Scholarlink

<p>The growing complexity and volume of water quality data—driven by environmental challenges, and high-resolution monitoring technologies—has necessitated the development of advanced analytical frameworks for water quality monitoring (WQM). In response, hybrid approaches that integrate machine learning (ML) with classical analytical methods have emerged as promising tools for handling multidimensional data and improving forecasting accuracy. This paper presents a systematic review of such hybrid approaches in WQM, based on a bibliometric analysis of 4,590 publications indexed in the Scopus database. The analysis reveals a sharp increase in relevant publications over the past decade, reflecting a growing shift toward integrated analytical frameworks. Co-occurrence analysis of author keywords identifies four principal thematic clusters reflecting major integration types: (1) unsupervised ML combined with multivariate statistical analysis (MSA) for pollution source identification; (2) supervised ML integrated with regression and remote sensing for environmental monitoring; (3) neural networks coupled with time-series models for predictive modeling; and (4) AI methods combined with spatial and signal processing techniques for comprehensive water quality assessment. Additional analyses show that statistical methods and neural networks are most frequently employed, whereas trend and frequency-based techniques remain underutilized. Five refined queries further reveal methodological overlaps and gaps, highlighting dominant and emerging hybridization patterns. By examining the complementary strengths of ML-classical method integrations, this review underscores their role in environmental decision-making, and identifies current methodological developments as well as future research opportunities, particularly with respect to improving interpretability, adaptability, and computational efficiency. Overall, the study provides actionable insights for advancing the practical application of hybrid approaches in WQM and for supporting sustainable and scalable water quality management.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Hybrid Approaches in Water Quality Monitoring: A Study of the Integration of Machine Learning and Classical Analytical Methods

  • 期刊论文

  • 发表日期：

    2026年3月28日

查看全文
  • Christine Saab,
  • Gérard-Philippe Zéhil

摘要

The growing complexity and volume of water quality data—driven by environmental challenges, and high-resolution monitoring technologies—has necessitated the development of advanced analytical frameworks for water quality monitoring (WQM). In response, hybrid approaches that integrate machine learning (ML) with classical analytical methods have emerged as promising tools for handling multidimensional data and improving forecasting accuracy. This paper presents a systematic review of such hybrid approaches in WQM, based on a bibliometric analysis of 4,590 publications indexed in the Scopus database. The analysis reveals a sharp increase in relevant publications over the past decade, reflecting a growing shift toward integrated analytical frameworks. Co-occurrence analysis of author keywords identifies four principal thematic clusters reflecting major integration types: (1) unsupervised ML combined with multivariate statistical analysis (MSA) for pollution source identification; (2) supervised ML integrated with regression and remote sensing for environmental monitoring; (3) neural networks coupled with time-series models for predictive modeling; and (4) AI methods combined with spatial and signal processing techniques for comprehensive water quality assessment. Additional analyses show that statistical methods and neural networks are most frequently employed, whereas trend and frequency-based techniques remain underutilized. Five refined queries further reveal methodological overlaps and gaps, highlighting dominant and emerging hybridization patterns. By examining the complementary strengths of ML-classical method integrations, this review underscores their role in environmental decision-making, and identifies current methodological developments as well as future research opportunities, particularly with respect to improving interpretability, adaptability, and computational efficiency. Overall, the study provides actionable insights for advancing the practical application of hybrid approaches in WQM and for supporting sustainable and scalable water quality management.