Análises seqüenciais de agrupamentos aplicadas ao estudo de partículas atmosféricas

Química Nova

Cláudia Flores Braga²  Elba Calesso Teixeira²  Daniela Migliavacca²  Fabiana G. Carvalho²  Jandyra Fachel¹ 

[1] ,Fundação Estadual de Proteção AmbientalPorto Alegre RS

关键词: atmospheric particles;    cluster analysis;    urban air pollution;   

DOI  :  10.1590/S0100-40422002000600004

来源: SciELO

【 摘 要 】

The present paper has as objective to apply a sequential Cluster Analysis to the atmospheric particles: Hierarchical Cluster Analysis followed by Nonhierarchical Cluster Analysis. The hierarchical cluster analysis results were used as start point for the nonhierarchical cluster analysis as an agglomerative technique. These particles were taken from two areas of the metropolitan region of Porto Alegre, Charqueadas and Sapucaia do Sul., from may /97 to may/98, using a High Volume Sampler (Hi-Vol). Around 10,000 particles were analysed by Scanning Electron Microscope with Energy-Dispersive X-Ray microanalysis (SEM-EDS). The Hierarchical Cluster Analysis allowed the identification of five groups of particles, whose amounts were differentiated according to the summer and the winter campaigns. The abundance of each type of particles inside each group according to the different sections was verified by the Nonhierarchical Cluster Analysis, resulting in information about the emissions sources. The groups of particles of Si/Al and Si and of Fe/Zn and Fe for Charqueadas were more significant in section 2 and 3 (NW and W wind directions) and in section 1 (SE wind direction), evidencing the influence of the coal power plant and steel industry, respectively located in these quadrants. In Sapucaia do Sul the data were more heterogeneous, causing a certain difficulty to identify the source as anthropogenic. Nevertheless the group of particles containing Fe was found in sectors of NW/W wind directions which shows the influence of the steel plant.

【 授权许可】

CC BY-NC   
 All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License