Scholarlink

<p>Understanding land use and land cover (LULC) patterns and their changes is crucial for monitoring environmental dynamics, guiding sustainable resource use, and supporting sound land management decisions. However, in Southeastern Ethiopia, timely and detailed assessments of LULC changes remain limited, posing challenges for effective land governance. This study addresses this gap by expolaring LULC dynamics in the Welmel watershed using Google Earth Engine (GEE) and evaluating the performance of various machine learning classifiers. LULC classification employed stratified random sampling across eight land cover types, using 2,116 samples per year. A 70/30 split was applied for training and validation to ensure robust and reliable model performance. The accuracy of four machine learning algorithms including CART, SVM, GTB, and RF was evaluated on the GEE platform for LULC classification. Landsat 5-TM, 7-ETM+, and 8-OLI surface reflectance images (30&#xa0;m resolution), along with spectral indices and topographic data, were used to improve classification performance. The accuracy of the LULC classification was assessed using overall accuracy, Kappa coefficient, producer accuracy, and consumer accuracy. The analysis revealed that the RF algorithm achieved the highest overall accuracy of 88.9% and a Kappa coefficient of 0.87. In contrast, the SVM displayed lower overall accuracy and Kappa coefficient values of 74.47% and 0.72, respectively. Based on performance, the RF classifier was selected for final LULC change detection, indicating that 60.67% of the watershed experienced significant changes during the study period. Key changes included shrubland to grazing land (6.3%) and woodland to cultivated land (6.2%), driven mainly by population growth and agricultural expansion, with important impacts on ecosystem health and land sustainability. The findings emphasize the need for targeted land use policies that promote sustainable agricultural practices, forest conservation, and climate-adaptive resource management to mitigate land degradation and support the resilience of the Welmel watershed.</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

Machine learning-based analysis of land use and land cover trends in southeastern Ethiopia using Google Earth Engine

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年8月26日

查看全文
  • Tesfaye Bogale,
  • Sileshi Degefa,
  • Gemedo Dalle,
  • Gebeyehu Abebe

摘要

Understanding land use and land cover (LULC) patterns and their changes is crucial for monitoring environmental dynamics, guiding sustainable resource use, and supporting sound land management decisions. However, in Southeastern Ethiopia, timely and detailed assessments of LULC changes remain limited, posing challenges for effective land governance. This study addresses this gap by expolaring LULC dynamics in the Welmel watershed using Google Earth Engine (GEE) and evaluating the performance of various machine learning classifiers. LULC classification employed stratified random sampling across eight land cover types, using 2,116 samples per year. A 70/30 split was applied for training and validation to ensure robust and reliable model performance. The accuracy of four machine learning algorithms including CART, SVM, GTB, and RF was evaluated on the GEE platform for LULC classification. Landsat 5-TM, 7-ETM+, and 8-OLI surface reflectance images (30 m resolution), along with spectral indices and topographic data, were used to improve classification performance. The accuracy of the LULC classification was assessed using overall accuracy, Kappa coefficient, producer accuracy, and consumer accuracy. The analysis revealed that the RF algorithm achieved the highest overall accuracy of 88.9% and a Kappa coefficient of 0.87. In contrast, the SVM displayed lower overall accuracy and Kappa coefficient values of 74.47% and 0.72, respectively. Based on performance, the RF classifier was selected for final LULC change detection, indicating that 60.67% of the watershed experienced significant changes during the study period. Key changes included shrubland to grazing land (6.3%) and woodland to cultivated land (6.2%), driven mainly by population growth and agricultural expansion, with important impacts on ecosystem health and land sustainability. The findings emphasize the need for targeted land use policies that promote sustainable agricultural practices, forest conservation, and climate-adaptive resource management to mitigate land degradation and support the resilience of the Welmel watershed.