Spectral analysis of vehicle pollutants at traffic intersection in Hong Kong.

This paper reports an investigation on the periodic variation of pollutant levels at a typical traffic intersection of Hong Kong. Carbon monoxide, carbon dioxide and particulate matters (PM) were measured respectively and the measured data show periodic variations with the traffic signal intervals. The power spectral density (PSD) approach was used to inspect the trends and periodic oscillations of measured pollutants. Singular spectrum analysis was applied to decompose the measured data into statistically significant non-linear trends and oscillations in the process. From the results, most of the trends tend to increase due to the upcoming rush hour during the experiment. In addition, all the oscillations changed regularly with a period of 136 s, which is coincident with the traffic signal period and the frequency calculated by using PSD. The trends, together with the oscillations, collectively explain the most percentage of the variability of the data in the time series and provide the principal components of the data in understanding the periodic variation of the pollutant concentration. It can be deduced that vehicle emission is the major contributor to the air pollution in downtown area and pedestrians should be more alerted when crossing the busy traffic intersections. [ABSTRACT FROM AUTHOR]