Abstract of PhD Thesis

Evaluation of background concentrations of air pollutants in Ireland and the development of guidelines for local assessment

Aoife Donnelly, Trinity College Dublin(2011)

The accuracy of local air quality modelling studies is significantly influenced by the background concentration adopted. The aim of this research was to produce a set of best practice guidelines for determining background concentrations and adding these to modelled output to improve overall results. An in-depth review and statistical analysis of monitored data at a range of background sites in Ireland was carried out, focusing on NO2 and PM10. Seasonal, diurnal and spatial variations were observed at all sites owing to both meteorological and anthropogenic factors. A non-parametric circular kernel regression model was developed to quantify wind speed/wind direction effects, resulting in significant improvement over data-binning, particularly where monitoring data are limited. The model is capable of separating true and spurious peaks and estimating values for missing data points. Using air mass history modelling, transboundary effects were found to contribute to seasonal variations and short-term concentration fluctuations, with European and stagnant air masses producing consistently elevated concentration levels.

The major output from the research is a set of hierarchical guidelines for the determination of background concentrations for a range of situations (including limit value comparison and environmental impact assessment), encompassing a novel method for predicting concentrations based solely on an estimate of the annual mean. Whilst being specific to rural or urban background sites, account is taken of diurnal and seasonal variations and variation with wind speed and direction. Empirically developed correction factors allow calculation of an annual mean concentration based on short-term monitoring. The recommended methods of background concentration determination and addition in local air quality modelling studies represent a large improvement over commonly applied methods.