Asymmetric multiscale behavior in PM2.5 time series: Based on asymmetric MS-DFA

Abstract

Abstract Particulate matter with an aerodynamic diameter of 2.5 mm or less (PM2.5) is one of the most serious air pollution, considered most harmful for people by World Health Organisation. In this paper, we utilized the asymmetric multiscale detrended fluctuation analysis (A-MSDFA) method to explore the existence of asymmetric correlation properties for PM2.5 daily average concentration in two USA cities (Fresno and Los Angeles) and two Chinese cities (Hong Kong and Shanghai), and to assess the properties of these asymmetric correlations. The results show the existences of asymmetric correlations, and the degree of asymmetric for two USA cities is stronger than that of two Chinese cities. Further, most of the local exponent β ( n ) are smaller than 0.5, which indicates the existence of anti-persistent long-range correlation for PM2.5 time series in four cities. In addition, we reanalyze the asymmetric correlation by the A-MSDFA method with secant rolling windows of different sizes, which can investigate dynamic changes in the multiscale correlation for PM2.5 time series with changing window size. Whatever window sizes, the correlations are asymmetric and display smaller asymmetries at small scales and larger asymmetries at large scales. Moreover, the asymmetries become increasingly weaker with the increase of window sizes.