Water Quality Modelling of River Mahanadi using Principal Component Analysis (PCA) and Multiple Linear Regression (MLR)

Abstract

Surface water quality is one of the critical environmental concerns of the globe and water quality management is top priority worldwide. In India, River water quality has considerably deteriorated over the years and there is an urgent need for improving the surface water quality. The present study aims at use of multivariate statistical approaches for interpretation of water quality data of Mahanadi River in India. Monthly water quality data pertaining to 16 parameters collected from 12 sampling locations on the river by Central Water Commission (CWC) and Central Pollution Control Board (CPCB) is used for the study. Cluster analysis (CA), is used to group the sampling locations on the river into homogeneous clusters with similar behaviour. Principal component analysis (PCA) is quite effective in identifying the critical parameters for describing the water quality of the river in dry and monsoon seasons. PCA and Factor Analysis (FA) was effective in explaining 69 and 66% of the total cumulative variance in the water quality if dry and wet seasons respectively. Industrial and domestic wastewaters, soil erosion and weathering, soil leaching organic pollution and natural pollution were identified as critical sources contribution to pollution of river water. However, the quantitative contributions were variable based on the season. Results of multiple linear regression (MLR) are effective in explaining the factor loadings and source contributions for most water quality parameters. The study results indicate suitability of multivariate statistical approaches to design and plan sampling and sampling programs for controlling water quality management programs in river basins.

Keywords

• water pollution
• multivariate analysis
• cluster analysis
• factor analysis
• receptor modelling