基于地理权重回归的长白山区云冷杉二元材积式构建  

Application of Geographically Weighted Regression to Establishment of Standard Volume Model of a Spruce-fir Mixed Stand in Changbai Mountain

作　　者：孔雷[1] 杨华[1] 亢新刚[1] 高延[2] 冯启祥[2] 王卓辉 

机构地区：[1]省部共建森林培育与保护教育部重点实验室(北京林业大学),北京100083 [2]吉林汪清林业局

出　　处：《东北林业大学学报》2011年第7期38-41,共4页Journal of Northeast Forestry University

基　　金：林业公益性行业科研专项(20080427)

摘　　要：以采用长白山北麓皆伐样地的59株样木为研究对象,利用地理权重回归模型(GWR)和最小二乘法(OLS)比较并构建二元材积公式。从数据的空间分析角度,研究了模型的胸径、树高与材积的关系。结果显示:GWR模型的拟合优度、预测能力都要高于OLS模型。GWR模型的变量参数可以反映出在样地内的空间分布规律及其稳定性,从而进一步揭示林木之间的竞争关系。GWR反映局部信息的能力为OLS所不及。优选模型的赤池信息量(AIC)最小。优选模型的稳定性、拟合优度、预测能力和残差的结果也均最优。A standard volume model for 59 sample trees in a clear cutting plot in the north of Changbai Mountains was established by applying Geographical Weighted Regression （GWR） methc, d and Ordinary Least Square（OLS） method. The relationships between diameter at breast height, tree height and individual volume were analyzed from the perspective of spatial analysis. Results show that the GWR model is superior to the OLS model in goodness of fit and prediction ability. The GWR model parameters could reflect the spatial distribution rule of trees in the sample plot and their stability, which fnrther reveal the competitive relationship between trees. With the spatial analysis of data, the GWR method has the potential to reveal the local patterns in the spatial distribution of a parameter, which would be ignored by the OLS approach. The optimal model with the smallest Akaike information criterion（AIC） was chosen. And the residual, stability, goodness of fit and prediction ability of the optimal model are ＇also the best.

关 键 词：地理权重回归模型 最小二乘法 二元材积 空间分析 

分 类 号：S757.2[农业科学—森林经理学]