Fractal behaviour of benzene concentration near refinery, traffic junctions and residential locations in India

Abstract

Due to its carcinogenicity, benzene, a hazardous chemical created both naturally and via human activity, poses a major threat to human health. Even at low concentrations, short-term exposure can cause severe health issues. The study of temporal variations in the time series of benzene concentration can be useful to policymakers in air quality assessment and management. Fractal behaviour in benzene concentration at various locations with different land-use characteristics and prevalent sources is studied using detrended fluctuation analysis (DFA) and multifractal detrended fluctuation analysis (MF-DFA). The temporal characteristics of benzene concentration have not been studied earlier and it is of interest to assess the influence of land-use characteristics and dominant sources on the temporal intrinsic properties of the time series. 24 h benzene concentration time series during 2019–2021 at 10 locations is therefore selected for the study. Preliminary analysis using autocorrelation function, power spectral analysis and DFA suggested the existence of long-range correlations in benzene time series. Multifractal analysis pointed towards the dominance of small fluctuations that frequently occur in the benzene concentration time series at most of the locations. The analysis of the origin of multifractality revealed that removing the linear temporal correlations weakens the multifractality and broad probability density function also partly explains the multifractality in benzene time series. The knowledge of the fractality of the time series is useful in developing the prediction model which can account for the linear temporal correlations along with the multifractality to predict the APCs. The results can aid in improving air quality evaluations and the perceptive of the mechanisms regulating the evolution of air pollution time series.