Spatial variability of water quality in the upstream Bhadra river, tributary of river Tungabhadra, in the Western Ghats of India: Application of multivariate statistical techniques

Abstract

Water quality of river water examined at 23 stations in the upstream of Bhadra River basin from Kudrekmukh National Park (KNP) to Bhadra Wildlife Sanctuary (BWS) in India. Spatial cluster analysis performed on 23 sampling stations revealed three clusters depending on the similarities in the water quality variables that could delineate the number of sampling stations required for optimal sampling. Sample cluster analysis classified the 24 environmental variables in the data set into seven clusters depending on the similarities in the water quality variables. Principal component analysis of water produced 7 components, which accounted for 88.15 % of total variance. Factor analysis was performed on principal components extracted seven variance factors (VFs) after rotation with all VFs having eigenvectors of high (> 0.70) and moderate (> 0.50) loadings. The VF 1 accounted for 37.93 % of total variance with 9 eigenvectors (loading > 0.70) such as conductivity, total alkalinity, total hardness, Ca, Mg, Na, K, chloride and silicate. On total eigenvectors generated on 24 water quality data sets of river water were classified into four types <em>viz</em>. safe (normal), low polluted, medium polluted and highly polluted waters with corresponding total eigenvectors of < 0.50, > 0.50 - < 1.00, > 1.00 - < 1.50 and > 1.50, respectively. Total factor score produced for 23 sampling stations revealed a total of five types of sampling stations with safe, low, medium, high and very high levels of pollution with corresponding total factor scores of less than zero (< 0.0), > 0.0 - < 1.50), (> 1.50 - < 3.00), (> 3.00 - < 4.50) and (> 4.50), respectively.