Vegetation mapping of Yunnan Province by integrating remote sensing, field observations, and models  

作　　者：Mingjian XIAHOU Mingchun PENG Zehao SHEN Qingzhong WEN Chongyun WANG Yannan LIU Qiuyuan ZHANG Lei PENG Changyuan YU Xiaokun OU Jingyun FANG 

机构地区：[1]Key Laboratory of Ministry of Education for Earth Surface Processes,College of Urban&Environmental Sciences,Peking University,Beijing 100871,China [2]School of Ecology and Environment,Yunnan University,Kunming 650504,China [3]Yunnan Institute of Forest Inventory and Planning,Kunming 650051,China

出　　处：《Science China Earth Sciences》2025年第3期836-849,共14页中国科学(地球科学英文版)

基　　金：supported by the Major Program for Basic Research Project of Yunnan Province (Grant No. 202101BC070002);the Second Comprehensive Scientific Expedition of the Tibetan Plateau (Grant No. 2019QZKK04020101)。

摘　　要：Vegetation maps are crucial for ecologists and decision-makers, providing essential information on the spatial distribution of various vegetation types to support ecosystem exploration and management. Despite advancements in Earth observation and machine learning enabling large-scale vegetation mapping, creating detailed and accurate maps in biodiversity hotspots remains challenging due to significant environmental heterogeneity and frequent human disturbances. The lack of sufficient ground-based data and complex climate-vegetation interactions further limits mapping accuracy. In this study, we developed an integrated framework for multi-source data fusion to enhance vegetation mapping and validation in Yunnan Province, a global biodiversity hotspot region in Southwest China. The mapping process involved four key steps:(1) vegetation classification using random forest and Landsat imagery,(2) boundary calibration based on a locally calibrated static climatevegetation model,(3) patch correction with independent forest inventory data, and(4) validation using adequate field observations. This approach enabled the mapping of 17 vegetation types and 44 subtypes in Yunnan Province(1:50000), categorized based on the growth-form composition of dominant species of the community. The overall accuracies were 0.747 and0.710 for natural vegetation types and subtypes, and 0.905 and 0.891 for artificial types and subtypes. This high-resolution map enhances our understanding of vegetation distribution and ecological complexity in this region, offering valuable insights for policymakers to support conservation efforts and sustainable management strategies.

关 键 词：Vegetation mapping Biodiversity hotspot region Multi-source data Data-fusion framework Ecological diversity 

分 类 号：F42[经济管理—产业经济]