Visibility graphs of ground-level ozone time series: A multifractal analysis

R. Carmona-Cabezas,A.B. Ariza-Villaverde,E. Gutierrez de Rave,F.J. Jimenez-Hornero

DOI: https://doi.org/10.1016/j.scitotenv.2019.01.147

2023-11-17

Abstract:A recent method based on the concurrence of complex networks and multifractal analyses is applied for the first time to explore ground-level ozone behavior. Ozone time series are converted into complex networks for their posterior analysis. The searched purpose is to check the suitability of this transformation and to see whether some features of these complex networks could constitute a preliminary analysis before the more thorough multifractal formalism. Results show effectively that the exposed transformation stores the original information about the ozone dynamics and gives meaningful knowledge about the time series. Based on these results, the multifractal analysis of the complex networks is performed. Looking at the physical meaning of the multifractal properties (such as fractal dimensions and singularity spectrum), a relationship between those and the degree distribution of the complex networks is found. In addition to all the promising results, this novel connection between time series and complex networks can deal with both stationary and non-stationary time series, overcoming one of the main limitations of multifractal analysis. Therefore, this technique can be regarded as an alternative to give supplementary information within the study of complex signals.

Atmospheric and Oceanic Physics

What problem does this paper attempt to address?

The problem this paper attempts to address is exploring the time series behavior of ground-level ozone by combining complex network and multifractal analysis methods. Specifically, the authors convert the time series of ground-level ozone into a 2D complex network (i.e., visibility graph) and perform multifractal analysis through these networks to examine whether this transformation can retain the information of the original time series and whether certain features of the network (such as degree distribution) can provide preliminary information for subsequent multifractal analysis. Additionally, this method can handle both stationary and non-stationary time series, overcoming a major limitation of traditional multifractal analysis.

The main objective of the paper is to validate the applicability of multifractal analysis on visibility graphs and to explore the relationship between the degree distribution of visibility graphs and multifractal parameters (such as generalized fractal dimensions and singularity spectra). Through this method, the researchers hope to provide a new supplementary tool for the study of complex signals.