广西巨尾桉人工林蓄积量估测及样地尺度效应研究  

Stand Volume Estimation of Eucalyptus grandis×E.urophylla Plantation and Scale Effect in Guangxi Province

作　　者：张国飞 岳彩荣 李春干[2] 杜湘[3] 晏娅萍 朱泊东 陈湫丽 Zhang Guofei;Yue Cairong;Li Chungan;Du Xiang;Yan Yaping;Zhu Bodong;Chen Qiuli(College of Forestry,Southwest Forestry University,Kunming Yunnan 650233,China;College of Forestry,Guangxi University,Nanning Guangxi 530004,China;Kunming Metallurgy College,Kunming Yunnan 650033,China;Yunnan Construction Investment First Survey and Design Co.Ltd.,Kunming Yunnan 650033,China)

机构地区：[1]西南林业大学林学院,云南昆明650233 [2]广西大学林学院,广西南宁530004 [3]昆明冶金高等专科学校,云南昆明650033 [4]云南建投第一勘察设计有限公司,云南昆明650033

出　　处：《西南林业大学学报（自然科学）》2023年第3期120-126,共7页Journal of Southwest Forestry University：Natural Sciences

基　　金：国家自然科学基金项目(31860182,31260156,41571372)资助;云南省教育厅科学研究基金项目(2018JS330)资助。

摘　　要：在广西高峰林场巨尾桉人工林内设置20个(30 m×30 m)方形大样地,每个方形样地中分成9个小样地(10 m×10 m),由4个小样地组成中样地(20 m×20 m),采用参数模型和机器学习算法探索林分平均高和密度等变量估测林分巨尾桉人工林蓄积量。结果表明:巨尾桉人工林林分蓄积量与林分平均高、每公顷株树呈显著正相关。在构建巨尾桉人工林蓄积量的所有模型中,基于林分疏密度和林分平均高构建的变参模型(R2=0.9973,RMSE=4.64 m^(3)/hm^(2))优于基于林分平均高和每公顷株数构建的随机森林模型(R2=0.9617,RMSE=18.53 m^(3)/hm^(2));2个林分蓄积量估测模型的3组测试样地的Pearson残差落主要在[-2,2]带状区域中,2个林分蓄积量模型可以在巨尾桉人工林林分不同面积尺度样地(100、400、900 m^(2))应用。基于林分密度和平均高的蓄积量模型与实测蓄积量的相关性较高(R2在0.9200~0.9973),反演误差值较好(RMSE在4.64~25.16 m^(3)/hm^(2));林分疏密度对林分蓄积量变动的解释能力好于每公顷株数;在100 m^(2)样地尺度上基于林分密度(每公顷株树、林分疏密度)和林分平均高拟合的林分蓄积量估测模型在400 m^(2)和900 m^(2)尺度上有较好的一致性。Twenty quadrats(30 m×30 m)were set up in Eucalyptus grandis×E.urophylla plantation in Gaofeng Forest Farm of Guangxi.Each quadrat was divided into 9 plots(10 m×10 m),and 4 plots constituted a medium plot(20 m×20 m).Parametric models and machine Learning algorithms were used to estimate stand volume of E.grandis×E.urophylla plantation with mean height and stand density.Result shows that stand density indicators,including number of per hectare,stand density,were positively correlated with the stand volume of E.grandis×E.urophylla plantation.Among all stand volume models of E.grandis×E.urophylla plantation,the variable parameter model(R2=0.9973,RMSE=4.64 m^(3)/hm^(2))based on stand density and average height was better than the RF model(R2=0.9617,RMSE=18.53 m^(3)/hm^(2))based on number of per hectare and average height;3 sets of standard residuals of test samples of stand volume estimation model all fell in the[−2,2]strip region,indicating that the 2 models had good adaptability in different area scales(100,400 m^(2) and 900 m^(2)).The correlation between the models and the stand volume of E.grandis×E.urophylla plantation were well(R2=0.9200-0.9973).Root Mean Square Error(RMSE)were well(RMSE=4.64-25.16 m^(3)/hm^(2)).Compared with number of per hectare,stand density had a better ability to explain changes in stand volume.The stand volume estimation models,based on stand density and average height in the scale of 100 m^(2) plots,were well adapted to the medium and big sample plots(400 m^(2) and 900 m^(2)).

关 键 词：巨尾桉 人工林 疏密度 蓄积量 尺度效应 

分 类 号：S757[农业科学—森林经理学]