101732. DOI URL
[36]	JANŽA M. Optimization of well field management to mitigate groundwater contamination using a simulation model and evolutionary algorithm[J]. Science of the Total Environment, 2022,807: 150811.
[37]	CHINNAMGARI S K. R Machine Learning Projects: implement supervised, unsupervised, and reinforcement learning techniques using R 3.5[M]. Birmingham: Packt Publishing Ltd, 2019.
[38]	HU Y, LUO S, HAN L, et al. Deep supervised learning with mixture of neural networks[J]. Artificial intelligence in medicine, 2020,102: 101764.
[39]	胡义明, 陈腾, 罗序义, 等. 基于机器学习模型的淮河流域中长期径流预报研究[J]. 地学前缘, 2022, 29(3): 284-291. DOI
[40]	CUNNINGHAM P, CORD M, DELANY S J. Supervised learning[M]// Machine learning techniques for multimedia:case studies on organization and retrieval. Berlin, Heidelberg: Springer Berlin Heidelberg, 2008: 21-49.
[41]	COLLINGWOOD L, WILKERSON J. Tradeoffs in accuracy and efficiency in supervised learning methods[J]. Journal of Information Technology & Politics, 2012, 9(3): 298-318.
[42]	YANG X, SONG Z, KING I, et al. A survey on deep semi-supervised learning[J]. IEEE Transactions on Knowledge and Data Engineering, 2022, 35(9): 8934-8954. DOI URL
[43]	曲昭伟, 吴春叶, 王晓茹. 半监督自训练的方面提取[J]. 智能系统学报, 2019, 14(4): 635-641.
[44]	SONG Z, YANG X, XU Z, et al. Graph-based semi-supervised learning: a comprehensive review[J]. IEEE Transactions on Neural Networks and Learning Systems, 2022, 34(11): 8174-8794. DOI URL
[45]	DING S, ZHU Z, ZHANG X. An overview on semi-supervised support vector machine[J]. Neural Computing and Applications, 2017, 28(5): 969-978.
[46]	DIKE H U, ZHOU Y, DEVEERASETTY K K, et al. Unsupervised learning based on artificial neural network:a review[C]//2018 IEEE International Conference on Cyborg and Bionic Systems (CBS). Shenzhen: IEEE, 2018: 322-327.
[47]	NANGA S, BAWAH A T, ACQUAYE B A, et al. Review of dimension reduction methods[J]. Journal of Data Analysis and Information Processing, 2021, 9(3): 189-231. DOI URL
[48]	KISI O, HEDDAM S, PARMAR K S, et al. Integration of Gaussian process regression and K means clustering for enhanced short term rainfall runoff modeling[J]. Scientific Reports, 2025, 15(1): 7444. DOI
[49]	ARABZADEH R, KHOLOOSI M M, BAZRAFSHAN J. Regional hydrological drought monitoring using principal components analysis[J]. Journal of Irrigation and Drainage Engineering, 2016, 142(1): 04015029. DOI URL
[50]	BALDI P. Autoencoders, unsupervised learning, and deep architectures[C]// Proceedings of ICML workshop on unsupervised and transfer learning. Washington: JMLR Workshop and Conference Proceedings, 2012: 37-49.
[51]	XIA X, PAN X, LI N, et al. GAN-based anomaly detection: a review[J]. Neurocomputing, 2022,493: 497-535.
[52]	HE G, CHEN Y, LI Z, et al. Exploring denoising diffusion probabilistic model for daily streamflow gap filling in Central Asia typical watersheds[J]. Journal of Hydrology: Regional Studies, 2024,52: 101701.
[53]	LI S E. Deep reinforcement learning[M]//Reinforcement learning for sequential decision and optimal control. Singapore: Springer Nature Singapore, 2023: 365-402.
[54]	KÄGE L, MILIĆ V, ANDERSSON M, et al. Reinforcement learning applications in water resource management: a systematic literature review[J]. Frontiers in Water, 2025,7: 1537868.
[55]	HOURFAR F, BIDGOLY H J, MOSHIRI B, et al. A reinforcement learning approach for waterflooding optimization in petroleum reservoirs[J]. Engineering Applications of Artificial Intelligence, 2019,77: 98-116.
[56]	LUO J, MA X, JI Y, et al. Review of machine learning-based surrogate models of groundwater contaminant modeling[J]. Environmental Research, 2023,238: 117268.
[57]	BURNETT W C, TANIGUCHI M, OBERDORFER J. Measurement and significance of the direct discharge of groundwater into the coastal zone[J]. Journal of Sea Research, 2001, 46(2): 109-116. DOI URL
[58]	BAI Z, LIU Q, LIU Y. Groundwater potential mapping in hubei region of china using machine learning,ensemble learning,deep learning and automl methods[J]. Natural Resources Research, 2022, 31(5): 2549-2569. DOI
[59]	DI CURZIO D, CASTRIGNANÒ A, FOUNTAS S, et al. Multi-source data fusion of big spatial-temporal data in soil, geo-engineering and environmental studies[J]. Science of the Total Environment, 2021, 788: 147842. DOI URL
[60]	THIELE S T, LORENZ S, KIRSCH M, et al. Multi-scale,multi-sensor data integration for automated 3-D geological mapping[J]. Ore Geology Reviews, 2021,136: 104252.
[61]	KITANIDIS P K, LEE J. Principal component geostatistical approach for large-dimensional inverse problems[J]. Water resources research, 2014, 50(7): 5428-5443. PMID
[62]	WONG P M, CHOI S, NIU Y. A comparision of pca/ica for data preprocessing in a geoscience application[J]. IEEE Transactions on Geoscience and Remote Sensing, 2002, 40(5): 410-420.
[63]	MO S, ZABARAS N, SHI X, et al. Integration of adversarial autoencoders with residual dense convolutional networks for estimation of non‐Gaussian hydraulic conductivities[J]. Water Resources Research, 2020, 56(2): e2019WR026082.
[64]	WU H, O’MALLEY D, GOLDEN J, et al. Learning to regularize with a variational autoencoder for hydrogeologic inverse analysis[C]// Nevada: Geological Society of America Abstracts. 2021, 53: 366458.
[65]	SUN R, PAN B, DUAN Q. Learning distributed parameters of land surface hydrologic models using a Generative Adversarial Network[J]. Water Resources Research, 2024, 60(7): e2024WR037380.
[66]	THANH-TUNG H, TRAN T. Catastrophic forgetting and mode collapse in GANs[C]//2020 international joint conference on neural networks (ijcnn). Glasgow: IEEE, 2020: 1-10.
[67]	FENG L, MO S, SUN A Y, et al. Deep learning-based geological parameterization for history matching CO₂ plume migration in complex aquifers[J]. Advances in Water Resources, 2024,193: 104833.
[68]	JIA W, SUN M, LIAN J, et al. Feature dimensionality reduction: a review[J]. Complex & Intelligent Systems, 2022, 8(3): 2663-93.
[69]	CUI Z, CHEN Q, LIU G. Characterization of subsurface hydrogeological structures with convolutional conditional neural processes on limited training data[J]. Water Resources Research, 2022, 58(12): e2022WR033161.
[70]	KHAN S, SERANI A, DIEZ M, et al. Physics-informed feature-to-feature learning for design-space dimensionality reduction in shape optimisation[C]// Tennessee:AIAA scitech 2021 forum. 2021: 1235.
[71]	DI FEDERICO G, DURLOFSKY L J. Latent diffusion models for parameterization of facies-based geomodels and their use in data assimilation[J]. Computers & Geosciences, 2025,194: 105755.
[72]	GONG G, MATTEVADA S, O’BRYANT S E. Comparison of the accuracy of kriging and IDW interpolations in estimating groundwater arsenic concentrations in Texas[J]. Environmental research, 2014,130: 59-69.
[73]	MACHADO F, MINE M, KAVISKI E, et al. Monthly rainfall-runoff modelling using artificial neural networks[J]. Hydrological Sciences Journal-Journal des Sciences Hydrologiques, 2011, 56(3): 349-361. DOI URL
[74]	RODRíGUEZ R, PASTORINI M, ETCHEVERRY L, et al. Water-quality data imputation with a high percentage of missing values: a machine learning approach[J]. Sustainability, 2021, 13(11): 6318. DOI URL
[75]	HE L, CHEN S, LIANG Y, et al. Infilling the missing values of groundwater level using time and space series: case of Nantong City,east coast of China[J]. Earth Science Informatics, 2020, 13(4): 1445-1459. DOI
[76]	VU M, JARDANI A, MASSEI N, et al. Reconstruction of missing groundwater level data by using Long Short-Term Memory (LSTM) deep neural network[J]. Journal of Hydrology, 2021,597: 125776.
[77]	ZHANG J, HU L, SUN J, et al. Reconstructing groundwater storage changes in the North China Plain using a numerical model and GRACE data[J]. Remote Sensing, 2023, 15(13): 3264. DOI URL
[78]	YOKOO K, ISHIDA K, NAGASATO T, et al. Reconstruction of groundwater level at Kumamoto, Japan by means of deep learning to evaluate its increase by the 2016 earthquake[C]// IOP Conference Series: Earth and Environmental Science. Volgograd: IOP Publishing, 2021, 851(1): 012032.
[79]	MO S, ZHONG Y, FOROOTAN E, et al. Hydrological droughts of 2017—2018 explained by the Bayesian reconstruction of GRACE (‐FO) fields[J]. Water Resources Research, 2022, 58(9): e2022WR031997.
[80]	ALI S, LIU D, FU Q, et al. Improving the resolution of GRACE data for spatio-temporal groundwater storage assessment[J]. Remote Sensing, 2021, 13(17): 3513. DOI URL
[81]	SEO J Y, LEE S I. Predicting changes in spatiotemporal groundwater storage through the integration of multi-satellite data and deep learning models[J]. IEEE Access, 2021,9: 157571-157583.
[82]	HU Z, TANG S, MO S, et al. Water storage changes (2003—2020) in the Ordos Basin,China,explained by GRACE data and interpretable deep learning[J]. Hydrogeology Journal,2024, 32(1): 307-320. DOI
[83]	王焰新, 马腾, 李义连, 等. 天然劣质地下水修复与示范年度报告[J]. 科技资讯, 2016, 14(10): 169-170.
[84]	KURWADKAR S, KANEL S R, NAKARMI A. Groundwater pollution: occurrence,detection,and remediation of organic and inorganic pollutants[J]. Water Environment Research, 2020, 92(10): 1659-1668. DOI URL
[85]	WANG Y, LI J, MA T, et al. Genesis of geogenic contaminated groundwater: as,F and I[J]. Critical Reviews in Environmental Science and Technology, 2021, 51(24): 2895-2933. DOI URL
[86]	MARíN CELESTINO A E, MARTíNEZ CRUZ D A, OTAZO SáNCHEZ E M, et al. Groundwater quality assessment: an improved approach to K-means clustering,principal component analysis and spatial analysis: a case study[J]. Water, 2018, 10(4): 437. DOI URL
[87]	LEE K J, YUN S T, YU S, et al. The combined use of self-organizing map technique and fuzzy c-means clustering to evaluate urban groundwater quality in Seoul metropolitan city,South Korea[J]. Journal of Hydrology, 2019,569: 685-697.
[88]	SONG C, CUI J, CUI Y, et al. Integrated STL-DBSCAN algorithm for online hydrological and water quality monitoring data cleaning[J]. Environmental Modelling & Software, 2025,183: 106262.
[89]	KARUNANIDHI D, RAJ M R H, ROY P D, et al. Integrated machine learning based groundwater quality prediction through groundwater quality index for drinking purposes in a semi-arid river basin of south India[J]. Environmental Geochemistry and Health, 2025, 47(4): 119. DOI
[90]	CHEN T, ZHANG H, SUN C, et al. Multivariate statistical approaches to identify the major factors governing groundwater quality[J]. Applied Water Science, 2018, 8(7): 215. DOI
[91]	ZHI W, APPLING A P, GOLDEN H E, et al. Deep learning for water quality[J]. Nature water, 2024, 2(3): 228-241. DOI
[92]	BEDI S, SAMAL A, RAY C, et al. Comparative evaluation of machine learning models for groundwater quality assessment[J]. Environmental Monitoring and Assessment, 2020, 192(12): 776. DOI PMID
[93]	LI P, NAZ I, ASLAM R W, et al. Groundwater quality assessment for rangeland dynamic: integration of multicriteria decision analysis with remote sensing data[J]. Rangeland Ecology & Management, 2025, 102: 110-127. DOI URL
[94]	CAO H, XIE X, WANG Y, et al. Predicting geogenic groundwater fluoride contamination throughout China[J]. Journal of Environmental sciences, 2022,115: 140-148.
[95]	DACHANUWATTANA S, YU W, SEPEHRNOORI K. An efficient MCMC history matching workflow using fit-for-purpose proxies applied in unconventional oil reservoirs[J]. Journal of Petroleum Science and Engineering, 2019,176: 381-395.
[96]	OLADYSHKIN S, CLASS H, NOWAK W. Bayesian updating via bootstrap filtering combined with data-driven polynomial chaos expansions: methodology and application to history matching for carbon dioxide storage in geological formations[J]. Computational Geosciences, 2013, 17(4): 671-687. DOI URL
[97]	ASHER M J, CROKE B F, JAKEMAN A J, et al. A review of surrogate models and their application to groundwater modeling[J]. Water Resources Research, 2015, 51(8): 5957-5973. DOI URL
[98]	PANG M, DU E, ZHENG C. Contaminant transport modeling and source attribution with attention‐based graph neural network[J]. Water Resources Research, 2024, 60(6): e2023WR035278.
[99]	TANG H, DURLOFSKY L J. Graph network surrogate model for subsurface flow optimization[J]. Journal of Computational Physics, 2024,512: 113132.
[100]	WANG N, ZHANG D, CHANG H, et al. Deep learning of subsurface flow via theory-guided neural network[J]. Journal of Hydrology, 2020,584: 124700.
[101]	SOLGI R, LOAICIGA H A, KRAM M. Long short-term memory neural network (LSTM-NN) for aquifer level time series forecasting using in-situ piezometric observations[J]. Journal of Hydrology, 2021,601: 126800.
[102]	NAN T, CAO W, WANG Z, et al. Evaluation of shallow groundwater dynamics after water supplement in North China Plain based on attention-GRU model[J]. Journal of Hydrology, 2023,625: 130085.
[103]	HE T, CHANG H, ZHANG D. Identification of physical processes and unknown parameters of 3D groundwater contaminant problems via theory-guided U-net[J]. Stochastic Environmental Research and Risk Assessment, 2024, 38(3): 869-900. DOI
[104]	MO S, ZABARAS N, SHI X, et al. Deep autoregressive neural networks for high‐dimensional inverse problems in groundwater contaminant source identification[J]. Water Resources Research, 2019, 55(5): 3856-3881. DOI URL
[105]	CHEN E, ANDERSEN M S, CHANDRA R. Deep learning framework with Bayesian data imputation for modelling and forecasting groundwater levels[J]. Environmental Modelling & Software, 2024,178: 106072.
[106]	LI B, RODELL M, KUMAR S, et al. Global GRACE data assimilation for groundwater and drought monitoring: advances and challenges[J]. Water Resources Research, 2019, 55(9): 7564-7586. DOI URL
[107]	BENZ S A, BAYER P, BLUM P. Global patterns of shallow groundwater temperatures[J]. Environmental Research Letters, 2017, 12(3): 034005. DOI URL
[108]	MO S, ZHONG Y, FOROOTAN E, et al. Bayesian convolutional neural networks for predicting the terrestrial water storage anomalies during GRACE and GRACE-FO gap[J]. Journal of Hydrology, 2022,604: 127244.
[109]	WANG L, ZHANG Y. Filling GRACE data gap using an innovative transformer-based deep learning approach[J]. Remote Sensing of Environment, 2024,315: 114465.
[110]	SHEN C, APPLING A P, GENTINE P, et al. Differentiable modelling to unify machine learning and physical models for geosciences[J]. Nature Reviews Earth & Environment, 2023, 4(8): 552-567.
[111]	FERNANDEZ-MARTINEZ J L, FERNANDEZ-MUNIZ Z. The curse of dimensionality in inverse problems[J]. Journal of Computational and Applied Mathematics, 2020,369: 112571.
[112]	ZHOU H, GóMEZ-HERNáNDEZ J J, LI L. Inverse methods in hydrogeology: evolution and recent trends[J]. Advances in Water Resources, 2014,63: 22-37.
[113]	AMINI S, MOHAGHEGH S. Application of machine learning and artificial intelligence in proxy modeling for fluid flow in porous media[J]. Fluids, 2019, 4(3): 126. DOI URL
[114]	TANG M, LIU Y, DURLOFSKY L J. A deep-learning-based surrogate model for data assimilation in dynamic subsurface flow problems[J]. Journal of Computational Physics, 2020,413: 109456.
[115]	LIMA M M, EMERICK A A, ORTIZ C E. Data-space inversion with ensemble smoother[J]. Computational Geosciences, 2020, 24(3): 1179-1200. DOI
[116]	LIAO Z, MI X, PANG Q, et al. History archive assisted niching differential evolution with variable neighborhood for multimodal optimization[J]. Swarm and Evolutionary Computation, 2023,76: 101206.
[117]	MA X, ZHANG K, ZHANG L, et al. Data-driven niching differential evolution with adaptive parameters control for history matching and uncertainty quantification[J]. Spe Journal, 2021, 26(2): 993-1010. DOI URL
[118]	沈焕锋, 张良培. 地球表层特征参量反演与模拟的机理-学习耦合范式[J]. 中国科学: 地球科学, 2023, 53(3): 546-560.
[119]	SECCI D, GODOY V A, GóMEZ-HERNáNDEZ J J. Physics-Informed Neural Networks for solving transient unconfined groundwater flow[J]. Computers & Geosciences, 2024,182: 105494.
[120]	WANG N, CHANG H, ZHANG D. Theory-guided auto-encoder for surrogate construction and inverse modeling[J]. Computer Methods in Applied Mechanics and Engineering, 2021,385: 114037.
[121]	YAZDANI S, TAHANI M. Data-driven discovery of turbulent flow equations using physics-informed neural networks[J]. Physics of Fluids, 2024, 36(3):035107. DOI URL
[122]	SONG W, SHI L, HU X, et al. Reconstructing the unsaturated flow equation from sparse and noisy data: leveraging the synergy of group sparsity and physics‐informed deep learning[J]. Water Resources Research, 2023, 59(5): e2022WR034122.
[123]	JIANG S, ZHENG Y, WANG C, et al. Uncovering flooding mechanisms across the contiguous United States through interpretive deep learning on representative catchments[J]. Water Resources Research, 2022, 58(1): e2021WR030185.
[124]	薛禹群. 论地下水超采与地面沉降[J]. 地下水, 2012,(6): 1-5.
[125]	YEH W W. Optimization methods for groundwater modeling and management[J]. Hydrogeology Journal, 2015, 23(6): 1051-65. DOI URL
[126]	ZHANG D, GUO P. Integrated agriculture water management optimization model for water saving potential analysis[J]. Agricultural Water Management, 2016,170: 5-19.
[127]	GORELICK S M, ZHENG C. Global change and the groundwater management challenge[J]. Water Resources Research, 2015, 51(5): 3031-51. DOI URL
[128]	NAGHDI S, BOZORG-HADDAD O, KHORSANDI M, et al. Multi-objective optimization for allocation of surface water and groundwater resources[J]. Science of the Total Environment, 2021,776: 146026.
[129]	WANG Z, YANG Y, WU J, et al. Multi-objective optimization of the coastal groundwater abstraction for striking the balance among conflicts of resource-environment-economy in Longkou City,China[J]. Water Research, 2022,211: 118045.
[130]	CHEN Y, LIU G, HUANG X, et al. Groundwater remediation design underpinned by coupling evolution algorithm with deep belief network surrogate[J]. Water Resources Management, 2022, 36(7): 2223-2239. DOI
[131]	DU J, SHI X, MO S, et al. Deep learning based optimization under uncertainty for surfactant-enhanced DNAPL remediation in highly heterogeneous aquifers[J]. Journal of Hydrology, 2022,608: 127639.
[132]	QIU W, MA T, WANG Y, et al. Review on status of groundwater database and application prospect in deep-time digital earth plan[J]. Geoscience Frontiers, 2022, 13(4): 101383. DOI URL
[133]	LI J, HONG D, GAO L, et al. Deep learning in multimodal remote sensing data fusion: a comprehensive review[J]. International Journal of Applied Earth Observation and Geoinformation, 2022,112: 102926.
[134]	ALI A S A, JAZAEI F, CLEMENT T P, et al. Physics-informed neural networks in groundwater flow modeling: advantages and future directions[J]. Groundwater for Sustainable Development, 2024,25: 101172.
[135]	BASAGAOGLU H, CHAKRABORTY D, LAGO C D, et al. A review on interpretable and explainable artificial intelligence in hydroclimatic applications[J]. Water, 2022, 14(8): 1230. DOI URL
[136]	GAD A G. Particle swarm optimization algorithm and its applications: a systematic review[J]. Archives of Computational Methods in Engineering, 2022, 29(5): 2531-2561 DOI

归纳核心	问题	数据任务类型	机器学习模型	特点
要素刻画 (自监督任务)	含水层结构刻画	以图像数据为主, 向多模态数据融合方向发展	主成分分析(PCA)	特征值降维
			卷积自编码器(CAE)	比PCA更适用于高维问题
			变分自编码器(CVAE)	比CAE更适用于非高斯储层刻画
			对抗神经网络	比自编码结构生成结构更保真
			对抗自编码器(AAE)	适用于非高斯储层刻画
			深度扩散模型(Diffusion)	多模态数据融合更灵活且更保真
	水位水量动态描述	序列数据	全链接神经网络(FCN)	适用问题灵活度高
			随机森林(RF)	可用于极度缺失数据填补
			克里金-支持向量机(UK-SVR)	相比于单一模型精度更高
			循环神经网络(RNN)	处理长时序问题的填补
			长短期记忆神经网络(LSTM)	比RNN捕捉长时序依赖关系更高
			Transformer	比LSTM精度高和计算高效
	水质因子聚类分析	离散数据	K均值算法(K-means)	实现简单,计算高效
			模糊均值聚类(FCM)	允许一个样本多个类
			密度聚类(DBSCAN)	适用于噪声数据聚类
			卷积神经网络(CNN)	评估和填补区域水质数据
过程建模 (监督任务)	正演预测	图像数据与序列数据	卷积神经网络(CNN)	构建高维数据间的映射关系
			残差卷积神经网络(ResNet)	相比于CNN精度更高
			U型神经网络(U-net)	网络结构提高预测精度
			图神经网络(GNN)	非结构网格处理
			贝叶斯神经网络(BNN)	可以提供模型不确定性
			长短期记忆神经网络(LSTM)	适用于水位、水量和水质序列数据预测
			门控循环神经网络(GRU)	相比于LSTM训练时间短
			Transformer	精度高和计算高效
	反演模拟	图像数据、序列数据与优化问题相结合	卷积+循环神经网络(CNN+LSTM)	构建端到端替代模型
			模拟退火,差分进化,粒子群算法	智能优化算法,实现简单无需梯度
			深度数据同化算法(DA(DL))	更适用于非高斯场反演
			多模态优化算法(MultiModal)	避免集合算法的“集合崩溃”
	机制挖掘	图像数据与序列数据	内嵌物理神经网络(PINN)	提高网络物理意义和精度
			理论引导神经网络(TGNN)	克服PINN高维采样难题
			深度稀疏框架(DeepGS)	对物理方程进行重构
管理优化 (混合任务)	水位调控与水量开采分配	图像数据、序列数据与优化问题相结合	非支配排序遗传算法Ⅱ(NSGA-II)	多目标优化效果表现优异
	水位调控与水量开采分配	图像数据、序列数据与优化问题相结合	强化学习方法(RL)	非线性、非平稳优化效果好
	地下水水质治理	图像数据、序列数据与优化问题相结合	复合策略优化算法(SCE-UA)	同时优化修复井位置和修复液注入策略
			深度置信网络(DBN)+粒子群算法(PSO)	节约优化时间的同时达到治理要求
			卷积神经网络+非支配排序遗传算法Ⅱ	大幅节约优化时间

归纳核心	问题	数据任务类型	机器学习模型	特点
要素刻画 (自监督任务)	含水层结构刻画	以图像数据为主, 向多模态数据融合方向发展	主成分分析(PCA)	特征值降维
			卷积自编码器(CAE)	比PCA更适用于高维问题
			变分自编码器(CVAE)	比CAE更适用于非高斯储层刻画
			对抗神经网络	比自编码结构生成结构更保真
			对抗自编码器(AAE)	适用于非高斯储层刻画
			深度扩散模型(Diffusion)	多模态数据融合更灵活且更保真
	水位水量动态描述	序列数据	全链接神经网络(FCN)	适用问题灵活度高
			随机森林(RF)	可用于极度缺失数据填补
			克里金-支持向量机(UK-SVR)	相比于单一模型精度更高
			循环神经网络(RNN)	处理长时序问题的填补
			长短期记忆神经网络(LSTM)	比RNN捕捉长时序依赖关系更高
			Transformer	比LSTM精度高和计算高效
	水质因子聚类分析	离散数据	K均值算法(K-means)	实现简单,计算高效
			模糊均值聚类(FCM)	允许一个样本多个类
			密度聚类(DBSCAN)	适用于噪声数据聚类
			卷积神经网络(CNN)	评估和填补区域水质数据
过程建模 (监督任务)	正演预测	图像数据与序列数据	卷积神经网络(CNN)	构建高维数据间的映射关系
			残差卷积神经网络(ResNet)	相比于CNN精度更高
			U型神经网络(U-net)	网络结构提高预测精度
			图神经网络(GNN)	非结构网格处理
			贝叶斯神经网络(BNN)	可以提供模型不确定性
			长短期记忆神经网络(LSTM)	适用于水位、水量和水质序列数据预测
			门控循环神经网络(GRU)	相比于LSTM训练时间短
			Transformer	精度高和计算高效
	反演模拟	图像数据、序列数据与优化问题相结合	卷积+循环神经网络(CNN+LSTM)	构建端到端替代模型
			模拟退火,差分进化,粒子群算法	智能优化算法,实现简单无需梯度
			深度数据同化算法(DA(DL))	更适用于非高斯场反演
			多模态优化算法(MultiModal)	避免集合算法的“集合崩溃”
	机制挖掘	图像数据与序列数据	内嵌物理神经网络(PINN)	提高网络物理意义和精度
			理论引导神经网络(TGNN)	克服PINN高维采样难题
			深度稀疏框架(DeepGS)	对物理方程进行重构
管理优化 (混合任务)	水位调控与水量开采分配	图像数据、序列数据与优化问题相结合	非支配排序遗传算法Ⅱ(NSGA-II)	多目标优化效果表现优异
	水位调控与水量开采分配	图像数据、序列数据与优化问题相结合	强化学习方法(RL)	非线性、非平稳优化效果好
	地下水水质治理	图像数据、序列数据与优化问题相结合	复合策略优化算法(SCE-UA)	同时优化修复井位置和修复液注入策略
			深度置信网络(DBN)+粒子群算法(PSO)	节约优化时间的同时达到治理要求
			卷积神经网络+非支配排序遗传算法Ⅱ	大幅节约优化时间