Comparison of two air pollution models applied to assess population exposure in Leipzig, Germany

摘要

Objectives: We compared the exposure assignment of two air pollution models to all addresses in the city of Leipzig. Methods: Predicted concentrations of nitrogen dioxide (NO2) and fine particulate matter (PM2.5) for 2010 were assigned to all addresses in Leipzig (n=69,662) from two air pollution models with different spatial resolutions: fine spatial resolution (100x100 m) hybrid land use regression models developed for West Europe within the ELAPSE project and models with 2x2 km resolution from German Environmental Agency (UBA) developed for all Germany using the optimal interpolation technique. Results: Modelled annual mean concentrations of N02 at all addresses were 26.9 ± 4.9 (11.8-56.2) ug/m3 and 20.7 ± 1.0 (15.8-25.6) ug/m3 for ELAPSE and UBA models, respectively. PM2.5 concentrations were 18.6 ± 0.8 (14.5-21.1) ug/m3 and 17.0 ± 0.5 (15.6-17.7) ug/m3 for ELAPSE and UBA models, respectively. The range of the ELAPSE exposures was much larger than for the UBA for both pollutants. Pearson's correlation coefficients between ELAPSE and UBA exposures were 0.29 and 0.50 for NO2 and PM2.5, respectively. NO2 ELAPSE exposures correlated poorly with NO2 UBA, possibly explained by the fact that NO2 is a locally varying pollutant, strongly associated with traffic, which is better picked up by the ELAPSE model (100x100m) than by the UBA model ((2x2km). On the other side, PM2.5 ELAPSE exposures correlated better with PM2.5 UBA as PM2.5 is a more regional pollutant so less variation at the local scale and hence a better correlation. Conclusion: We assigned two different models developed for Western Europe and Germany to a very defined spatial area in Germany. The model with higher spatial resolution fitted better to explain within city variation of pollutants related to local sources to describe exposures of residents of Leipzig and the LIFE (The Leipzig Research Centre for Civilization Diseases) cohort.