Lead levels in the Zagreb air

This paper presents the results of one-year (1999) lead measurements in samples of suspended particulate matter collected at five sites in the city of Zagreb, Croatia (Center, North, West, Industrial suburb I and Industrial suburb II). Samples were analyzed using flame atomic absorption spectrometry (Welz, 1986). The highest lead concentrations were recorded in the city center because of dense traffic. Our aim was to investigate the characteristic patterns of annual variation in lead concentrations in the city. In addition to standard descriptive statistics, the data were analyzed using the functional principal components analysis (Ramsey and Silverman, 1997). A smooth function was fitted to data on lead concentrations for each site. Typical temporal variation in lead concentrations showed high winter and low summer concentrations with a lower peak in spring. The spring peak varied in intensity and shifted along the time axis among the sites. Functions were represented in the Fourier basis with 7 principal basis functions. The first three principal component (PCA) functions accounted for 99.5% of variability (Fig 1). The solid curve represents mean concentration and + (-) represents the effect of adding (subtracting) a principal component. The first PCA function accounted for 84.2% of variability and corresponded to the overall level of lead. The greatest variability between the sites was found by heavily weighting spring concentrations. Sites with positive score on this function had higher concentrations of lead throughout the year and the higher spring peak. The Center's high score on the first PCA function indicated its high overall lead concentrations. The second PCA function accounted for 10.7% of variability. The West's score was the most negative on the second PCA function, indicating a time shift to the left (i.e. the beginning of the year). The third PCA function accounted for 4.6% of variability. Sites with a positive score on this function had larger difference between the summer and winter concentrations. Both industrial suburb sites had negative scores on this component, while the North and the West scored positive. The three dominant modes of annual variation appear to be the overall lead concentration level, the time shift, and the summer/winter range. References 1. Ramsay J.O. and Silverman B.W. (1997) Functional Data Analysis, Springer-Verlag Inc., New York. 2. Welz B. (1986) Atomic Absorption Spectrometry, John Whiley & Sons, New York.