基于Sentinel-2的互助北山林场森林蓄积量反演研究  

The Inversion of Forest Stock Volume in Huzhu Beishan Forest Farm Based on Sentinel-2 Images

作　　者：邓平 顾天江 杜凯 Deng Ping;Gu Tianjiang;Du Kai(Qinghai Provincial Key Laboratory of Physical Geography and Environmental Process,College of Geographical Science,Qinghai Normal University,Xining 810008,China;Key Laboratory of Tibetan Plateau Land Surface Processes and E cological Conservation(Ministry of Education),Qinghai Normal University,Xining 810008,China;Southern Qilian Mountain Forest Ecosystem Observatio n and Research Station of Qinghai Province,Huzhu 810500,China)

机构地区：[1]青海师范大学地理科学学院,青海省自然地理与环境过程重点实验室,西宁810008 [2]青海师范大学,青藏高原地表过程与生态保育教育部重点实验室,西宁810008 [3]青海祁连山南坡森林生态系统国家定位观测研究站,互助810500

出　　处：《青海科技》2024年第3期27-33,共7页Qinghai Science and Technology

基　　金：青海师范大学中青年科研基金资助项目(KJQN2022002),青海省自然科学基金项目青年项目(2024-ZJ-960)。

摘　　要：森林蓄积量是评估森林健康和固碳能力的关键指标,利用遥感技术能够实现区域森林蓄积量的快速估测。本研究以青海省海东市互助北山林场为研究区域,基于森林样地调查数据和Sentinel-2影像,利用提取出的单波段反射率、植被指数、纹理因子和地形因子等筛选出具有代表性的遥感特征,运用随机森林算法实现研究区森林蓄积量反演。主要结果如下:(1)基于皮尔逊相关性分析,确定B11、B10、B6、B7、EVI、SAVI、NIRv、ExGR、MNLI、B12为本研究森林蓄积量反演的遥感特征因子。(2)利用随机森林算法能够有效反演研究区森林蓄积量。反演结果的决定系数R2为0.73,均方根误差RMSE为1.65m^(3)/hm^(2)。(3)互助北山林场乔木林平均蓄积量为134.4m3/hm2,在空间分布上,中东部、东南部和北部部分区域森林蓄积量较高。Forest stock is a key indicator for assessing forest health and carbon sequestration capacity,and the use of remote sensing technology can achieve rapid estimation of regional forest stock.In this study,the Huzhu Beishan Forest Farm in Haidong City,Qinghai Province was taken as the study area.Based on the forest sample plot survey data and Sentinel-2 images,representative remote sensing features were screened out using extracted single-band reflectance,vegetation indices,texture factor and topography factor,and the Random Forest Algorithm was applied to realize the inversion of the forest stock in the study area.The main results are as follows:(1)Based on the Pearson correlation analysis,B11,B10,B6,B7,EVI,SAVI,NIRv,ExGR,MNLI,and B12 are identified as the remote sensing feature factors for forest storage inversion in this study.(2)The random forest algorithm can effectively invert the forest stock in the study area.The coefficient of determination R²of the inversion result was 0.73,and the root mean square error RMSE was 1.65m³/hm2.(3)The average storage volume of arbor forest in Huzhu Beishan Forest Farm was 134.4m^(3)/hm^(2),and in the spatial distribution,the forest storage volume in the east-central,south-eastern,and part of the northern area was higher.

关 键 词：祁连山南麓 Sentinel-2 特征选择 随机森林 蓄积量反演 

分 类 号：S758.51[农业科学—森林经理学] S771.8[农业科学—林学]