详细信息
基于线性混合模型的落叶松人工林节子大小预测模型 被引量:16
Models to Predict Knot Size for Larch Plantation Using Linear Mixed Model
文献类型:期刊文献
中文题名:基于线性混合模型的落叶松人工林节子大小预测模型
英文题名:Models to Predict Knot Size for Larch Plantation Using Linear Mixed Model
作者:陈东升[1] 李凤日[2] 孙晓梅[1] 贾炜玮[2]
第一作者:陈东升
机构:[1]中国林业科学研究院林业研究所国家林业局林木培育重点实验室;[2]东北林业大学林学院
年份:2011
卷号:47
期号:11
起止页码:121-128
中文期刊名:林业科学
外文期刊名:Scientia Silvae Sinicae
收录:CSTPCD;;Scopus;北大核心:【北大核心2008】;CSCD:【CSCD2011_2012】;
基金:林业公益性行业科研专项经费项目(201004026;201104027);中央高校基本科研业务经费专项资助项目(DL09BA10;DL09CA11)
语种:中文
中文关键词:落叶松人工林;节子直径;节子角度;节子长度;混合效应模型
外文关键词:larch plantation; knot diameter; knot angle; knot length; mixed effect model
分类号:S758.5
摘要:以黑龙江省东北部落叶松人工林为研究对象,应用19块标准地中95株解析木(每块标准地选取5株)的节子剖析数据,采用线性混合效应模型理论建立落叶松节子各因子(节子直径、着生角度、长度)的预估模型。结果表明:节子直径、着生角度和长度都随着树木胸径的增大而增大;节子直径随着着生高度的增大先增大后减小;节子角度随着着生高度的增大而逐渐减小;节子长度随着直径的增大而增大。与固定效应模型相比,考虑混合效应所建立的节子大小预测模型其参数估计更为准确,残差分布更加均匀,模型拟合精度明显得到提高(R2约提高了0.3)。独立样本的检验结果表明:各模型的预估精度均在90%以上,说明所建模型可以很好地描述落叶松人工林节子变化规律。
Using knot analysis data of 95 sample trees in 19 sample plots of larch(Larix spp.) plantation in northeast Heilongjiang Province(5 sample trees were selected in each sample plot), the models for predicting knot variables, including knot diameter, angle and length, were developed based on the linear mixed effect model. The results showed that the knot diameter, angle and length increased as DBH increasing. When knot height increased, knot diameter increased firstly, and then decreased, and knot angle decreased gradually. The knot length increased with knot diameter increasing. Comparing with fixed effects model, the mixed effect based knot size models can obtain more accurate parameter estimates,the residual distribution of the model was more uniform, and the fitting precision of models were obviously improved(the R2 value improved about 0.3).The results of independent validation showed that prediction precision of each model reached above 90%. Therefore, knot models developed in this paper can suitably describe the knots variation for larch plantation.
参考文献:
正在载入数据...