Characterization and Classification of Hydrochemistry using Multivariate Graphical and Hydrostatistical Techniques

Kumar Manoj, Ghosh

2013 · 15 citations · 6 references

Concepts

Abstract

Multivariate data analyses systems provide simultaneous inspection of several variables both in space and time. This study was undertaken to present the utility of multivariate data analyses models in characterization and classification of surface water chemistry. A case study was developed, where hydrochemistry of some water bodies (ponds) in the SantiniketanBolpur-Sriniketan zone, Birbhum district, West Bengal, India, was investigated followed by the assessment of impact of the well known annual fair Pous Mela on the chemistry of surface waters present near the vicinity of the fair ground. Santiniketan has historical importance in India and the world, made famous by versatile Nobel Laureate Gurudev Rabindra Nath Tagore. The Pous Mela, which was started by Maharshi Debendranath Tagore in 1894, is an annual fair organized on the campus of Visva-Bharati University in the month of December and is visited by thousands of people every year. Water samples collected, both spatially and temporally, were analyzed for sixteen parameters and multivariate Piper trilinear diagram was constructed for examination of their chemistry. Two multivariate data analyses techniques viz. agglomerative hierarchical cluster analysis (AHCA) and discriminant analysis (DA) were further applied for intelligent interpretation of water quality data matrix. Piper diagram classified all water samples into ‘Mixed Ca 2+ —Na + —HCO3 Type’. No change in water type was recorded temporally suggesting the ionic stability of water bodies with respect to Na + , K + , Ca 2+ , Mg 2+ , HCO3 , Cl and SO4 2and further indicating the impact of the Pous Mela on major ion chemistry of nearby waters as negligible. However, hydrochemical data analyses revealed organic wastes along with PO4 3as the most serious pollutants. AHCA grouped sampling sites into three groups viz. having less anthropogenic influence, medium anthropogenic influence and high anthropogenic influence. Application of DA on data matrix uncovered four predictor variables namely BOD, pH, Cl and total hardness as the most relevant discriminating parameters between three groups. This study showed that the multivariate models are highly effective in elucidating and illustrating the hydrochemical profile and framework assessment which can be applied on large scale for better interpretation of water chemistry and management of water resources of any

References

6