Kumar Naresh
Atmos Environ (1994). 2009 Feb;43(5):1153. doi: 10.1016/j.atmosenv.2008.10.055.
This article offers an optimal spatial sampling design that captures maximum variance with the minimum sample size. The proposed sampling design addresses the weaknesses of the sampling design that Kanaroglou et al. (2005) used for identifying 100 sites for capturing population exposure to NO(2) in Toronto, Canada. Their sampling design suffers from a number of weaknesses and fails to capture the spatial variability in NO(2) effectively. The demand surface they used is spatially autocorrelated and weighted by the population size, which leads to the selection of redundant sites. The location-allocation model (LAM) available with the commercial software packages, which they used to identify their sample sites, is not designed to solve spatial sampling problems using spatially autocorrelated data. A computer application (written in C++) that utilizes spatial search algorithm was developed to implement the proposed sampling design. This design was implemented in three different urban environments - namely Cleveland, OH; Delhi, India; and Iowa City, IA - to identify optimal sample sites for monitoring airborne particulates.
本文提供了一种最优空间抽样设计，该设计能以最小样本量捕获最大方差。所提出的抽样设计解决了Kanaroglou等人（2005年）用于在加拿大多伦多确定100个捕获人群NO₂暴露位点的抽样设计的弱点。他们的抽样设计存在诸多弱点，无法有效捕获NO₂的空间变异性。他们使用的需求面在空间上具有自相关性，并按人口规模加权，这导致选择了冗余位点。他们用于确定样本位点的商业软件包中可用的位置分配模型（LAM）并非设计用于使用空间自相关数据解决空间抽样问题。开发了一个利用空间搜索算法的计算机应用程序（用C++编写）来实施所提出的抽样设计。该设计在三个不同的城市环境中实施——即俄亥俄州克利夫兰市；印度德里；以及爱荷华州艾奥瓦城——以确定监测空气中颗粒物的最优样本位点。
