欢迎访问林业科学,今天是

林业科学 ›› 2013, Vol. 49 ›› Issue (8): 65-74.doi: 10.11707/j.1001-7488.20130810

• 论文与研究报告 • 上一篇    下一篇

基于混合效应模型的杉木单木冠幅预测模型

符利勇1, 孙华1,2   

  1. 1. 中国林业科学研究院资源信息研究所 北京 100091;
    2. 中南林业科技大学林业遥感信息工程研究中心 长沙 410004
  • 收稿日期:2012-11-08 修回日期:2013-03-23 出版日期:2013-08-25 发布日期:2013-08-17
  • 通讯作者: 孙华
  • 基金资助:

    林业公益性行业重点项目"我国典型森林类型健康经营关键技术研究"(201004002); 林业公益性行业科研专项"林分结构与生长模拟技术研究"(201104028)。

Individual Crown Diameter Prediction for Cunninghamia lanceolata Forests Based on Mixed Effects Models

Fu Liyong1, Sun Hua1,2   

  1. 1. Research Institute of Forest Resources Information Techniques, CAF Beijing 100091;
    2. Research Center of Forestry Remote Sensing and Information Engineering, Central South University of Forestry and Technology Changsha 410004
  • Received:2012-11-08 Revised:2013-03-23 Online:2013-08-25 Published:2013-08-17

摘要:

以湖南省黄丰桥国有林场103块样地共2 461株杉木为例,建立单木冠幅模型。由于所调查数据是在不同立地条件下相同样地中重复观察得到,数据间存在明显相关性,为解决此问题,将考虑立地指数和样地对冠幅生长的随机影响,即建立嵌套2水平非线性混合冠幅模型。从12个常用林分模型中选出较好的冠幅直径模型作为构建混合模型的基础模型。除胸高直径外,还考虑其他17个林分或树木因子对冠幅的影响。通过指标AIC(akaike information criterion)和对数似然确定最佳形式参数随机效应组合类型,用指数函数、幂函数以及常数加幂函数3种形式的残差方差模型消除异方差,最后对混合模型和传统回归模型进行比较及评价。结果表明: 逻辑斯蒂形式的冠幅直径模型[模型(13)]拟合效果较好,选择为基础模型; 胸径、冠底高、树高和样地优势高是影响冠幅的主要因子; 幂函数消除异方差效果最好; 与立地指数相比,立地指数与样地的嵌套效应对冠幅影响更大; 模型(15)的嵌套2水平比总体平均水平和立地指数水平预测精度高,相比于模型(13)有明显改进。本文主要为方法研究,对于其他树种可以用相似方法构建冠幅模型。

关键词: 杉木, 单木冠幅模型, 嵌套2水平非线性混合模型, 异方差

Abstract:

An individual crown diameter model was developed based on a data set consisting of 2 461 Cunninghamia lanceolata from 103 plots located in Huangfengqiao state-owned forest farm in Hunan Province. Because of the problem of high correlation among observations taken from the same sample plot located in different levels of site index, the random effects of site index and sample plots to crown diameter were considered, namely, developing nested two levels nonlinear mixed effects canopy model. Base model that used to develop mixed model was selected from 12 commonly used models in forest. In addition to diameter at breast height, the effects of other 12 stand or trees factors to crown diameter were considered. The best random effects combination for formal parameters was determined by indexes of AIC (akaike information criterion) and logarithm likelihood. Three residual variance models of Exponential function, power function and constant plus function were used to decline the heteroskedasticity. Mixed effects model and traditional regression model were compared and evaluated finally. Results showed that crown diameter-diameter model of logistic formal had a better fit effect and was selected as base model; diameter at breast height, under branch height, height and dominant height of plot were significant factors in crown diameter model; power function had a better ability to decline the heteroskedasticity; comparing to site index, the nested effects of site index and plot played more important role in crown diameter model; the prediction efficiency of the nested two level of model (15) was higher than site index level and population average level, and obviously improved comparing with model (13). This article was mainly emphasize method research, it can be used similar method that proposed in this study to built canopy with model for other tree species.

Key words: Cunninghamia lanceolata, individual crown diameter model, nested two level nonlinear mixed effects model, heteroskedasticity

中图分类号: