Comparison of Bayesian, k-Nearest Neighbor and Gaussian process regression methods for quantifying uncertainty of suspended sediment concentration prediction

Unraveling soil salinity on potentially toxic element accumulation in coastal Phragmites australis: A novel integration of multivariate and interpretable machine-learning models

Revealing the key mechanisms influencing the behavior of potentially toxic elements (PTEs) in soil-plant systems is of great significance for environmental protection and grassland development in coastal areas. This study utilized redundancy analysis to assess the effects of soil environmental variables on the concentrations and enrichment of various PTEs in the advantageous forage species Phragmites australis. Advanced models like PLS-PM and RF-SHAP quantitatively assessed soil salinity impacts. The main findings are as follows: (1) P. australis exhibited enrichment capacity for Cd, Cr, and Cu. (2) Soil pH, exchangeable potassium (aK), and exchangeable calcium (aCa) were key determinants of PTE distribution, with Cu being highly sensitive to these variables. (3) Significant interactions between soil electronic conductivity (EC) and pH, as well as between soil EC and aCa (p < 0.01). (4) A pH value of 8.30 and an aCa concentration of 4.4 g/kg were identified as critical thresholds affecting the Cu uptake. These results provide insights into PTE migration and management strategies for coastal grasslands.

Bayesian-optimized recursive machine learning for predicting human-induced changes in suspended sediment transport

The suspended sediment load (SSL) of a river is a key indicator of water resource management, river morphology, and ecosystem health. This study analyzes historical changes in SSL and evaluates machine learning (ML) models for SSL prediction in the Godavari River Basin. The dataset was divided into pre-1990 (1969-1990) and post-1990 (1990-2020) periods, revealing a significant decline in mean annual SSL from 136.85 to 62.38 million tons post-1990 due to anthropogenic influences such as dam construction and land-use/land-cover (LULC) changes. Despite a consistent seasonal distribution (~ 73% SSL contribution from monsoon months in both periods), there was a notable decline in median and peak SSL values, along with a narrowing interquartile range, indicating reduced sediment availability. The empirical cumulative distribution function (ECDF) further revealed shifts in sediment transport, with post-1990 SSL values surpassing pre-1990 levels at higher cumulative probabilities, suggesting altered sediment retention and release patterns. To improve SSL prediction, tree-based ML models were developed and evaluated using R2, RMSE, and MAE metrics. Among them, the extra trees regressor (ETR) demonstrated the highest predictive accuracy (R2 = 0.97 in training, 0.9 in testing) with the lowest errors, while the random forest regressor (RFR) and gradient boosting regressor (GBR) provided competitive results. The findings highlight the impact of human modifications on sediment transport and emphasize that ensemble tree-based models offer a robust solution for SSL prediction. This study provides valuable insights for river basin management and sustainable sediment transport modeling under changing hydrological conditions.

Assessing the performance of machine learning algorithms for analyzing land use changes in the Hyrcanian forests of Iran

Land use changes are of critical importance in understanding and managing environmental sustainability and resource utilization. Machine learning algorithms (MLAs) have emerged as powerful tools for analyzing and predicting land use changes, offering the potential to uncover patterns and trends that may not be readily apparent through traditional methods. This study is aimed at evaluating the efficiency of various MLAs (such as SVM, KNN, CART, Naïve Bayes, and Random Forest) in analyzing LULC changes in Northeast Iran. The analysis utilized the Google Earth Engine (GEE) to process satellite imagery spanning the years 1994 to 2021, covering a period of 27 years. Landsat 5 and TM sensor data from 1994 and 2001, as well as Landsat 8 and OLI sensor data from 2014 and 2021, were employed in this research. Additionally, post-processing tasks on the classified images were carried out using ArcGIS 10.8 software. Based on the validation results, it is evident that the Random Forest machine learning algorithm outperformed other algorithms. In contrast, for the years 2014 and 2021, the support vector machine algorithm had the highest accuracy of 85%, making it the most optimal choice during those years. The results indicated a decrease in rangeland, with a significant difference (34.04%) observed in 1994-2021. This decline could be attributed to factors such as rangeland degradation and a shift in LULC towards agriculture and orchards. Conversely, agricultural land had significant increases of 275.83%, 223.77%, and 61.97% in 2021 compared to 1994, 2001, and 2014, respectively. However, the area of forest lands decreased notably over the studied periods, with reductions of 81.66%, 64.21%, and 30.56% in 2021 compared to 1994, 2001, and 2014, respectively. The study results reveal distinction shifts in LULC patterns, indicating declines in rangelands and significant expansions in agricultural areas, which need to be considered in land use planning and environmental conservation programs.

Testing the EKC hypothesis for ecological and carbon intensity of well-being: The role of forest extent

The G-20 countries represent a considerable percentage of the global economy and are crucial in matters to do with support for environmental sustainability. The uniqueness of this study lies in determining the effects of forests on human well-being and environmental degradation with respect to G20, offering a unique perspective regarding the efforts to battle climate change. The study analyzed the impact of income, forest extent and education on ecological and carbon intensity of well-being following the Environmental Kuznets Curve (EKC) hypothesis. Based on annual data from 1990 to 2022 and employing the Method of Moments Quantile Regression, the results validate the presence of an inverted U-shaped relationship between GDP and environmental well-being which refers to the existence of EKC. Our results connect improved ecological and carbon intensity of well-being with expanding forest extent and improving education levels. Forest management combined with educational management work as an effective mechanism reducing environmental degradation while also positively contributing to human well-being. In addition, through these informed and rational decisions, policy makers can promote the environmental stability of forests.

Evaluation of the prediction effectiveness for geochemical mapping using machine learning methods: A case study from northern Guangdong Province in China

This study compares seven machine learning models to investigate whether they improve the accuracy of geochemical mapping compared to ordinary kriging (OK). Arsenic is widely present in soil due to human activities and soil parent material, posing significant toxicity. Predicting the spatial distribution of elements in soil has become a current research hotspot. Lianzhou City in northern Guangdong Province, China, was chosen as the study area, collecting a total of 2908 surface soil samples from 0 to 20 cm depth. Seven machine learning models were chosen: Random Forest (RF), Support Vector Machine (SVM), Ridge Regression (Ridge), Gradient Boosting Decision Tree (GBDT), Artificial Neural Network (ANN), K-Nearest Neighbors (KNN), and Gaussian Process Regression (GPR). Exploring the advantages and disadvantages of machine learning and traditional geological statistical models in predicting the spatial distribution of heavy metal elements, this study also analyzes factors affecting the accuracy of element prediction. The two best-performing models in the original model, RF (R2 = 0.445) and GBDT (R2 = 0.414), did not outperform OK (R2 = 0.459) in terms of prediction accuracy. Ridge and GPR, the worst-performing methods, have R2 values of only 0.201 and 0.248, respectively. To improve the models' prediction accuracy, a spatial regionalized (SR) covariate index was added. Improvements varied among different methods, with RF and GBDT increasing their R2 values from 0.4 to 0.78 after enhancement. In contrast, the GPR model showed the least significant improvement, with its R2 value only reaching 0.25 in the improved method. This study concluded that choosing the right machine learning model and considering factors that influence prediction accuracy, such as regional variations, the number of sampling points, and their distribution, are crucial for ensuring the accuracy of predictions. This provides valuable insights for future research in this area.

Application of Machine Learning Models for Early Detection and Accurate Classification of Type 2 Diabetes

Early detection of diabetes is essential to prevent serious complications in patients. The purpose of this work is to detect and classify type 2 diabetes in patients using machine learning (ML) models, and to select the most optimal model to predict the risk of diabetes. In this paper, five ML models, including K-nearest neighbor (K-NN), Bernoulli Naïve Bayes (BNB), decision tree (DT), logistic regression (LR), and support vector machine (SVM), are investigated to predict diabetic patients. A Kaggle-hosted Pima Indian dataset containing 768 patients with and without diabetes was used, including variables such as number of pregnancies the patient has had, blood glucose concentration, diastolic blood pressure, skinfold thickness, body insulin levels, body mass index (BMI), genetic background, diabetes in the family tree, age, and outcome (with/without diabetes). The results show that the K-NN and BNB models outperform the other models. The K-NN model obtained the best accuracy in detecting diabetes, with 79.6% accuracy, while the BNB model obtained 77.2% accuracy in detecting diabetes. Finally, it can be stated that the use of ML models for the early detection of diabetes is very promising.

Novel machine learning algorithms to predict the groundwater vulnerability index to nitrate pollution at two levels of modeling

The accurate mapping and assessment of groundwater vulnerability index are crucial for the preservation of groundwater resources from the possible contamination. In this research, novel intelligent predictive Machine Learning (ML) regression models of k-Neighborhood (KNN), ensemble Extremely Randomized Trees (ERT), and ensemble Bagging regression (BA) at two levels of modeling were utilized to improve DRASTIC-LU model in the Miryang aquifer located in South Korea. The predicted outputs from level 1 (KNN and ERT models) were used as inputs for ensemble bagging (BA) in level 2. The predictive groundwater pollution vulnerability index (GPVI), derived from DRASTIC-LU model was adjusted by NO₃-N data and was utilized as the target data of the ML models. Hyperparameters for all models were tuned using a Grid Searching approach to determine the best effective model structures. Various statistical metrics and graphical representations were used to evaluate the superior predictive performance among ML models. Ensemble BA model in level 2 was more precise than standalone KNN and ensemble ERT models in level 1 for predicting GPVI values. Furthermore, the ensemble BA model offered suitable outcomes for the unseen data that could subsequently prevent the overfitting issue in the testing phase. Therefore, ML modeling at two levels could be an excellent approach for the proactive management of groundwater resources against contamination.