Scholarlink

Water quality monitoring is crucial for the sustainable management of aquatic resources, particularly in regions like the Moulouya River Basin in Morocco, where environmental factors significantly impact water quality. This study utilizes advanced artificial intelligence (AI) techniques to model water quality based on key physicochemical parameters, including pH, temperature (T°C), electrical conductivity (EC), dissolved oxygen (DO), NH4, NO2, SO4, PO4, and biological oxygen demand (BOD5). Using a dataset of 66 samples collected from 22 stations over three periods in 2014, we implemented a structured methodology that encompassed data preprocessing, feature selection, the calculation of water quality indices such as the Pollution Index (PI), and the application of ensemble learning models, including Gradient Boosting, Adaptive Boosting, and a Stacking Ensemble approach, to predict water quality. The results indicated that the AdaBoost model provided the most accurate predictions, achieving a Mean Squared Error (MSE) of 0.0132 and an R2 of 0.9450. The study further explores integrating these models into a real-time monitoring system, underscoring their potential for proactive water quality management. This research advances the field of environmental AI by offering a robust framework for real-time water quality prediction and management.

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

Modeling Water Quality Based on Environmental Factors in the Moulouya River Basin: An Ensemble Learning Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Sara Bouziane,
  • Badraddine Aghoutane

摘要

Water quality monitoring is crucial for the sustainable management of aquatic resources, particularly in regions like the Moulouya River Basin in Morocco, where environmental factors significantly impact water quality. This study utilizes advanced artificial intelligence (AI) techniques to model water quality based on key physicochemical parameters, including pH, temperature (T°C), electrical conductivity (EC), dissolved oxygen (DO), NH4, NO2, SO4, PO4, and biological oxygen demand (BOD5). Using a dataset of 66 samples collected from 22 stations over three periods in 2014, we implemented a structured methodology that encompassed data preprocessing, feature selection, the calculation of water quality indices such as the Pollution Index (PI), and the application of ensemble learning models, including Gradient Boosting, Adaptive Boosting, and a Stacking Ensemble approach, to predict water quality. The results indicated that the AdaBoost model provided the most accurate predictions, achieving a Mean Squared Error (MSE) of 0.0132 and an R2 of 0.9450. The study further explores integrating these models into a real-time monitoring system, underscoring their potential for proactive water quality management. This research advances the field of environmental AI by offering a robust framework for real-time water quality prediction and management.