Integrated groundwater quality assessment using geochemical modelling and machine learning approach in Northern India

Area of Science:

• Environmental Science
• Hydrogeology
• Water Resource Management

Background:

• Groundwater is vital for drinking and agriculture but faces increasing contamination risks.
• Kasganj, Uttar Pradesh, India, relies heavily on groundwater resources.
• Assessing groundwater quality is crucial for public health and sustainable agriculture.

Purpose of the Study:

• To comprehensively assess groundwater quality in Kasganj, Uttar Pradesh, India.
• To evaluate the suitability of groundwater for drinking and irrigation purposes.
• To apply advanced machine learning techniques for hydrochemical analysis and water quality prediction.

Main Methods:

• Collected 115 groundwater samples from 23 locations for analysis of 12 water quality parameters.
• Utilized Water Quality Indexing (WQI) and Irrigation Water Quality Indexing (IWQI) techniques.
• Employed Piper diagram analysis, mineral saturation indices, and machine learning models (ANN, RF, XGB) for data interpretation and prediction.

Main Results:

• High levels of Total Dissolved Solids (TDS) and fluoride were detected, with fluoride exceeding WHO limits in many samples.
• 60.87% of groundwater samples were unfit for drinking based on WQI, and 26.08% were rated as relatively poor.
• Geochemical analysis indicated Ca-Mg-Cl hydrochemical facies and oversaturation of calcite, dolomite, and aragonite, linked to high TDS.
• Machine learning models, particularly Random Forest (RF), demonstrated high accuracy in predicting WQI.

Conclusions:

• Groundwater in the Kasganj area is critically polluted, posing significant risks to human health and agriculture.
• The study highlights the urgent need for targeted remediation strategies and effective water resource management plans.
• Advanced machine learning models offer a powerful tool for understanding hydrogeochemical processes and forecasting water quality.