A Groundwater Quality Assessment Model for Water Quality Index: Combining Principal Component Analysis, Entropy Weight Method, and Coefficient of Variation Method for Dimensionality Reduction and Weight Optimization, and Its Application.

Abstract

Groundwater underpins water supply for most of the world's regions, yet its sustainable utilization has been markedly compromised by inappropriate exploitation and a multitude of pollution sources. Water quality evaluation has emerged as an essential strategy to guarantee the optimized utilization and vigilant conservation of water resources. In this study, principal component analysis (PCA), entropy weight method (EWM), coefficient of variation method (CVM), and Water Quality Index (WQI) were used to construct an integrated WQI groundwater quality assessment model that integrates PCA-CVM-EWM for dimensionality reduction and weight optimization. Taking a village in Shandong Province, China, as an example, PCA identified seven evaluation indicators. The CVM-EWM were coupled to calculate comprehensive weights through the principle of minimum information entropy, followed by a comprehensive assessment of groundwater quality based on WQI values. The results indicated that Class III groundwater predominated in the study area, accounting for 74%, with localized pollution present. The hydrochemical type of the groundwater was primarily SO₄·HCO₃-Ca, significantly influenced by human activities. The coefficients of variation for Fe, Mn, and NH₄-N all exceeded 1. Compared to other methods, the optimized WQI model demonstrated superior performance in the selection of evaluative indicators, weight distribution, and comprehensive water quality assessment, showing a distinct advantage for water quality data with numerous hydrochemical indicators and substantial coefficients of variation. The findings provided a scientific reference for diagnosing groundwater quality issues and formulating preventive and control measures. PRACTITIONER POINTS: A comprehensive water quality index evaluation model was constructed. Optimized steps for selecting indicators and assigning weights for the water quality index model. Selection of evaluation indicators based on indicator correlation analysis. The variability of hydrochemical data is considered.