Classification and receptor modeling of wet precipitation data from central Austria (1984-1993)
Vasil Simeonov, Hans Puxbaum, Stefan L. Tsakovski, Costel Sârbu, M. Kalina · Environmetrics · 1999
Various classification procedures (hierarchical and non-hierarchical crisp and fuzzy clustering, cross-classification fuzzy clustering) and principal component analysis (varimax rotation) combined with receptor modeling are applied to a data set consisting of wet deposition loads of major ions from five sampling sites in Central Austria collected in a period of 10 years (1984–1993). The data classification with respect to the sampling sites as objects has shown that a distinct separation between sites located at the northern Alpine rim (Reutte, Kufstein and Haunsberg) and in the inner Alpine region (Innervillgraten and Werfenweng) is achieved indicating in this way the role of the geographical disposition of the sampling sites. Both cluster and principal components analysis proved that three conditional (latent) factors are responsible for the data structure called in this study ‘anthropogenic’, ‘crustal’ and ‘mixed salt’—(NH + 4 SO 2− 4 NO − 3 H+), (Ca2+, Mg2+), (K+, Na+, Cl−), respectively. The source apportioning model has indicated that anthropogenic gaseous emissions might be responsible for the first factor, natural crustal influences and Sahara events for the second and sea salt and manure for the third. Copyright © 1999 John Wiley & Sons, Ltd.