Management of air quality monitoring using principal component and cluster analysis -: Part II:: CO, NO2 and O3

The aim of this study was to evaluate the performance of two statistical methods, principal component analysis (PCA) and cluster analysis (CA), for the management of air quality monitoring network (AQMN) of Oporto Metropolitan Area (Oporto-MA). The specific objectives were: (i) to identify city areas with similar air pollution behaviours; and (ii) to locate emission sources. The statistical methods were applied to the mass concentrations of carbon monoxide (CO), nitrogen dioxide (NO2) and ozone (O-3), collected in the AQMN of Oporto-MA from January 2003 to December 2005. It was demonstrated that for each pollutant the monitoring sites are grouped into different classes based on their air pollution behaviour. The sites were divided: (i) into three different groups for CO and for NO2 and (ii) into two groups for O-3. It was also found that several monitoring sites covered city areas characterized by the same specific air pollution behaviour, suggesting then an ineffective management of the air quality-monitoring system. The redundant equipment should be transferred to other monitoring sites allowing enlargement of the monitored area. The conclusions obtained with the statistical methods were supported by the location of main emission sources through the analysis of the wind direction. Four main emission sources of CO and NO2 were located. Additionally, it was concluded that the sea wind had an important contribution towards the increase in the O-3 concentration. For all pollutants, two sites were always coupled in one group due to the different air pollution behaviour presented in the analysed period. (C) 2007 Elsevier Ltd. All rights reserved.