Scholarlink

<p>An up-to-date vegetation map is essential for characterizing current ecosystem conditions in the source regions of the Yangtze and Yellow Rivers (SRYYR) under ongoing climate change. Here, we release a region-specific 30 m vegetation type dataset that represents typical conditions in the early 2020s. The map was created using systematic ground survey data from 1,168 georeferenced sites collected from 2020 to 2025. A random forest classifier was trained using Landsat imagery and environmental predictors to generate the vegetation type map. Sample representativeness was quantified by comparing predictor distributions between survey sites and the study area. The Euclidean distance ranged from 2.99% to 8.85% and Pearson’s r from 0.59 to 0.97. The final map had an overall accuracy of 79.56% and a Cohen’s kappa of 0.74. Unlike existing plateau-wide products, this dataset explicitly delineates alpine swamp meadows (ASMs) and improves the spatial representation of alpine shrublands (ASHs), providing an updated baseline for permafrost–vegetation coupling and related ecohydrological processes in the SRYYR.</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

A 30 m vegetation map for the source regions of the Yangtze and Yellow Rivers from 2020–2025 ground surveys

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月14日

查看全文
  • Defu Zou,
  • Xiaodong Wu,
  • Guojie Hu,
  • Tonghua Wu,
  • Qiu Jin,
  • Liangang Chen,
  • Yanjun Bian,
  • Cong Wang,
  • Yuting Lan,
  • Lin Zhao,
  • Zhibin Li

摘要

An up-to-date vegetation map is essential for characterizing current ecosystem conditions in the source regions of the Yangtze and Yellow Rivers (SRYYR) under ongoing climate change. Here, we release a region-specific 30 m vegetation type dataset that represents typical conditions in the early 2020s. The map was created using systematic ground survey data from 1,168 georeferenced sites collected from 2020 to 2025. A random forest classifier was trained using Landsat imagery and environmental predictors to generate the vegetation type map. Sample representativeness was quantified by comparing predictor distributions between survey sites and the study area. The Euclidean distance ranged from 2.99% to 8.85% and Pearson’s r from 0.59 to 0.97. The final map had an overall accuracy of 79.56% and a Cohen’s kappa of 0.74. Unlike existing plateau-wide products, this dataset explicitly delineates alpine swamp meadows (ASMs) and improves the spatial representation of alpine shrublands (ASHs), providing an updated baseline for permafrost–vegetation coupling and related ecohydrological processes in the SRYYR.