Statistical Evaluation of Influence of Hydrochemical Parameters on Public Water Supply in Ibadan, Nigeria

Abstract

The recent trends in the global socio-economic and technological developments have exerted undue pressure on water quality and supply. A robust model that predicts water supply from hydrochemical parameters could be useful in water resources management. Most of the existing models used bivariate analysis or simple multivariate analysis that treats each variable separately, but combined parameters could be better at capturing many related water properties and hence have stronger predictive value. In this work, exploratory factor analysis was used to investigate the relationships between fourteen measured hydrochemical parameters and the volume of water pumped for public water supply from Asejire waterworks in Ibadan. Representative monthly samples of raw water were collected and analyzed for their physical and chemical properties. Principal Component Analysis (PCA) was used to derive five novel latent variables which represent combined properties of water. The results showed that, with the exception of colour, no other single hydrochemical parameter was significantly associated with the amount of water pumped for public supply. Interestingly, one of the novel latent variables ‘Opaque Index’ was significantly associated with water supply. Additionally, though colour as a single parameter had a predictive value, but it explained only 6.7 % of the variation in the amount of water pumped, whereas the Opaque Index explained 25.5 % of the variation in the amount of water pumped. This therefore suggests that the Opaque Index latent variable has a stronger predictive power than any of the individual measured parameters and hence could a useful metric for effective water resources management.

Keywords: Hydrochemical, Latent variable, Prediction, Structural equation modelling