Multivariate statistics applied to water quality in different hydrographic microbase environments

Abstract

Studies on the water quality behavior are important, so the objective of this work was to group the studied waters regarding the similarity and to select the physical-chemical characteristics to explain the variability of water quality in four micro-basins. Four micro-basins with different soil uses were selected: pasture, forest regeneration, forest and coffee; under different environments: lentic and lotic environments, springs and groundwater. The collections occurred between February 2014 and December 2014, being analyzed: total coliforms and thermotolerant; dissolved oxygen (OD); total nitrogen (Nt); PO43-; turbidity; temperature; pH; Biochemical Oxygen Demand (BOD); electrical conductivity (EC); total solids (TS); dissolved solids (SD); suspended solids (SS); and the metals calcium, magnesium and iron. Multivariate statistical analysis techniques were used, through cluster analysis (AA) and principal component analysis (PCA). In AA, four distinct groups were formed in the rainy season and three in the dry season. The difference between the environments was the main factor of influence in the segregation of the groups. From the PCA, 4 main components were selected, which explained 73.09% of the total data variance. The selected variables were CE, turbidity, magnesium, iron, SD, Nt, BOD, pH and thermotolerant coliforms.