Principal Component Analysis for Water Quality Assessment of the Ganga River in Uttar Pradesh, India

Аннотация

The present study is an attempt to apply principal component analysis (PCA) for spatial assessment of water quality parameters that are responsible for water quality deterioration in the Ganga river at four cities of Uttar Pradesh. 48 water samples of the Ganga river from Allahabad, Mirzapur, Shahzadpur and Varanasi in Uttar Pradesh were collected during January to December in 2013. Data were analyzed for assessment of 18 different physicochemical and biological water quality parameters. These variables were examined using PCA to define and standardize the parameters mainly responsible for water quality variance in Ganga river at four selected cities. PCA highlighted anthropogenic effect and industrial effect as two main significant components which explain more than 99.32% of the variance, accounting for 64.47 and 34.85% respectively of the total variance of Ganga water quality at the four cities. Results revealed that total dissolved solids, total alkalinity, total hardness were the important parameters in assessing variations in Ganga water quality during October to April (post-monsoon months) and turbidity, suspended solids are the important parameters in assessing variations in Ganga water quality during June to September (monsoon months). Ca²⁺, Cl^–, ${\text{SO}}_4^{2-}$, temperature, fluoride, pH, Fe, Cl^–, were found to be non-principal water quality parameters. Principal component analysis produced three significant main components explaining more than 82.9% of the variance (anthropogenic and industrial effect) that present 57.1, 13.8 and 12% respectively of the total variance of water quality in Ganga river at the four selected cities. The result reveals that turbidity, dissolved oxygen and biochemical oxygen demand are the parameters that are most important in assessing variations of water quality. Water quality index based on eight parameters (turbidity, DO, BOD, COD, pH, TS, TSS and TDS) calculated for all four cities were found to range from medium to bad. Thus, this study illustrates the usefulness of multivariate statistical techniques for analysis, interpretation of complex data sets and understanding spatial variations in water quality for effective river water quality management.