Sulphur dioxide in Europe: Statistical relationships between emissions and measured concentrations

Air pollution in Europe is the topic of a wide range of studies involving different disciplines. However most of the analyses presented in the literature focus on the study of monitored concentrations with few articles investigating the relationship between trends in concentrations and reported emissions. In this paper a statistical analysis of sulphur dioxide (SO2SO2) concentrations monitored in Europe from 1990 to 2001 uses additive models to test the relationships with emissions from the same and neighbouring countries. This nonparametric approach allows data to determine the shape of trends with time, correctly models the serial correlation and dependence on covariates, and provides model selection procedures and valid statistical inference. The analysis showed that own country emissions adequately represented the trend over time at all but a few of the 112 sites across Europe, but there was clear statistical evidence that change in concentration was not proportionally related to changes in emissions at almost all sites. For 11 sites with collocated meteorological data, covariates were constructed for neighbouring countries’ emissions accounting for distance, wind direction and wind speed. Even with this simple approach, the statistically significant effect of neighbouring countries’ emissions on monitored concentrations was shown for 6 out of the 11 sites. This confirms the benefit of access to meteorological data within these analyses and the possible improvements if emissions data were available on a finer spatial and temporal scale. The successful application of these statistical methods in this case study extends the statistical toolbox available for air pollution studies.