412
M.e. Gennaro· S. Angelino
for suspended matter, a clear grouping can not be evidenced in the biplot of Fig. 22.2b.
PC1 explains only 32% of the total variance, while PC2 contributes for an additive 22%
and probably explains the composition of the biological material, the main carrier of
the CBs. Figure 22.2a also confirms, as expected, that the anthropogenic contaminants
in deeper waters have however their source in sea surface as confirmed by the generally higher concentration in intermediate-deeper waters than in the bottom ones. Since
the composition of the CBs mixture in surface waters in this Japan region is quite variable, as shown by the plot of PC1 loading vs. PC2, correspondingly large differences
were observed in the different water masses in the vertical profiles. Similar behaviour
was observed for NoPHs too. The behaviour was explained by the taking place of mixing processes of different water bodies.
22.2.1.3
Marine Aerosol
In the North Sea, in the southern bight, aerosol samples were collected from an aircraft, in order to characterize aerosol particles for their content in metals (Cr, In, Pb,
Ni, V, Sn, Cd, Cu) (Van Malderen et al. 1996). The input of heavy metals in the North
Sea significantly increased due to the growing industrial activity and car traffic in
European countries. Wet and dry deposition of pollutants from the atmosphere is a
dominant transport system for a number of elements. To assess heavy metal concentrations and size distribution and to quantify the deposition fluxes of these elements,
electron-probe X-ray microanalysis (EPXMA) was used, because it requires shorter
sampling times (minutes rather than hours) with respect to the use of other analytical techniques. Thus it is possible to study short time variations in the atmospheric
composition (for instance induced by changes of wind direction) and to overcome
some problems of airborne sampling from aeroplanes. The aircraft flew tracks of about
110 km at six different heights, more or less evenly spaced between sea level and the
inversion layer. The lowest track, 10-30 m above the sea level, was intended to assess
particle resuspension by sea spray: airborne particulate matter was collected on
0.4 jlm-filters. 96 samples were taken during 16 flights and 45000 particles were
analysed. More than 5000 were found to contain significant concentrations of at least
one of the following metals: Cr, Pb and In. Significant differences in abundance were
evidenced in the aerosols, depending on the origin of the air masses: the highest values for Cr, Pb and In were always found under south-eastern wind directions. The
chemometric treatment of the data set made use of a combination of hierarchical (dendrogram) HCA, non hierarchical (NHCA) and fuzzy clustering analysis (FCA). The
dendrogram offers the advantage of an easy visualisation of the results but presents
risks of misclassification, since it assumes a hierarchical structure (unlikely for atmospheric data sets). On the other hand, since NHCA methods would require a prior
knowledge of the data structure, the results of HCA as the centroids for the NHCA
were used. Both HCA and NHCA are hard clustering methods, according to which each
sample can belong to only one cluster and can therefore lead to misclassifications when
overlapping clusters are present. On the contrary, in FCA the possibility for each object to belong to each cluster is considered and its probability evaluated.
The results of the FCA individuated 5-7 major particle types. Those containing
Cr, Pb or In can be, directly or indirectly, connected to emission processes of metal-
M.e. Gennaro· S. Angelino
for suspended matter, a clear grouping can not be evidenced in the biplot of Fig. 22.2b.
PC1 explains only 32% of the total variance, while PC2 contributes for an additive 22%
and probably explains the composition of the biological material, the main carrier of
the CBs. Figure 22.2a also confirms, as expected, that the anthropogenic contaminants
in deeper waters have however their source in sea surface as confirmed by the generally higher concentration in intermediate-deeper waters than in the bottom ones. Since
the composition of the CBs mixture in surface waters in this Japan region is quite variable, as shown by the plot of PC1 loading vs. PC2, correspondingly large differences
were observed in the different water masses in the vertical profiles. Similar behaviour
was observed for NoPHs too. The behaviour was explained by the taking place of mixing processes of different water bodies.
22.2.1.3
Marine Aerosol
In the North Sea, in the southern bight, aerosol samples were collected from an aircraft, in order to characterize aerosol particles for their content in metals (Cr, In, Pb,
Ni, V, Sn, Cd, Cu) (Van Malderen et al. 1996). The input of heavy metals in the North
Sea significantly increased due to the growing industrial activity and car traffic in
European countries. Wet and dry deposition of pollutants from the atmosphere is a
dominant transport system for a number of elements. To assess heavy metal concentrations and size distribution and to quantify the deposition fluxes of these elements,
electron-probe X-ray microanalysis (EPXMA) was used, because it requires shorter
sampling times (minutes rather than hours) with respect to the use of other analytical techniques. Thus it is possible to study short time variations in the atmospheric
composition (for instance induced by changes of wind direction) and to overcome
some problems of airborne sampling from aeroplanes. The aircraft flew tracks of about
110 km at six different heights, more or less evenly spaced between sea level and the
inversion layer. The lowest track, 10-30 m above the sea level, was intended to assess
particle resuspension by sea spray: airborne particulate matter was collected on
0.4 jlm-filters. 96 samples were taken during 16 flights and 45000 particles were
analysed. More than 5000 were found to contain significant concentrations of at least
one of the following metals: Cr, Pb and In. Significant differences in abundance were
evidenced in the aerosols, depending on the origin of the air masses: the highest values for Cr, Pb and In were always found under south-eastern wind directions. The
chemometric treatment of the data set made use of a combination of hierarchical (dendrogram) HCA, non hierarchical (NHCA) and fuzzy clustering analysis (FCA). The
dendrogram offers the advantage of an easy visualisation of the results but presents
risks of misclassification, since it assumes a hierarchical structure (unlikely for atmospheric data sets). On the other hand, since NHCA methods would require a prior
knowledge of the data structure, the results of HCA as the centroids for the NHCA
were used. Both HCA and NHCA are hard clustering methods, according to which each
sample can belong to only one cluster and can therefore lead to misclassifications when
overlapping clusters are present. On the contrary, in FCA the possibility for each object to belong to each cluster is considered and its probability evaluated.
The results of the FCA individuated 5-7 major particle types. Those containing
Cr, Pb or In can be, directly or indirectly, connected to emission processes of metal-