Artificial intelligence and analytical techniques for groundwater quality assessment: a systematic review

Abstract

Groundwater constitutes a vital component of global freshwater resources, particularly in arid and semiarid regions where surface water availability is constrained. However, increasing anthropogenic pressures, alongside natural geochemical processes, have progressively degraded groundwater quality, raising significant concerns regarding its suitability for drinking, irrigation, and domestic applications. Consequently, the development of accurate and efficient assessment methods has become essential for sustainable water resource management. This study investigates a comprehensive and systematic review of recent advancements in artificial intelligence (AI)-based techniques for groundwater quality assessment that support evidence-based decision-making in sustainable water resource management. A systematic literature search was conducted in Scopus, Web of Science, and Science Direct to identify studies published between 2020 and 2025, resulting in the inclusion of 55 eligible studies. The analysis synthesizes evidence from these studies, systematically classifies the applied techniques into eight methodological categories: three primary AI computational modeling categories (DL, ML and ensemble and hybrid model and five complementary methodological categories), and evaluates the relative performance, strengths, limitations, and representative applications. The findings indicate that ensemble learning techniques, particularly random forest, gradient boosting models, and different hybrid models, consistently outperform conventional statistical approaches in terms of predictive accuracy and robustness. Furthermore, deep learning models, including long short-term memory (LSTM) networks and convolutional neural networks (CNNs), demonstrate superior capability in capturing complex nonlinear and spatiotemporal relationships within hydrochemical datasets. Hybrid approaches further enhance model performance by integrating complementary methodological strengths. A practical guidance for selecting appropriate computational techniques is proposed by the classification framework and comparative synthesis provided in this review that highlights the future research directions for groundwater quality assessment.