欢迎访问《中国农学通报》,

中国农学通报 ›› 2021, Vol. 37 ›› Issue (4): 31-37.doi: 10.11924/j.issn.1000-6850.casb2020-0023

• 林学·园艺·园林 • 上一篇    下一篇

基于混合效应的湖北杉木冠长率模型

袁慧1(), 杜超群1, 赵小涛2, 肖志明3, 赵虎1()   

  1. 1湖北省林业科学研究院,武汉 430075
    2阳新县自然资源和规划局,湖北阳新 435200
    3阳新县自然资源和规划局龙港自然资源所,湖北阳新 435200
  • 收稿日期:2020-04-21 修回日期:2020-10-21 出版日期:2021-02-05 发布日期:2021-01-25
  • 通讯作者: 赵虎
  • 作者简介:袁慧,男,1989年出生,内蒙古赤峰人,助理研究员,硕士,研究方向:森林培育。通信地址:430075 湖北省武汉市东湖新技术开发区森林大道枫林路39号 湖北省林业科学研究院,Tel:027-86952156,E-mail: 953836298@qq.com
  • 基金资助:
    “十三五”国家重点研发计划任务“湖北杉木速生材高效培育技术研究”(2016YFD0600302-4)

A Crown Ratio Model for Cunninghamia lanceolata Plantations Based on Mixed Effect in Hubei

Yuan Hui1(), Du Chaoqun1, Zhao Xiaotao2, Xiao Zhiming3, Zhao Hu1()   

  1. 1Hubei Academy of Forestry, Wuhan 430075
    2Natural Resources and Planning Bureau of Yangxin County, Yangxin Hubei 435200
    3Longgang Natural Resources Office of Yangxin County Bureau of Natural Resources and Planning, Yangxin Hubei 435200
  • Received:2020-04-21 Revised:2020-10-21 Online:2021-02-05 Published:2021-01-25
  • Contact: Zhao Hu

摘要:

为研究冠长率与林分因子之间的生长关系,分析不同因素对冠长率的影响,对湖北全省杉木栽植区域进行标准地调查,利用181块调查样地的715株优势木和平均木数据建立冠长率模型。从树木及立地因子中选出与冠长率相关性显著的树高、胸径、高径比、每公顷株数和大于对象木断面积5个因子作为变量,利用Logistics方程构建冠长率基础模型,并以此建立关于样地的单水平混合效应模型。混合模型所有参数及其组合形式均作为随机效应进行拟合,通过比较模型AICBIC和似然比显著性检验,与无随机参数模型相比,含有随机参数af的混合模型拟合效果最优。进行独立样本检验,混合模型的均方根误差和平均误差均有所降低,分别为-0.0095、0.1175,且残差范围变小。与基础模型相比,混合模型的R2由0.3231提高到0.5417,RMSEBias均有所减小。基于模型的拟合优度及检验指标,混合模型能够显著提高模型的拟合精度,可以用于树冠长度的预测。

关键词: 冠长率, 杉木, Logistic方程, 混合效应, 模型

Abstract:

The paper aims to study the relationship between crown ratio and stand factors, and analyze the effects of different factors on crown ratio. 715 dominant trees and average trees in 181 sample plots were established in Chinese fir planting areas in Hubei Province. Tree height, diameter at breast height, height to diameter ratio, stand density and basal area per hectare for trees larger than the subject tree closely associated with crown ratio were selected among tree and site factors. The basic model of crown ratio was constructed by adding correlation factors as variables in the form of logistic equation, and the single-level mixed effect model of sample plot was further established. By comparing the AIC, BIC and likelihood ratio test, the mixed model with random parameters a and f has the best fitting effect compared with the model without random parameters. The results of independent sample test show that the root mean square errors and average error of the mixed model are reduced, respectively -0.0095 and 0.1175, and the residual range is smaller. Compared with the basic model, R2 of the mixed model increased from 0.3231 to 0.5417, but RMSE and Bias decreased. Based on the goodness of fit and test index of the model, the mixed model can significantly improve the fitting accuracy of the model, and can be used to predict the crown length.

Key words: crown ratio, Cunninghamia lanceolata, logistic equation, mixed effect, model

中图分类号: