Prioritizing geochemical drivers of groundwater quality and health risks in coastal aquifers of Bangladesh using machine learning algorithms.

This study explores the use of machine learning (ML)techniquesto predict Fourier-transform infrared (FTIR) intensities of productsfrom the thermal cracking of Athabasca bitumen, aiming to developa reliable soft-sensor. The ultimate goal is to obtain the FTIR spectraof the thermally cracked products online to reduce process time fromslow physical measurements. Various ML models, including Linear Regression(LinR), partial least squares regression (PLSR), support vector regression(SVR), K-nearest neighbors (k-NN), random forest(RF), and gradient boosting regression (GBR), were implemented toenhance the predictive accuracy and efficiency of FTIR spectroscopy,aiming to reduce the need for traditional physical measurements whichare often slow compared to the rapid predictions offered by ML techniques.To assess the model’s generalization capabilities, with respectto model predictions, the models were trained and tested across fourdifferent scenarios with varying temperature data obtained from visbreakingexperiments performed on Athabasca Bitumen at temperatures rangingfrom 25 to 420 °C with reaction times ranging from 15 min to27 h. Scenario 1 included all 61,740 data points utilizing an 80/20train-test split with 10-fold cross-validation (CV). Scenario 2 involvedtraining on temperatures of 25, 350, and 400 °C and testing on300, 380, and 420 °C. Scenario 3 involved training on temperaturesof 350, 380, and 400 °C and testing on 25, 300, and 420 °C.Finally, Scenario 4 involved training on temperatures of 25, 300,350, and 380 °C and testing on 400 and 420 °C. Bayesianoptimization was employed for hyperparameter tuning to identify theoptimal configurations for each model. The results indicate that ensemblemethods, particularly GBR, consistently achieved the highest predictiveaccuracy (R2) and lowest root mean squarederror (RMSE) across all scenarios. In Scenario 1, GBR achieved a predictionaccuracy of 99.66%. Scenario 2 highlighted the models’ abilityto generalize across varying temperatures, with both RF and GBR achievingsimilar performance with high prediction accuracies of around 94%.Scenario 3, characterized by significant temperature variability,demonstrated the robustness of GBR, which outperformed RF and k-NN with a predictive accuracy of 92.15%. Scenario 4, focusingon high-temperature predictions from low-temperature training data,showed that GBR still performed robustly with a predictive accuracyof 80.40%. The study concludes that GBR models, particularly thosewith well-tuned hyperparameters, are highly effective in predictingFTIR intensities, outperforming other techniques like RF, k-NN, LinR, and PLSR. The integration of advanced ML techniquesand Bayesian optimization significantly enhances the capability topredict FTIR spectra, providing a reliable soft-sensor as an alternativeto traditional physical experimentation methods. This approach notonly saves time and resources but also ensures consistent and high-qualitypredictive performance in chemical analysis and monitoring.

This study proposes an innovative framework integrating geographic information systems (GISs), water quality index (WQI) analysis, and advanced machine learning (ML) models to evaluate the prevalence and impact of organic and inorganic pollutants across the urban–industrial confluence zones (UICZ) surrounding the National Capital Territory (NCT) of India. Surface water samples (n = 118) were systematically collected from the Gautam Buddha Nagar, Ghaziabad, Faridabad, Sonipat, Gurugram, Jhajjar, and Baghpat districts to assess physical, chemical, and microbiological parameters. The application of spatial interpolation techniques, such as kriging and inverse distance weighting (IDW), enhances WQI estimation in unmonitored areas, improving regional water quality assessments and remediation planning. GIS mapping highlighted stark spatial disparities, with industrial hubs, like Faridabad and Gurugram, exhibiting WQI values exceeding 600 due to untreated industrial discharges and wastewater, while rural regions, such as Jhajjar and Baghpat, recorded values below 200, reflecting minimal anthropogenic pressures. The study employed four ML models—linear regression (LR), random forest (RF), Gaussian process regression (GPR), and support vector machines (SVM)—to predict WQI with high precision. SVM_Poly emerged as the most effective model, achieving testing CC, RMSE, and MAE values of 0.9997, 11.4158, and 5.6085, respectively, outperforming RF (0.9925, 29.8107, 21.7398) and GPR_PUK (0.9811, 68.4466, 54.0376). By leveraging machine learning models, this study enhances WQI prediction beyond conventional computation, enabling spatial extrapolation and early contamination detection in data-scarce regions. Sensitivity analysis identified total suspended solids as the most critical predictor influencing WQI, underscoring its relevance in monitoring programs. This research uniquely integrates ML algorithms with spatial analytics, providing a novel methodological contribution to water quality assessment. The findings emphasize the urgency of mitigating the fate and transport of organic and inorganic pollutants to protect Delhi’s hydrological ecosystems, presenting a robust decision-support system for policymakers and environmental managers.

Water quality degradation of freshwater bodies is a concern worldwide, particularly in Africa, where data are scarce and standard water quality monitoring is expensive. This study explored the use of remote sensing imagery and machine learning (ML) algorithms as an alternative to standard field measuring for monitoring water quality in large and remote areas constrained by logistics and finance. Six machine learning (ML) algorithms integrated with Landsat 8 imagery were evaluated for their accuracy in predicting three optically active water quality indicators observed monthly in the period from August 2016 to April 2022: turbidity (TUR), total dissolved solids (TDS) and Chlorophyll a (Chl-a). The six ML algorithms studied were the artificial neural network (ANN), support vector machine regression (SVM), random forest regression (RF), XGBoost regression (XGB), AdaBoost regression (AB), and gradient boosting regression (GB) algorithms. XGB performed best at predicting Chl-a, with an R2 of 0.78, Nash–Sutcliffe efficiency (NSE) of 0.78, mean absolute relative error (MARE) of 0.082 and root mean squared error (RMSE) of 9.79 µg/L. RF performed best at predicting TDS (with an R2 of 0.79, NSE of 0.80, MARE of 0.082, and RMSE of 12.30 mg/L) and TUR (with an R2 of 0.80, NSE of 0.81, and MARE of 0.072 and RMSE of 7.82 NTU). The main challenges were data size, sampling frequency, and sampling resolution. To overcome the data limitation, we used a K-fold cross validation technique that could obtain the most out of the limited data to build a robust model. Furthermore, we also employed stratified sampling techniques to improve the ML modeling for turbidity. Thus, this study shows the possibility of monitoring water quality in large freshwater bodies with limited observed data using remote sensing integrated with ML algorithms, potentially enhancing decision making.

Prediction of water quality indexes with ensemble learners: Bagging and boosting

Evaluation of groundwater chemistry and its related health hazard risk for humans is a prerequisite remedial measure. The semi-urban region in southern India was selected to measure the groundwater quality to know the human health risk valuation for different age groups of adults and children through oral intake and skin contact with elevated concentrations of fluoride ([Formula: see text]) and nitrate ([Formula: see text]) groundwater. Groundwater samples were collected from the semi-urban region for pre- and post-rainfall periods and resolute its major ion chemistry. The pH values showed the water is alkaline to neutral in nature. Total dissolved solid (TDS) ranged from 201 to 3612mg/l and 154 to 3457mg/l. However, [Formula: see text] concentration ranges from 0.28 to 5.48mg/l and 0.21 to 4.43mg/l; and NO3- ranges from 0.09 to 897.28mg/l and 0.0 to 606.10mg/l elevating the drinking water standards of [Formula: see text] in 32% and 38% samples and for [Formula: see text] about 62% and 38% during pre- and post-rainfall seasons, respectively. The fluoride-bearing minerals are the main sources of elevated concentrations of [Formula: see text] and excessive use of chemical fertilizers as the chief source of NO3- concentration in the aquifer regime. Water quality index (WQI) ranged from 18.3 to 233 and 12.97 to 219.14; 20% and 22% showed poor water quality for pre- and post-rainfall seasons with WQI ≥ 200. Piper plot suggests that 46% and 51% of samples signify carbonate water type ([Formula: see text]), and 32% and 28% of groundwater samples show ([Formula: see text]) type water for pre- and post-rainfall seasons respectively. Gibbs' plot suggests the dominance of water-rock interaction in the aquifer system. Further, the principal component analysis (PCA) revealed three and four components which explain 74.85% and 79.30% of the variance in pre- and post-rainfall seasons with positive loading of EC, TDS, Ca2+, Na+, Mg2+, K+, [Formula: see text], Cl-, and [Formula: see text] due to mineral weathering and water-rock interactions altering the chemistry for an elevated concentration of [Formula: see text] and [Formula: see text] in groundwater. Cluster analyses of chemical variables observed four clusters with a linkage distance of 5 to 25 with a linkage between different variables displaying predominant ion exchange, weathering of silicate and fluoride-rich minerals, salinization of the water, and a high value of [Formula: see text] concentration, resulting from fertilizers. The hazard quotient (HQ) through ingestion (HQing) and dermal (HQder) pathways of F- and NO3- was observed higher than its acceptable limit of 1.0 for different age groups indicating the non-carcinogenic effect on human health. Effective strategic measures like defluoridation, denitrification, safe drinking water supply, sanitary facilities, and rainwater harvesting structures are to be implemented in the area for improvement of human health conditions and also bring awareness to the local community about the health hazard effects of using high concentrated [Formula: see text] and [Formula: see text] water for daily uses.

Groundwater is an essential resource for global drinking and agricultural practices, but it is increasingly threatened by contamination. A comprehensive study was conducted for groundwater quality at 23 different locations in Kasganj, Uttar Pradesh, India, utilizing state-of-the-art Water Quality Indexing (WQI) and Irrigation Water Quality Indexing (IWQI) techniques. A total of one hundred fifteen groundwater samples were analyzed for twelve water quality aspects: pH, total dissolved solids (TDS), total alkalinity, total hardness, calcium (Ca2⁺), magnesium (Mg2⁺), sodium (Na⁺), potassium (K⁺), chloride (Cl−), bicarbonate (HCO₃−), Sulphate (SO42−), Nitrate (NO3−), and fluoride (F−). The results revealed that the TDS levels were alarmingly high, spanning 252 to 2054 ppm with an average of 942 ppm. Similarly, fluoride levels, ranging from 0.21 to 3.80 ppm (average 1.55 ppm), exceeded the World Health Organization’s permissible limit of 1.5 ppm. Strong correlations among fluoride levels, alkalinity, pH, Na⁺, and HCO₃⁻ point to geochemical interactions causing pollution. Piper diagram analysis divided most samples into Ca–Mg–Cl hydrochemical facies, a classification indicating the dominant ions in the water. Mineral saturation indices indicated dolomite, calcite, and aragonite oversaturation, which means these minerals are present in excess, potentially due to the water’s high TDS levels. With WQI scores ranging from 63.64 to 221.18, WQI results were concerning: 60.87% of samples were judged unfit for drinking, and 26.08% were relatively poor. These findings raise serious health concerns for the affected populations. Variations in IWQI indicators—Na%, SAR, MH, and KL ratio—informed irrigation fit for different sites. The use of advanced machine learning models (ANN, RF, XGB) for hydrochemical facies analysis, geochemical modeling, and predictive WQI in the sampled area makes the current study unique. To enhance forecast accuracy and support water management, Machine Learning models (Random Forest (RF), Artificial Neural Network (ANN), and Extreme Gradient Boosting (XGB), were used. The outcomes are indicated by better performance by RF with minimum error values (RMSE: 5.97, MSE: 35.69, MAE: 5.49) and a high R2 value of 0.951. ANN followed closely with an R2 of 0.957, while XGB achieved an R2 of 0.831. The performance by RF was the best in WQI prediction among the models tested. The results reveal critical groundwater pollution in the Kasganj area, emphasizing the immediate requirement of focused remedial action and effective water management plans.

Water is one of most critical assets for a healthy life, covering around 70% of the Earth's surface. However, rapid industrialization and urbanization have led to an alarming rate of water quality degradation, resulting in horrible diseases. Traditional techniques for evaluating water quality are expensive, time-consuming, and require laboratory and statistical tests, making real-time monitoring impractical. Therefore, there is a need for a quicker and more workable solution to address the catastrophic consequences of poor water quality. To this end, a particular index called Water Quality Index (WQI) is defined to complete Water Quality Class (WQC) and Water Quality (WQ) and use the WQI. So, in this research work, we explore several supervised machine learning methods to calculate WQI. Suggested approach uses four input parameters: Total Dissolved Solids (TDS), pH, turbidity, and temperature. The most effective algorithms for forecasting the WQI are gradient boosting and polynomial regression with degree of 2 with learning rate of 0.1. However, these algorithms have Mean Absolute Errors (MAE) of 1.8074 and 2.8957. Conversely, a Multi-Layer Perceptron (MLP) classifies WQC most accurately, with accuracy of 0.8796 and a configuration of (3, 7). Proposed approach confirms it’s prospective for use in water quality detecting systems by attaining rational accuracy with inadequate set of parameters. Therefore, it is crucial to develop more accurate methods to monitor and cope with WQ to ensure the sustainability of life on Earth.

Supply chain (SC) efficacy and efficiency can be severely hampered by supplier delays in orders, especially in the fast-paced business environment of today. Effective risk reduction necessitates the identification of suppliers who are prone to delays and the precise prediction of future interruption. Accurately predicting availability dates is therefore a key factor in successfully executing logistics operations. By leveraging machine learning (ML) techniques, organizations can proactively identify high-risk suppliers, anticipate delays, and implement proactive measures to minimize their impact on manufacturing processes and overall SC performance. This study explores and utilizes various regression and classification ML algorithms to predict future delayed delivery, determine the status of order deliveries, and classify suppliers according to their delivery performance. The employed models include K-Nearest Neighbors (KNN) Random Forest (RF) Classifier and Regression, Gradient Boosting (GB) Regression and Classifier, Linear Regression (LR), Decision Trees(DT) Classifier and Regression, Logistic Regression and Support Vector Machine (SVM) Based on real data, our experiments and evaluation metrics including Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE) demonstrate that the ensemble based regression algorithms (RF Regression and GB Regression) provide the best generalization error and outperforms all other regression models tested. Similarly, Logistic regression and GB Classifier outperforms other classification algorithms according to precision, recall, and F1-score metrics. The knowledge obtained from this study could aid in the proactive identification of high-risk suppliers and the application of proactive actions to increase resilience in the face of unanticipated disruptions, in addition to increasing SC efficiency and decreasing manufacturing disturbances.

Enhancing biomass Pyrolysis: Predictive insights from process simulation integrated with interpretable Machine learning models

The pollution in Dhaka's navigable waterways, including the Buriganga, Balu, Tongi Khal, and Turag rivers, is a significant concern due to rapid industrial and urban expansion. Industrial discharges, domestic sewage and inadequate waste management are the primary sources of this pollution, degrading water quality and threatening aquatic ecosystems. This study aimed to predict the Water Quality Index (WQI) of these rivers using fourteen machine learning (ML) models: Decision Tree Regression, Linear Regression, Ridge Regression, Stochastic Gradient Descent (SGD) Regressor, Extreme Gradient Boosting (XGB) Regressor, Light Gradient Boosting Machine (GBM) Regressor, Elastic Net Regressor, Support Vector Regression (SVM), Random Forest Regression, Bayesian Ridge Regressor, Artificial Neural Network (ANN), AdaBoost Regressor, CatBoost Regressor and Extra Trees Regressor. The objective was to evaluate and compare these models to identify the most effective predictive method for WQI, enabling efficient environmental monitoring and management of urban waterways. Among the evaluated ML models, ANN and Random Forest Regressor performed the best. The ANN model demonstrated superior predictive capability, achieving a Root Mean Squared Error (RMSE) of 2.34, a Mean Absolute Error (MAE) of 1.24, a Nash–Sutcliffe Efficiency (NSE) of 0.97, and a Coefficient of Determination (R2) of 0.97. Furthermore, an Adjusted R2 value of 0.965 further confirmed its ability to capture complex patterns in water quality data with remarkable accuracy. These findings emphasize the importance of using AI modeling techniques, specifically ANN and Random Forest Regression, to improve the accuracy of WQI forecasts for the waterways. This study contributes to the field of environmental science by offering a novel integration of feature selection techniques with ML models to enhance efficiency and cost-effectiveness of water quality monitoring. Unlike previous studies, this research specifically addresses the challenges of urban waterways in Dhaka, Bangladesh, a region significantly impacted by industrial and urban pollution. To our knowledge, this is the first study to apply such a comprehensive range of ML models to predict the WQI of Dhaka’s four major rivers. By providing a reliable methodology for WQI estimation, this study supports informed decision-making and proactive measures to protect vital water resources.

In this work, a combination of an acoustic emission (AE) technique and a machine learning (ML) algorithm (Random Forest (RF) and Gradient Boosting Regressor (GBR)) is developed to characterize the particle size distribution in gas-solid fluidized bed reactors. A theoretical approach to explain the generation of acoustic emission signal in gas-solid flows is presented. An AE signal is generated in gas-solid fluidized beds due to the collision and friction between fluidized particles as well as between particles and the bed inner wall. The generated AE signal is in the form of an elastic wave with frequencies >100 KHz and it propagates through the gas-solid mixture. An inversion algorithm is used to extract the information about the particle size starting from the energy of the AE signal. The advantages of this AE technique are that it is a cheap, sensitive, non-intrusive, radiation-free, suitable for on-line measurements. Combining this AE technique with ML algorithms is beneficial for applications to industrial settings, reducing the cost of signal post-processing. Experiments were conducted in a pseudo-2D flat fluidized bed with four glass bead samples, with sizes ranging from 100 μm to 710 μm. AE signals were recorded with a sampling frequency of 5 MHz. The AE signal post-processing and data preparation for the ML process are explained. For the ML process, the AE frequency, AE energy and particle collision velocity data sets were divided into training (60%), cross-validation (20%) and test sets (20%). Two ensemble ML approaches, namely Random Forest and Gradient Boosting Regressor, are applied to predict particle sizes based on the AE signal features. The combination of these two models results in a coefficient of determination (R2) value greater than 0.9504.

Machine learning and deep learning for mineralogy interpretation and CO2 saturation estimation in geological carbon Storage: A case study in the Illinois Basin

Groundwater quality assessment is crucial for ensuring safe drinking water and sustainable resource management. However, traditional monitoring methods involving extensive sampling and laboratory analysis are time-consuming and costly. The present study proposes an efficient approach for predicting groundwater quality in Madhya Pradesh, India using data-driven models and an entropy-weighted water quality index (EWQI). A large spatiotemporal dataset of different parameters of groundwater quality like pH, total dissolved solids (TDS), calcium (Ca2⁺), total hardness (TH), nitrate (NO₃⁻), sodium (Na⁺), chloride (Cl⁻), potassium (K⁺), sulfate (SO₄2⁻), magnesium (Mg2⁺), and fluoride (F⁻) from the year (2003-2023) across Madhya Pradesh was analysed. All advanced data-driven models such as Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Artificial Neural Network (ANN), and Support Vector Machine (SVM) were developed to predict the EWQI using easily measurable parameters pH, TH and TDS. The individual ability of the models was assessed using statistical analysis with the coefficient of determination (R2), mean absolute error (MAE), and root mean square error (RMSE). During the training phase, all models such as RF, SVM, XGBoost, and ANN proved excellent predictive capabilities, achieving an R2 value exceeding 0.90 while maintaining minimal errors when pH, TH, and TDS were considered as input variables. The overall outcomes confirmed that the data-driven models could accurately estimate the EWQI, closely matching the actual values with an R2 greater than 0.90. This finding highlights the model's ability to predict a reliable overview of water quality for a small area using easily measurable parameters.Keywords: Groundwater, Data-driven, Drinking water, Water Quality Index, Machine learning

Given its availability amid the increasing scarcity of surface freshwater, groundwater has become a vital and increasingly relied-upon resource, especially in semi-arid and arid regions. Thus, to ensure groundwater complies with standards before use, continuous monitoring and comprehensive quality assessment are essential. This study aimed to assess the quality of groundwater (GW) in the Skikda aquifer, northeastern Algeria, for irrigation using irrigation water quality indices (IWQIs), multivariate statistical analysis,and machine learning algorithms (MLAs): Random Forest regression (RF), Extreme Gradient Boosting regression (XGBR), and Adaptive Boosting Regression (ABR), integrated with SHAP analysis. Forty-four groundwater samples were collected from the study area during summer and winter seasons andanalysed for temperature, pH, electrical conductivity (EC), turbidity, total dissolved solids (TDS), and concentrations of calcium (Ca²⁺), magnesium (Mg²⁺),sodium (Na⁺),potassium (K⁺), chloride (Cl⁻),bicarbonate (HCO₃⁻), sulfate (SO₄²⁻), and nitrate (NO₃⁻).The dominating hydrochemical facies in the study area were Mg-Ca-SO4, accompanied by the Sodium-Chloride (Na-Cl).Principle Component Analysis (PCA) for summer and winter datasets identified four key components suggesting a strong correlation between variables and factors, with PCA indicating that geochemical processes, such as rock0water interaction and dissolution of evaporite minerals, control the groundwater’s chemical composition.Groundwater quality for irrigation varied across the samples, with most exhibiting moderate to high constraints based on IWQI. Sodium Adsorption Ratio (SAR) and Permeability Index (PI) suggested excellent to good water quality,while Sodium Percent (Na%) and Soluble Sodium Percentage (SSP) indicate a small but significant fraction of inappropriate samples.Magnesium Hazard (MH) and SSP indicated that most samples were safe.Compared to winter, summer samples showed slightly poorer quality (higher Na%, SSP, and lower IWQI), likely due toevaporative solute concentration. Random Forest (RF) modelshowed superior predictive accuracy for all Water Quality Indices (WQIs), with strong validation results for both seasons. These results highlight RF's effectiveness in predicting WQIs and highlight the influence of seasonal geochemical processes on groundwater quality, requiring the development of management strategies for sustainable irrigation.

Fluoride contamination in groundwater sources in Southwestern Nigeria: Assessment using multivariate statistical approach and human health risk