Geochemistry of natural and anthropogenic fall-out (aerosol and precipitation) collected from the NW Mediterranean: two different multivariate statistical approaches
Introduction
Statistical analysis in atmospheric studies is a well-known and well-developed field (Morandi et al., 1987; Hopke, 1988; Van Malderen et al., 1992). This study presents an example of the use of cluster analysis (CA), Q-Factor analysis (Q-FA) and discriminant analysis (DA) in atmospheric geochemistry, in an attempt to classify mixed aerosol and precipitation samples on the basis of their major element and trace metal chemistry.
Atmospheric inputs to the Mediterranean are of particular interest, in view of the different areas of aerosol production on the north and south belts.
The principal source of natural “crustal” material in the Mediterranean is the Sahara, while industrialised and semi-industrialised Northern Europe is a source of anthropogenic “background” material to the atmosphere. These inputs have an important influence on both the mineralogical and chemical composition of the Mediterranean aerosol. Previous works have shown that the bulk aerosol in the lower troposphere in this area has two main components: sea salt aerosol and mineral aerosol (Bergametti et al., 1989; Correggiari et al., 1989; Chester et al., 1993a).
These aerosols are removed from the atmosphere to the sea surface by a combination of “dry” (i.e., not involving aqueous phase) and “wet” (precipitation scavenging) depositional modes, and these modes are geochemically different with respect to the solutions with which the aerosols come into contact. Estimates of dust transport and deposition in the Mediterranean have been made by the Group of Experts on the Scientific Aspects of Marine Pollution (GESAMP, 1989) using a model. The temporal variability of dust deposition rates may vary greatly because of the strong seasonal cycle in dust concentrations and the large seasonal variability of precipitation rates (Prospero, 1996).
The present study is based on a collection of 6-month aerosol samples and one-year precipitation samples at a remote site. An overall sampling program was applied to investigate the geochemistry of natural (Saharan dust) and anthropogenic (background) aerosols and to separate local versus distant inputs to the Western Mediterranean (Guerzoni et al., 1997). If and how crust-enriched precipitation is similar to desert-derived aerosol was also verified.
Two different ways of treating the data sets were planned, as follows:
- 1.
a rank transformation was applied to the data and subsequently CA on the ranked data, followed by DA on the CA groups;
- 2.
the data matrix was first reduced on the basis of Q-FA on the ranked data; then CA was applied to samples, with factor loading as new variables, and then DA on the CA groups.
The aims of this paper were to present the geochemical composition of insoluble aerosol and suspended particulate in precipitation and to show two different multivariate statistical approaches to the data sets; and then to translate statistical results into meaningful geological statements. The contributions of local versus distant sources were particularly highlighted.
Section snippets
Sampling
Chemical data of the insoluble fraction of aerosols collected at a land-based coastal station in SE Sardinia, Western Mediterranean (9° long E, 39° lat N) (Fig. 1) are presented, together with measurements on mineral particulate deposited with precipitation. A total of 86 samples (55 aerosol; 31 rain) were collected during the period October 1990/October 1991, and 69 of these (55 aerosol; 14 rain) are presented here. The remaining 17 rain samples were not analysed due to the low concentration
Measurements in air and rain
Table 1 lists arithmetic and geometric means for insoluble fractions in aerosol and rain. The overall means are 6 μg m−3 and 26 mg l−1, with significant differences between samples collected from winds from the south, called Saharan, and samples from all other directions, called Background. This classification was also carried out prior to chemical analysis, using a suite of meteorological data, together with back-trajectories. Table 2 shows the wind component of particle concentrations in
Conclusions
Concurrent sampling of aerosol and precipitation is useful in comparing contents of the insoluble fraction of air particulate.
The statistical methods used in this study are well known, and commonly used in situations in which the interplay of many sources of variation do not allow an objective and comprehensive method of describing physical phenomena. Divergent results are obtained by applying two different combinations of classification and statistical methods to geochemical data.
In the first
Acknowledgements
This paper is IGM-CNR No. 1083. We thank F. Visin and G. Quarantotto for chemistry work, and G. Walton for revision of the English text. We also thank Prof. P. Mantovan and Dr A. Pastore for assistance in the review of the statistical procedures.
References (23)
- et al.
Factors controlling the solubilities of trace metals from non-remote aerosols deposited to the sea surface by the “dry” deposition mode
Marine Chem.
(1993) - et al.
Saharan dust inputs to the W Mediterranean Sea: depositional patterns, geochemistry and sedimentological implications
Deep-Sea Res. II
(1997) Target transformation factor analysis as an aerosol mass apportionment method: a review and sensitivity study
Atmos. Environ.
(1988)- et al.
Ordinal principal component analysis. Theory and application
Computational Statistics and Data Analysis
(1998) - et al.
Development of a modified factor analysis/multiple regression model to apportion suspended particulate matter in a complex urban airshed
Atmos. Environ.
(1987) - Bergametti, G., Goms, L., Remoudaki, E., Desbois, M., Martin, D. and Buat-Menard, P., 1989. Present-day transport and...
- Bücher, A., 1989. Fallout of Saharan dust in the northern Mediterranean region. In: Leinen, M., Sarnthein, M. (Eds.),...
- et al.
Defining the chemical character of aerosols from the atmosphere of the Mediterranean Sea and surrounding regions
Oceanol. Acta
(1993) - et al.
Dust deposition in the Central Mediterranean (Tyrrhenian and Adriatic Seas): relationship with marine sediments and riverine input
Terra Nova
(1989) - GESAMP, 1989. The atmospheric input of trace species to the world oceans. Reports and Studies...
Fluxes of soluble and insoluble metals and nutrients from the atmosphere to the Central Mediterranean Sea
Water Poll. Res. Rep.
Cited by (18)
Chemistry of dry and wet atmospheric deposition over the Balearic Islands, NW Mediterranean: Source apportionment and African dust areas
2020, Science of the Total EnvironmentCitation Excerpt :The second case study verses on the deposition of nutrients (N-NH4+sol, Pins, K+sol, Feins) (Claustre et al., 2002; Desboeufs et al., 2018; Guerzoni et al., 1999; Guieu et al., 2010), and secondary aerosol components with agricultural interest (N-NO3−sol, S-SO42−sol) (Pieri et al., 2010), throughout the campaign, with especial attention to the main peak occurred in January–February 2012. Prior studies have investigated the mineralogy and the elemental ratios of atmospheric dust to trace African dust source regions (Desboeufs et al., 2018; Gelado-Caballero et al., 2012; Marconi et al., 2014; Meskhidze et al., 2005; Morales-Baquero et al., 2013; Moreno et al., 2006; Scheuvens et al., 2013; Stuut et al., 2009; Marconi et al., 2014; Molinaroli et al., 1999; Moreno et al., 2006). The following ratios are very useful to split those contributions from other crustal sources: Catotal/Al, K/Al, Mg/Al, Fe/Al, Na/Al, P/Al, (Catotal + Mg)/Fe.
Passive monitoring of particulate matter and gaseous pollutants in Fogo Island, Cape Verde
2018, Atmospheric ResearchCitation Excerpt :The relative order of concentrations of elements with crustal origin in settled dust samples was Si > Al > Ca > Na > Mg > K > Fe > Ti > P. Among these, Si was always present, regardless of the sample, although in varying concentrations. These elements were selected since they are used as fingerprints of the Saharan/Sahel dusts by several authors (e.g. Chiapello et al., 1997; Formenti et al., 2001; Molinaroli et al., 1999; Scheuvens et al., 2013). Pearson correlations disclosed significant results at the 0.05 level for the sets S/Mg (0.556), S/P (0.452), Fe/Ti (0.509) and at the 0.01 level for the pairs Si/Ca (0.580), Fe/K (0.592), Fe/Na (0.592), K/Na (1.000).
Damage monitoring on carbonate stones: Field exposure tests contributing to pollution impact evaluation in two Italian sites
2017, Construction and Building MaterialsBulk composition of northern African dust and its source sediments - A compilation
2013, Earth-Science ReviewsApplication of discriminant analysis with clustered data to determine anthropogenic metals contamination
2009, Science of the Total EnvironmentMultidisciplinary characterization of the middle Holocene eolian deposits of the Elsa River basin (central Italy)
2009, Quaternary InternationalCitation Excerpt :Prodi and Fea (1979) found particle sizes of 0.2–1 μm in recent depositions of Saharan dust in Sardinia. In addition, although the geochemistry of Saharan dust can be rather variable, our geochemical data differ significantly from those reported for Sahara dust (Chester et al., 1984; Orange et al., 1993; Molinaroli et al., 1999; Guerzoni et al., 1999; Moreno et al., 2006); especially the average metal content is larger in our samples. Another possible source of eolian components in soils is volcanic ash.