Multifractal characterisation of particulate matter ([formula omitted]) time series in the Caribbean basin using visibility graphs

Abstract

A good knowledge of pollutant time series behavior is fundamental to elaborate strategies and construct tools to protect human health. In Caribbean area, air quality is frequently deteriorated by the transport of African dust. In the literature, it is well known that exposure to particulate matter with an aerodynamic diameter of 10 μm or less (PM10) have many adverse health effects as respiratory and cardiovascular diseases. To our knowledge, no study has yet performed an analysis of PM10 time series using complex network framework. In this study, the so-called Visibility Graph (VG) method is used to describe PM10 dynamics in Guadeloupe archipelago with a database of 11 years. Firstly, the fractal nature of PM10 time series is highlighted using degree distribution for all data, low dust season (October to April) and high dust season (May to September). Thereafter, a profound description of PM10 time series dynamics is made using multifractal analysis through two approaches, i.e. Rényi and singularity spectra. Achieved results are consistent with PM10 behavior in the Caribbean basin. Both methods showed a higher multifractality degree during the low dust season. In addition, multifractal parameters exhibited that the low dust season has the higher recurrence and the lower uniformity degrees. Lastly, centrality measures (degree, closeness and betweenness) highlighted PM10 dynamics through the year with a decay of centrality values during the high dust season. To conclude, all these results clearly showed that VG is a robust tool to describe times series properties.