不同森林类型树高空间分布特征及影响因素  

Spatial Distribution Characteristics and Influencing Factors of Tree Height in Different Forest Types

作　　者：潘子阳 崔哲霖 徐雁南[1] 孙心雨 Pan Ziyang;Cui Zhelin;Xu Yannan;Sun Xinyu(Collaborative Innovation Center of Sustainable Forestry in Southern China of Jiangsu Province(Nanjing Forestry University),Nanjing 210037,P.R.China;Jiangsu Vocational College of Agriculture and Forestry)

机构地区：[1]南方现代林业协同创新中心(南京林业大学),南京210037 [2]江苏农林职业技术学院

出　　处：《东北林业大学学报》2025年第5期31-40,152,共11页Journal of Northeast Forestry University

基　　金：国家重点研发计划课题(2019YFD1100404);江苏省林业科技创新与推广项目(LYKJ[2021]14)。

摘　　要：树高是森林资源调查中反应森林生长状况的重要因子,也是林木资产评估、生物量计算和生态服务功能评价的关键指标。以安徽省黄山区的Sentinel-2时序影像为数据源,使用SG滤波后提取的统计特征,结合随机森林与交叉验证特征递归消除方法(RFECV),获取黄山区森林类型分类栅格;结合机载激光雷达数据提取的树高栅格,分析各森林类型的空间分布特征;利用提取的地形参数和森林结构参数,通过广义加性模型(GAM)分析常见地形参数和冠层参数对树高的影响。结果表明:(1)根据时序统计特征,随机森林模型对森林类型分类总体精度为0.91,Kappa系数为0.89。(2)各森林类型树高的空间自相关性均随距离增加而显著减弱,研究区竹林树高低值主要聚集在西北方,针叶林分布稀疏未呈现聚集现象,阔叶林为研究区的主要森林类型,其树高高值沿东北-西南方向集中分布。(3)在广义加性模型(GAM)中,点云偏态和冠层孔隙率在各森林类型占据较高的相对重要性,坡度、坡向和高程的相对重要性之和不超过40%,冠层孔隙率和平均曲率在多数森林类型中与树高呈显著线性相关关系。Tree height is an important factor reflecting the growth status of forests in forest resource surveys,and it is also a key indicator for the assessment of forest tree assets,biomass calculation,and evaluation of ecological service functions.Taking the Sentinel-2 time-series images in Huangshan District,Anhui Province as the data source,the statistical features extracted after SG filtering are used.Combined with the Random Forest and Recursive Feature Elimination with Cross-Validation(RFECV)method,the classified raster of forest types in Huangshan District is obtained.By combining with the tree height raster extracted from airborne LiDAR data,the spatial distribution characteristics of each forest type are analyzed.Using the extracted topographic parameters and forest structure parameters,the influence of common topographic parameters and canopy parameters on tree height is analyzed through the Generalized Additive Model(GAM).The results show that:(1)According to the time-series statistical features,the overall accuracy of the Random Forest model for forest type classification is 0.91,and the Kappa coefficient is 0.89.(2)The spatial autocorrelation of tree height of each forest type decreases significantly with the increase of distance.The low tree height values of bamboo forests in the study area mainly gather in the northwest.The distribution of coniferous forests is sparse and there is no clustering phenomenon.Broad-leaved forests are the main forest type in the study area,and their high tree height values are concentrated along the northeast-southwest direction.(3)In the Generalized Additive Model(GAM),the point cloud skewness and canopy porosity occupy a relatively high relative importance in each forest type.The sum of the relative importance of slope,aspect,and elevation does not exceed 40%.The canopy porosity and average curvature have a significant linear correlation with tree height in most forest types.

关 键 词：森林类型 树高分布 机载激光雷达 机器学习模型 

分 类 号：S757.4[农业科学—森林经理学]