10th International Aerosol Conference September 2 - September 7, 2018 America's Center Convention Complex St. Louis, Missouri, USA
Abstract View
Long-term Trends in Particle Number Size-distributions and New Particle Formation Observed at San Pietro Capofiume, Italy
TUOMO NIEMINEN, Jorma Joutsensaari, Ville Leinonen, Santtu Mikkonen, Taina Yli-Juuti, Pasi Miettinen, Annele Virtanen, Kari Lehtinen, Ari Laaksonen, Stefano Decesari, Leone Tarozzi, M. Cristina Facchini, University of Eastern Finland
Abstract Number: 1387 Working Group: Remote/Regional Atmospheric Aerosol
Abstract Atmospheric aerosols have a large impact on air quality, human health and even the global climate. Both the aerosol number concentration and their size have influence on the climatic effects of aerosols. Formation and growth of secondary aerosol particles is a major source of atmospheric aerosols, and has been a subject of active research during the past two decades. However, only few long-term datasets (over 10 years of measurements) of atmospheric aerosol number size distributions exist1,2,3,4.
In this work, we characterize trends in aerosol number size-distributions and in new particle formation (NPF) at San Pietro Capofiume in Po Valley, Northern Italy. Number size distributions of 3–630 nm particles have been measured there continuously since March 2002 with a twin-DMPS setup. The site is influenced by emissions of local anthropogenic pollutants (such as SO2) as well as long-range transport from Central and Eastern Europe, and thus can provide information on the impact of anthropogenic activity on aerosol size-distributions and NPF.
The particle number size-distribution data was classified into NPF event, non-event and undefined days5. The formation and growth rates of nucleation mode particles (defined here as particles of 3–25 nm in diameter) were calculated based on the time-evolution of the measured number size-distribution data. To quantify the trends in particle concentrations, we have used two methods. The first method fits the concentration time series as a sum of constant linear trend and seasonally varying component. The statistical significance of the fitted trend is estimated by performing the fitting multiple times with bootstrap sampling. The second method is a Dynamic Linear Model (DLM)6, where the particle number concentration timeseries is decomposed into level, trend, seasonality, and noise. These components are allowed to change as functions of time, and the magnitude of this change is modelled and estimated7.
There has been a longer break in the measurements from November 2010 until June 2011, during which time the twin-DMPS setup was in maintenance. Before this break, statistically significant decreasing linear trends are observed in nucleation, Aitken and accumulation mode particle concentrations. The largest decrease (–10%/year) occurs in the nucleation mode particle concentration. After the measurement break, only the nucleation mode particle concentration shows a statistically significant trend (decreasing –4.9%/year). The DLM method indicates that these changes occurred in 2008–2009. The particulate mass concentrations (PM10 and PM2.5) have also been reported to decrease throughout the Po Valley region, and it is attributed to decreases both in primary emissions and in precursors of secondary inorganic aerosol emissions mostly from vehicular traffic8. However, unlike the change in the trends of the particle number concentrations after 2007–2008 seen in our study, the PM concentrations seem to have continued their decrease until at least 2014.
The annual frequency of NPF event occurrence did not show any clear trend, varying between 20% and 40% of the days in a year. Both the formation and growth rates of nucleation mode particles had a decreasing trend of –3%/year and –2%/year, respectively. This would indicate that even though the sink for the newly formed particles has decreased (due to decrease in PM concentrations), a simultaneous decrease in precursor vapour emissions (sulphur dioxide, ammonia, amines, organics) has also occurred.
[1] Hamed, A. et al. (2010). Atmos. Chem. Phys. 10, 1071–1091. [2] Asmi, E. et al. (2011). Atmos. Chem. Phys. 11, 12959–12972. [3] Kyrö, E.-M. et al. (2014). Atmos. Chem. Phys. 14, 4383–4396. [4] Nieminen, T. et al. (2014). Boreal Env. Res. 19 (suppl. B), 191–214. [5] Kulmala, M. et al. (2012). Nat. Protocols 7, 1651–1667. [6] Durbin, T. J. and Koopman, S. J. (2012). Time Series Analysis by State Space Methods. Oxford University Press. [7] Mikkonen, S. et al. (2015). Stoch. Environ. Res. Risk Assess. 29, 1521–1529. [8] Bigi, A. and Ghermandi, G. (2016). Atmos. Chem. Phys. 16, 15777–15788.