欢迎访问林业科学,今天是

林业科学 ›› 2014, Vol. 50 ›› Issue (6): 34-41.doi: 10.11707/j.1001-7488.20140605

• 论文与研究报告 • 上一篇    下一篇

基于异速参数概率分布的立木地上生物量估算

黄兴召, 陈东升, 孙晓梅, 张守攻   

  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 北京 100091
  • 收稿日期:2013-07-09 修回日期:2013-09-22 出版日期:2014-06-25 发布日期:2014-07-07
  • 基金资助:

    林业公益性行业科研专项经费项目(201104027)。

Estimation of Above-Ground Tree Biomass Based on Probability Distribution of Allometric Parameters

Huang Xingzhao, Chen Dongsheng, Sun Xiaomei, Zhang Shougong   

  1. Key Laboratory of Tree Breeding and Cultivation of State Forestry Administration Research Institute of Forestry, CAF Beijing 100091
  • Received:2013-07-09 Revised:2013-09-22 Online:2014-06-25 Published:2014-07-07
  • Contact: 张守攻

摘要:

对收集的80篇文献中304个地上部分生物量(M)和胸径(D)的异速生物量模型lnM=a+blnD数据集研究发现:模型参数ab符合二元正态分布;参数ab之间、参数b和纬度间呈负相关,并依此相关关系应用联立方程组建立参数ab随纬度变化的通用模型。以实测的北亚热带高山区日本落叶松地上部分生物量数据对新建的通用模型、最小二乘法和贝叶斯方法拟合生物量的适用性进行研究, 结果表明:虽然通用模型的拟合精度最低(R2为0.892),但可以根据植物生长的纬度实现无实测样地情况下的生物量估算。在拟合样本≥50时,最小二乘法和贝叶斯方法的拟合效果无显著差异;当拟合样本<50时,贝叶斯方法的拟合效果优于最小二乘法。因此在建模样本<50时,建议使用贝叶斯方法估测立木地上部分生物量。

关键词: 异速生物量模型, 参数, 概率分布, 贝叶斯方法

Abstract:

Allometric biomass equations are widely used to predict above-ground biomass in forest ecosystems. It found the distribution of the parameters a and b of the allometry between above-ground biomass (M) and diameter at breast height(D), lnM=a + blnD, well approximated by a bivariate normal from analysis a data of 304 functions of 80 papers. ANOVA was tested to parameters in seven genera. In contrast to the parameter a, there was significant difference in parameter b. There were negative correlation between the parameter a and b, the parameter b and latitude. From this negative correlation, simultaneous-equation was used to build general model for parameters which were changed by latitude. Three methods which include established general model, minimum-least-square regression and Bayesian approach were used to fitting the above-ground biomass of Larix kaempferi in sub-tropical alpine area. The result showed that general model was the lowest precise quantifications(R2=0.892), but it could estimate the biomass where forest situated in latitude without samples. With sample size was more than 50, both Bayesian method and minimum-least-square regression was no significant difference in the mean absolute error. And it was less than 50, Bayesian method was better than minimum-least-square regression. Therefore, it was suggested that Bayesian method was used to estimate above-ground biomass when the sample size was less than 50.

Key words: allometric biomass equations, parameters, probability distribution, Bayesian method

中图分类号: