| 摘要: |
| 【目的】通过建模来研究高山松单木木材碳密度在纵向（从树干基部到树梢）和径向（从木材髓心向外）的变化规律。【方法】对云南省香格里拉市10株高山松样木进行树干解析，测定不同树高位置和部位的高山松木材碳密度，使用单因素方差分析法分析高山松单木木材碳密度变化规律；采用非线性混合效应模型技术，以纵向的树高效应和径向的部位效应作为随机效应，构建高山松单木木材碳密度混合效应模型。【结果】高山松单木木材碳密度在纵向和径向的变异均极显著，且高山松木材碳密度呈现从树干基部到树梢逐渐增加、从木材髓心向外逐渐减小的变化趋势；与基础模型相比，将树高效应、部位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，且考虑部位随机效应的混合效应模型具有最佳的拟合表现；所有模型的预估精度均在97%以上，且考虑部位随机效应的混合效应模型和两水平混合效应模型的预估精度高于98%，尤其是两水平混合效应模型的预估精度达到98.04%。【结论】高山松单木木材碳密度随部位和树高变化差异显著，且考虑部位和树高随机效应的两水平混合效应模型在模拟高山松单木木材碳密度时具有更高的拟合精度和预估精度。 |
| 关键词: 高山松 木材碳密度 混合效应模型 |
| DOI： |
| 分类号: |
| 基金项目:国家林业公益性行业科研专项（201404309）；国家自然科学基金项目（31560209） |
|
| Wood carbon density variation of Pinus densata and establishment of mixed effects models |
|
XIONG Hexian,WEI Anchao,XU Hui,et al
|
| Abstract: |
| 【Objective】Models were established to investigate wood carbon density variation of Pinus densata at longitudinal and radial directions.【Method】Ten sampling trees of P.densata were investigated by stem analysis in Shangri-La,Yunnan,and wood carbon density at different parts and different tree heights was measured.Then,the variation of carbon density was analyzed by ANOVA,and wood carbon density models were constructed by mixed effects model technology considering random effects of height and part.【Result】The variations of wood carbon density were extremely significant among different parts and heights.It gradually increased from base to top,and decreased from pith to outward.Both single level mixed effect models considering random effect of different parts and heights and two level mixed effect models improved the fitting precision,and the mixed effect model considering the random effect of parts had the best fitting performance.The prediction precisions of all models were above 97%,and the values of single level mixed effect model considering random effect of parts and two-level mixed effect model were more than 98%.Especially,the precision of the two level mixed effect model had the highest prediction accuracy of 98.04%.【Conclusion】The variations of wood carbon density were extremely significant among different parts and heights,and the mixed effect model considering random effects of tree height and part had better fitting and predicting precision. |
| Key words: |
