Predictive modeling of selected trace elements in groundwater using hybrid algorithms of iterative classifier...

Area of Science:

• Environmental Science
• Hydrogeology
• Analytical Chemistry

Background:

• Trace element (TE) pollution in groundwater poses significant risks to human health globally.
• Arsenic (As), Barium (Ba), and Rubidium (Rb) are naturally occurring TEs in groundwater due to water-rock interactions.
• Campania Plain (CP) aquifers in South Italy are a focus for studying TE groundwater contamination.

Purpose of the Study:

• To predict the concentrations of naturally occurring trace elements (As, Ba, Rb) in groundwater.
• To evaluate the performance of various regression algorithms, including standalone iterative classifier optimizer (ICO) and hybrid models (AR-ICO, ASC-ICO, BA-ICO).
• To identify the most effective input variables and predictive models for TE concentration forecasting.

Main Methods:

• Collected 244 groundwater samples from CP wells for analysis of electrical conductivity, pH, major ions, and selected TEs.
• Randomly divided the dataset into training (70%) and evaluation (30%) subsets for model development.
• Utilized ICO, AR-ICO, ASC-ICO, and BA-ICO algorithms to predict TE concentrations, optimizing input variable combinations.

Main Results:

• Prediction of As and Ba concentrations strongly depends on bicarbonate (HCO₃^-), while Rubidium (Rb) prediction is most influenced by sodium (Na⁺).
• Models incorporating all available input variables demonstrated the highest predictive power.
• The hybrid ASC-ICO model outperformed others in predicting As and Ba, while ASC-ICO and BA-ICO showed higher accuracy for Rb prediction.

Conclusions:

• Hybrid machine learning models, especially ASC-ICO, offer a robust approach for predicting trace element pollution in groundwater resources.
• Understanding the relationship between major ions and TE concentrations is key for accurate predictive modeling.
• Effective prediction of TEs like As, Ba, and Rb is vital for safeguarding groundwater quality and public health.