不跨越边界基于Horvitz-Thompson估计量的分层自适应群团抽样  

Stratified Adaptive Cluster Sampling Based on the Adjusted Horvitz-Thompson Estimator with the Network Nonoverlapping Strata

作　　者：朱光玉[1,2] 唐守正[1] 雷渊才[1] 

机构地区：[1]中国林业科学研究院资源信息研究所,北京100091 [2]中南林业科技大学林业遥感信息工程研究中心,长沙410004

出　　处：《林业科学》2010年第7期1-6,共6页Scientia Silvae Sinicae

基　　金：国家自然基金项目(30510103195);科技部社会公益研究专项(2005DIB5J142);国家林业局"948"引进项目(2006-4-34);中南林业科技大学人才引进项目(104-0086)

摘　　要：提出一种新的不跨越边界的基于Horvitz-Thompson估计量的分层自适应群团抽样方法,给出其估计量公式,并证明其无偏性。将9种抽样方法(简单随机抽样、分层简单随机抽样、基于修正Hansen-Hurwitz估计量的ACS、基于修正Horvitz-Thompson估计量的ACS、跨越边界基于Hansen-Hurwitz估计量的ACSI、跨越边界基于Hansen-Hurwitz估计量的ACSII、不跨越边界基于修正Hansen-Hurwitz估计量的ACS、跨越边界基于修正Horvitz-Thompson估计量的ACS、不跨越边界基于修正Horvitz-Thompson估计量的ACS)应用于中国乌兰布和沙漠边缘植被花棒密度调查,并对比9种抽样方法精度。结果表明:不跨越边界的基于Horvitz-Thompson估计量的分层自适应群团抽样的效果最佳;分层的抽样方法比不分层的抽样方法效率要高。In this paper,a new stratified adaptive cluster sampling method is proposed based on the adjusted Horvitz-Thompson estimator with the network nonoverlapping strata. The mean and variance estimator are proved to be unbiased. Nine sampling methods were used for estimating the density of Hedysarum scoparium in Ulanbuh Desert edge in China, simple random sampling,stratified simple random sampling,adaptive cluster sampling based on the adjusted Hansen-Hurwitz estimator,adaptive cluster sampling based on the adjusted Horvitz-Thompson estimator,stratified adaptive cluster sampling I （ACSI） based on the adjusted Hansen-Hurwitz estimator with the network overlapping strata,stratified adaptive cluster sampling II （ACSII） based on the adjusted Hansen-Hurwitz estimator with the network overlapping strata,stratified adaptive cluster sampling based on the adjusted Horvitz-Thompson estimator with the network overlapping strata,stratified adaptive cluster sampling based on the adjusted Horvitz-Thompson estimator with the network nonoverlapping strata. The precisions of the nine sampling methods were compared. The result showed that the new stratified adaptive cluster sampling method based on the adjusted Horvitz-Thompson estimator with the network nonoverlapping strata was more effective than the others,and the sampling in stratified populations was more effective than the sampling in non-stratified populations.

关 键 词：花棒 自适应群团抽样 分层自适应群团抽样 Horvitz-Thompson估计量 Hansen-Hurwitz估计量 模拟抽样 乌兰布和沙漠 

分 类 号：S757.2[农业科学—森林经理学]