Accurate Prediction of Soil Heavy Metal Pollution Using an Improved Machine Learning Method: A Case Study in the Pearl River Delta, China

Spatial-machine learning framework for rapid identification of soil cadmium risk in high geochemical background areas

Rapid and accurate identification of cadmium (Cd) risk remains challenging in agricultural lands with high geochemical background levels. While predicting soil Cd mobility using the risk assessment code (RAC) is essential for evaluating ecological risks at regional scales, traditional prediction methods struggle to achieve high spatial prediction accuracy because of complex influencing factors and spatial heterogeneity. This study investigated the spatial distribution patterns of soil Cd mobility in karst regions under the influence of anthropogenic activities and natural background conditions. Our analysis revealed that areas of very high risk were predominantly concentrated in black shale formations and mining zones, reflecting the spatial heterogeneity of soil available Cd. Geographically weighted regression analysis demonstrated both negative and positive local correlation coefficients between soil properties and RAC values, suggesting complex spatial interactions. Incorporating these spatial relationships as covariates in the random forest model resulted in an enhanced prediction accuracy (R2 = 0.96) compared to the non-spatial approach (R2 = 0.80). The machine learning model with integrated spatial information developed in this study provides an improved framework for identifying soil Cd risks and understanding Cd geochemical behaviors, supporting the development of targeted pollution prevention and control strategies in areas with high geochemical background.

Accurate prediction of spatial distribution of soil heavy metal in complex mining terrain using an improved machine learning method

Accurate prediction of heavy metals (HMs) spatial distribution in mining areas is crucial for pollution management. However, predicting the spatial distribution of HMs remains a significant challenge in mining areas with complex terrain and variable contaminant transport pathways. This study aims to optimize the spatial prediction of arsenic (As) distribution in the Shimen realgar mining area, the largest in Asia, by integrating machine learning models with kriging interpolation and feature selection techniques. The results show that the Random Forest (RF) model achieved the best performance in predicting soil As concentration, with an R2 of 0.84 for the test data. Incorporating environmental variables improved the spatial prediction accuracy, with RF (R2 = 0.76, RMSE = 24.68 mg/kg) and Random Forest Regression Kriging (RFRK) (R2 = 0.78, RMSE = 23.46 mg/kg) outperforming ordinary kriging and geographically weighted regression kriging. Importance analysis and recursive feature elimination further optimized the model, leading to a 5 % increase in R2 and a reduction of RMSE by 8 %-12.4 %. The optimized RFRK model accurately captured the spatial distribution of As in the mining area, revealing the outward diffusion pattern of As from the smelting plant. The findings highlight the critical role of feature selection in improving prediction accuracy in highly polluted and complex terrain regions, an aspect that has often been overlooked in previous studies. This study provides a practical framework for spatial prediction of contaminants in similar areas, enhancing the understanding of pollution distribution.

Prediction of zinc, cadmium, and arsenic in european soils using multi-end machine learning models

Heavy metal contamination in soil is a major environmental and public health concern, especially in regions with substantial industrial and agricultural activities. Conventional predictive models often focus on single contaminants, limiting their utility for comprehensive environmental monitoring. This study addressed these limitations by developing an advanced multi-end ensemble convolutional neural network model capable of simultaneously predicting the concentrations of cadmium, arsenic, and zinc in European soils. A comprehensive dataset with 18 diverse factors was prepared, including soil properties, climatic factors, and anthropogenic activities. Moreover, the model compared four ensemble learning techniques in contamination prediction, including simple averaging, snapshot ensembles, integrated stacking, and separate stacking. Among these, the separate stacking model with random forest regressor meta-model achieved the highest accuracy, with a mean spared error of 0.0378, a mean absolute error of 0.0785, and a coefficient of determination of 0.79 in the testing phases. Sensitivity analysis highlighted farming area, road length, nitrogen content, and mean annual temperature as key factors influencing metal concentrations. To enhance accessibility, a GUI-based web application was developed, allowing users to enter relevant factors and receive real-time predictions of contamination levels. This application empowers stakeholders, such as environmental regulators and policymakers, to make informed, data-driven decisions for targeted remediation. These findings underscore the critical role of integrated machine learning approaches in environmental science, offering a powerful tool for identifying contamination hotspots, supporting soil health management, and promoting sustainable land use.

Advanced three-dimensional prediction model based on stable machine learning for soil pollution: A case study from a contaminated site in Southern China

With over five million contaminated sites worldwide, accurately characterizing the three-dimensional (3D) distribution of soil contamination is critical for effective risk assessment and site remediation. However, current 3D interpolation methodologies often fail to simultaneously account for spatial correlation and spatial heterogeneity, both of which are critical for capturing the complex spatial structure of subsurface contamination. This study developed a refined 3D interpolation model that integrates site characteristics, spatial position, spatial correlation, and spatial heterogeneity to simulate site contamination and quantify prediction uncertainty. The proposed machine learning (ML) model achieved high predictive performance, with coefficient of determination (R2) values above 0.73 for four heavy metals (HMs). To enhance model generalizability, a stability analysis framework was developed alongside a novel model selection strategy based on random dataset partitioning and random ordering of input covariates, and 1000 random simulations could provide a reliable basis for model screening. This study introduces a new, precise 3D spatial interpolation method. Owing to the easy accessibility of its covariates, it offers high versatility, making a significant contribution to site assessment and remediation efforts.

Global Potential Risk of Thallium in Topsoil: A Cropland-Focused Quantification Framework

Thallium (Tl) pollution from natural and anthropogenic sources is increasingly recognized for its environmental and health risks, with localized threats in polluted areas despite low global background levels. Utilizing over 20,000 topsoil Tl measurements with 21 related environmental variables, a CatBoost classification model (AUC = 0.89, recall = 0.80, balanced accuracy = 0.84) was applied to predict whether global topsoil Tl concentrations exceeding 1 mg/kg, identifying both known and unreported hotspots. A CatBoost regression model (R2 = 0.62) further predicted Tl concentration distributions, highlighting regional variations. This study reveals that high-risk areas are highly overlapped with anthropogenic factors (mining activities and land cover) and geological conditions (mineralized zones, lithology, and geological structures), collectively influencing 14.81% of the model outputs. By integrating cropland cover maps with our predictions, we found that approximately 9.9% of the world's cropland has a greater than 47% probability of Tl concentrations exceeding 1 mg/kg, particularly in South America (34.7%), Asia (12.3%), and Africa (10.8%). These findings underscore the need for heightened attention to soil Tl testing in high-risk croplands to ensure agricultural safety.

Applications of machine learning in potentially toxic elemental contamination in soils: A review

Soil contamination by potentially toxic elements (PTEs) poses substantial risks to the environment and human health. Traditional investigational methods are often inadequate for large-scale assessments because they are time-consuming, costly, and have a limited accuracy. Machine learning (ML) techniques have emerged as promising tools in environmental studies because of their superiority in processing high-dimensional and unstructured data. However, critical evaluations of contemporary ML applications and methods in PTEs content, distribution, and identification remain scarce. To address this research gap, this study reviews applications of ML to soil PTEs contamination including content prediction, spatial distribution, source identification, and other related tasks. Hyperspectral data combined with ML methods can predict the content of PTEs in large-scale areas at a low cost. In addition, ML algorithms that integrate environmental covariates offer superior performance in spatial predictions compared with traditional geostatistical methods. Moreover, ML techniques incorporated with receptor models provide important advances in the quantitative identification and apportioning of PTE sources, thereby supporting effective environmental management and risk assessment. Based on the frequency of the variables used, we propose that soil pH, soil organic matter (SOM), industrial activities, soil texture, and other relevant factors are key environmental variables that enhance the accuracy of predictions regarding the spatial distribution and source identification of PTEs. From these findings, ML techniques, through their powerful data processing capabilities, provide new perspectives and tools for the efficient assessment and management of soil PTEs contamination.

Coupling a process-driven model with geographic information system enables the spatial predication of cadmium in paddy soils

Policies on the management of paddy fields are usually made at a broad scale and from a long-term perspective, while predicting the spatial extent of cadmium (Cd) contamination in paddy soils remains challenging. In this study, we developed a process-driven spatial model to quantify the transport of Cd in paddy soils and validated it against observed data from a 10-year regional investigation in southern China. Using a geographic information system and Monte Carlo simulation, the model was then applied to evaluate the effectiveness of different remediation strategies for contaminated paddy fields at field-to-regional scales in a 100-year period. In the last decade, atmospheric emissions have accounted for 43.5 % of the total Cd input in local paddy soils. However, the local clean air act failed to mitigate Cd contamination in 99.8 % of study area over the period of 2020-2120 because straw return became the dominant contributor to Cd inputs. Improving aerosol emission reductions by 3 % per year, stopping straw return to soil, and cleaning irrigation channels would take approximately 30 years (2020-2050) to protect 95 % of local rice production from causing an excessive human Cd kidney burden, especially in the paddy fields located in mining-affected areas.

Using machine learning to predict soil lead relative bioavailability

Although the relative bioavailability (RBA) can be applied to assess the effects of Pb on human health, there is no definition and no specific data of Pb-RBA to different soil sources and endpoints in vivo. In this study, we estimated the Pb-RBA from different soil sources and endpoints based on machine learning. The Pb-BAc and Pb-RBA in soils were found to be mostly in the range of 20-80 %, which is different from the USEPA Pb-RBA of 60 % in soils. The mean Pb-RBA for different biological endpoints in vivo predicted using the RF model were 49.94 ± 18.65 % for blood; 60.15 ± 26.62 %, kidney; 60.90 ± 21.51 %, liver; 50.70 ± 17.56 %, femur; and 62.89 ± 16.64 % as a combined measure. Pb-RBA of shooting range soils was 88.21 ± 16.92 % (mean), spiked/aged soils 77.11 ± 14.05 % and certified reference materials 73.70 ± 20.31 %; agricultural soil 68.28 ± 18.93 %, urban soil 64.36 ± 21.82 %, mining/smelting soils 53.99 ± 17.66 %, and industrial soils 47.71 ± 20.35 %. This study is first to define the Pb-RBA according to various soil sources and endpoints in vivo with the objective of providing more accurate Pb-RBA data for soil lead risk assessment.

High-accuracy spatial prediction of soil pollutants and their speciation in strong human-affected areas

Strong human activities greatly challenge the high-accuracy spatial prediction of soil pollutants and their speciation. This study first determined three auxiliary variables of soil total arsenic (TA) in a typical strong human-affected area, namely in-situ portable X-ray fluorescence (PXRF) TA calibrated by robust geographically weighted regression (RGWR), atmospheric deposition information simulated by atmospheric diffusion model (AERMOD), and land-use types. Then, robust residual cokriging with the above three auxiliary variables (RRCoK-RCPXRF/AD/LUT) was proposed to spatially predict soil TA. Finally, RGWR-robust ordinary kriging (RGWR-ROK) with the RRCoK-predicted soil TA was proposed to spatially predict soil As(III). The results show that: (i) RGWR obtained a higher spatial calibration accuracy (RI = 64.78%) for in-situ PXRF TA than the basic geographically weighted regression and traditionally-used ordinary least squares; (ii) The effect of auxiliary variables and model robustness on the prediction accuracy of soil TA is significant (RI > 14.33%); (iii) RRCoK-RCPXRF/AD/LUT achieved a higher prediction accuracy (RI = 58.87%) for soil TA than the other six traditional models; and (iv) RGWR-ROK achieved a higher prediction accuracy (RI = 55.26%) for soil As(III) than the other three traditional models. Therefore, this study provided a cost-effective solution for high-accuracy spatial prediction of soil pollutants and their speciation in strong human-affected areas.

Natural factor-based spatial prediction and source apportionment of typical heavy metals in Chinese surface soil: Application of machine learning models

Predicting the natural distribution of heavy metals (HMs) in soil is important to understand the potential risk of pollution. However, suitable technologies are still lacking for wide scale due to the large spatial heterogeneity. In this study, we developed machine learning models for predicting natural contents of five typical HMs in soil, including As, Cd, Cr, Hg and Pb. It was found that the optional random forest (RF) model had the best performance with the R2 up to 0.64. Based on this model, potential distribution of the five HMs explored that elevated contents were mainly concentrated in the southwest and south central of China. Feature analysis illustrated that importance of natural factors followed the order of geological attributes > soil properties > climatic conditions > ecological functions. In particular, lithology of the parent material dominated the content of metals, with the contributions of 18-25%. Moreover, soil properties of pH, cation exchange capacity, profile depth of soil and vegetation coverage had different influences on HMs, due to the variability in the properties of different HMs. This study developed a mapping relationship between natural factors and soil HMs by data science method, which may provide instructive information for pollution control and planning decisions.

Identification of driving factors for heavy metals and polycyclic aromatic hydrocarbons pollution in agricultural soils using interpretable machine learning

This study integrated data-driven interpretable machine learning (ML) with statistical methods, complemented by knowledge-driven discrimination diagrams, to identify the primary driving factors of heavy metal (HM) and polycyclic aromatic hydrocarbon (PAH) contamination in agricultural soils influenced by complex sources in a rapidly industrializing region of a megacity in southern China. First, the statistical characteristics of the concentrations of HMs and PAHs, and their correlations with the environmental covariates were explored. Three ML models and a statistical model comprising multiple environmental variable predictors were developed and assessed to predict the concentration of HMs in the agricultural soil. The Shapley Additive Explanations (SHAP) tool was introduced to reveal the influences of the main driving factors on pollutant concentrations. In addition, knowledge-based discrimination diagrams were adopted to discriminate the potential sources of the PAHs. Our findings indicated that Cd, Hg and Cu could be effectively predicted by the LightGBM and RF models. The identification of pollution drivers revealed that traffic emission, industry activity and irrigation significantly contributed to the pollution of Cd, Hg, Cu and high-ring PAHs in the study area, while the soil nature properties including SOM and pH also played crucial roles in influencing the HM and PAH concentrations. This work introduced an innovative approach to leverage ML for understanding complex urban soil pollution, thereby setting a precedent for data-driven environmental protection strategies to mitigate the pollution of HMs and PAHs. Future research is encouraged to optimize the models, enhance the prediction accuracy, and incorporate a broader range of influential parameters.

Machine learning combined with geodetector to predict the spatial distribution of soil heavy metals in mining areas

An accurate understanding of the spatial distribution of soil heavy metals (HMs) is crucial for the effective prevention of soil pollution and remediation strategies. Traditional machine learning models often overlook the spatially stratified heterogeneity inherent to environmental data, which can impair predictive accuracy. Therefore, we combined the Geodetector model (GDM) with machine learning models. The factor detection results were used to screen covariates to consider the local spatial heterogeneity of model features. The interaction detection results were used to construct spatially stratified covariates to consider the spatially stratified heterogeneity of model features. The results showed that covariate screening largely avoided the introduction of redundant features. The constructed spatially stratified covariates improved the predictive performance of the model (both the R-squared (R2) and root mean square error (RMSE) of different models were optimized). Among these, the XGB model exhibited the best performance. Analysis of the factors influencing Pb and Cr revealed that the interaction between pH and NDVI was the main determinant of Pb spatial distribution (q = 0.3516, XGB Importance Score = 93). In contrast, the interaction between DEM and pH (q = 0.7156, XGB Importance Score = 121) as well as the distance to waste piles (q = 0.6390, XGB Importance Score = 66), were the main driving factors for the spatial distribution of Cr. The current work provides an improved approach for interrogating the factors that influence HM distribution in soil. This study offers valuable insights into the spatial distribution of soil HMs. The proposed methodology can be applied in future soil pollution assessments and environmental management strategies, thus contributing to more precise pollution prevention and remediation efforts.

Machine learning models with innovative outlier detection techniques for predicting heavy metal contamination in soils

Machine learning (ML) models for accurately predicting heavy metals with inconsistent outputs have improved owing to dataset outliers, which influence model reliability and accuracy. A comprehensive technique that combines machine learning and advanced statistical methods was applied to assess data outlier's effects on ML models. Ten ML models with three outlier detection methods predicted Cr, Ni, Cd, and Pb in Narayanganj soils. XGBoost with density-based spatial clustering of applications with noise (DBSCAN) improved model efficacy (R2). The R2 of Cr, Ni, Cd, and Pb was considerably enhanced by 11.11 %, 6.33 %, 14.47 %, and 5.68 %, respectively, indicating that outliers affected the model's HM prediction. Soil factors affected Cr (80 %), Ni (72.61 %), Cd (53.35 %), and Pb (63.47 %) concentrations based on feature importance. Contamination factor prediction showed considerable contamination for Cr, Ni, and Cd. LISA revealed Cd (55.4 %), Cr (49.3 %), and Pb (47.3 %) as the significant pollutant (p < 0.05). Moran's I index values for Cr, Ni, Cd, and Pb were 0.65, 0.58, 0.60, and 0.66, respectively, indicating strong positive spatial autocorrelation and clusters with similar contamination. Finally, this work successfully assessed the influence of data outliers on the ML model for soil HM contamination prediction, identifying crucial regions that require rapid conservation measures.

Systematic bibliographic analysis of heavy metal remediation

Heavy metals pose a significant threat to human health, with contaminated water sources linked to severe conditions, including gastric cancer. Consequently, the effective remediation of heavy metals is crucial. This study employs a bibliographic analysis to examine key methodologies, leading organizations, and prominent countries involved in heavy metal remediation. By systematically reviewing around 1,000 records, the paper identifies the most critical remediation techniques and provides a comprehensive overview of current practices in the field. Additionally, the study explores prospects, emphasizing the potential of emerging technologies such as big data and machine learning to enhance remediation efforts. It highlights recent advancements, identifies significant trends, such as the growing use of bioremediation and nanotechnology, and addresses critical challenges in the remediation landscape, including regulatory hurdles and technological limitations. By making stronger connections between the identified trends and their implications for future research, this comprehensive analysis aims to provide valuable insights and guide the development of improved strategies for mitigating the impact of heavy metal contamination, ultimately safeguarding public health.

Applications of geographically weighted machine learning models for predicting soil heavy metal concentrations across mining sites

The accurate prediction of soil heavy metal contamination is crucial for the effective environmental management of abandoned mining areas. However, conventional machine learning models (CMLMs) often fail to account for the spatial heterogeneity of soil contamination, which limits their predictive accuracy. This study evaluated the performance of geographically weighted machine learning models (GWMLMs) in predicting soil Cd and Pb concentrations in abandoned mines in the Republic of Korea. We compared two GWMLMs (Geographically Weighted Random Forest and Geographically Weighted Extreme Gradient Boosting) with four CMLMs (Random Forest, Gradient Boosting, Light Gradient Boosting, and extreme Gradient Boosting). The data used in this study included soil samples from six abandoned mining sites with various geographical and soil input variables. The results showed that the GWMLMs consistently outperformed the CMLMs in predicting heavy metal contamination. For Cd predictions, GWMLMs exhibited on average 0.02 lower root mean square error and mean absolute error values, with a 0.26 increase in R2 values compared to CMLMs. Similarly, for Pb predictions, the GWMLMs showed 0.18 and 0.13 lower root mean square error and mean absolute error values, respectively, and a 0.17 increase in R2 relative to the CMLMs. The findings demonstrate the usefulness of GWMLMs for predicting the spatial distribution of soil heavy metals. SHapley Additive exPlanations analysis exhibited elevation and distance from abandoned mining sites as the most influential factors in predicting both Cd and Pb concentrations. This study highlights the value of GWMLMs that incorporate spatial heterogeneity into CMLMs for enhancing prediction accuracy and providing crucial insights for environmental management in mining-impacted regions.

Heavy metals in roadside soil along an expressway connecting two megacities in China: Accumulation characteristics, sources and influencing factors

Transportation is widely recognized as a significant contributor to heavy metal (HM) pollution in roadside soils. A better understanding of HM pollution in soils near expressways is crucial, particularly given the rapid expansion of expressway transportation in China in recent years. In this study, 329 roadside topsoil samples were collected along the Beijing-Tianjin Expressway, which connects two megacities in China. Chemical analysis showed that HM concentrations in the soil samples were generally below national limits. The mean pollution index (Pi) values for As, Cr, Cu, Ni, Pb, and Zn ranged from 0.94 to 1.01, while Cd and Hg exhibited slightly higher mean Pi values of 1.19 and 1.13, respectively. The Nemerow integrated pollution index values for all samples ranged from 0.71 to 4.97, with a mean of 1.26. This suggests a slight enrichment of HM above natural background levels, especially for Cd and Hg. Source apportionment using positive matrix factorization revealed that natural sources contributed the most to soil HMs (64.51 %), followed by agricultural sources (19.15 %), traffic sources (9.77 %), and industrial sources (6.57 %). The Shapley additive explanation analysis, based on the random forest model, identified soil organic carbon, deep soil HM content, altitude, total soil K2O, urbanization composite impact index, and total soil P as primary influencing factors. This indicates that the impact of transportation on roadside soils along the Beijing-Tianjin Expressway is currently relatively limited. The prominent influence of soil properties and altitude underscored the importance of "transport" and "receptor" in the soil HMs accumulation process at the local scale. These findings provide critical data and a scientific basis for decision-makers to develop policies for expressway design and roadside soil protection.

Determining the priority control sources of heavy metals in the roadside soils in a typical industrial city of North China

Heavy metals (HMs) in roadside soils (RS) are closely related to urban ecosystem and human health, but priority control sources and eco-health risks of HMs remain unclear. We explored pollution sources of HMs using positive matrix factorization (PMF), scanning electron microscopy with energy dispersive spectroscopy (SEM-EDS), and random forest (RF), and assessed their source-specific eco-health risks. The combination of PMF, SEM-EDS and RF indicated that pollution sources of HMs were coal combustion (27.99 %), construction materials (30.15 %), and traffic emissions (41.86 %). SEM-EDS revealed RS particles were identified by a combination of physical (spherical, porous rounded, crustal flake, crustal rounded and irregular particles) and elemental surface characteristics (O, C, Si, Al, Fe, Ca, Mg, K, Zn, Cu, and Mn). RF highlighted road network density, residential density, and industrial density had significant importance influence on pollution sources. Approximately 72.35 % of soil samples were at a low ecological risk with traffic emissions being the major contributor. Non-carcinogenic risks had a minimal effect, but carcinogenic risks were at a cautionary level with coal combustion being the highest contributor. Overall, coal combustion and traffic emissions were regarded as priority control sources of HMs. These findings provided effective guidance for soil pollution prevention and risk control.

Single-cell atlases reveal leaf cell-type-specific regulation of metal transporters in the hyperaccumulator Sedum alfredii under cadmium stress

Hyperaccumulation in plants is a complex and dynamic biological process. Sedum alfredii, the most studied Cd hyperaccumulator, can accumulate up to 9000 mg kg-1 Cd in its leaves without suffering toxicity. Although several studies have reported the molecular mechanisms of Cd hyperaccumulation, our understanding of the cell-type-specific transcriptional regulation induced by Cd remains limited. In this study, the first full-length transcriptome of S. alfredii was generated using the PacBio Iso-Seq technology. A total of 18,718,513 subreads (39.90 Gb) were obtained, with an average length of 2133 bp. The single-cell RNA sequencing was employed on leaves of S. alfredii grown under Cd stress. A total of 12,616 high-quality single cells were derived from the control and Cd-treatment samples of S. alfredii leaves. Based on cell heterogeneity and the expression profiles of previously reported marker genes, seven cell types with 12 transcriptionally distinct cell clusters were identified, thereby constructing the first single-cell atlas for S. alfredii leaves. Metal transporters such as CAX5, COPT5, ZIP5, YSL7, and MTP1 were up-regulated in different cell types of S. alfredii leaves under Cd stress. The distinctive gene expression patterns of metal transporters indicate special gene regulatory networks underlying Cd tolerance and hyperaccumulation in S. alfredii. Collectively, our findings are the first observation of the cellular and molecular responses of S. alfredii leaves under Cd stress and lay the cornerstone for future hyperaccumulator scRNA-seq investigations.

Environment-compatible heavy metal risk prediction method created with multilevel ensemble learning

Accurate health risk prediction (HRP) is an effective means of reducing the hazards of heavy metal (HM) exposure. It can address the drawbacks of lag and passivity faced by health risk assessment. This study innovatively proposed an HRP method, MEL-HR, based on multilevel ensemble learning (MEL) technology and environment compatibility. We conducted point and interval prediction experiments on health risks using 490 sets of data covering 17 environment factors. The point prediction results indicated that when the model predicts HI and TCR, the R2 values were 0.707 and 0.619, respectively. For P5, P50, and P95 in interval prediction, the R2 values of the model were 0.706, 0.703, and 0.672 for HI, and that for TCR were 0.620, 0.607, and 0.616, respectively. The analysis of feature importance indicated that, in addition to HM factors, longitude, mining area coefficient, and soil organic matter were key environmental factors affecting the MEL-HR model. Comparative experiments showed that compared to soil HMs-based MEL-HR, environment compatibility-based MEL-HR has improved the accuracy for HI and TCR by 19.83 % and 40.36 % for the point prediction and 22.06 % and 40.01 % for interval prediction. This study can provide technical support for targeted and resilient prevention and control of health risks.

Artificial intelligence in microplastic detection and pollution control

The rising prevalence of microplastics (MPs) in various ecosystems has increased the demand for advanced detection and mitigation strategies. This review examines the integration of artificial intelligence (AI) with environmental science to improve microplastic detection. Focusing on image processing, Fourier transform infrared spectroscopy (FTIR), Raman spectroscopy, and hyperspectral imaging (HSI), the review highlights how AI enhances the efficiency and accuracy of these techniques. AI-driven image processing automates the identification and quantification of MPs, significantly reducing the need for manual analysis. FTIR and Raman spectroscopy accurately distinguish MP types by analyzing their unique spectral features, while HSI captures extensive spatial and spectral data, facilitating detection in complex environmental matrices. Furthermore, AI algorithms integrate data from these methods, enabling real-time monitoring, traceability prediction, and pollution hotspot identification. The synergy between AI and spectral imaging technologies represents a transformative approach to environmental monitoring and emphasizes the need to adopt innovative tools for protecting ecosystem health.

Risk hotspots and influencing factors identification of heavy metal(loid)s in agricultural soils using spatial bivariate analysis and random forest

Heavy metal(loid)s (HMs) in agricultural soils not only affect soil function and crop security, but also pose health risks to residents. However, previous concerns have typically focused on only one aspect, neglecting the other. This lack of a comprehensive approach challenges the identification of hotspots and the prioritization of factors for effective management. To address this gap, a novel method incorporating spatial bivariate analysis with random forest was proposed to identify high-risk hotspots and the key influencing factors. A large-scale dataset containing 2995 soil samples and soil HMs (As, Cd, Cr, Cu, Mn, Ni, Pb, Sb, and Zn) was obtained from across Henan province, central China. Spatial bivariate analysis of both health risk and ecological risks revealed risk hotspots. Positive matrix factorization model was initially used to investigate potential sources. Twenty-two environmental variables were selected and input into random forest to further identify the key influencing factors impacting soil accumulation. Results of local Moran's I index indicated high-high HM clusters at the western and northern margins of the province. Hotspots of high ecological and health risk were primarily observed in Xuchang and Nanyang due to the widespread township enterprises with outdated pollution control measures. As concentration and exposure frequency dominated the non-carcinogenic and carcinogenic risks. Anthropogenic activities, particularly vehicular traffic (contributing ∼37.8 % of the total heavy metals accumulation), were the dominant sources of HMs in agricultural soils. Random forest modeling indicated that soil type and PM2.5 concentrations were the most influencing natural and anthropogenic variables, respectively. Based on the above findings, control measures on traffic source should be formulated and implemented provincially; in Xuchang and Nanyang, scattered township enterprises with outdated pollution control measures should be integrated and upgraded to avoid further pollution from these sources.

Utilizing machine learning to evaluate heavy metal pollution in the world's largest mangrove forest

The world's largest mangrove forest (Sundarbans) is facing an imminent threat from heavy metal pollution, posing grave ecological and human health risks. Developing an accurate predictive model for heavy metal content in this area has been challenging. In this study, we used machine learning techniques to model sediment pollution by heavy metals in this vital ecosystem. We collected 199 standardized sediment samples to predict the accumulation of eleven heavy metals using ten different machine learning algorithms. Among them, the extremely randomized tree model exhibited the best performance in predicting Fe (0.87), Cr (0.89), Zn (0.85), Ni (0.83), Cu (0.87), Co (0.62), As (0.68), and V (0.90), achieving notable R2 values. On the other hand, the random forest outperformed for predicting Cd (0.72) and Mn (0.91), whereas the decision tree model showed the best performance for Pb (0.73). The feature attribute analysis identified FeV, CrV, CuZn, CoMn, PbCd, and AsCd relationships resembled with correlation coefficients among them. Based on the established models, the prediction of the contamination factor of metals in sediments showed very high Cd contamination (CF ≥ 6). The Moran's I index for Cd, Cr, Pb, and As were 0.71, 0.81, 0.71, and 0.67, respectively, indicating strong positive spatial autocorrelation and suggesting clustering of similar contamination levels. Conclusively, this research provides a comprehensive framework for predicting heavy metal sediment pollution in the Sundarbans, identifying key areas needing urgent conservation. Our findings support the adoption of integrated management strategies and targeted remedial actions to mitigate the harmful effects of heavy metal contamination in this vital ecosystem.

Prediction of soil heavy metal contents in urban residential areas and the strength of deep learning: A case study of Beijing

Predicting soil heavy metal (SHM) content is crucial for understanding SHM pollution levels in urban residential areas and guide efforts to reduce pollution. However, current research indicates low SHM prediction accuracy in urban areas. Therefore, we employed a deep learning method (fully connected deep neural network) alongside four other methods (muti-layer perceptron, radial basis function neural network, multiple stepwise linear regression, and Kriging interpolation) to predict SHM content in the urban residential areas of Beijing and demonstrated the strength of deep learning in improving prediction accuracy. We found the contents of the evaluated heavy metals (Cd, Cu, Pb, and Zn) exhibited significant correlations with numerous other soil physicochemical properties and environmental factors. The prediction accuracy for Cu, Pb, and Zn contents was relatively high across different methods. Notably, deep learning showed considerable strength in predicting the contents of the four heavy metals, with the R2 for the test set of the model ranging from 0.75 to 0.91. Compared to other methods, deep learning achieved markedly higher prediction accuracy according to different accuracy evaluation indicators (e.g., deep learning showed increases in the cumulative R2 of the four heavy metals ranging from 53.16 % to 187.36 % compared to other methods). Our study indicates that deep learning can significantly improve SHM content prediction accuracy in urban areas and is highly applicable in urban residential areas with complex environmental influences.

Advancing source apportionment of soil potentially toxic elements using a hybrid model: a case study in urban parks, Beijing, China

Identifying the source-specific health risks of potentially toxic elements (PTE) in urban park soils is essential for human health protection. However, previous studies have mostly focused on the deterministic source-specific health risks, ignoring the health risk assessment from a probabilistic perspective. To fill this gap, we developed a hybrid model that incorporated machine learning (ML) interpretability into positive matrix factorization (PMF) and probability health risk assessment (PHRA) based on the Monte Carlo simulation. The results indicated that concentrations of soil PTEs except for Mn and Sb were significantly higher than their corresponding background values. Random forest (RF) was regarded as the best ML model to identify key drivers for As, Cd, Cr, Cu, Ni, Pb, and Zn, with R2 > 0.60, but was less effective for other soil PTEs (R2 < 0.49). Specifically, the contributions of the four potential pollution sources were mixed sources, traffic emission, fuel combustion, and building materials, with contribution rate of 24.88%, 30.56%, 28.99%, and 15.56%, respectively. Fuel combustion contributed the most to non-carcinogenic for children (39.45%), male (43.84%), and female (43.76%), and the non-carcinogenic risk could be considered negligible for human. However, building materials was the major contributor to carcinogenic risk for children (36.1%), male (44.9%), and female (43.2%). The integration of the RF model with PMF and PHRA improved the accuracy of the results by identifying and quantifying the specific sources of each soil PTE using the relative importance analysis from the RF model. The results of this study assisted in providing efficient strategies for risk management and control of soil PTEs in Beijing parks.

Spatial heterogeneity: Necessary and feasible for revealing soil trace elements pollution, sources, risks, and their links

The source diversity and health risk of trace elements (TEs) in soil make it necessary to reveal the relationship between pollution, source, and risk. However, neglect of spatial heterogeneity restricts the reliability of existing identification methods. In this study, spatial heterogeneity is proposed as a necessary and feasible factor for accurately dissecting the pollution-source-risk link of soil TEs. A comprehensive framework is developed by integrating positive matrix factorization, Geodetector, and risk evaluation tools, and successfully applied in a mining-intensive city in northern China. Overall, the TEs are derived from natural background (28.5 %), atmospheric deposition (25.6 %), coal mining (21.8 %), and metal industry (24.1 %). The formation mechanism of heterogeneity for high-variance TEs (Se, Hg, Cd) is first systematically deciphered by revealing the heterogeneous source-sink relationship. Specifically, Se is dominated (76.5 %) by heterogeneous coal mining (q=0.187), Hg is determined (92.6 %) by the heterogeneity of metal mining (q=0.183) and smelting (q=0.363), and Cd is caused (50.9 %) by heterogeneous atmospheric deposition (q>0.254) co-influenced by the terrains and soil properties. Highly heterogeneous sources are also noteworthy for their potential to pose extreme risks (THI=1.122) in local areas. This study highlights the necessity of integrating spatial heterogeneity in pollution and risk assessment of soil TEs.

Evaluation of fluoride emissions and pollution from an electrolytic aluminum plant located in Yunnan province

The monitoring and evaluation of fluoride pollution are essentially important to make sure that concentrations do not exceed threshold limit, especially for surrounding atmosphere and soil, which are located close to the emission source. This study aimed to describe the atmospheric HF and edaphic fluoride distribution from an electrolytic aluminum plant located in Yunnan province, on which the effects of meteorological conditions, time, and topography were explored. Meanwhile, six types of solid waste genereted from different electrolytic aluminum process nodes were characterized to analyze the fluoride content and formation characteristics. The results showed that fluoride in solid waste mainly existed in the form of Na3AlF6, AlF3, CaF2, and SiF4. Spent electrolytes, carbon residue, and workshop dust are critical contributors to fluoride emissions in the primary aluminum production process, and the fluorine content is 17.14 %, 33.30 %, and 31.34 %, respectively. Unorganized emissions from electrolytic aluminum plants and solid waste generation are the primary sources of fluoride in the environment, among which the edaphic fluoride content increases most at the sampling sites S1 and S7. In addition, the atmospheric HF concentration showed significant correlations with wind speed, varying wildly from March to September, with daily average and hourly maximum HF concentrations of 4.32 μg/m3 and 9.0 μg/m3, respectively. The results of the study are crucial for mitigating fluorine pollution in the electrolytic aluminum industry.

Improved mapping of heavy metals in agricultural soils using machine learning augmented with spatial regionalization indices

The accurate spatial mapping of heavy metal levels in agricultural soils is crucial for environmental management and food security. However, the inherent limitations of traditional interpolation methods and emerging machine-learning techniques restrict their spatial prediction accuracy. This study aimed to refine the spatial prediction of heavy metal distributions in Guangxi, China, by integrating machine learning models and spatial regionalization indices (SRIs). The results demonstrated that random forest (RF) models incorporating SRIs outperformed artificial neural network and support vector regression models, achieving R2 values exceeding 0.96 for eight heavy metals on the test data. Hierarchical clustering for feature selection further improved the model performance. The optimized RF models accurately predicted the heavy metal distributions in agricultural soils across the province, revealing higher levels in the central-western regions and lower levels in the north and south. Notably, the models identified that 25.78 % of agricultural soils constitute hotspots with multiple co-occurring heavy metals, and over 6.41 million people are exposed to excessive soil heavy metal levels. Our findings provide valuable insights for the development of targeted strategies for soil pollution control and agricultural soil management to safeguard food security and public health.

Accurate prediction of spatial distribution of soil potentially toxic elements using machine learning and associated key influencing factors identification: A case study in mining and smelting area in southwestern China

Accurate prediction of spatial distribution of potentially toxic elements (PTEs) is crucial for soil pollution prevention and risk control. Achieving accurate prediction of spatial distribution of soil PTEs at a large scale using conventional methods presents significant challenges. In this study, machine learning (ML) models, specially artificial neural network (ANN), random forest (RF), and extreme gradient boosting (XGB), were used to predict spatial distribution of soil PTEs and identify associated key factors in mining and smelting area located in Yunnan Province, China, under the three scenarios: (1) natural + socioeconomic + spatial datasets (NS), (2) NS + irrigation pollution index (IPI) datasets, (3) NS + IPI + deposition (DEPO) datasets. The results highlighted the combination of NS+IPI+DEPO yielded the highest predictive accuracy across ML models. Particularly, XGB exhibited the highest performance for As (R2 =0.7939), Cd (R2 =0.6679), Cu (R2 =0.8519), Pb (R2 =0.8317), and Zn (R2 =0.7669), whereas RF performed the best for Ni (R2 =0.7146). The feature importance and Shapley additive explanation (SHAP) analysis revealed that DEPO and IPI were the pivotal factors influencing the distribution of soil PTEs. Our findings highlighted the important role of DEPO in spatial distribution prediction of soil PTEs, which has often been ignored in previous studies.

Geochemical baseline establishment and accumulation characteristics of soil heavy metals in Sabaochaqu watershed at the source of Yangtze River, Qinghai-Tibet Plateau

The establishment of soil geochemical baseline and heavy metal pollution assessment in the Qinghai-Tibet Plateau is of great significance for guiding environmental management in the high-cold and high-altitude regions. A total of 126 topsoil samples (0-20 cm) were collected and the contents of Cu, Pb, Zn, Ni, Cr, Cd, As and Hg were determined in the Sabaochaqu basin of the Tuotuo River, the source of the Yangtze River, in the Tibetan Plateau. The baseline values of 8 heavy metals were determined by mathematical statistics, iterative 2times standard deviation method, cumulative frequency and reference element standardization, and the soil heavy metal pollution in the study area was assessed by enrichment factor method and pollution index method. The results showed that the average contents of As, Cd, Cr, Cu, Hg, Ni, Pb and Zn were 31.84, 0.29, 66.07, 17.35, 0.021, 27.86, 49.35 and 88.56 mg/kg, respectively. Baseline values were 22.24, 0.217, 64.16, 15.69, 0.0191, 26.46, 34.91, and 68.62 mg/kg, respectively. There is a great difference between the baseline value of soil heavy metals in study area and the Xizang soil background value, especially the baseline value of Cd was 2.68 times of its background value. The results of the pollution evaluation based on the baseline values showed that the 8 heavy metals were slightly enriched, and the overall pollution status was light pollution, and measures should be taken to control and manage them. The research results can provide a reference value for the evaluation of soil heavy metal pollution in the source region of the Yangtze River, and also provide a theoretical basis for the construction of soil heavy metal baseline values in similar high-cold and high-altitude regions.

Predictive analysis and risk assessment of potentially toxic elements in Beijing gas station soils using machine learning and two-dimensional Monte Carlo simulations

Gas stations not only serve as sites for oil storage and refueling but also as locations where vehicles frequently brake, significantly enriching the surrounding soil with potentially toxic elements (PTEs). Herein, 117 topsoil samples from gas stations were collected in Beijing to explore the impact of gas stations on PTE accumulation. The analysis revealed that the average Pollution Index (PI) values for Cd, Hg, Pb, Cu, and Zn in the soil samples all exceeded 1. The random forest (RF) model, achieving an AUC score of 0.95, was employed to predict PTE pollution at 372 unsampled gas stations. Additionally, a Positive Matrix Factorization (PMF) model indicated that gas station operations and vehicle emissions were responsible for 70 % of the lead (Pb) enrichment. Probabilistic health risk assessments showed that the carcinogenic risk (CR) and noncarcinogenic risk (NCR) for PTE pollution to adult females were the highest, at 0.451 and 1.61E-05 respectively, but still within acceptable levels. For adult males at contaminated sites, the Pb-associated CR and NCR were approximately twice as high as those at uncontaminated sites, with increases of 107 % and 81 %, respectively. This study provides new insights for managing pollution caused by gas stations.

Perspectives on Advancing Multimodal Learning in Environmental Science and Engineering Studies

The environment faces increasing anthropogenic impacts, resulting in a rapid increase in environmental issues that undermine the natural capital essential for human wellbeing. These issues are complex and often influenced by various factors represented by data with different modalities. While machine learning (ML) provides data-driven tools for addressing the environmental issues, the current ML models in environmental science and engineering (ES&E) often neglect the utilization of multimodal data. With the advancement in deep learning, multimodal learning (MML) holds promise for comprehensive descriptions of the environmental issues by harnessing data from diverse modalities. This advancement has the potential to significantly elevate the accuracy and robustness of prediction models in ES&E studies, providing enhanced solutions for various environmental modeling tasks. This perspective summarizes MML methodologies and proposes potential applications of MML models in ES&E studies, including environmental quality assessment, prediction of chemical hazards, and optimization of pollution control techniques. Additionally, we discuss the challenges associated with implementing MML in ES&E and propose future research directions in this domain.

Exploring the environmental risks and seasonal variations of potentially toxic elements (PTEs) in fine road dust in resource-based cities based on Monte Carlo simulation, geo-detector and random forest model

The environmental pollution caused by mineral exploitation and energy consumption poses a serious threat to ecological security and human health, particularly in resource-based cities. To address this issue, a comprehensive investigation was conducted on potentially toxic elements (PTEs) in road dust from different seasons to assess the environmental risks and influencing factors faced by Datong City. Multivariate statistical analysis and absolute principal component score were employed for source identification and quantitative allocation. The geo-accumulation index and improved Nemerow index were utilized to evaluate the pollution levels of PTEs. Monte Carlo simulation was employed to assess the ecological-health risks associated with PTEs content and source orientation. Furthermore, geo-detector and random forest analysis were conducted to examine the key environmental variables and driving factors contributing to the spatiotemporal variation in PTEs content. In all PTEs, Cd, Hg, and Zn exhibited higher levels of content, with an average content/background value of 3.65 to 4.91, 2.53 to 3.34, and 2.15 to 2.89 times, respectively. Seasonal disparities were evident in PTEs contents, with average levels generally showing a pattern of spring (winter) > summer (autumn). PTEs in fine road dust (FRD) were primarily influenced by traffic, natural factors, coal-related industrial activities, and metallurgical activities, contributing 14.9-33.9 %, 41.4-47.5 %, 4.4-8.3 %, and 14.2-29.4 % to the total contents, respectively. The overall pollution and ecological risk of PTEs were categorized as moderate and high, respectively, with the winter season exhibiting the most severe conditions, primarily driven by Hg emissions from coal-related industries. Non-carcinogenic risk of PTEs for adults was within the safe limit, yet children still faced a probability of 4.1 %-16.4 % of unacceptable risks, particularly in summer. Carcinogenic risks were evident across all demographics, with children at the highest risk, mainly due to Cr and smelting industrial sources. Geo-detector and random forest model indicated that spatial disparities in prioritized control elements (Cr and Hg) were primarily influenced by particulate matter (PM10) and anthropogenic activities (industrial and socio-economic factors); variations in particulate matter (PM10 and PM2.5) and meteorological factors (wind speed and precipitation) were the primary controllers of seasonal disparities of Cr and Hg.

Machine learning-based prediction of cadmium pollution in topsoil and identification of critical driving factors in a mining area

Mining activities have resulted in a substantial accumulation of cadmium (Cd) in agricultural soils, particularly in southern China. Long-term Cd exposure can cause plant growth inhibition and various diseases. Rapid identification of the extent of soil Cd pollution and its driving factors are essential for soil management and risk assessment. However, traditional geostatistical methods are difficult to simulate the complex nonlinear relationships between soil Cd and potential features. In this study, sequential extraction and hotspot analyses indicated that Cd accumulation increased significantly near mining sites and exhibited high mobility. The concentration of Cd was estimated using three machine learning models based on 3169 topsoil samples, seven quantitative variables (soil pH, Fe, Ca, Mn, TOC, Al/Si and ba value) and three quantitative variables (soil parent rock, terrain and soil type). The random forest model achieved marginally better performance than the other models, with an R2 of 0.78. Importance analysis revealed that soil pH and Ca and Mn contents were the most significant factors affecting Cd accumulation and migration. Conversely, due to the essence of controlling Cd migration being soil property, soil type, terrain, and soil parent materials had little impact on the spatial distribution of soil Cd under the influence of mining activities. Our results provide a better understanding of the geochemical behavior of soil Cd in mining areas, which could be helpful for environmental management departments in controlling the diffusion of Cd pollution and capturing key targets for soil remediation.

A novel prediction approach driven by graph representation learning for heavy metal concentrations

The potential risk of heavy metals (HMs) to public health is an issue of great concern. Early prediction is an effective means to reduce the accumulation of HMs. The current prediction methods rarely take internal correlations between environmental factors into consideration, which negatively affects the accuracy of the prediction model and the interpretability of intrinsic mechanisms. Graph representation learning (GraRL) can simultaneously learn the attribute relationships between environmental factors and graph structural information. Herein, we developed the GraRL-HM method to predict the HM concentrations in soil-rice systems. The method consists of two modules, which are PeTPG and GCN-HM. In PeTPG, a graphic structure was generated using graph representation and communitization technology to explore the correlations and transmission paths of different environmental factors. Subsequently, the GCN-HM model based on the graph convolutional neural network (GCN) was used to predict the HM concentrations. The GraRL-HM method was validated by 2295 sets of data covering 21 environmental factors. The results indicated that the PeTPG model simplified correlation paths between factor nodes from 396 to 184, reducing by 53.5 % graph scale by eliminating the invalid paths. The concise and efficient graph structure enhanced the learning efficiency and representation accuracy of downstream prediction models. The GCN-HM model was superior to the four benchmark models in predicting the HM concentration in the crop, improving R2 by 36.1 %. This study develops a novel approach to improve the prediction accuracy of pollutant accumulation and provides valuable insights into intelligent regulation and planting guidance for heavy metal pollution control.

Improving prediction of soil heavy metal(loid) concentration by developing a combined Co-kriging and geographically and temporally weighted regression (GTWR) model

The study of heavy metal(loid) (HM) contamination in soil using extensive data obtained from published literature is an economical and convenient method. However, the uneven distribution of these data in time and space limits their direct applicability. Therefore, based on the concentration data obtained from the published literature (2000-2020), we investigated the relationship between soil HM accumulation and various anthropogenic activities, developed a hybrid model to predict soil HM concentrations, and then evaluated their ecological risks. The results demonstrated that various anthropogenic activities were the main cause of soil HM accumulation using Geographically and temporally weighted regression (GTWR) model. The hybrid Co-kriging + GTWR model, which incorporates two of the most influential auxiliary variables, can improve the accuracy and reliability of predicting HM concentrations. The predicted concentrations of eight HMs all exceeded the background values for soil environment in China. The results of the ecological risk assessment revealed that five HMs accounted for more than 90% of the area at the "High risk" level (RQ ≥ 1), with the descending order of Ni (100%) = Cu (100%) > As (98.73%) > Zn (95.50%) > Pb (94.90%). This study provides a novel approach to environmental pollution research using the published data.

Major influencing factors identification and probabilistic health risk assessment of soil potentially toxic elements pollution in coal and metal mines across China: A systematic review

Deposition of potentially toxic elements (PTEs) in soils due to different types of mining activities has been an increasingly important concern worldwide. Quantitative differences of soil PTEs contamination and related health risk among typical mines remain unclear. Herein, data from 110 coal mines and 168 metal mines across China were analyzed based on 265 published literatures to evaluate pollution characteristics, spatial distribution, and probabilistic health risks of soil PTEs. The results showed that PTE levels in soil from both mine types significantly exceeded background values. The geoaccumulation index (Igeo) revealed metal-mine soil pollution levels exceeded those of coal mines, with average Igeo values for Cd, Hg, As, Pb, Cu, and Zn being 3.02-15.60 times higher. Spearman correlation and redundancy analysis identified natural and anthropogenic factors affecting soil PTE contamination in both mine types. Mining activities posed a significant carcinogenic risk, with metal-mine soils showing a total carcinogenic risk an order of magnitude higher than in coal-mine soils. This study provides policymakers a quantitative foundation for developing differentiated strategies for sustainable remediation and risk-based management of PTEs in typical mining soils.

Boosted simultaneous removal of chlortetracycline and Cu (II) by Litchi Leaves Biochar: Influence of pH, ionic strength, and background electrolyte ions

The coexistence of heavy metals and antibiotics in the environment always results in greater toxicity compared to the individual precursors. Therefore, efficient and economic technology for the simultaneous removal of antibiotics and heavy metals is essential. Herein, litchi leaves biochar carbonized at 550 °C (L550) demonstrated high efficiency in co-removal of CTC (1838.1 mmol/kg) and Cu (II) (1212.9 mmol/kg) within wide range of pH (pH 4-7). Ionic strength obviously enhanced the Cu (II) removal but showed no significant effect on CTC removal. Although Al3+ and HPO42- decreased the adsorption capacities of CTC and Cu (II) on L550, the coexistence of Na+, K+, Mg2+, Cl-, NO3-, CO32- and SO42- showed a negligible effect on the simultaneous removal of CTC and Cu (II). Moreover, the adsorption capacities of CTC and Cu (II) on L550 were excellent in the river water, tap water, and lake water. In addition to electrostatic interactions, ion exchange governed Cu (II) adsorption, while surface complexation played a key role in CTC adsorption on L550. Our results demonstrated that litchi leaves biochar could be a promising adsorbent for remediating multi-contaminated environments.