Source apportionment advances using polar plots of bivariate correlation and regression statistics
Source apportionment advances using polar plots of bivariate correlation and regression statistics
复制标题
DOI：
10.1016/j.atmosenv.2016.09.016
复制
发表时间：
2016-11-01
影响因子：
5
通讯作者：
Carslaw, David C.
中科院分区：
文献类型：
--
作者：
Grange, Stuart K.;Lewis, Alastair C.;Carslaw, David C.
本文概述了增强型双变量极图的发展，它允许使用成对统计来比较两种污染物的浓度，以探索大气污染物的来源。新方法将提供源特征信息的双变量极图与提供两种污染物如何相互关联的信息的成对统计相结合。实施的配对统计方法包括加权皮尔逊相关和两种线性回归方法的斜率。该发展使用高斯核来局部加权风速-方向表面上的统计计算，并结合可变比例。通过使用联合王国伦敦两个监测点一年(2013年)的常规空气质量数据，展示了增强极地图的应用实例。伦敦的例子表明，双变量极图、相关和回归技术的组合可以提供对空气污染源特征的相当深入的了解，如果只使用散点图和平均极图进行分析，将会遗漏这些特征。具体地说，使用相关性和斜率作为成对统计，分离了长距离输送过程，并量化了路边监测点的黑碳(BC)对PM2.5的贡献。并对其更广泛的应用和未来的发展进行了讨论。(C)爱思唯尔有限公司出版的2016年。
This paper outlines the development of enhanced bivariate polar plots that allow the concentrations of two pollutants to be compared using pair-wise statistics for exploring the sources of atmospheric pollutants. The new method combines bivariate polar plots, which provide source characteristic information, with pair-wise statistics that provide information on how two pollutants are related to one another. The pair-wise statistics implemented include weighted Pearson correlation and slope from two linear regression methods. The development uses a Gaussian kernel to locally weight the statistical calculations on a wind speed-direction surface together with variable-scaling. Example applications of the enhanced polar plots are presented by using routine air quality data for two monitoring sites in London, United Kingdom for a single year (2013). The London examples demonstrate that the combination of bivariate polar plots, correlation, and regression techniques can offer considerable insight into air pollution source characteristics, which would be missed if only scatter plots and mean polar plots were used for analysis. Specifically, using correlation and slopes as pair-wise statistics, long-range transport processes were isolated and black carbon (BC) contributions to PM2.5 for a kerbside monitoring location were quantified. Wider applications and future advancements are also discussed. (C) 2016 Published by Elsevier Ltd.