Multi-way data analysis of soils irrigated with wastewater - A case study

Data set of soils irrigated with wastewater was analyzed using two- and three-way PCA models to assess the soil/sub-soil contamination, and mobility of contaminants in soil profiles. Although two-way PCA provided information about association between wastewater-related contaminants and their accumulation in upper soil layers, due to three-way data structure, it could not extract information about general behaviour patterns for the sampling sites. Three-way data analysis was performed using PARAFAC and Tucker3 models validated for stability and goodness of fit. A two-component PARAFAC model, explaining 36.23% of data variance, allowed interpretation of the data information. The first component was related with general characteristics of soils with little vertical variation, whereas the second component is related with the sites and depth contaminated with metals. The interpretation of core elements (Tucker3) revealing interactions among components of different modes allowed inferring more realistic information about the contamination pattern of soils both along the horizontal and vertical coordinates as compared to traditional data analysis techniques. It concluded that soils at sites closer to the wastewater outlet point are more heavily contaminated with metals. The study further concluded that wastewater-related heavy metals confined to upper soil profiles are more accessible to the crop/vegetation plants uptake grown in the region and their hyper-accumulation in edible parts of crops and fodder vegetation may pose risk to the consumers. (c) 2006 Elsevier B.V. All rights reserved.