测绘通报 ›› 2025, Vol. 0 ›› Issue (8): 43-47.doi: 10.13474/j.cnki.11-2246.2025.0807

• 学术研究 • 上一篇    下一篇

多源遥感数据融合的大湄公河次区域森林地上生物量制图

袁丽莉1, 杨欣慰2, 李梦华1,3,4, 陈玉权5, 唐伯惠1,3,4   

  1. 1. 昆明理工大学国土资源工程学院, 云南 昆明 650093;
    2. 中国科学院空天信息创新研究院, 北京 100094;
    3. 云南省定量遥感重点实验室(筹), 云南 昆明 650093;
    4. 云南省山地灾害空天地一体化智慧监测国际联合实验室, 云南 昆明 650093;
    5. 云南骏驰信息技术有限公司, 云南 昆明 650000
  • 收稿日期:2025-05-12 出版日期:2025-08-25 发布日期:2025-09-02
  • 通讯作者: 李梦华。E-mail:menghuali@kust.edu.cn E-mail:menghuali@kust.edu.cn
  • 作者简介:袁丽莉(2000—),女,硕士生,主要从事定量遥感等研究工作。E-mail:yuanlili@stu.kust.edu.cn
  • 基金资助:
    国家自然科学基金(42404036;42230109)

Forest aboveground biomass mapping in the Greater Mekong Subregion using multi-source remote sensing data fusion

YUAN Lili1, YANG Xinwei2, LI Menghua1,3,4, CHEN Yuquan5, TANG Bohui1,3,4   

  1. 1. Faculty of Land Resources Engineering, Kunming University of Science and Technology, Kunming 650093, China;
    2. Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100094, China;
    3. Yunnan Key Laboratory of Quantitative Remote Sensing, Kunming 650093, China;
    4. Yunnan International Joint Laboratory for Integrated Sky-Ground Intelligent Monitoring of Mountain Hazards, Kunming 650093, China;
    5. Yunnan Junchi Information Technology Co., Ltd., Kunming 650000, China
  • Received:2025-05-12 Online:2025-08-25 Published:2025-09-02

摘要: 准确估算森林地上生物量密度对于推动森林的可持续经营至关重要。本文以大湄公河次区域(GMS)为研究对象,基于星载全球生态系统动态调查(GEDI)数据、Sentinel-1、Sentinel-2和其他辅助数据提取了52种特征变量,通过应用LightGBM机器学习模型,绘制1 km分辨率的GMS森林地上生物量密度地图。结果表明,LightGBM模型的R2=0.65、RMSE=38.11 Mg/hm2、EA=72.03%。在整个研究区范围内,生物量密度范围为15.16~423.87 Mg/hm2。本文得到的生物量产品与GEDI L4B产品有较高的相关系数(R2=0.52,RMSE=61.91 Mg/hm2)。总而言之,开放获取的地球观测数据(EO)在估算森林地上生物量方面具有较大的潜力。

关键词: 森林地上生物量, GEDI, 多源遥感数据, 机器学习, 大湄公河次区域

Abstract: Accurate estimation of forest aboveground biomass density is crucial for advancing sustainable forest management.This study focuses on the Greater Mekong Subregion (GMS)and utilizes spaceborne global ecosystem dynamics investigation(GEDI),Sentinel-1,Sentinel-2,and auxiliary datasets to extract 52 feature variables.By applying the LightGBM machine learning model,a 1 km resolution forest aboveground biomass density map of the GMS is generated.The results indicate that the LightGBM model achieved R2=0.65,RMSE=38.11 Mg/hm2,and EA=72.03%.Across the study area,biomass density ranged from 15.16 to 423.87 Mg/hm2.The derived biomass product demonstrated strong correlation with the GEDI L4B product (R2=0.52,RMSE=61.91 Mg/hm2).In conclusion,open-access earth observation (EO)data exhibits significant potential for estimating forest aboveground biomass.

Key words: aboveground biomass density, GEDI, remote sensing data, machine learning, greater mekong subregion

中图分类号: