Geochemical characterization of an abandoned mine site: A combined positive matrix factorization and GIS approach compared with principal component analysis

详细信息	查看全文 | 推荐本文 |

• 作者：Sara Comero^a ; ^b ; ^{sara.comero@jrc.ec.europa.eu} ; Diego Servida^a ; Luisa De Capitani^a ; Bernd Manfred Gawlik^b
• 关键词：Abandoned mine ; Principal component analysis ; Positive matrix factorization ; Kriging ; Gromo (Lombardy ; Italy)
• 刊名：Journal of Geochemical Exploration
• 出版年：2012
• 期刊代码：44_03756742
• 类别：ear
• 出版时间：July, 2012
• 卷：118
• 期：Complete
• 页码：30-37
• 文件大小：1039 K

摘要

Statistical methods are increasingly used for geochemical characterization of contaminated sites. The geochemical characteristics of the abandoned Coren del Cuc矛 mine dump (Upper Val Seriana, Italy) were modelled by principal component analysis (PCA) and positive matrix factorization (PMF) of 56 soil samples analyzed for 11 elements and pH. PCA and PMF were used to investigate how different approaches deal with the preset type of data. PCA was performed on two data subsets鈥攕amples inside and outside the dump鈥攔ecognized by cluster analysis. PMF was performed on the whole data set. However, a GIS-based approach was combined with PMF for better factor resolution. Three main principal components (PCs) were identified inside the dump: (i) the local ore mineralization; (ii) the background/regional metal content of rocks; and (iii) the variability of Cd. Two main PCs were obtained outside the dump: (i) the background/regional metal content of rocks; and (ii) the local ore elements. Five factors were determined by PMF: (i) two background geo-morphological characteristics of the area outside the dump; (ii) a source of mineralization situated inside the waste disposal area; and (iii) two different geochemical anomaly zones. PMF was found to be useful for estimating the number and composition of sources or processes that govern data characterized by heterogeneous behavior. In contrast to the application of PCA, no data pre-treatments procedures are needed to apply PMF.