Prediction of Dichloroethene Concentration in the Groundwater of a Contaminated Site Using XGBoost and LSTM

Abstract

Chlorinated aliphatic hydrocarbons (CAHs) are widely used in agriculture and industries and have become one of the most common groundwater contaminations. With the excellent performance of the deep learning method in predicting, LSTM and XGBoost were used to forecast dichloroethene (DCE) concentrations in a pesticide-contaminated site undergoing natural attenuation. The input variables included BTEX, vinyl chloride (VC), and five water quality indicators. In this study, the predictive performances of long short-term memory (LSTM) and extreme gradient boosting (XGBoost) were compared, and the influences of variables on models’ performances were evaluated. The results indicated XGBoost was more likely to capture DCE variation and was robust in high values, while the LSTM model presented better accuracy for all wells. The well with higher DCE concentrations would lower the model’s accuracy, and its influence was more evident in XGBoost than LSTM. The explanation of the SHapley Additive exPlanations (SHAP) value of each variable indicated high consistency with the rules of biodegradation in the real environment. LSTM and XGBoost could predict DCE concentrations through only using water quality variables, and LSTM performed better than XGBoost.

Enhanced Insights into Effluent Prediction in Wastewater Treatment Plants: Comprehensive Deep Learning Model Explanation Based on SHAP

“…This result is consistent with other reports of predicting water quality using ML algorithms. For example, in predicting dichloroethylene in groundwater by Xia et al, the accuracy of LSTM was significantly higher than that of XGBoost in all sampling points . It has been demonstrated in the field of water resource management that LSTM networks have reliable performance, often outperforming ANN and traditional ML models .…”

“…For example, in predicting dichloroethylene in groundwater by Xia et al, the accuracy of LSTM was significantly higher than that of XGBoost in all sampling points. 46 It has been demonstrated in the field of water resource management that LSTM networks have reliable performance, often outperforming ANN and traditional ML models. 22 Both LSTM and GRU exhibited better performance in predicting, but LSTM outperformed TN and TP prediction.…”

A Critical Review of the Modelling Tools for the Reactive Transport of Organic Contaminants

“…In the context of the numerous studies on water quality, machine learning can predict the amount and fate of pollutants, taking into account complex processes and interactions between different control parameters. Particular attention is paid to organic pollutants and the assessment of contaminated sites, including through the application of image recognition technology [533][534][535][536][537][538][539].…”

“…Xia et al [538] used Long Short-Term Memory (LSTM) and Extreme Gradient Boosting (XGBoost) to predict DCE degradation at the contaminated site. The variables from 3 months of data were used for training and prediction of DCE concentrations.…”

Water-Quality Assessment and Pollution-Risk Early-Warning System Based on Web Crawler Technology and LSTM

“…Among these, long short-term memory (LSTM) networks have a good performance in processing data with time series characteristics [ 18 ]. They addresses the issues of gradient explosions and the gradient disappearance of conventional recurrent networks; LSTM networks are presently used in many fields, such as trend predictions of COVID-19 [ 19 ], sudden change simulations in financial markets [ 20 ] and water quality prediction [ 21 , 22 ]. This research has verified that the LSTM model has better prediction accuracy than the traditional numerical models.…”