Strategies to Enhance the Interpretation of Single-Particle Ambient Aerosol Data

New instruments are beginning to reveal the chemical complexity of atmospheric aerosol particles. Exploitation of the plethora of information being made accessible through aerosol particle spectrometry and other techniques requires new strategies for data interpretation. This paper demonstrates and evaluates several analysis methods used to exploit this single-particle high-time-resolution data. In the first part of this study, Standard Reference Material (SRM) particulate matter samples were analyzed by an Aerosol Time-of-Flight Mass Spectrometer (ATOFMS) in order to evaluate the use of a modified, logarithm based, method of clustering mass spectra using the Adaptive Resonance Theory (ART-2a) algorithm. In the second part of this study, data obtained from the ATOFMS during the four seasons of 2007 were interpreted using a variety of approaches so as to elucidate the nature and sources of particles influencing the great lakes region of North America. This dataset is believed to represent the longest time-span of single-particle data ever analyzed in a study of this nature. These mass spectra were clustered into 21 different particle types using the supervised log-transformed ART-2a algorithm. Both long-term seasonal trends and high-time-resolution temporal patterns of particle type concentrations were examined. Source identification was supported by comparison with known source samples. Potential source contribution functions were used to identify source regions. This paper describes and evaluates these approaches to data interpretation using examples from the ambient air study to illustrate the methodology and highlight the findings. Furthermore, these ambient examples demonstrate how the application of these strategies enhances the interpretation of single-particle ambient aerosol data.