Change Points Detection and Trend Analysis to Characterize Changes in Meteorologically Normalized Air Pollutant Concentrations
Identifying changes in ambient air pollution levels and establishing causation is a research area of strategic importance to assess the effectiveness of air quality interventions. A major challenge in pursuing these objectives is represented by the confounding effects of the meteorological conditions which easily mask or emphasize changes in pollutants concentrations. In this study, a methodological procedure to analyze changes in pollutants concentrations levels after accounting for changes in meteorology over time was developed. The procedure integrated several statistical tools, such as the change points detection and trend analysis that are applied to the pollutants concentrations meteorologically normalized using a machine learning model. Data of air pollutants and meteorological parameters, collected over the period 2013–2019 in a rural area affected by anthropic emissive sources, were used to test the procedure. The joint analysis of the obtained results with the available metadata allowed providing plausible explanations of the observed air pollutants behavior. Consequently, the procedure appears promising in elucidating those changes in the air pollutant levels not easily identifiable in the original data, supplying valuable information to identify an atmospheric response after an intervention or an unplanned event.