Scholarlink

<p>This study investigates the prediction of key irrigation water quality parameters—Sodium Adsorption Ratio (SAR), Magnesium Adsorption Ratio (MAR), Percent Sodium (%Na), Permeability Index (PI), and Kelly’s Index (KI) using hydrochemical data as input features. Groundwater quality data from 272 samples collected in the central western part of Haryana, India, were analyzed. To improve predictive accuracy, four machine learning models were employed: Random Forest (RF), Support Vector Regression (SVR), M5P, and Linear Regression (LR). Additionally, Principal Component Analysis (PCA) was conducted to identify underlying correlations between water quality parameters and to assess the influence of natural processes and anthropogenic activities, such as rock weathering and improper irrigation practices. The results indicated that PI and Magnesium Hazard (MH) values exceeded permissible limits for irrigation, while SAR, %Na, and KI were within acceptable ranges. The SVR and M5P models outperformed RF and LR in predictive accuracy, as supported by uncertainty analysis, which showed lower uncertainty for the former models. The findings highlight the potential of machine learning models, particularly SVR and M5P, to support decision-makers in managing irrigation water quality, offering a robust tool for sustainable water resource management in agricultural locations.</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

Analysis and prediction of groundwater quality using machine learning algorithm for irrigation purposes

  • 期刊论文

  • 发表日期：

    2025年3月22日

查看全文
  • Hemant Raheja,
  • Arun Goel,
  • Mahesh Pal

摘要

This study investigates the prediction of key irrigation water quality parameters—Sodium Adsorption Ratio (SAR), Magnesium Adsorption Ratio (MAR), Percent Sodium (%Na), Permeability Index (PI), and Kelly’s Index (KI) using hydrochemical data as input features. Groundwater quality data from 272 samples collected in the central western part of Haryana, India, were analyzed. To improve predictive accuracy, four machine learning models were employed: Random Forest (RF), Support Vector Regression (SVR), M5P, and Linear Regression (LR). Additionally, Principal Component Analysis (PCA) was conducted to identify underlying correlations between water quality parameters and to assess the influence of natural processes and anthropogenic activities, such as rock weathering and improper irrigation practices. The results indicated that PI and Magnesium Hazard (MH) values exceeded permissible limits for irrigation, while SAR, %Na, and KI were within acceptable ranges. The SVR and M5P models outperformed RF and LR in predictive accuracy, as supported by uncertainty analysis, which showed lower uncertainty for the former models. The findings highlight the potential of machine learning models, particularly SVR and M5P, to support decision-makers in managing irrigation water quality, offering a robust tool for sustainable water resource management in agricultural locations.