建立湖南楠木次生林断面积生长预测模型,为林分的生长预测和经营决策提供理论依据和科学指导。以湖南省1989—2014年6期森林资源连续清查样地中的楠木次生林为对象,以5个有生物学意义的理论模型构建断面积生长基础模型;在此基础上,加入与林分生长密切相关的立地指数,构建立地指数模型;为提高模型预测精度,加入含林分密度的随机参数,构建湖南楠木次生林断面积生长混合效应模型。选择5个有生物学意义的理论模型中确定系数(R²=0.2410)和预估精度(P=98.8213%)最高、残差平方和最小(SSE=2986.1492)的Logistic生长模型作为基础模型;在参数b₁处加入立地指数,构建了湖南楠木次生林立地指数模型,其确定系数提升到0.331 2;以林分密度划分密度等级,构建以林分密度为随机效应的混合效应模型,结果显示,在参数b₄处加入随机参数的模拟,其AIC和BIC值最小,拟合效果显著优于其他模拟(ρ<0.0001);相比基础模型,混合效应模型的ME,MAE,RME,RMAE值均明显降低,确定系数提升到0.946 2,预测精度有所提升。所构建的混合效应模型能够提升林分断面积生长的预估精度,消除林分不同密度等级间的差异,为湖南楠木次生林的合理经营提供了科学依据。

Establishing basal area growth model of Phoebe zhennan secondary forest in Hunan province to provide theoretical basis and scientific guidance for its growth prediction and management decision-making.Taking Phoebe zhennan secondary forest in the National Forest Inventory sample of Hunan province from 1989 to 2014 as the object,the basic model of basal area growth was constructed based on 5 theoretical models with biological significance.On this basis,the site index closely related to stand growth is added to construct the site index model.In order to improve the prediction accuracy of the model,the mixed effect model of the growth of Phoebe zhennan secondary forest in Hunan area is constructed by adding the random parameters including stand density.The Logistic growth model with the highest coefficient (R²=0.241 0) and prediction accuracy (P=98.821 3%) and the smallest sum of residual squares (SSE=2 986.149 2) was selected as the basic model;the site index was added to parameter b₁,and the site index model of Hunan Phoebe zhennan secondary forest was constructed,the coefficient of determination was increased to 0.3312.The mixed effect model was constructed by dividing the stand density into different grades,the results showed that the AIC and BIC values of the simulation with random parameters added to parameter b₄ were the lowest,and the fitting effect was significantly better than other simulation (p< 0.000 1);compared with the basic model,the ME,MAE,RME and RMAE values of the mixed effect model were significantly reduced,and the R² was increased to 0.946 2,and the prediction accuracy was improved.The mixed effect model can improve the prediction accuracy of stand area growth,eliminate the differences between different density grades,and provide scientific basis for the reasonable management of Phoebe zhennan secondary forest in Hunan province.