Association analysis between air pollutants and meteorological factors based on multiplex visibility graphs: A case study of PM10 in Hong Kong

Revealing the associations between pollutants and meteorological factors is important for understanding air pollution processes. For PM10 and meteorological factors in Hong Kong, the monthly mean values of some basic measures (average degree, average edge overlap, and Pearson’s correlation coefficient of degree sequences) of multiplex visibility graphs (MVGs) constructed from their monthly time series can effectively capture their seasonal variations and associations. The average degrees of VGs can distinguish the PM10 in urban and rural areas. The metric results of two-layer MVGs transformed from PM10 and individual meteorological factor disclose the strongest negative association with temperature, compared with pressure, wind speed, and relative humidity, and the most pronounced seasonal characteristics of the influence of pressure with its positive association with PM10. The average edge overlaps of three-layer MVGs mapped from PM10 and two meteorological factors show weaker (stronger) combined effect on PM10 when they possess opposite (similar) effects on PM10. Taking temperature and pressure with opposite effects on PM10 at Yuen Long as an example, the differences between the maximum and minimum monthly mean values of the average edge overlaps of their two-layer MVGs and three-layer MVGs are 0.0450, 0.0663, and 0.0220, respectively. This means that the combined effect of temperature and pressure on PM10 at Yuen Long is weaker than the individual effect of either one. Our findings suggest that MVGs can provide meaningful insights into the associations between pollutants and meteorological factors, enabling the local government to formulate appropriate pollution regulatory policies by considering the current meteorological data.