南京马尾松根径与胸径多元混合效应模型研建

摘要

In order to describe the relation between ground diameters and diameters at breast height (DBH) ofPinus massoniana Lamb trees in Nanjing, related mathematical models were established with 11 monadic models (linear equation, power equation, hyperbolic equation and so on) and multivariate nonlinear mixed effect model based on the data of 531Pinus massoniana Lamb trees from 28 20 m×20 m square plots. The results reveal that the correlation indexes of all monadic models were more than 0.8, the fit goodness of monadic linear model was best (R2=0.919,SEE=1.8548), and the values ofTRB,E andP(TRB=-1.34%,E=1.33%, P=7.734%) were least in all monadic models. The adaptability test shows that the monadic linear model has more widespread suitability (E=1.125%,P=7.4645%). The established multivariate nonlinear mixed effect model has higher accuracy andift goodness compared to monadic linear model (R2=0.9504,SEE=1.7973,TRB=0.32%,E=0.313%,P=5.6428%). The suitability test reveals that the multivariate nonlinear mixed effect model has more widespread application (E=-0.0888%,P=5.3594%), which indicates the multivariate mixed effect model canift more precisely the relation between ground diameters (d0) and DBHs ofPinus massoniana Lamb trees in Nanjing than monadic linear model .%为了建立南京地区马尾松Pinus massoniana根径与胸径的相关关系模型，基于28块20 m×20 m的方形样地的531株马尾松根径和胸径数据，分别采用线性、乘幂、双曲线、三次曲线等11种一元模型和多元非线性混合效应模型拟合了马尾松根径和胸径的相关关系。结果表明，所有一元模型的相关指数均大于0.8，其中，一元线性模型的拟合优度最高（R2=0.919，SEE=1.8548），总相对偏差、平均相对误差和平均相对误差绝对值最小（TRB=-1.34%，E=1.33%，P=7.734%）。适应性检验结果显示，一元线性模型有较好的适用性（E=1.125%，P=7.4645%）。与一元线性模型相比，多元非线性混合效应模型有更高的拟合优度和精度（R2=0.9504， SEE=1.7973，TRB=0.32%，E=0.313%，P=5.6428%）。模型适应性检验显示，多元非线性混合效应模型有更强的适用性（E=-0.0888%，P=5.3594%），从而表明，与一元线性模型相比，多元非线性混合效应模型能更精确地拟合马尾松根径（d0）和胸径的相关关系。