欢迎访问林业科学,今天是

林业科学 ›› 2008, Vol. 44 ›› Issue (3): 40-44.doi: 10.11707/j.1001-7488.20080311

• 论文 • 上一篇    下一篇

航空像片冠幅与地面直径的线性混合模型

郎璞玫   

  1. (中国林业科学研究院资源信息研究所 北京100091)
  • 收稿日期:2006-09-29 修回日期:1900-01-01 出版日期:2008-03-25 发布日期:2008-03-25

Linear Mixed Model of Aerial Photo Crown Width and Ground Diameter

Lang Pumei   

  1. (Institute of Forest Resources Information Techniques, CAF Beijing 100091)
  • Received:2006-09-29 Revised:1900-01-01 Online:2008-03-25 Published:2008-03-25

摘要:

大量采集航空像片冠幅x与地面树木直径y的相关资料,指出一群由树冠“亮点集"组成的航空像片图像是“冠幅"检测的必要条件,并从专业的角度论证航片冠幅x与树木直径y应满足带截距的线性相关关系。由于树冠密度的随机干扰,使得冠幅x与直径y不满足等方差条件,所以必须在原来固定参数线性模型的基础上引入随机效应参数。本文采用“样地"作为随机效应的构造变量,“树冠类型"为组变量,它们的叉积构造“随机效应"参数设计矩阵,从而构造出航空像片冠幅x与树木直径y的线性混合模型,由此获得总体y的最优无偏估计,线性混合模型的相关系数由一元线性模型的0.57平均提高到0.72。线性混合模的实质是在固定参数方程上迭加随机“噪声"。由于数据经过标准化处理,带有随机挠动的预测方程参数与航空像片比例尺无关。

关键词: 树冠冠幅, 线性混合模型, 最优无偏估计, 随机挠动的预测方程

Abstract:

The paper pointed out that crown “bright dots" constitute of image were prerequisite to test crown width, supported by plentyly of aerial photo crown width x and ground tree diameter y, and then demonstrated the intercept linear relation between aerial photo crown width x and ground tree diameter y . As random disturbance of tree crown density, crown width x and ground tree diameter y were dissatisfied equal variance. It was necessary to introduce random effect parameter based on original parameter linear model. The paper let “plot" as random effect structure parameter,“tree crown type" as group variable, their cross product as “random effect" parameter design matrix.Aerial photo crown width x and ground tree diameter y linear mixed model were constructed and optimization unbias estimation.Linear mixed model Rwas 0.72, and general linear model R was 0.57 The essence of linear mixed model was fixed parameters equation add random disturbance. Random disturbance forecast equation parameters with were independent on aerial photo scale as standardized data.

Key words: crown width, linear mixed model, optimization unbias estimation, random disturbance estimation equation