Simplifying aerosol size distributions modes simultaneously detected at four monitoring sites during SAPUSS
- 1Institute of Environmental Assessment and Water Research (IDǼA) Consejo Superior de Investigaciones Científicas (CSIC), C/ Jordi Girona 18–26, 08034 Barcelona, Spain
- 2Department of Astronomy and Meteorology, Faculty of Physics, University of Barcelona, C/ Martí i Franquès 1, 08028 Barcelona, Spain
- 3National Centre for Atmospheric Science Division of Environmental Health & Risk Management School of Geography, Earth & Environmental Sciences, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK
- 4Department of Environmental Sciences/Center of Excellence in Environmental Studies, King Abdulaziz University, Jeddah, 21589, Saudi Arabia
- *now at: Institute of Marine Sciences (ICM) Consejo Superior de Investigaciones Científicas (CSIC), Pg Marítim de la Barceloneta 37–49, 08003 Barcelona, Spain
Abstract. The analysis of aerosol size distributions is a useful tool for understanding the sources and the processes influencing particle number concentrations (N) in urban areas. Hence, during the one-month SAPUSS campaign (Solving Aerosol Problems by Using Synergistic Strategies, EU Marie Curie Action) in autumn 2010 in Barcelona (Spain), four SMPSs (Scanning Mobility Particle Sizer) were simultaneously deployed at four monitoring sites: a road side (RSsite), an urban background site located in the city (UBsite), an urban background site located in the nearby hills of the city (Torre Collserola, TCsite) and a regional background site located about 50 km from the Barcelona urban areas (RBsite). The spatial distribution of sites allows study of the aerosol temporal variability as well as the spatial distribution, progressively moving away from urban aerosol sources. In order to interpret the data sets collected, a k-means cluster analysis was performed on the combined SMPS data sets. This resulted in nine clusters describing all aerosol size distributions from the four sites. In summary there were three main categories (with three clusters in each category): "Traffic" (Traffic 1, "Tclus_1" – 8%; Traffic 2, "Tclus_2" – 13%; and Traffic 3, "Tclus_3" – 9%) "Background Pollution" (Urban Background 1, "UBclus_1" – 21%; Regional Background 1, "RBclus_1" – 15%; and Regional Background 2, "RBclus_2" – 18%) and "Special Cases" (Nucleation, "NUclus" – 5%; Regional Nitrate, "NITclus" – 6%; and Mix, "MIXclus" – 5%). As expected, the frequency of traffic clusters (Tclus_1–3) followed the order RSsite, UBsite, TCsite, and RBsite. These showed typical traffic modes mainly distributed at 20–40 nm. The urban background sites (UBsite and TCsite) reflected also as expected urban background number concentrations (average values, N = 1.0 × 104 cm−3 and N = 5.5 × 103 cm−3, respectively, relative to 1.3 × 104 cm−3 seen at RSsite). The cluster describing the urban background pollution (UBclus_1) could be used to monitor the sea breeze circulation towards the regional background study area. Overall, the RBsite was mainly characterised by two different regional background aerosol size distributions: whilst both exhibited low N (2.7 × 103 for RBclus_1 and 2.2 × 103 cm−3 for RBclus_2), RBclus_1 had average PM10 concentrations higher than RBclus_2 (27 vs. 23 μg m−3). As regards the minor aerosol size distribution clusters, the "Nucleation" cluster was observed during daytime, whilst the "Regional Nitrate" was mainly seen at night. The ninth cluster ("Mix") was the least well defined and likely composed of a number of aerosol sources.
When correlating averaged values of N, NO2 and PM (particulate mass) for each k-means cluster, a linear correlation between N and NO2 with values progressively increasing from the regional site RBsite to the road site RSsite was found. This points to vehicular traffic as the main source of both N and NO2. By contrast, such an association does not exist for the case of the nucleation cluster, where the highest N is found with low NO2 and PM.
Finally, the clustering technique allowed study of the impact of meteorological parameters on the traffic N emissions. This study confirms the shrinking of freshly emitted particles (by about 20% within 1 km in less than 10 min; Dall'Osto et al., 2011a) as particles are transported from the traffic hot spots towards urban background environments. Additionally, for a given well-defined aerosol size distribution (Tclus_2) associated with primary aerosol emissions from road traffic we found that N5–15 nm concentrations can vary up to a factor of eight.
Within our measurement range of SMPSs (N15–228 nm) and Condensation Particle Counters (CPCs, N>5 nm), we found that ultrafine particles within the range 5–15 nm in urban areas are the most dynamic, being a complex ensemble of primary evaporating traffic particles, traffic tailpipe new particle formation and non-traffic new particle formation.