文献类型：期刊文献

中文题名：基于混合模型的落叶松树高生长模型

英文题名：Predicting Models of Tree Height Growth for Larch Based on Mixed Model

作者：陈东升[1] 孙晓梅[1] 李凤日[2]

第一作者：陈东升

机构：[1]国家林业局林木培育重点实验室(中国林业科学研究院林业研究所);[2]东北林业大学

年份：2013

卷号：41

期号：10

起止页码：60-64

中文期刊名：东北林业大学学报

外文期刊名：Journal of Northeast Forestry University

收录：CSTPCD;;北大核心:【北大核心2011】；CSCD:【CSCD2013_2014】；

基金：林业公益性行业科研专项(201104027)

语种：中文

中文关键词：落叶松人工林;种和杂种;树高生长;混合模型

外文关键词：Larch plantation; Species and hybrids; Tree height growth; Mixed model

分类号：S758.52

摘要：以辽宁省大孤家林场不同种和杂种落叶松试验林为研究对象,应用理查德基础模型构建非线性混合模型,考虑落叶松种和杂种之间的差异为随机效应,采用R软件进行模拟,选择模型收敛及其AIC和BIC值最小的混合模型作为最优模型。在此基础上考虑树高生长数据的时间序列相关性,来提高拟合的精确度,并分析不同种及杂种间树高生长的差异性,以期为建立精细到种的林分生长和收获模型奠定理论基础。结果表明:考虑种间效应的混合模型模拟精度优于传统的回归模型方法,残差分布明显改善;ARMA(2,2)自回归误差结构矩阵模型在解释树高生长的时间序列相关性时不仅提高了混合模型的模拟精度,而且能够很好地表达连续观测数据间误差分布情况;因此混合模型提高了模型的精度和通用性,并且模型中每个参数都有特定的数学含义。通过参数分析得出随机因子对树高生长的影响程度,其中以日81×长白落叶松杂种树高生长表现最具优势。
An experiment was conducted to develop nonlinear mixed model based on Richard growth model for different kinds of larch species and hybrid in Dagujia Tree Farm of Liaoning Province. The variation of larch species and hybrids were con sidered as random effect, and the model performance was evaluated by utilizing information criterion statistics （AIC and BIC） with R software simulation. Considering time series correlation of tree height growth, we improved the precision of model and analyzed the variation of high growth between larch species and hybrid. The study was theoretical basis for es-tablishing the fine stand growth and harvest model. The mixed model of interspecies interaction is better than traditional re gression model and the residual distribution obviously is improved. The time series correlation of tree growth was interpre ted based on ARMA （2,2） regression error structure matrix model. It can improve the accuracy of the simulation and ex press the error distribution of continuous observation data. Therefore, the mixed model can not only improve the precision of the model, but the model parameters can have specific mathematical means to analyze the random factor for trees high growth influence, and the hybrid Larix kaempferi81xLarix gmelinii is best in tree height growth.