Elsevier

Atmospheric Environment

Volume 62, December 2012, Pages 303-317
Atmospheric Environment

Spatial variation of PM2.5, PM10, PM2.5 absorbance and PMcoarse concentrations between and within 20 European study areas and the relationship with NO2 – Results of the ESCAPE project

https://doi.org/10.1016/j.atmosenv.2012.08.038Get rights and content

Abstract

The ESCAPE study (European Study of Cohorts for Air Pollution Effects) investigates relationships between long-term exposure to outdoor air pollution and health using cohort studies across Europe. This paper analyses the spatial variation of PM2.5, PM2.5 absorbance, PM10 and PMcoarse concentrations between and within 20 study areas across Europe.

We measured NO2, NOx, PM2.5, PM2.5 absorbance and PM10 between October 2008 and April 2011 using standardized methods. PMcoarse was determined as the difference between PM10 and PM2.5. In each of the twenty study areas, we selected twenty PM monitoring sites to represent the variability in important air quality predictors, including population density, traffic intensity and altitude. Each site was monitored over three 14-day periods spread over a year, using Harvard impactors. Results for each site were averaged after correcting for temporal variation using data obtained from a reference site, which was operated year-round.

Substantial concentration differences were observed between and within study areas. Concentrations for all components were higher in Southern Europe than in Western and Northern Europe, but the pattern differed per component with the highest average PM2.5 concentrations found in Turin and the highest PMcoarse in Heraklion. Street/urban background concentration ratios for PMcoarse (mean ratio 1.42) were as large as for PM2.5 absorbance (mean ratio 1.38) and higher than those for PM2.5 (1.14) and PM10 (1.23), documenting the importance of non-tailpipe emissions. Correlations between components varied between areas, but were generally high between NO2 and PM2.5 absorbance (average R2 = 0.80). Correlations between PM2.5 and PMcoarse were lower (average R2 = 0.39). Despite high correlations, concentration ratios between components varied, e.g. the NO2/PM2.5 ratio varied between 0.67 and 3.06.

In conclusion, substantial variability was found in spatial patterns of PM2.5, PM2.5 absorbance, PM10 and PMcoarse. The highly standardized measurement of particle concentrations across Europe will contribute to a consistent assessment of health effects across Europe.

Highlights

► We used one method to measure PM2.5, PM2.5 absorbance & PM10 in 20 European areas. ► We studied contrasts of these metrics and PMcoarsewithin andbetweenall 20 areas.► Concentrationswerehigher in Southern than in Western and Northern European areas.► Within-area contrasts varied by area andwerelarger for PM2.5 absorbance & PMcoarse. ► Concentration ratios of particle metrics and NO2varied significantly across areas.

Introduction

Human exposure to ambient levels of air pollution is a risk to public health (Brunekreef and Holgate, 2002; Pope and Dockery, 2006; WHO, 2006). It has been estimated that, in Europe, exposure to outdoor and traffic-related air pollution has greater adverse effects on public health in the long term, versus the short term (Künzli et al., 2000). Epidemiological studies have suggested associations of long-term exposure to current air pollution levels and particularly cardio-respiratory health (Brunekreef and Holgate, 2002; Rückerl et al., 2011). Most studies have found associations of health with particulate matter characterized as the mass concentration of particles smaller than 2.5 or 10 μm (PM2.5 or PM10) and nitrogen dioxide (NO2) (Brunekreef and Holgate, 2002). In the EU, the air quality limit values for PM10 and NO2 are still exceeded frequently (European Environment Agency (EEA) 2009; Velders and Diederen 2009), raising significant public concern. Less information is available concerning the exceedance of the new PM2.5 guideline (Brunekreef and Maynard, 2008).

Early epidemiological studies compared air pollution concentrations and health outcomes between cities, and have mostly ignored within city variability. Long-term city pollution levels were often characterized by a single (averaged) concentration, based on a limited number of monitors per city, e.g. the American Six Cities Study (Dockery et al., 1993) the American Cancer Society (ACS) study (Pope et al., 2002) and the ECRHS study in Europe (Götschi et al., 2005). Multiple recent studies have shown significant intra-urban spatial contrasts (Beelen et al., 2007; Hoek et al., 2002a; Jerrett et al., 2005b). Land Use Regression (LUR) modelling has been used frequently to explain these spatial contrasts, using predictor variables derived from Geographic Information Systems (GIS) (Hoek et al., 2008; Jerrett et al., 2005a). Several epidemiological studies have since made use of estimated within-urban pollution contrasts based on LUR models (e.g. Morgenstern et al., 2007; Beelen et al., 2008) often focussing mainly on motorized traffic as an important source of intra-urban air pollution contrast (HEI, 2010).

Significant variability of PM10 concentrations between cities across Europe has been reported based upon routine monitoring data (European Environment Agency (EEA) 2009), a series of research projects (Putaud et al., 2004; Van Dingenen et al., 2004) and a wintertime study in 14 European cities (Hoek et al., 1997). The lowest concentrations were generally found in Northern Europe and the highest in Southern and Eastern Europe. Spatial variation of PM2.5 across Europe is less well characterized because it is not routinely measured in most monitoring networks. Nevertheless, significant north–south gradients have been reported for PM2.5 based on research projects (Van Dingenen et al., 2004) and a purpose designed network consisting of 21 urban background stations across Europe (Götschi et al., 2005; Hazenkamp-von Arx et al., 2004). Within-urban contrasts have been characterized in various studies (Monn, 2001; Hoek et al., 2002b), but there are few. The interpretation of spatial contrasts of PM concentrations is limited by differences in site selection, and differences of sampling and analysis methods including different correction factors used to compensate for sampling losses of volatile components between countries and network operators (European Environment Agency (EEA) 2009; Van Dingenen et al., 2004). Furthermore, there is limited information on the spatial patterns of the coarse fraction, except from a few research projects (Van Dingenen et al., 2004; Puustinen et al., 2007). Yet, there is increasing epidemiological evidence of the adverse health effects of coarse particles (Brunekreef and Forsberg, 2005).

Ambient concentrations of PM2.5, PM10, particle composition, NO2 and NOx were measured in the framework of the ESCAPE project (European Study of Cohorts for Air Pollution Effects; www.escapeproject.eu). The objective of ESCAPE is to investigate the health effects of long-term exposure to ambient air pollution in 36 study areas across Europe. Individual exposure estimates for cohort subjects will be assigned based on predictions of land-use regression (LUR) models (Hoek et al., 2008), which are developed based on the air pollution measurements and geographic predictors. The study areas were selected because of the availability of informative cohort studies in these areas. We decided on performing study-specific sampling as most existing monitoring networks have insufficient density to capture small-scale spatial variation; locations may not be representative for human, residential exposures, or do not measure all components of interest routinely (e.g. PM2.5, PM2.5 absorbance). While NO2 and NOx were measured in all 36 areas, particulate matter was measured in 20 out of 36 areas. In each of the 20 areas, PM measurements were made at 20 sites using a standardized protocol using identical gravimetric samplers.

The aim of this paper is to assess the spatial contrasts of PM2.5, PM2.5 absorbance, PM10, and PMcoarse within and between areas. A second aim is to assess the variability of differences between regional background, urban background and street locations across Europe, as a likely source of within-area variability. The third aim is to assess the variability in concentrations ratios and correlations between the various particle metrics and NO2 across Europe. A detailed analysis of the spatial contrast for NO2 and NOx will be reported separately (Cyrys et al., 2012).

Section snippets

Sampling design

Particulate matter was measured in 20 study areas (Fig. 1). In each study area, 20 sites were measured. Sampling campaigns were conducted over an entire year, Measurements took place in all study areas between October 2008 and April 2011. Participating centres used identical sampling protocols and common criteria for the selection of sampling sites. Furthermore, they employed the same equipment and all samples were analysed centrally at one laboratory (IRAS, Utrecht University). Calculation of

Quality control

Detection limits were low for all centres for both PM10 (0.7–4.0 μg m−13) and PM10 absorbance (0.04–0.10 × 10−5 m−1). Only 3 samples (2 for PM2.5 and 1 for PM10) from Helsinki/Turku were below the detection limit (Online supplement C). We retained the original values. Reproducibility was good in most areas, as coefficients of variation (CV) varied between 2% and 7% for PM10, and between 2% and 5% for PM10 absorbance (Online supplement C). CV values for PM10 in Helsinki/Turku and Manchester were

Discussion

We found significant concentration differences for PM2.5, PM2.5 absorbance, PM10 and PMcoarse across 20 European study areas. Especially for PM2.5 absorbance and PMcoarse we saw significant contrasts within study areas. For PM2.5 absorbance, the contrast within study areas was larger than between study areas. Concentrations at street sites were higher than at urban background sites for all components, but the ratios differed widely across the study areas. PM10 and PM2.5 concentrations generally

Conclusion

We found clear spatial contrasts between PM2.5, PM2.5 absorbance, PM10 and PMcoarse, both within and between the 20 study areas. While there were large differences in background concentration between areas for all components, within-area contrasts were particularly clear for PM2.5 absorbance and PMcoarse. Significant differences were found between urban background sites and street sites for most study areas. High street/background ratios for PMcoarse specifically indicate the importance of

Competing interest

The authors declare they have no competing financial interest.

Acknowledgements

We thank Meng Wang, Marjan Tewis and Geert de Vrieze (IRAS, Utrecht University, The Netherlands) for their help in filter weighing, reflection measurements and data management. Furthermore, we thank all those who were responsible for air pollution measurements, data management and project supervision in all study areas: Haytham Alhamwi, Sònia Alvarez, Giovanna Berti, Laura Bouso, Simone Bucci, Ennio Cadum, Glòria Carrasco, Guido Fischer, Francesco Forastiere, Marco Gilardetti, Erwan Gremaud,

References (41)

Cited by (404)

View all citing articles on Scopus
View full text