Skip to content

Interpretation of organic components from Positive Matrix Factorization of aerosol mass spectrometric data

by I M Ulbrich, M R Canagaratna, Q Zhang, D R Worsnop, J L Jimenez
Atmos. Chem. Physics ()
  • ISSN: 1680-7316


The organic aerosol (OA) dataset from an Aerodyne Aerosol Mass Spectrometer\n(Q-AMS) collected at the Pittsburgh Air Quality Study (PAQS) in September\n2002 was analyzed with Positive Matrix Factorization (PMF). Three\ncomponents - hydrocarbon-like organic aerosol OA (HOA), a highly-oxygenated\nOA (OOA-1) that correlates well with sulfate, and a less-oxygenated,\nsemi-volatile OA (OOA-2) that correlates well with nitrate and chloride\n- are identified and interpreted as primary combustion emissions,\naged SOA, and semivolatile, less aged SOA, respectively. The complexity\nof interpreting the PMF solutions of unit mass resolution (UMR) AMS\ndata is illustrated by a detailed analysis of the solutions as a\nfunction of number of components and rotational forcing. A public\nweb-based database of AMS spectra has been created to aid this type\nof analysis. Realistic synthetic data is also used to characterize\nthe behavior of PMF for choosing the best number of factors, and\nevaluating the rotations of non-unique solutions. The ambient and\nsynthetic data indicate that the variation of the PMF quality of\nfit parameter (Q, a normalized chi-squared metric) vs. number of\nfactors in the solution is useful to identify the minimum number\nof factors, but more detailed analysis and interpretation are needed\nto choose the best number of factors. The maximum value of the rotational\nmatrix is not useful for determining the best number of factors.\nIn synthetic datasets, factors are ``split{''} into two or more components\nwhen solving for more factors than were used in the input. Elements\nof the ``splitting{''} behavior are observed in solutions of real\ndatasets with several factors. Significant structure remains in the\nresidual of the real dataset after physically-meaningful factors\nhave been assigned and an unrealistic number of factors would be\nrequired to explain the remaining variance. This residual structure\nappears to be due to variability in the spectra of the components\n(especially OOA-2 in this case), which is likely to be a key limit\nof the retrievability of components from AMS datasets using PMF and\nsimilar methods that need to assume constant component mass spectra.\nMethods for characterizing and dealing with this variability are\nneeded. Interpretation of PMF factors must be done carefully. Synthetic\ndata indicate that PMF internal diagnostics and similarity to available\nsource component spectra together are not sufficient for identifying\nfactors. It is critical to use correlations between factor and external\nmeasurement time series and other criteria to support factor interpretations.\nTrue components with <5% of the mass are unlikely to be retrieved\naccurately. Results from this study may be useful for interpreting\nthe PMF analysis of data from other aerosol mass spectrometers. Researchers\nare urged to analyze future datasets carefully, including synthetic\nanalyses, and to evaluate whether the conclusions made here apply\nto their datasets.

Author-supplied keywords

Cite this document (BETA)

Readership Statistics

1 Reader on Mendeley
by Discipline
100% Environmental Science
by Academic Status
100% Student > Doctoral Student

Sign up today - FREE

Mendeley saves you time finding and organizing research. Learn more

  • All your research in one place
  • Add and import papers easily
  • Access it anywhere, anytime

Start using Mendeley in seconds!

Sign up & Download

Already have an account? Sign in