Scholarlink

<p>Rapid land use and land cover (LULC) changes driven by global population growth and industrial activities, such as coal mining, pose significant challenges to ecosystems, necessitating robust monitoring for sustainable environmental management. This study leverages Google Earth Engine (GEE) to classify LULC around a major open-cast coal mining complex in Chhattisgarh, India, using Sentinel-2 (10&#xa0;m resolution) and Landsat-7 (30&#xa0;m resolution) imagery from January to October of 2013 and 2023. Three machine learning classifiers, Random Forest (RF), Support Vector Machine (SVM), and Classification and Regression Trees (CART), were applied to categorise LULC into five classes: mining_area, water, built-up_area, agriculture, and vegetation. Supervised classification and accuracy assessments revealed RF as the top performer, achieving overall accuracies of 0.84 for Sentinel-2 and 0.81 for Landsat-7, compared to 0.80 and 0.61 for SVM and 0.79 and 0.80 for CART, respectively. Sentinel-2’s finer resolution and red-edge bands enhanced accuracy by 3–5%, particularly for fine-scale features like mining area boundaries. Key LULC transitions included a 31% shift from agriculture to built-up areas, reflecting mining-induced urbanisation and dust deposition. These findings highlight RF’s efficacy for LULC mapping in mining regions and the advantage of high-resolution imagery for environmental monitoring. By integrating multi-resolution data, this study offers a framework for policymakers to address mining-related environmental degradation in Chhattisgarh, supporting data-driven sustainable land management and climate change mitigation efforts.</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

Investigation of Land Use and Land Cover Using Google Earth Engine Machine Learning Algorithms Surrounding a Major Open Cast Coal Mining Complex, Chhattisgarh (India)

  • 期刊论文

  • 发表日期：

    2025年12月19日

查看全文
  • Abhimanyu Kumar Gond,
  • Aarif Jamal,
  • Tarun Verma

摘要

Rapid land use and land cover (LULC) changes driven by global population growth and industrial activities, such as coal mining, pose significant challenges to ecosystems, necessitating robust monitoring for sustainable environmental management. This study leverages Google Earth Engine (GEE) to classify LULC around a major open-cast coal mining complex in Chhattisgarh, India, using Sentinel-2 (10 m resolution) and Landsat-7 (30 m resolution) imagery from January to October of 2013 and 2023. Three machine learning classifiers, Random Forest (RF), Support Vector Machine (SVM), and Classification and Regression Trees (CART), were applied to categorise LULC into five classes: mining_area, water, built-up_area, agriculture, and vegetation. Supervised classification and accuracy assessments revealed RF as the top performer, achieving overall accuracies of 0.84 for Sentinel-2 and 0.81 for Landsat-7, compared to 0.80 and 0.61 for SVM and 0.79 and 0.80 for CART, respectively. Sentinel-2’s finer resolution and red-edge bands enhanced accuracy by 3–5%, particularly for fine-scale features like mining area boundaries. Key LULC transitions included a 31% shift from agriculture to built-up areas, reflecting mining-induced urbanisation and dust deposition. These findings highlight RF’s efficacy for LULC mapping in mining regions and the advantage of high-resolution imagery for environmental monitoring. By integrating multi-resolution data, this study offers a framework for policymakers to address mining-related environmental degradation in Chhattisgarh, supporting data-driven sustainable land management and climate change mitigation efforts.