Source apportionment of submicron organic aerosol at an urban background site in Zürich

Transcription

1 Materials Science & Technology Source apportionment of submicron organic aerosol at an urban background site in Zürich 11 th ETH-Conference on Combustion Generated Nanoparticles Valentin Lanz (PhD Student, Empa Dübendorf) R. Alfarra, U. Baltensperger, B. Buchmann, C. Hüglin, and A. Prévôt

2 Overview Introduction: Atmospheric aerosols Methods: Receptor modeling = CS X i i + E i Results: Zürich-Kaserne Conclusions

3 Atmospheric (organic) aerosols Impact on health (climate, ecosystems, visibility,.)

4 Atmospheric (organic) aerosols Impact on health (climate, ecosystems, visibility,.) Composition: ammonium 10% Organic matter 24% (primary & secondary) EC 9% nitrate 17% unknwon 13% sulfate 17% mineral dust 7% trace elements: 2% PM 2.5 (Zürich-Kaserne; Hueglin et al., 2005)

5 Aerosol measurements Aersol mass spectrometer (Q-AMS, Aerodyne Inc.) Non-refractory, submicron (~PM 1 ) aerosol Online, (semi-)quantitative High time-resolution (sec - minutes) Nitrate Equivalent Mass Concentration (µg m -3 ) Organic mass fragments Organic MS ( m/z) m/z

6 Receptor modeling Receptor models (linear mixing models) = CS X i i + E i X i : multivariate observation (m-vector) measured AMS spectrum (270 dimensional) C: loadings (m x p-matrix), C 0 aerosol source profile (270 x p-matrix) S i : scores (p-vector), S 0 p latent aerosol source activities No assumptions about source distributions Optional: meteorology/source composition Types CMB - chemical mass balance: PMF - positive matrix factorization: chemical signature of sources known unknown

7 Receptor modeling II reconstructed data matrix measured data matrix (X) ˆ = - ˆ X CS X X = E modeled source profiles modeled time series of source activities factors/sources =

8 AMS sampling site: Zürich, urban background Urban background site (Zürich-Kaerne) Public backyard 500 meters from Zurich main station 1 mile from lakeside Site of the Swiss National Air Pollution Monitoring Network (NABEL; Ancillary data to validate math. model: - meteo (radiation, wind direction, temperature ) - gas-phase (CO, NO x, O 3, SO 2, VOCs, ) - aerosol (PM 10, OC, EC)

9 Zürich, Summer July 4 August 2005 Photochemical episodes ( summer smog ) Positive matrix factorization (PMF) (Lanz et al., 2007, Atmos. Chem. Phys.)

10 Zürich, Summer 2005 Source profiles (C matrix): Verification by independently measured AMS profiles 1. SOA, type I ex. 1 PMF computed 2. SOA, type II R 2 =0.96 measured (literature) 3. Diesel fuel ex Charbroiling PMF computed 5. Wood burning R 2 = Minor source measured (literature)

11 Zürich, Summer 2005 Source contributions: 66% of submicron AMS-organics are SOA SOA = secondary organic aerosols (i.e. formed by gas-to-particle reactions in the atmosphere) a. SOA, OOA, type I [µg (44%) m -3 ] AMS particle-sulphate [µg m -3 ] b. SOA, OOA, typeii II [µg (22%) m -3 ] dat AMS particle-nitrate [µg m -3 ] temperature [ C] POA (=primary organic aerosols) account for 34% and can directly be attributed to sources (10% wood burning, 7% fuel combustion, 13% charbroiling, <4% food cooking)

12 Zürich, Winter to 23 January Temperature inversions 4 Temperature [ C] ( winter smog ) 0 Hybrid model: CMB x PMF (Lanz et al., ES&T, in review) -4-8 PM 10 [µg m -3 ] ( ) ( ) AMS-organics [µg m -3 ] O 3 [ppb] Date

13 Zürich, Winter 2006 Source contributions: a. Fuel combustion (7%) b. Wood burning (38%) (emission rel. levoglucosan!) Primary organic aerosols (POA): 45% c. SOA (55%)

14 Diesel combustion Wood burning aerosol Secondary organic aerosol (May 2005) (July 2005) (January 2006) 88% Härkingen Rural/motorway 66% 10% 55% 38% Zürich-Kaserne Urban/background 12% 7% 45% 7% (March 2005) 54% 23% Roveredo Alpine/motorway 13%

15 Source apportionment of SOA Apportionment of SOA to emission sources: winter 2006 POA (45%) Fossil fuel burning VOC OVOC Wood combustion SOA (55%)

16 Source apportionment of SOA Apportionment of SOA to emission sources: winter 2006 POA (45%) Fossil fuel burning VOC OVOC Wood combustion SOA (55%) Szidat et al. (2006a, 2006b, 2007): 14 C, OC/EC fossil non-fossil ~17% ~38%

17 Conclusions Ambient (AMS) organic aerosols can be attributed to sources by linear mixing models (PMF). SOA accounted for 50-90% of the organic aerosol SOA is dominant both in summer (photochemistry) and in winter (accumulation, low temperatures) POA accounted for 10-50% of organic aerosol Residential areas/winter: POA dominated by wood burning 14 C measurements give further insights into SOA source attribution (fossil/non-fossil) (Winter 2006): Fuel combustion contributes more to SOA (17%) than POA (7%) Wood burning contributes equally to SOA (38%) and POA (38%)

18 Thanks to Sönke Szidat, Miriam Wehrli, Lukas Wacker, Alexandre Caseiro, Hans Puxbaum, Silke Weimer, Robert Gehrig, Ulrike Lohmann, Pentti Paatero, Swiss Federal Office for the Environment (FOEN), Kanton Zürich, ACCENT,