Application of absolute principal component analysis to size distribution data: identification of particle origins T. W. Chan1,* and M. Mozurkewich1 1Department of Chemistry and Centre for Atmospheric Chemistry, York University, Toronto, Ontario, Canada *now at: Environment Canada, Toronto, Ontario, Canada
Abstract. Absolute principal component analysis can be applied, with suitable
modifications, to atmospheric aerosol size distribution measurements. This
method quickly and conveniently reduces the dimensionality of a data set.
The resulting representation of the data is much simpler, but preserves
virtually all the information present in the original measurements. Here we
demonstrate how to combine the simplified size distribution data with trace
gas measurements and meteorological data to determine the origins of the
measured particulate matter using absolute principal component analysis. We
have applied the analysis to four different sets of field measurements that
were conducted at three sites in southern Ontario. Several common factors
were observed at all the sites; these were identified as photochemically
produced secondary aerosol particles, regional pollutants (including
accumulation mode aerosol particles), and trace gas variations associated
with boundary layer dynamics. Each site also exhibited a factor associated
specifically with that site: local industrial emissions in Hamilton (urban
site), processed nucleation mode particles at Simcoe (polluted rural site),
and transported fine particles at Egbert (downwind from Toronto).
Citation: Chan, T. W. and Mozurkewich, M.: Application of absolute principal component analysis to size distribution data: identification of particle origins, Atmos. Chem. Phys., 7, 887-897, doi:10.5194/acp-7-887-2007, 2007.