南京林业大学学报(自然科学版) ›› 2020, Vol. 44 ›› Issue (2): 117-124.doi: 10.3969/j.issn.1000-2006.201812012

• 研究论文 • 上一篇    下一篇

华北落叶松人工林直径分布预测模型构建

周泽宇1(), 杨绕华2, 张玉珍1, 黄选瑞1, 张志东1, 王冬至1,*(), 李大勇2   

  1. 1.河北农业大学林学院,河北省林木种植资源创新与保护实验室, 河北 保定 071000
    2.河北省木兰围场国有林场管理局, 河北 承德 068450
  • 收稿日期:2018-12-06 修回日期:2019-02-27 出版日期:2020-03-30 发布日期:2020-04-01
  • 通讯作者: 王冬至
  • 基金资助:
    国家重点研发计划(2017YFD0600403);河北省教育厅资助科研项目(QN2018125);国家林业公益性行业科研专项项目(20150430304)

Prediction model construction of diameter distribution of Larix principis-rupprechtii plantation

ZHOU Zeyu1(), YANG Raohua2, ZHANG Yuzhen1, HUANG Xuanrui1, ZHANG Zhidong1, WANG Dongzhi1,*(), LI Dayong2   

  1. 1. College of Forestry Agricultural University of Hebei, Forest Resources Innovation and Protection Laboratory of Hebei, Baoding 071000, China
    2. Mulan Weichang State-owned Forest Farm Administration Bureau of Hebei Province, Chengde 068450, China
  • Received:2018-12-06 Revised:2019-02-27 Online:2020-03-30 Published:2020-04-01
  • Contact: WANG Dongzhi

摘要:

【目的】基于不同直径分布预测模型(Weibull分布模型、Gamma分布模型、Lognormal分布模型),构建包含华北落叶松林分因子的直径分布线性混合效应模型,有助于分析直径分布对林分因子动态变化的响应。【方法】利用塞罕坝华北落叶松人工林标准地调查数据,应用最大似然估计法(Maximum Likelihood Estimation, MLE)估计模型参数,通过K-S(Kolmogorov-Smirnov)检验、C-V(Cramer-von Mises)检验、A-D(Anderson-Darling)检验对模型适用性进行检验,基于最优模型构建华北落叶松人工林直径分布线性混合效应模型。【结果】塞罕坝华北落叶松人工林直径分布最优模型为Weibull分布;基于最优模型,构建了包含优势高、断面积、对数密度的线性混合效应模型,当3个参数随机效应方差-协方差结构和误差项结构均为对角矩阵结构[UN(1)]时,模型的拟合效果最好。包含位置、尺度、形状3参数随机效应项模型的决定系数R2分别为0.895、0.888、0.801,均方误差(MSE)分别为5.365、1.724、1.151,均方根误差(RMSE)分别为2.316、1.313、1.073,拟合结果均较好。【结论】线性混合效应模型具有较好的预测直径分布能力,可为精准预测华北落叶松人工林直径分布提供理论依据和技术参数。

关键词: 华北落叶松, 动态直径分布模型, 最大似然估计, 线性混合模型

Abstract:

【Objective】 Based on different diameter distribution prediction models(Weibull distribution, Gamma distribution, Lognormal distribution), linear mixed-effects model including the main stand factors are constructed to help to reflect the response of diameter distribution to stand dynamics. 【Method】 Using the survey data of typical Larix principis-rupprechtii plantation, the model parameters were estimated by using the Maximum Likelihood Estimation (MLE), and the adaptability of the established model was evaluated via K-S (Kolmogorov-Smirnov)test, C-V (Cramer-von Mises) test, and A-D (Anderson-Darling) test. Based on the optimal model, a linear mixed-effects model of diameter distribution of Larix principis-rupprechtii plantation was constructed. 【Result】 The optimal model for the diameter distribution ofLarix principis-rupprechtii plantation was the 3-parameter Weibull distribution. Based on the optimal model, a linear mixed-effect model with inputs of stand dominant height, stand basal area and logarithmic density was constructed and the mixed-effect model had the best fitting result as the variance-covariance structure of random-effect and the error term variance-covariance structure was both Diagonal matrix UN(1). The R 2 of linear mixed-effect model of Weibull distribution of parameter a, b, and c were estimated at 0.895, 0.888, and 0.801 respectively. The MSE were at 5.365, 1.724, 1.151 and RMSE were at 2.316, 1.313, 1.073. the fitting result was relatively accurate. 【Conclusion】 Linear mixed-effects model has the capability to accurately predict stand diameter distribution. The model can provide theoretical basis and technical parameters for accurately predicting the diameter distributions ofLarix principis-rupprechtii plantation.

Key words: Larix principis-rupprechtii, dynamic diameter distribution model, Maximum Likelihood Estimation(MLE), linear mixed-effect model

中图分类号: