基于气候因子的湖南省主要树种组林分生长率模型研建  

Developing stand level growth rate models for dominant tree species groups based on climatic factors in Hunan Province

作　　者：杜志 陈振雄 贺东北 刘紫薇 孙华[2] 黄鑫 王金池 DU Zhi;CHEN Zhenxiong;HE Dongbei;LIU Ziwei;SUN Hua;HUANG Xin;WANG Jinchi(Central South Academy of Inventory and Planning of NFGA,Changsha 410014,Hunan,China;Research Center of Forestry Remote Sensing&Information Engineering,Central South University of Forestry&Technology,Changsha 410004,Hunan,China)

机构地区：[1]国家林业和草原局中南调查规划院,湖南长沙410014 [2]中南林业科技大学林业遥感信息工程研究中心,湖南长沙410004

出　　处：《中南林业科技大学学报》2025年第1期8-17,25,共11页Journal of Central South University of Forestry & Technology

基　　金：湖南省科技创新领军人才项目(2023RC1065)。

摘　　要：【目的】建立基于气候因子的湖南省主要树种组林分生长率混合效应模型,为产出主要林分蓄积量年度变化、支撑省域森林资源年度监测提供理论依据。【方法】基于湖南省第六、七、八、九次全国森林资源连续清查固定样地的调查成果,选取杉木类、马尾松类、栎类、其他硬阔、其他软阔、针叶混、阔叶混、针阔混8个主要树种组共3 036个样地数据,在传统林分蓄积量生长率模型基础上,通过相关性分析引入气候因子构建再参数化模型,并考虑样地效应引入自相关矩阵和异方差函数,构建主要树种组的林分生长率混合效应模型。以十折交叉验证法对3种模型进行检验。【结果】在各树种组林分生长率的模型上,杉木类、马尾松类引入最冷月均温度气候因子,栎类、其他硬阔、其他软阔、针叶混、阔叶混5个树种组引入最热月均温度气候因子,而针阔混树种组引入年平均降水量气候因子,分别构建再参数化模型,并确定了最优随机参数组合后构造混合效应模型。8个主要树种组的基础模型、再参数化模型和混合效应模型拟合效果均逐步提升,栎类树种组的决定系数从基础模型的0.857 9增加到混合效应模型的0.965 8,杉木类、马尾松类和针阔混3个树种组的模型决定系数都提高了0.05以上,均达到0.93以上;针阔混树种组的平均百分标准误差从基础模型的14.59%到再参数化模型的14.32%和混合效应模型的8.22%。各树种组混合效应模型中除其他软阔的决定系数为0.895 6外,其他7个树种组的决定系数均超过了0.900 0,针阔混达到了0.980 1。【结论】引入气候因子的主要树种组林分生长率混合效应模型对于林分蓄积量变化具备较好的拟合效果,能有效支撑省域尺度森林资源蓄积量的年度更新。【Objective】Mixed-effect model of stand level growth rate models for dominant tree species groups was established based on climatic factors in Hunan Province,which provides a theoretical basis for the annual variation stand volume and monitoring of forest resources.【Method】Based on the data of the sixth,seventh,eighth and ninth national forest resources continuous inventory sample plots in Hunan Province,eight dominant tree species groups of Chinese fir group,Pinus group,Quercus spp.,other hardwood broadleaved,other softwood broadleaved,mixed coniferous,mixed broadleaved,mixed coniferous and broadleaved,with a total of 3036 sample plots were selected,based on the traditional stand volume growth rate model,the climate factors were introduced through correlation analysis to construct a re-parameterization model.Meanwhile,the effect of the sample plots was considered,and the autocorrelation matrix and heteroskedasticity function were introduced to construct the mixed-effect model.The three models were tested by the ten-fold cross-validation method.【Result】The coldest monthly mean temperature was introduced into the Chinese fir group and Pinus group,the hottest monthly average temperature was introduced into the Quercus spp.,other hardwood broadleaved species groups,other softwood broadleaved species groups,mixed coniferous species groups,and mixed broadleaved species groups,and the annual average precipitation were introduced into the mixed coniferous and broad-leaved species groups.The re-parameterization model was constructed,and the optimal combination of random parameters was determined to construct a mixed-effect model.The results showed that the fitting effect of the basic model,the re-parameterization model and the mixed-effect model were gradually improved,among which the determination coefficient of the Pinus group increased from 0.8579 in the basic model to 0.9658 in the mixed-effect model,and Chinese fir group,Pinus group,mixed coniferous and broad-leaved species groups increased more than 0.05

关 键 词：林分生长率 主要树种组 气候因子 再参数化模型 混合效应模型 

分 类 号：S758.52[农业科学—森林经理学]