Detrended fluctuation analysis of particle number concentrations on roadsides in Hong Kong

摘要

Particulate matter at unban intersection is originated from a variety of sources and influenced by plenty of meteorological conditions. It frequently presents nonlinear patterns. To deep identify these nonlinear characteristic, statistical methods including power-law distribution, autocorrelation function and detrended fluctuation analysis are applied to investigate it. The experiment was carried out in various seasons in Hong Kong and the number concentrations of particles larger than 0.3 μm were collected. Based on the measurements database, autocorrelations were calculated for every particle size group and particles less than 5 μm show periodic variation. The obtained period is consistent with traffic signal period. Additionally, the power-law distribution for every particle size group is investigated and the results show that the concentration distributions for particle between 0.5 and 5 μm exhibit asymmetric forms while the concentration distributions for particle between 0.3 and 0.49 μm present symmetrical trend with the far tails. Furthermore, the detrended fluctuation analysis was applied to discover the long-term variation. The Hurst parameters for every particle size group were determinate by means of fractal analysis and these results provide fundamental insights into the nature of long-term correlations.