基于混合效应的湖南杉木人工林平均高和优势木平均高相关关系模型  

Correlation model of average height and dominant average height of Cunninghamia lanceolata plantations in Hunan province based on mixed effect

作　　者：杨子铎 李新建 朱光玉[1] 刘洪娜 YANG Ziduo;LI Xinjian;ZHU Guangyu;LIU Hongna(College of Forestry,Central South University of Forestry&Technology,Changsha 410004,Hunan,China;Central South Forest Inventory and Planning Institute of State Forestry Administration,Changsha 410014,Hunan,China)

机构地区：[1]中南林业科技大学林学院,湖南长沙410004 [2]国家林业和草原局中南林业调查规划设计院,湖南长沙410014

出　　处：《中南林业科技大学学报》2022年第3期62-71,共10页Journal of Central South University of Forestry & Technology

基　　金：国家自然科学基金项目(31570631);湖南省教育厅项目(18C0277);国家级大学生创新训练项目(202110538003)。

摘　　要：【目的】考虑湖南杉木人工林的立地条件、林分密度和林分年龄,基于混合效应构建最适宜的林分平均高与优势木平均高相关关系模型,为湖南杉木人工林的立地质量评价与可持续经营提供理论依据。【方法】以湖南省内设置的512块杉木样地数据为基础,基于基础线性模型,利用数量化方法Ⅰ、k-means聚类和线性混合效应回归建模方法,构建基于混合效应的湖南杉木林分平均高和优势木平均高相关关系模型。【结果】1)采用数量化方法Ⅰ筛选出坡位、坡形、土壤类型、土壤厚度、株数密度和土壤松紧度是对高比值(林分优势木平均高和林分平均高之比)影响显著的因子;2)构建线性混合效应模型可以显著提高建模精度,立地类型作为随机效应加在参数b上时确定系数最高,相对于基础线性模型,其确定系数从0.790158提高到0.914348。3)构建基于聚类的线性混合效应模型可以进一步提高建模精度,立地类型组作为随机效应加在参数b上时确定系数最高,确定系数达到了0.9435128,相对于基础线性模型,确定系数提高了19.41%,平均绝对误差降低了53.89%,均方根误差降低了48.11%;4)添加株数密度和林分年龄作为固定效应对模型精度提升不明显。【结论】林分密度和林分年龄对林分平均高与优势木平均高相关关系无显著影响;基于立地效应的混合模型可以反映平均高与优势木平均高相关关系变化的整体趋势,也可以体现立地条件差异对两者相关关系的影响,模型精度更高,适用性更好。【Objective】Considering the site conditions,stand density and stand age of Cunninghamia lanceolata plantations in Hunan Province,the most suitable correlation model between the average stand height and dominant average height was established based on mixed effect in this study,in order to provide a theoretical basis for the site quality evaluation and sustainable management of C.lanceolata plantations in Hunan province.【Method】Based on the data of 512 C.lanceolata sample plots in Hunan province and the basic linear model,this article used the quantitative method I,k-means clustering and linear mixed effect regression modeling methods to construct the correlation model of average height and dominant average height of C.lanceolata stands in Hunan province.【Result】1)Slope position,slope shape,soil type,soil thickness,plant density and soil tightness were selected as significant factors for the high ratio(the ratio of average stand dominant height to average stand height)by using quantitative method I.2)The linear mixed effect model was built,which could significantly improve the modeling accuracy,and site type was added to parameter b as a random effect.Compared with the basic linear model,the coefficient of certainty increased by 0.914348.3)When site type group was added to parameter b as a random effect,the coefficient of determination was the highest,reaching 0.9435128.Compared with the basic linear model,the coefficient of determination was increased by 19.41%,the mean absolute error was reduced by 53.89%,and the root mean square error was reduced.4)Adding plant number density and stand age as the fixed effects did not significantly improve the accuracy of the model.【Conclusion】Stand density and stand age had no significant effect on the correlation between stand average height and dominant average height.The hybrid model based on site effect can reflect not only the overall trend of the correlation between the mean height and the dominant mean height,but also the impact of site condition differen

关 键 词：杉木 林分优势木平均高 林分平均高 线性混合模型 

分 类 号：S791.27[农业科学—林木遗传育种]