Multidimensional modeling of aerosol monitoring data Aleksander Astel a, * , Vasil Simeonov b , Heidi Bauer c , Hans Puxbaum c a Pomeranian Academy, Biology and Environmental Protection Institute, Environmental Chemistry Research Unit, 22a Arciszewskiego Str., 76-200 Slupsk, Poland b Chair of Analytical Chemistry, Faculty of Chemistry, University of Sofia “St. Kl. Okhridski”, 1164 Sofia, J. Bourchier Blvd. 1, Bulgaria c Institute of Chemical Technologies and Analytics, Vienna University of Technology,1060 Vienna, Getreidemarkt 9, Austria Chemical profiles of particulate matter were evaluated by Tucker3 chemometric procedure to identify major sources of pollution in Lower Austria and to assess their spatiotemporal impact. article info Article history: Received 11 March 2010 Received in revised form 1 July 2010 Accepted 6 July 2010 Keywords: Particulate matter Chemometrics Air pollution N-way PCA Chemical composition abstract The present study deals with the application of N-way factor analysis for modeling and interpretation of a three-dimensional environmental data set acquired from monitoring of particulate matter (PM) collected at four different sampling locations in Lower Austria region (Central Europe). In the study the Tucker3 algorithm for N-way modeling was used. It was statistically validated that the Tucker3 model offered having the dimensionality [222] is appropriate for correct interpretation of the relationships between chemical parameters, sampling locations and sampling period. The Tucker3 model allowed to distinguish three major sources of pollution in the region of interest conditionally named “soil dust”, “combustion” and “street dust” latent factors as responsible for chemical profile of PM and to identify seasonal variability. Additionally, some specificity of the sampling locations was also pointed out. Ó 2010 Elsevier Ltd. All rights reserved. 1. Introduction Assessment of the air pollution at sites of interest requires conse- quent and constant monitoring of carefully selected parameters. This usually results in a data set with complex and multi-way structure. For instance, it considers a data set obtained as a result of monitoring of air pollution described by several parameters measured at different sampling sites over a long time period. Such data can be seen as 3-way data array arranged as sampling site  parameters  time and thus explored with approaches dealing with 3-way data structure. The most popular in chemometrics to model such data are PARAFAC, PARAFAC2 and Tucker3 (Henrion, 1994; Geladi, 1989; Smilde et al., 2004; Bro, 2006). The Tucker3 method has been successfully used in environ- mental studies for modeling the bio-geochemical interactions in the surface waters of the Gulf of Trieste (Barbieri et al., 1999), for assessing the time variability of freshwaters related to a municipal water supply (Barbieri et al., 2002), for analyzing the relationship of heavy-metal exposure of soils in different plant cultivations at several East-German regions (Stanimirova et al., 2006), for modeling and exploration of environmental n-way data from bulk precipitation quality control (Astel and Ma1ek, 2008) etc. In the present study it has been of significant interest to monitor the concentration of different chemical components in the aerosol samples collected on a daily basis at four sampling sites scattered in the Austrian province Lower Austria (Central Europe) during spring, summer, and winter and, further, to interpret possible relationships between sampling locations, sampling parameters and season of sampling. It could make possible to identify different sources of pollution with specific source profiles and to distinguish seasonal changes in the air quality of the region of interest. The aim of the present study is to model the monitoring data set from the region by the use of the chemometric N-way method (Tucker3 model) in order to reveal the interrelationships between chemical parameters, sampling locations and sampling period. It could be of use for risk assessment and risk management of the air quality of the sampling area. 2. Experimental 2.1. Sampling procedure and sampling sites Aerosol collection was performed at four sites in central and eastern Lower Austria on a daily basis (0e24 h) in the period between 1st January and 30th June 2005 in accordance to PM10 sampling procedures standardized in EN 12341 (1999). PM10 samples were collected in parallel with DIGITEL DA80H High Volume samplers (Digitel Elektronik AG, Hegnau, Switzerland) at 30.5 m 3 h 1 and auto- mated Leckel Low Volume samplers (Sven Leckel GmbH, Berlin, Germany) at a flow rate of 2.3 m 3 h 1 . High volume samples were collected on Quartz fiber filters (Pallflex, Tissuequarz 2500QAT-UP Ø 150 mm) and used for gravimetric analysis of * Corresponding author. E-mail address: astel@apsl.edu.pl (A. Astel). Contents lists available at ScienceDirect Environmental Pollution journal homepage: www.elsevier.com/locate/envpol 0269-7491/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.envpol.2010.07.003 Environmental Pollution 158 (2010) 3201e3208