Scholarlink

<p>Groundwater depletion has become a critical global issue, particularly in hard rock terrains, where rapid urbanization, over-extraction, and unregulated land use practices have exacerbated both water scarcity and contamination. These challenges highlight the urgent need for accurate identification of Groundwater Potential Zones (GWPZs), essential for sustainable groundwater resource management. Mapping these zones is a complex task due to the intricate interplay of various hydrogeological factors, including topography, geology, and climate, as well as the limited availability of reliable groundwater data. Traditional methods often struggle to address these complexities, necessitating advanced approaches that can integrate multiple environmental and geological variables for more precise and effective groundwater assessments. This study aims to fill this gap by applying machine learning (ML) algorithms—Generalized Linear Model (GLM), Support Vector Machine (SVM), and Random Forest (RF)—to estimate groundwater potential in Ranchi. To assess the comparative effectiveness of the ML models, a multi-criteria decision-making (MCDM) method, specifically the Rank Sum model, was also employed. The study identifies critical factors influencing groundwater occurrence, including slope, drainage density, geomorphology, lithology, soil texture, rainfall, land use/land cover, and lineament density. Well discharge and groundwater depth data were used as the primary training and testing datasets, with a random sampling ratio of 70:30 for model training and validation. The performance of the models was evaluated using metrics such as Area under the Curve (AUC), recall, accuracy, and precision. The AUC-based accuracy levels for the Rank Sum, SVM, GLM, and RF models were 76%, 81%, 90%, and 96%, respectively, demonstrating superior predictive performance by the ML models over the MCDM approach. The findings suggest that the developed models can provide valuable insights into the spatial and temporal dynamics of groundwater potential, contributing to effective groundwater resource management in regions with similar terrain.</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

Geospatial and Machine Learning Technique Based Zonation of Groundwater Potential: A Case Study of Ranchi District, Jharkhand, India

  • 期刊论文

  • 发表日期：

    2025年7月20日

查看全文
  • Rani Kumari,
  • Akhouri Pramod Krishna

摘要

Groundwater depletion has become a critical global issue, particularly in hard rock terrains, where rapid urbanization, over-extraction, and unregulated land use practices have exacerbated both water scarcity and contamination. These challenges highlight the urgent need for accurate identification of Groundwater Potential Zones (GWPZs), essential for sustainable groundwater resource management. Mapping these zones is a complex task due to the intricate interplay of various hydrogeological factors, including topography, geology, and climate, as well as the limited availability of reliable groundwater data. Traditional methods often struggle to address these complexities, necessitating advanced approaches that can integrate multiple environmental and geological variables for more precise and effective groundwater assessments. This study aims to fill this gap by applying machine learning (ML) algorithms—Generalized Linear Model (GLM), Support Vector Machine (SVM), and Random Forest (RF)—to estimate groundwater potential in Ranchi. To assess the comparative effectiveness of the ML models, a multi-criteria decision-making (MCDM) method, specifically the Rank Sum model, was also employed. The study identifies critical factors influencing groundwater occurrence, including slope, drainage density, geomorphology, lithology, soil texture, rainfall, land use/land cover, and lineament density. Well discharge and groundwater depth data were used as the primary training and testing datasets, with a random sampling ratio of 70:30 for model training and validation. The performance of the models was evaluated using metrics such as Area under the Curve (AUC), recall, accuracy, and precision. The AUC-based accuracy levels for the Rank Sum, SVM, GLM, and RF models were 76%, 81%, 90%, and 96%, respectively, demonstrating superior predictive performance by the ML models over the MCDM approach. The findings suggest that the developed models can provide valuable insights into the spatial and temporal dynamics of groundwater potential, contributing to effective groundwater resource management in regions with similar terrain.