基于树干解析的高山松天然林单木木材生物量生长模型

以云南省香格里拉市两块典型样地内的10株高山松样木为研究对象,基于树干解析测定和计算其单木木材生物量生长及木材生物量生长率,采用非线性混合效应模型技术,分别考虑了样地效应和样木效应,将所有不同随机参数组合的模型进行拟合并分析模型的方差和协方差结构,构建其生物量生长及生物量生长率混合效应模型。结果表明：考虑样地效应、样木效应作为随机效应的单水平混合效应模型和两水平混合效应模型均提高了模型的拟合精度,其中考虑两水平随机效应的混合效应模型具有最佳的拟合表现,具有最低的AIC和BIC值。考虑两水平混合效应在生物量生长量及生物量生长率模型构建中预估精度最高,分别达93.05%和89.83%;考虑样木效应...

Full description

Saved in:
Bibliographic Details
Published in广东农业科学 Vol. 44; no. 1; pp. 66 - 75
Main Author 魏安超 熊河先 胥辉 李超 闾妍宇 张博 冷燕 欧光龙
Format Journal Article
LanguageChinese
Published 西南林业大学西南地区生物多样性保育国家林业局重点实验室,云南昆明,650224 2017
Subjects
木材生物量生长
树干解析
生物量生长率
非线性混合效应模型
高山松
生物量生长率
非线性混合效应模型
高山松
木材生物量生长
树干解析
Online AccessGet full text
ISSN1004-874X
DOI10.16768/j.issn.1004-874X.2017.01.010

Cover

More Information
Summary:以云南省香格里拉市两块典型样地内的10株高山松样木为研究对象,基于树干解析测定和计算其单木木材生物量生长及木材生物量生长率,采用非线性混合效应模型技术,分别考虑了样地效应和样木效应,将所有不同随机参数组合的模型进行拟合并分析模型的方差和协方差结构,构建其生物量生长及生物量生长率混合效应模型。结果表明：考虑样地效应、样木效应作为随机效应的单水平混合效应模型和两水平混合效应模型均提高了模型的拟合精度,其中考虑两水平随机效应的混合效应模型具有最佳的拟合表现,具有最低的AIC和BIC值。考虑两水平混合效应在生物量生长量及生物量生长率模型构建中预估精度最高,分别达93.05%和89.83%;考虑样木效应的混合效应模型次之,分别为88.34%和88.74%;考虑样地效应的混合模型预估精度均最低,分别为83.99%和67.27%;而一般回归模型的预估精度仅87.00%和87.11%。
Bibliography:44-1267/S
Taking 10 Pinus densata sampling trees at two plots located in Shangri-La city of Yunnan province as the research object, we measured and calculated single wood biomass growth and wood biomass growth rates based on stem analysis. Considering random effect of the plot effect and tree effect, the biomass growth and growth rate models were constructed by nonlinear mixed effect model technology, and all the different random parameter combinations were fitted and the variance and covariance structures of the models were analyzed. The results showed that, considering random effect of plot effect and tree effect model as the single-level mixed effect model and two-level mixed effect model, the fitting precision of the models was improved, especially two-level mixed effect model had the best fitting performance with the lowest values for AIC and BIC. Both biomass growth and growth rates two-level mixed effect models had the highest prediction accuracy, and the values reached 93.05% and 89.83%; the second- be
ISSN:1004-874X
DOI:10.16768/j.issn.1004-874X.2017.01.010