Scholarlink

<p>Continuous forest monitoring using machine learning algorithms is more reliable for land use planning and management. In this study, we assess forest cover change, associated carbon stocks, and future predictions in southwestern Ethiopia using Landsat 8 Operational Land Imager and thermal infrared sensor satellite data of three epochs (2018, 2022, 2025). Multi-temporal land use land cover (LULC) classifications were performed using three machine learning models: classification and regression trees (CART), random forest (RF), and support vector machine (SVM). Classification accuracies were evaluated using overall accuracy. The results show that the RF algorithm offers the most consistent and reliable performance, with greater than 92% overall accuracy. LULC analyses reveal a significant and stable increase in forest areas during the period, while agricultural and grassland areas show varying levels of variability depending on the classifier. Forest cover assessment shows that the net forest gains are much higher than losses. Forest cover increased markedly between 2018 and 2025, by 15.4%, 23.3%, and 23.4% according to the CART, RF, and SVM models, respectively. The result also reveals that the carbon sequestration potential increases as a result of the Green Legacy Initiative (GLI), which increases forest cover. Projections for 2035 indicate that forest areas and carbon sinks are expected to continue increasing; however, the results vary depending on the algorithm used (CART, RF, and SVM). Therefore, this study quantitatively demonstrates the spatial and climatic impacts of afforestation and restoration activities under the Ethiopian GLI; it provides a scientific basis for sustainable forest management and for combating climate change.</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

Integration of machine learning and geospatial technologies for predicting forest cover change and carbon stock under Ethiopian Green Legacy Initiative

  • 期刊论文

  • 发表日期：

    2026年6月22日

查看全文
  • Gedefa Adugna Tamene,
  • Mitiku Badasa Moisa,
  • Indale Niguse Dejene,
  • Debelo Negeri Begna,
  • Fedhasa Benti Chalchissa,
  • Zenebe Reta Roba,
  • Harison Kiplagat Kipkulei,
  • Kiros Tsegay Deribew,
  • Mengistu Muleta Gurmessa,
  • Aqil Tariq,
  • Mehmet Ali Çelik,
  • Dessalegn Obsi Gemeda

摘要

Continuous forest monitoring using machine learning algorithms is more reliable for land use planning and management. In this study, we assess forest cover change, associated carbon stocks, and future predictions in southwestern Ethiopia using Landsat 8 Operational Land Imager and thermal infrared sensor satellite data of three epochs (2018, 2022, 2025). Multi-temporal land use land cover (LULC) classifications were performed using three machine learning models: classification and regression trees (CART), random forest (RF), and support vector machine (SVM). Classification accuracies were evaluated using overall accuracy. The results show that the RF algorithm offers the most consistent and reliable performance, with greater than 92% overall accuracy. LULC analyses reveal a significant and stable increase in forest areas during the period, while agricultural and grassland areas show varying levels of variability depending on the classifier. Forest cover assessment shows that the net forest gains are much higher than losses. Forest cover increased markedly between 2018 and 2025, by 15.4%, 23.3%, and 23.4% according to the CART, RF, and SVM models, respectively. The result also reveals that the carbon sequestration potential increases as a result of the Green Legacy Initiative (GLI), which increases forest cover. Projections for 2035 indicate that forest areas and carbon sinks are expected to continue increasing; however, the results vary depending on the algorithm used (CART, RF, and SVM). Therefore, this study quantitatively demonstrates the spatial and climatic impacts of afforestation and restoration activities under the Ethiopian GLI; it provides a scientific basis for sustainable forest management and for combating climate change.