Accurate groundwater level (GWL) prediction is crucial in groundwater resource management. Currently, it relies mainly on physics-based models for prediction and quantitative analysis. However, physics-based models used for prediction often have errors in structure, parameters, and data, resulting in inaccurate GWL predictions. In this study, machine learning algorithms were used to correct the prediction errors of physics-based models. First, a MODFLOW groundwater flow model was created for the Hutuo River alluvial fan in the North China Plain. Then, using the observed GWLs from 10 monitoring wells located in the upper, middle, and lower parts of the alluvial fan as the test standard, three algorithms—random forest (RF), extreme gradient boosting (XGBoost), and long short-term memory (LSTM)—were compared for their abilities to correct MODFLOW’s predicted GWLs of these 10 wells under two sets of feature variables. The results show that the RF and XGBoost algorithms are not suitable for correcting predicted GWLs that exhibit continuous rising or falling trends, but the LSTM algorithm has the ability to correct them. During the prediction period, the LSTM2 model, which incorporates additional source–sink feature variables based on MODFLOW’s predicted GWLs, can improve the Pearson correlation coefficient (PR) for 80% of wells, with a maximum increase of 1.26 and a minimum increase of 0.02, and can reduce the root mean square error (RMSE) for 100% of the wells with a maximum decrease of 1.59 m and a minimum decrease of 0.17 m. And it also outperforms the MODFLOW model in capturing the long-term trends and short-term seasonal fluctuations of GWLs. However, the correction effect of the LSTM1 model (using only MODFLOW’s predicted GWLs as a feature variable) is inferior to that of the LSTM2 model, indicating that multiple feature variables are superior to a single feature variable. Temporally and spatially, the greater the prediction error of the MODFLOW model, the larger the correction magnitude of the LSTM2 model.

In previous investigations, the demarcation of capture zones within a specific research area predominantly relied on a singular method, leading to pronounced limitations and uncertainties. To address this challenge, an extensive field survey was conducted, focusing on the systematic classification of water sources in the Linfen City region. Building upon this classification, an intricate fusion of a hydrogeological analysis and formulaic methodology was employed. This integrated approach, coupled with independent numerical simulation methods, was applied to delineate recharge areas for both alluvial fan pore water in piedmont regions and exposed karst water in small- to medium-sized water sources. Simultaneously, precise spatial interpolation was carried out on water quality monitoring data from supply wells within the water source area for the year 2020. This meticulous analysis facilitated the determination of the spatial distribution characteristics of hydrochemical elements. To assess the water quality within the capture zone, the class III groundwater quality standards of China were employed as a pivotal tool for validating the results of the delineation of water source recharge areas. In the final analysis, a comparative study between the integrated coupling method and numerical simulation outcomes revealed the successful delineation of the boundaries for the water supply areas of Tumen and Caojiapo in Linfen City, covering areas of 5.5 km² and 22.29 km², respectively. Simultaneously, the combination of the three methods accurately outlined the boundary of the Hexi water supply area, encompassing an area of 2.5224 km². These results vividly illustrate that the amalgamation of various methodologies proves more beneficial for the precise delineation of capture zones, particularly in diverse types and scales of groundwater sources. The synergy exhibited by these three methods underscores their collective efficacy, providing a more comprehensive and intuitive delineation of the recharge areas for small- to medium-sized water sources. Consequently, these findings significantly enhance the practical application value of the study and hold promise in making substantial contributions to local groundwater security and management initiatives.

Carbon Tetrachloride (CCL4) is a colorless, volatile, and toxic liquid. Once it pollutes groundwater, it will not only destroy the ecological environment but also negatively affect the functioning of the human liver. An industrial city in eastern China has been contaminated with carbon tetrachloride (CCL4). Due to the complex hydrogeological conditions, it is difficult to determine the pollution source by a single hydrochemical analysis. In order to solve the traceability problem, in this work we established a traceability system by combining hydrochemical analysis, backward tracing, and forward transport, and analyzed the pollution distribution, pollution-source location and pollution-transport characteristic in groundwater, which provided technical support for CCL4-pollution control of groundwater in this area. (1) Groundwater samples were analyzed using the gas chromatography/mass spectrometry method. Through the pollution concentration-field, we identified the location of the pollution center and the concentration value in the northeast and southwest, screening out the monitoring wells exceeding the criteria: HF#1, DJ#19, DJ#7, SS#4, ZF#2, and DY#3. (2) Backward tracing over the past 30 years and 50 years was conducted through MODPATH. Potential polluting factories were identified by comparing the capture area with the historical distribution of factories. (3) Forward solute-transport was performed at the location of these potentially polluted factories. The distribution characteristics of pollutants in limestone layers and the Quaternary layer were analyzed using a simulated concentration-field for August 2012 and October 2016. (4) Comparing the simulated concentration with the observed concentration, the source of CCL4 pollution was inferred to be the textile mill in the 1990s, the steelworks in the 1970s, and the machine-tool plant in the 1970s and 1990s. Based on the concentration–duration curve at the pollution source, the transport characteristics and the transport speed of the pollutants in the study area were analyzed. This work not only successfully found the location of CCL4-pollution sources, but also helped the local government to analyze the year of pollutant release and recognize the transport pattern of CCL4 in aquifers.