科研之友
微信
新浪微博
Facebook
分享链接
摘要
Multivariate statistical techniques including cluster analysis and principal components analysis were applied on 22 variables consisted of 3 physicochemical parameters, 8 major ions and 11 trace elements. Samples were collected from the south Rhodope multilayered coastal aquifer in north Greece which is facing saltwater intrusion and anthropogenic contamination over the last 35 years. Cluster analysis grouped the variables into five main groups while principal components analysis revealed four distinct hydrochemical processes in the aquifer system, explaining 84.5 % of the total variance between the variables. The identified processes correspond to, saltwater intrusion and subsequent reverse cation exchange, the presence of deep connate groundwater masses, application of fertilizers in shallow wells and anthropogenic contamination with heavy metals nearby an improperly constructed landfill. The wells categorized with the above techniques were grouped and five constituent ratios Na/ Cl, (Mg + Ca)/Cl, Ca/(HCO3 + SO4), Ca/SO4 and Ca/Mg were utilized to identify the ones which enable the more accurate distinction between the group cases. The results of stepwise discriminant analysis showed that the calculated classification function can distinguish almost 80 % of groundwater samples with the Na/ Cl ratio being the most statistically significant grouping variable. All the aforementioned statistical models managed to successfully identify numerous hydrochemical processes in a complex multilayered aquifer system and to explicitly attribute them for every investigated well, allowing a deeper insight into groundwater chemical characteristics with the use of an optimized smaller number of variables.
- 出版日期2014-11
