Identification of groundwater pollution sources based on multivariate statistical approach

Comprehensive and joint applications of factor analysis and multivariate linear regression were carried out for identification of groundwater pollution sources. In the factor analysis, four factors were extracted based on eigenvalue (larger than 1), which represented four potential pollution sources, such as mineralization effect, fecal pollution, mineral dissolution and natural pollution. The factor scores of each sampling site were analyzed using the inverse distance weighting method, and the results show that the influencing degree by various pollution sources differed among the sampling sites. The results of multivariate linear regression based on factor scores show that the contribution rates of the 4 factors on water quality were 42.63%, 29.23%, 22.40% and 5.74%. This study indicates that the multivariate approach is useful and effective for identification of groundwater pollution sources.