摘要
精确估算单位面积森林蓄积量,是测算林地碳汇能力、合理确定林木采伐量、防治水土流失的技术基础,对促进林地资源的精细化、智慧化管理有重要作用和现实意义。在我国“双碳”目标驱动下,森林蓄积量的精准监测已成为生态管理的核心需求。本文针对现有森林蓄积量估算方法的不足,以浙江东阳市为研究区,通过机载LiDAR点云数据和森林资源二类调查数据,选择松、杉、阔叶混、针叶混四种树种类型开展单位面积森林蓄积量精确估算技术研究。研究表明:通过激光雷达提取的高度变量和强度变量能很好表征森林结构特征,提出的基于激光雷达数据和AutoGlum集成模型能实现不同树种森林蓄积量的精确估算。与传统的线性回归模型(OLS)相比较,基于AutoGluon加权集成模型的蓄积量估算总精度在松、杉、阔叶混、针阔混等树种分别提升了13.80%、13.60%、18.79%、18.18%。本文提出的森林蓄积量精确估算方法,在山地丘陵等林区有一定的推广价值和借鉴作用。
Accurate estimation of forest stock volume per unit area serves as a critical technical foundation for measuring forest carbon sequestration capacity,determining rational timber harvest quotas,and preventing soil erosion.It plays a vital role in promoting the finegrained and intelligent management of forest resources.Driven by China’s“dual carbon”goals,the precise monitoring of forest stock volume has become a core requirement for ecological management.Addressing the limitations of existing forest stock volume estimation methods,this study focuses on Dongyang City in Zhejiang Province as the research area.Utilizing airborne LiDAR(Light Detection and Ranging)point cloud data and ClassⅡforest resource inventory data,we conducted research on precise estimation techniques for the stock volume per unit area across four vegetation types:masson pine(Pinus massoniana),Chinese fir(Cunninghamia lanceolata),broad-leaved mixed forest,and coniferous mixed forest.The research demonstrates that LiDAR-derived height and intensity variables effectively characterize forest structural attributes.The proposed AutoGluon-based weighted ensemble model,leveraging LiDAR data,enables accurate estimation of forest stock volume across different tree species.Compared to the traditional ordinary least squares(OLS)linear regression model,the overall estimation accuracy of the AutoGluon ensemble model increased significantly:by 13.80%for masson pine,13.60%for Chinese fir,18.79%for broad-leaved mixed forest,and 18.18%for coniferous mixed forest.The precise forest stock volume estimation method proposed in this study exhibits certain promotional value and offers reference significance for application in forested regions characterized by mountainous and hilly terrain.
作者
姚鸿文
祝锦霞
杜群
季碧勇
朱程昊
王柯
YAO Hongwen;ZHU Jinxia;DU Qun;JI Biyong;ZHU Chenghao;WANG Ke(Forest Resource Monitoring Center of Zhejiang Province,Zhejiang Hangzhou 310020,China;School of Public Policy and Management,Zhejiang University of Finance&Economics,Zhejiang Hangzhou 310018,China)
出处
《上海国土资源》
2025年第4期136-142,150,共8页
Shanghai Land & Resources
基金
浙江省“领雁”研发攻关计划项目(2022C02053)。
关键词
土地生态
“双碳”目标
林地资源
智慧管理
森林蓄积量
精确估算
机载LiDAR点云
集成学习算法
精细化管理
land ecology
“dual carbon”goals
forest land resources
intelligent management
forest stock volume
accurate estimation
airborne LiDAR point cloud
ensemble learning algorithm
refined management
