Spatial analysis and soft computational modeling for hazard assessment of potential toxic elements in potable...

Area of Science:

• Environmental Science
• Hydrogeology
• Analytical Chemistry

Background:

• Rapid urbanization and industrialization in Doon Valley have led to deteriorating groundwater quality.
• Potential Toxic Elements (PTEs) pose a significant threat to water resources and human health.

Purpose of the Study:

• To analyze the spatial distribution and associated hazards of six PTEs (Cr, Ni, As, Mo, Cd, Pb) in Doon Valley groundwater.
• To utilize machine learning algorithms for predicting water quality and identifying influential PTEs.

Main Methods:

• Groundwater samples were collected from various locations in Doon Valley.
• Concentrations of six PTEs were measured using Inductively Coupled Plasma Mass Spectrometry (ICP-MS).
• Machine learning models, including Radial Basis Function Neural Network (RBF-NN) and Multilayer Perceptron Neural Network (MLP-NN), were employed for prediction.

Main Results:

• Mean PTE concentrations were highest for Molybdenum (Mo), followed by Nickel (Ni), Lead (Pb), Arsenic (As), Chromium (Cr), and Cadmium (Cd).
• All analyzed PTE levels and computed risks were found to be below established safe limits.
• RBF-NN demonstrated high accuracy (R² 0.912-0.976) in predicting pollution indices and health risks, while MLP-NN showed higher precision for health risk indices (R² 0.887-0.995).

Conclusions:

• While current PTE levels in Doon Valley groundwater are within safe limits, ongoing monitoring is crucial.
• Machine learning models, particularly RBF-NN, are effective tools for predicting groundwater quality and assessing associated health risks.
• The study highlights the importance of understanding PTE distribution and utilizing advanced analytical and predictive techniques for water resource management.