Places in Switzerland with AMS Measurements

Size: px

Start display at page:

Download "Places in Switzerland with AMS Measurements"

Thomas Lewis
5 years ago
Views:

Source apportionment of the particulate organic mass during winter and summer in Zurich, Switzerland A.S.H. Prévôt, M.R. Alfarra, J. Sandradewi, S. Weimer, N. Perron, U.

Szi University of Berne, Switzerland Discussion of Aerodyne aerosol mass spectrometer and 1 C-a Thanks for various reference spectra of the Aerodyne AMS community Thanks to

1 Source apportionment of the particulate organic mass during winter and summer in Zurich, Switzerland A.S.H. Prévôt, M.R. Alfarra, J. Sandradewi, S. Weimer, N. Perron, U. Baltensperger Paul Scherrer Institut, Switzerland Valentin Lanz, C. Hueglin EMPA, Switzerland S. Szi University of Berne, Switzerland Discussion of Aerodyne aerosol mass spectrometer and 1 C-a Thanks for various reference spectra of the Aerodyne AMS community Thanks to BAFU, Kantons Lucerne, Graubünden, Tessin, Zürich, Oekoscience Places in Switzerland with AMS Measurements Härkingen (May) Zürich (Summer und Winter) Reiden (February) Payerne (June, January 7) JFJ (Winter, Summer) Massongex (Dezember ) Roveredo (March, December) 1

2 Average composition in Zürich in summer and winter (Aerosol mass spectrometer and aethalometer a) 13% 1% 1% 1% 11% 13% 3% Zürich (July) 3 μg/m 3 % Black Carbon Organic mass Nitrate Sulfate Ammonium 15% 15% Zürich (January) 9 μg/m 3 1 C analysis of EC in summer and winter in Zürich 1 C-analyses (Szi et al.,,, 7) of HiVol-Quartzfilter allow for the determination of the fossil and non-fossil carbon fractions of the EC, OC, and more fractions like water-soluble OC,.. 19% winter summer EC (fossil) EC (non-fossil) EC (fossil) EC (non-fossil) 1% 93%

Organic aerosol mass spectra in Zurich contribution to total orgnaics %

m/z 3 57 9 9 11 13 Zurich, summer 1 3 5 7 m/z 9 11 13 Positive Matrix

time i: 1 n X {n x m} = G {n x p} number of factors/ reduced dimensions

3 Organic aerosol mass spectra in Zurich contribution to total orgnaics % Zurich, winter Contribution to total organics % m/z Zurich, summer m/z Positive Matrix Factorization (PMF): Bilinear unmixing Mass peaks j: 1 m samples in time i: 1 n X {n x m} = G {n x p} number of factors/ reduced dimensions F {p x m} + rows source profiles E E {n {n x m} x m} columns source strengths 3

Positive Matrix Factorization of OM in Zürich summer % % 15% 9% Org NO3 SO NH Lanz et al., ACP (7) Temporal evolutions of OOA I and OOA II 1 15 1 5 temperature [ C] AMS particle-nitrate [µg m -3 ] 1.

4 Positive Matrix Factorization of OM in Zürich summer % % 15% 9% Org NO3 SO NH Lanz et al., ACP (7) Temporal evolutions of OOA I and OOA II temperature [ C] AMS particle-nitrate [µg m -3 ] OOA, type II [µg m -3 ] 3 OOA, type I [µg m -3 ] AMS particle-sulphate [µg m -3 ] C- analysis indicates that most SOA is non-fossil (probably a combination from oxiion of terpenes, sesquiterpenes, isoprene, and possibly from biomass burning emissions Lanz et al., ACP, 7

5 If contribution of various OA components do not vary much in time Positive Matrix Factorization or similar unmixing methods do not work: Example Zürich winter..5. m/z 57 (diesel) [µg m -3 ] m/z (oxygenated) [µg m -3 ] m/z (wood burning) [µg m -3 ] A solution could be to use chemical mass balance (CMB) methods or hybrid versions of CMB and PMF ij = gipfpj + p p' x g f + e ip' q ij x CMB (G unknown) unknown sources/ free factors (PMF) x known sources/ fixed factors (CMB) x PMF (G, F unknown) F jm= F ' jm ± a j ' F jm solved by the multilinear engine, ME (Paatero, 1999) a = { 1} - In Zurich winter we used HOA as the known source, OOA and wood burning could be additionally identified Lanz et al., ES&T, accepted 5

6 Härkingen (Mai) Zürich (Sommer und Winter) Rheintal (erste mobile Messungen Reiden (Februar) im Februar 7) Payerne (Juni, Januar 7) JFJ (Winter, Sommer) Massongex (Dezember ) Payerne (Juni, Januar 7) Massongex (Dezember ) Härkingen (Mai) Reiden (Februar) Roveredo (März, Dezember) Zürich (Sommer und Winter) Rheintal (erste mobile Messungen im Februar 7) JFJ (Winter, Sommer) Roveredo (März, Dezember) Sensitivity analysis toward the degree of freedom of the constraint HOA factor a =. a =. a =. 3% 13% Campaign average 5 % % 3% 5% 35% wood burning OOA HOA high PM episode (a =.) high PM episode (a =.) high PM episode (a =.) Highest concentra tions during campaign 3% % 53% 5% 9% % 3% 1 1% wood burning OOA HOA Lanz et al., ES&T, accepted calculated G wood burning (a=.) [µg m -3 ] CO [ppm] Plausibility of solution including 3 factors wood burning (modelled) vs CO (measured): R = OOA (modelled) vs. AMS-ammonium (measured) R =.7 OOA (a=.) [µg m -3 ] AMS-NH [µg m -3 ] calculated F norm. intensity norm. intensity m/z 9 oxygenated organic aerosol (OOA, type I) Roveredo MS: R =.93 chestnut MS: R =.7 wood burning aerosol 73 aged rural MS: R =.93 fulvic acid MS: R =.7 m/z HOA (a=.) 5 HOA (modelled) vs. NOx (measured): R =.7 3 NO x [ppb] norm. intensity hydrocarbon-like organic aerosol (HOA) m/z Lanz et al., ES&T, accepted

7 Look at residuals after statistical analysis Look at residuals after statistical analysis 7

Payerne (Juni, Januar 7) Massongex (Dezember ) Härkingen (Mai) Reiden (Februar) Zürich (Sommer und Winter) Rheintal (erste mobile Messungen im Februar 7) JFJ (Winter, Sommer) Roveredo (März,

8 Payerne (Juni, Januar 7) Massongex (Dezember ) Härkingen (Mai) Reiden (Februar) Zürich (Sommer und Winter) Rheintal (erste mobile Messungen im Februar 7) JFJ (Winter, Sommer) Roveredo (März, Dezember) Carbon apportionment using 1 C analysis Estimation of fossil and non-fossil SOA contribution OM EC Z_meanPM1 3% Wood burning Zürich (mean) 3% Traffic 1 OCnonfossil 5% 19% ECbb (biomass Secondary burning) % ECf (fossil) Secondary 1 (non-fossil) OCfossil 3% EC (fossil) EC (non-fossil) Use of AMS analysis : - wood burning 3% - HOA Assumptions : - only SOA, HOA and wood burning present - OM/OC= for wood burning and SOA and OM/OC=1. for HOA 1% Lanz et al., submitted to ES&T - RESULT : SOAnonfossil: 9% SOAfossil: 31% PSI mobile laboratory

9 Estimation of wood burning contribution to OM as a function of location Geschätzter Holzverbrennungsanteil am morning Morning wodburningfraction m*3/org *(Masse *3)/organics = Wood burning fraction (Alfarra et al. submitted 7)..7evening Evening wodburningfraction m*3/org Conclusion PMF and hybrid versions of PMF and CMB are very powerful tools for the source apportionment of organic aerosol mass spectra Be careful with PMF results : valie as much as possible (time traces with other indicators, comparison to reference spectra), look carefully at remaining structure in the residual. Question/concern: is the ionization efficiency the same for OOA1, OOA, HOA, wood burning organics?! The combination of 1 C analyses with AMS a is very promising. We would like to do that more often and would like to offer our cooperation in the analysis of Hi-Vol filters concerning 1 C Measurements expensive and time-consuming : Costs per analysis: 3 (costs to use the accelerator MS at ETH) dollars (including all the preparation, analysis, etc.) (typically -5 filters per campaign and station good enough: if EC and OC analyzed, one needs 1 analyses. A very limited amount we might be able to do for free.. We are interested in a limited amount of measurement campaign a including high temperature, considerable temperature variation and supposedly high SOA amounts. Idea: Assessment of OOA1, OOA spectra and their partitioning. 9

Supplement of Processing of biomass-burning aerosol in the eastern Mediterranean during summertime

Supplement of Processing of biomass-burning aerosol in the eastern Mediterranean during summertime Supplement of Atmos. Chem. Phys., 1, 793 87, 1 http://www.atmos-chem-phys.net/acp-1-793-1/ doi:1.519/acp-1-793-1-supplement Author(s) 1. CC Attribution 3. License. Atmospheric Chemistry and Physics Open