基于混合模型的红松人工林枝条动态研究

Branch Dynamics for the Korean Pine Plantation Based on Linear Mixed Mode

采用孟家岗林场79株人工红松4 987个枝条的枝解析数据,分别构建了人工红松枝条基径、枝长线性混合效应模型和枝条基径、枝长生长混合效应模型。利用SAS9.22统计软件对模型参数进行求解,并通过AIC、BIC及LRT对收敛的非线性模型之间的差异性进行显著性检验。结果表明:对于基径和枝长的线性混合效应模型,模型中所有参数的t检验均显著,参数的标准误差比较小,模型的稳定性很好。对于基径生长的非线性混合效应模型,在不考虑样地效应的条件下,对于模型参数b1和b3的组合,无论是AIC还是BIC都比较小。当考虑到样地效应的过程中,同样是参数b1和b3的组合形式取得最小的AIC和BIC,因此在考虑到样地效应时,这种参数的随机效应组合形式是最优模型。对于人工红松枝条枝长生长的非线性混合效应模型,当没有考虑样地效应时,通过比较不同参数的随机组合的随机效应,可以得知,当参数b1和b3组合的过程中,无论是AIC还是BIC都比较小。当考虑到样地效应的过程中,参数b3和b4的组合形式取得最小的AIC和BIC,因此在考虑到样地效应时,这种参数的随机效应组合形式是最优模型。

With the branch analysis data of 4 987 branches for 79 Korean pine plantations in Mengjiagang Forest Farm, we established the branch diameter and branch length of the linear mixed model and the branch diameter and branch length growth of the nonlinear mixed model. We used SASg. 22 to solve parameters of the model and used AIC, BIC and LRT to give significance test for the converged model. For branch diameter and branch length linear mixed model, all the parameters were significant for t test of the branch diameter and branch length and the standard error of the parameters was low, and ＇it performed well as for the stability of the models. For branch diameter growth nonlinear mixed model, the combination for b_1 and b_3 for the model performed well based on the low value for AIC and BIC, when the sample plots effect was not considered. When the sample plots effect was considered, the combination for b_1 and b_3 for the model performed well compared to the other combination form. For branch length growth nonlinear mixed model, the parameter combination for b_1 and b_3 of the model performed well based on the AIC and BIC when the sample plot effect was not considered, and if considered, the combination for b_3 and b_4 performed best for the model.