Scholarlink

<p>Land cover mapping is essential for analyzing environmental dynamics and the impact of human activities on land degradation and biodiversity. By integrating information on the Atlas pistachio tree, which plays a key role in combating desertification, particularly in arid areas, decision-makers can better guide their restoration and conservation strategies in the face of environmental challenges. To this end, our study focuses on mapping land use units. In particular, the spatial distribution of Atlas pistachio (<i>Pistacia atlantica</i> Desf) stands, using the Google Earth Engine (GEE) platform and machine learning algorithms, including random forest (RF), support vector machine (SVM), spectral gradient temporal-backtracking (SGTB), and spectral correlation (SC), was used. Sentinel-2A (COPERNICUS/S2) data for June 2023 with a 10-m resolution was utilized. The methodology involved preprocessing, normalization, creating a composite image using spectral indices (NDVI, NDWI, SAVI, GCI, GNDVI) and bands (‘B2’, ‘B3’, ‘B4’, ‘B8’), and a slope map. Training and validation points were coded based on image reflectance and ground truth verification. Classification assessment utilized metrics like confusion matrix, overall accuracy, producer’s accuracy, consumer’s accuracy, and kappa coefficient. The RF algorithm showed the highest accuracy (83.31%, kappa: 0.80). The land composition included 0.01% water bodies, 43.53% dunes, 42.70% pastures, 7.66% bare soil, 0.82% urban areas, and 0.77% agriculture, 2.06% other land cover types.</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

Land cover classification and Atlas pistachio mapping in El-Bayadh and Naâma Provinces: Using Sentinel-2 data and machine learning algorithms

  • 期刊论文

  • 发表日期：

    2025年8月5日

查看全文
  • Naimi Bendouina,
  • Lakhdar Guerine,
  • Kouider Hadjadj

摘要

Land cover mapping is essential for analyzing environmental dynamics and the impact of human activities on land degradation and biodiversity. By integrating information on the Atlas pistachio tree, which plays a key role in combating desertification, particularly in arid areas, decision-makers can better guide their restoration and conservation strategies in the face of environmental challenges. To this end, our study focuses on mapping land use units. In particular, the spatial distribution of Atlas pistachio (Pistacia atlantica Desf) stands, using the Google Earth Engine (GEE) platform and machine learning algorithms, including random forest (RF), support vector machine (SVM), spectral gradient temporal-backtracking (SGTB), and spectral correlation (SC), was used. Sentinel-2A (COPERNICUS/S2) data for June 2023 with a 10-m resolution was utilized. The methodology involved preprocessing, normalization, creating a composite image using spectral indices (NDVI, NDWI, SAVI, GCI, GNDVI) and bands (‘B2’, ‘B3’, ‘B4’, ‘B8’), and a slope map. Training and validation points were coded based on image reflectance and ground truth verification. Classification assessment utilized metrics like confusion matrix, overall accuracy, producer’s accuracy, consumer’s accuracy, and kappa coefficient. The RF algorithm showed the highest accuracy (83.31%, kappa: 0.80). The land composition included 0.01% water bodies, 43.53% dunes, 42.70% pastures, 7.66% bare soil, 0.82% urban areas, and 0.77% agriculture, 2.06% other land cover types.