A new, powerful technique to analyze single particle aerosol mass spectra using a combination of OPTICS and the fuzzy c-means algorithm

• Published in: Journal of aerosol science Vol. 98; pp. 1 - 14
• Main Authors: Reitz, Paul, Zorn, Sören R., Trimborn, Stefan H., Trimborn, Achim M.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.08.2016
• Subjects: Aerosol chemistry; C-means; Clustering; LAAPTOF; Single particle aerosol mass spectrometry; Visualization; Visualization; LAAPTOF; Clustering; Aerosol chemistry; Single particle aerosol mass spectrometry; C-means
• ISSN: 0021-8502; 1879-1964
• DOI: 10.1016/j.jaerosci.2016.04.003

Single particle aerosol mass spectrometers record two mass spectra for each individual particle, producing large amounts of data. The analysis of these spectra is typically performed using data mining algorithms like fuzzy c-means clustering. Here we present a new approach by applying the Ordering Points To Identify the Clustering Structure (OPTICS) algorithm in combination with fuzzy c-means clustering to single particle mass spectra. OPTICS treats spectra as points in n-dimensional space where each mass to charge ratio represents a dimension. The algorithm orders the spectra based on their density in this n-dimensional space. To demonstrate the strength of this combination of algorithms we applied it to an ambient dataset. The graphical representation of the results reflects chemical processing in the sampled aerosol. This allows for a detailed interpretation of the chemical evolution of the atmospheric aerosol. •2D illustration of the cluster structure.•Visualization of changes of the chemical composition of aerosol particles.•Ordering single particle mass spectra by degree of chemical processing.