欢迎访问林业科学,今天是

林业科学 ›› 2011, Vol. 47 ›› Issue (11): 106-113.doi: 10.11707/j.1001-7488.20111117

• 论文 • 上一篇    下一篇

立木生物量方程的优度评价和精度分析

曾伟生, 唐守正   

  1. 中国林业科学研究院资源信息研究所 北京100091
  • 收稿日期:2010-04-26 修回日期:2010-06-09 出版日期:2011-11-25 发布日期:2011-11-25
  • 通讯作者: 唐守正

Goodness Evaluation and Precision Analysis of Tree Biomass Equations

Zeng Weisheng, Tang Shouzheng   

  1. Research Institute of Forest Resources Information Techniques, CAF Beijing 100091
  • Received:2010-04-26 Revised:2010-06-09 Online:2011-11-25 Published:2011-11-25

摘要:

以建立东北落叶松和南方马尾松立木生物量模型为例,对模型的优度评价和精度分析进行专题研究。首先,在综合分析各种常用模型评价指标的基础上,提出6项基本评价指标,并通过对2个树种的生物量模型进行评价,证明是可行和有效的; 然后,基于线性回归估计的基本假设及其置信区间计算公式,提出适合立木生物量模型预估的条件均值和单一预估值置信区间的估计方法,并确定2个树种生物量模型的置信区间; 最后,通过蒙特卡罗模拟进行随机再抽样试验,结果表明采用检验样本进行适用性检验的做法不可取,应该利用全部样本(不分建模样本和检验样本)来建立模型,它能充分利用样本信息,使模型的预估误差最小。

关键词: 立木生物量, 相对生长方程, 拟合优度, 模型评价, 蒙特卡罗模拟, 预估精度, 置信区间

Abstract:

Taking the establishment of single-tree biomass equations for larch (Larix) of the northeast and Masson Pine (Pinus massoniana) of the south in China as the example, the goodness evaluation and precision analysis were studied in this paper. Firstly, based on the comprehensive analysis of various statistical indices for model evaluation, six essential evaluation indices were presented which proved to be feasible and effective through evaluating the tree biomass models of two species above. Secondly, according to basic assumptions of linear regression estimation and related formulas of confidence intervals for prediction, the approach to estimate confidence intervals of conditional mean and single predicted value of tree biomass models was provided, and the confidence intervals of tree biomass models of two species were determined. Finally, the Monte Carlo simulation was used for random resampling test, and the test results showed that examination of applicability based on validation samples was inadvisable, and the adequate procedure was to utilize all sample trees for constructing the model, not splitting into two groups for model construction and model validation, which could make full use of the information of entire sample and decreased the prediction error to the highest degree.

Key words: single-tree biomass, allometric equation, goodness-of-fit, model evaluation, Monte Carlo simulation, prediction precision, confidence interval

中图分类号: