Analysis of air quality time series of Hong Kong with graphical modeling

Identifying the interaction of air pollutants in each monitoring station and the inter-relationship between pollutants at different stations is an important issue in the management and control of air quality and pollutants. In this paper, a graphical model is utilized to analyze the air pollution in Hong Kong using the daily air pollution data from the three monitoring stations located at Tsuen Wan, Tap Mun, and Tung Chung in Hong Kong. The model follows broadly a spectral analytic method, for determining the edges of a graph. The method extends a graphical model for analyzing multivariate data to multivariate time series, and in our case, it is applied to time series over different spatial locations. We adopt the graphical model for time series to investigate the inter-relationship of air pollutants at the three stations in Hong Kong and the interaction between pollutants among the stations. The results obtained have good interpretations in terms of both geographical locations and chemistry. © 2016 John Wiley & Sons, Ltd.