Scholarlink

<p>Modelling land use and land cover changes resulting from mining at Grootegeluk coal mine has become essential for understanding the spatial and temporal impacts of mining on the surrounding geo-environment and for informing environmental management and rehabilitation strategies. Using Landsat satellite imagery spanning the years 1990, 1999, 2008, 2017, and 2025, the study employs advanced machine learning algorithms, namely eXtreme Gradient Boost (XGB) and Random Forest (RF), to classify and model land use and land cover (LULC) changes over the years. Post-classification change detection was applied to quantify transitions among major land cover classes—built-up, vegetation, and water—enabling the assessment of the extent and spatial distribution of environmental transformation resulting from mining operations and urban expansion. Furthermore, prediction modelling using the CA–Markov framework was conducted to project future LULC patterns for 2035, providing insights into anticipated landscape evolution under continued mining activity. Comparative analysis demonstrated that XGB outperformed RF, achieving higher spatial precision, smoother class boundaries, and superior accuracy (overall accuracy = 89%, Kappa = 0.7931), while effectively handling gaps in data resulting from the 2003 Landsat 7 Scan Line Corrector (SLC) failure. Change detection analysis showed consistent vegetation-to-built-up transitions, confirming extensive environmental disturbance. CA–Markov projections for 2035 suggest continued vegetation loss, increased fragmentation, and limited natural recovery. XGB also performed better in prediction modelling with an overall accuracy of 0.7931. The findings highlight the importance of integrating remote sensing and machine learning techniques for effective environmental monitoring, impact assessment, and sustainable mining management. Based on these outcomes, the study proposes continuous LULC assessment to support sustainable land use planning and environmental restoration in mining-affected regions. And suggests post-mining land restoration frameworks.</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

Modelling long-term land cover changes resulting from mining at Grootegeluk coal mine, Limpopo province, South Africa: implication for environmental management

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月28日

查看全文
  • Nothando Boikanyo Dhlongolo,
  • Paidamwoyo Mhangara,
  • Eskinder Gidey

摘要

Modelling land use and land cover changes resulting from mining at Grootegeluk coal mine has become essential for understanding the spatial and temporal impacts of mining on the surrounding geo-environment and for informing environmental management and rehabilitation strategies. Using Landsat satellite imagery spanning the years 1990, 1999, 2008, 2017, and 2025, the study employs advanced machine learning algorithms, namely eXtreme Gradient Boost (XGB) and Random Forest (RF), to classify and model land use and land cover (LULC) changes over the years. Post-classification change detection was applied to quantify transitions among major land cover classes—built-up, vegetation, and water—enabling the assessment of the extent and spatial distribution of environmental transformation resulting from mining operations and urban expansion. Furthermore, prediction modelling using the CA–Markov framework was conducted to project future LULC patterns for 2035, providing insights into anticipated landscape evolution under continued mining activity. Comparative analysis demonstrated that XGB outperformed RF, achieving higher spatial precision, smoother class boundaries, and superior accuracy (overall accuracy = 89%, Kappa = 0.7931), while effectively handling gaps in data resulting from the 2003 Landsat 7 Scan Line Corrector (SLC) failure. Change detection analysis showed consistent vegetation-to-built-up transitions, confirming extensive environmental disturbance. CA–Markov projections for 2035 suggest continued vegetation loss, increased fragmentation, and limited natural recovery. XGB also performed better in prediction modelling with an overall accuracy of 0.7931. The findings highlight the importance of integrating remote sensing and machine learning techniques for effective environmental monitoring, impact assessment, and sustainable mining management. Based on these outcomes, the study proposes continuous LULC assessment to support sustainable land use planning and environmental restoration in mining-affected regions. And suggests post-mining land restoration frameworks.