عنوان مقاله [English]
Canonical correlation analysis to determine the relationship between water quality parameters and heavy metals in water samples (Ramian City-Golestan Province)
Background and Objectives: Water is essential for all forms of life and its pollution is generally considered more critical than soil or air. Distinguishing and correlating among physical and chemical parameters and finding source of water contamination are major issues in monitoring water quality. Application of statistical methods in water quality monitoring can be useful to achieve this goal. Heavy metals are important water pollutants especially due to their bioaccumulation nature and toxicity. Anthropogenic and natural sources are the main entrance ways of heavy metals into water.
Materials and Methods: Water quality assessment with multivariate statistical methods is the main objective of this study. Statistical Principal Component Analysis (PCA) method was used to determine the major water quality parameters and Canonical Correlation Analysis (CCA) was employed to find the relationship between water quality parameters and heavy metal contents. Water samples from twenty three wells in Ramiyan district (Golestan province) were collected in 2012. Water quality parameters, such as; temperature (T), electrical conductivity (EC), pH, Ammonia (NH3), Fluoride (F-), Sulfate (SO42-), Nitrate (NO3-), Nitrite (NO2-), Phosphate (PO43-), Bicarbonate (HCO3-) and heavy metal contents including Zn, Cd, Pb, Cu, Ni and Co were used in this research. Heavy metal contents in water samples were determined by voltammetry method in Environmental Science Research Laboratory of University of Zanjan. Temperature, electrical conductivity in the samples was measured by a portable device in sampling sites.
Results: Principle Component Analysis identified six factors explaining 79.67% of total variance affecting water quality in the studied area. CCA also identified three canonic classes with correlation coefficients; 0.973, 0.795 and 0.624, respectively, suggesting that predictor variables (EC, HCO3-, NO2-, NO3-, pH & PO43-) and response variables (Zn, Co, Cd & Ni), were highly scored among all parameters in water quality assessment.
Conclusion: The results of canonical correlation analysis also show a significant correlation between the two categories of variables. The studied metals can be selected based on the main human activities in the area (agricultural activities in this region). The information obtained can be used to improve water quality monitoring and management program. By measuring relevant parameters, there is no need to measure all the physical and chemical properties of water, reducing considerable coasts of water analysis.
Keywords: Water quality, Principle component analysis, Multivariate analysis, Canonical correlation analysis