Structural Characteristics of Particulate Matter Time Series Observed in an Urban Environment

In this study, we used the fractal and multifractal analysis to explore the structural characteristics of РМ10 time series, among which self-similarity and invariance can be considered particularly important. The eXtreme Gradient Boosting method was used to fill in the missing data for multiscale multifractal analysis. The analysis has revealed self-similarity in РМ10 time series with a positively correlated structure which was stable over a study period. Small fluctuations of РМ10 levels were observed as a result of variations in local emissions and meteorological conditions. The uncoordinated and uncorrelated intervals in concentration time series were observed as a consequence of occasional pollution events in the areas dominated by industrial activities or as a consequence of the remote emission source activity when wind direction and speed were favorable.