摘要
【目的】分析林木竞争与马尾松冠幅的相关关系,构建含竞争指数的马尾松冠幅广义模型、广义可加模型和非线性混合效应模型,对比分析不同模型的拟合效果,选出最优的马尾松冠幅预测模型,为更全面地表达马尾松树冠特征以及科学经营和管理马尾松人工林提供理论依据。【方法】以贵州省开阳县楠木渡镇的2259株马尾松为研究对象,利用八邻域平移法消除边缘效应,采用Voronoi图确定竞争单元。采用Pearson相关性分析法,结合Hegyi竞争指数(CI1)、以大小比数为权重的交角竞争指数(CI2)、综合竞争指数(CI3),分析竞争与马尾松冠幅的相关关系。对冠幅基础模型进行扩展,构建含上述3种竞争指数的广义模型、广义可加模型和非线性混合效应模型,对比模型的拟合精度,分析竞争指数对模型的影响,同时采用十折交叉验证对模型的预测能力进行检验。【结果】1)3种竞争指数均与马尾松冠幅呈极显著的负相关关系(P<0.001);3种竞争指数之间都呈极显著的正相关关系(P<0.001),其中CI1与CI3的相关性更强。2)在基础模型的拟合中,Logistic模型拟合精度最高(R2为0.6105,RMSE为0.7429 m,MAE为0.5851 m,AIC为5075.895),选定其为最优基础模型。3)不管是加入竞争指数之前还是之后,模型的拟合精度从高到低都依次为非线性混合效应模型、广义可加模型、广义模型。4)添加竞争指数能提高3种模型的精度,广义模型和广义可加模型中都是添加CI3时模型效果最佳,非线性混合效应模型中则是添加CI1时模型效果最佳且优于其他模型。【结论】竞争效应对冠幅的生长具有明显的抑制作用。协变量、随机效应以及竞争指数的加入能提高马尾松冠幅模型的预测精度,但不同竞争指数在不同模型中表现能力不同。包含CI1的非线性混合效应模型相较于其余3种模型具有更好的预测能力,是最优的马尾松冠幅预测模型。
【Objective】To analyze the correlation between forest competition and crown width of Pinus massoniana,we constructed a generalized crown width model,generalized additive model and nonlinear mixed effect model with competition index to compare and analyze the fitting effects of different models,and select the optimal crown width prediction model of Pinus massoniana.It provides a theoretical basis for more fully expressing the crown characteristics of Pinus massoniana and scientific management and management of Pinus massoniana plantation.【Method】In this study,2259 Pinus massoniana in Nanmudu Town,Kaiyang County,Guizhou Province were selected as the research object.The edge effect was eliminated by eight-neighborhood translational method,and the competing units were determined by Voronoi diagram.The correlation between competition and crown width of Pinus massoniana was analyzed by Pearson correlation analysis,combined with Hegyi competition index(CI1),cross Angle competition index(CI2)and comprehensive competition index(CI3).The basic crown model was extended,and a generalized model,a generalized additive model and a nonlinear mixed effect model with the above three competition indices were constructed.The fitting accuracy of the models was compared,and the influence of the competition on the model was analyzed.At the same time,the prediction ability of the model was tested by 10-fold cross-validation.【Result】1)The three competition indexes are negatively correlated with the crown width of Pinus massoniana.(P<0.001).There is a very significant positive correlation between the three competition indexes(P<0.001),and the correlation between(CI1)and(CI3)was stronger;2)In the fitting of basic models,the logistic model has the highest fitting accuracy(R2=0.6105,RMSE=0.7429 m,MAE=0.5851 m,AIC=5075.895),and is selected as the optimal basic model;3)No matter before or after adding the competition index,the fitting accuracy of the model is nonlinear mixed effect model,generalized additive model and generalized model in order from high to low;4)The addition of competition index can improve the accuracy of the three models.In both the generalized model and the generalized additive model,the model effect is the best when CI3 is added,and the model effect is the best when CI1 is added to the mixed effect model and is better than other models.【Conclusion】There is a significant negative correlation between the three competition indexes and the crown width of Pinus massoniana,indicating that competition has a significant inhibitory effect on the growth of crown width.The addition of covariates and random effects can greatly improve the prediction accuracy of the canopy width model.Adding competition index can improve the prediction accuracy of the model,but different competition index has different performance ability in different models.Compared with the basic model,the generalized model and the generalized additive model,the nonlinear mixed effect model constructed in this study has higher fitting accuracy,and is the best model for predicting the crown width of Pinus massoniana.
作者
谭茜
谭伟
田昕
梅本清
樊隆高
TAN Xi;TAN Wei;TIAN Xin;MEI Benqing;FAN Longgao(College of Forestry,Guizhou University,Guiyang 550025,Guizhou,China;Forestry Information Engineering Research Center,Guizhou University,Guiyang 550025,Guizhou,China;Institute of Forest Resource Information Techniques,Chinese Academy of Forestry,Beijing 100091,China)
出处
《中南林业科技大学学报》
2025年第9期71-81,105,共12页
Journal of Central South University of Forestry & Technology
基金
国家重点研发计划项目(2023YFD2201705)。
关键词
马尾松
冠幅模型
竞争指数
非线性混合效应模型
广义可加模型
Pinus massoniana
crown width model
competition index
nonlinear mixed effect model
generalized additive model
作者简介
第一作者:谭茜(2474099797@qq.com),硕士研究生;通信作者:谭伟(wtan@gzu.edu.cn),教授,博士,博士研究生导师。
