Abstract

This paper proposes three mathematical approaches that can be used to study chemical atmospheric data related to air quality: a) Descriptive statistical analysis, b) Fourier analysis and c) Wavelet analysis. These techniques are used to improve understanding of SO2 pollution in the city of Talcahuano (Chile). The results show high SO2 dispersion values, an annual average standard deviation of 63.5 µg m-3 and a Pearson coefficient of variation of 1.59. The asymmetry and kurtosis coefficients (3.1 and 14.3, respectively) confirm the "non-Gaussianity" of the SO2 concentrations. The main fluctuations of the data studied are due to seasonal and daily cyclical components. Important annual irregularities are detected by the wavelet analysis of daily cycles, especially in winter. This cyclical component is fundamentally due to meteorological factors (because the zonal industrial emissions, mainly emitted by Oil Refining Plant, are roughly constant). Non-negligible correlations, between SO2 concentration and several meteorological variables (vertical wind, temperature, solar radiation, horizontal wind), were also verified (0.49, 0.40, 0.36 and 0.34, respectively). © 2009 Elsevier Ltd. All rights reserved.