Spatial variations and development of land use regression models of PAH, EC/OC, levoglucosan and oxidative potential of PM2.5 in European study areas

It has been shown that human exposure to air pollution causes adverse health effects. The most used indicator of air quality is the concentration of particulate matter with diameters smaller than 2.5 or 10 µm (PM2.5, PM10, respectively). PM is a chemically complex mixture and it has been suggested that observed adverse health effects depend on PM chemical composition. Most studies of PM composition have investigated short term health effects. Recently, the ESCAPE project (www.escapeproject.eu) contributed knowledge of adverse health effects of long term exposure to PM elemental constituents. Little is known about health effects caused by long-term exposure to organic PM components.

The thesis contributes to characterization of air quality by monitoring and modelling of organic components and oxidative potential. Its main goals were:

- Determination of the spatial contrast of the ambient concentration of organic components (EC/OC, PAH and hopanes/steranes, levoglucosan) and oxidative potential measured with dithiothreitol assay (OP DTT) within and between ten European study areas.
- Assessment of the relationship between those components with other components measured more routinely (NO2, NOx, PM2.5 mass and absorbance).
- Development and evaluation of Land use Regression models of spatial variation of the measured air pollutants.
Substantial spatial variability was found for all measured pollutants. EC, OC and OP DTT concentrations had similar patterns in Europe as the previously reported components: NO2, PM2.5 and PM absorbance. The highest concentrations for those pollutants were found in southern Europe. ΣPAH and levoglucosan exhibited a different trend. In southern and northern Europe ΣPAH concentrations were similar with the highest concentrations in Copenhagen, Rome and Athens. Levoglucosan concentrations (marker for wood burning) did not differ significantly between the four study areas. The contribution of wood-smoke calculated based on levoglucosan measurements to PM2.5 mass was 9 to 28% in the cold season.
Concentrations of all pollutants were higher at street locations. The highest median Street to urban background ratio was found for EC (1.62). The median S/UB ratio for ΣPAH and OC was lower (1.44 and 1.32, respectively) suggesting the importance of other sources in addition to local traffic.
The correlation between PM2.5and most components was moderate on average, suggesting they provide additional information on air quality.
The highest median explained variance (R2) was found for EC – 84%. PAH and OC LUR models had moderate performance. The lowest performance was for OP DTT – 30%. Levoglucosan models had moderate performance in all four areas, but models were non-specific as we did not have predictor variables associated with wood smoke emissions.
Traffic predictors were included in most models with exception to levoglucosan model.
In conclusion, substantial variability was found in spatial patterns of all investigates air pollutants both within and between European study areas. The application of this highly standardized measurement and modelling approach across different locations will contribute to a consistent assessment of air pollutant levels and potentially contribute to understanding health effects associated with them.