Determination of background concentrations for air quality models using spectral analysis and filtering of monitoring data

The use of background concentrations in air pollution modelling is usually a critical issue and a source of errors. The current work proposes an approach for the estimation of background concentrations using air quality measured data decomposed on baseline and short-term components. For this purpose, the spectral density was obtained for air quality monitoring data based on the Fourier series analysis. After, short-term fluctuations associated with the influence of local emissions and dispersion conditions were extracted from the original measurements using an iterative moving-average filter and taking into account the contribution of higher frequencies determined from the spectral analysis. The deterministic component obtained by the filtering is characterised by wider spatial and temporal representativeness than original monitoring data and is assumed to be appropriate for establishing the background values. This methodology was applied to define background concentrations of particulate matter (PM₁₀) used as input data for a local scale CFD model, and compared with an alternative approach using background concentrations provided by a mesoscale air quality modelling system. The study is focused on a selected domain within the Lisbon urban area (Portugal). The results present a better performance for the microscale model when initialised by decomposed time series and demonstrate the importance of the proposed methodology in reducing the uncertainty of the model predictions. The decomposition of air quality measurements and the removal of short-term fluctuations discussed in the work is a valuable technique to determine representative background concentrations.