Principal component analysis incorporated water quality index modeling for Dhaka-based rivers

Abstract

Principal component analysis (PCA) can reduce the subjectivity of Water quality index (WQI) models by reducing parametric dimension and has gained immense attention in exploring water quality among researchers. Therefore, this study focuses on developing a novel WQI model for 4 Dhaka-based rivers namely Buriganga, Turag, Balu, and Shitalakhya following PCA as a method for selecting and weighting water quality parameters. The dataset includes 12 water quality parameters from 19 sites of these rivers sourced from the Department of Environment (DoE), Bangladesh. Correlation analysis followed by PCA, was conducted to decrease the parameter count from 12 to 7. The Measure of Sampling Adequacy (MSA) was found to be 0.853 in the Kaiser-Meyer-Olkin (KMO) test and Bartlett’s test of sphericity was significant at an alpha level of 0.05 indicating the dataset was suitable for factor analysis. Sub-indexing was introduced with the quality rating curves provided by the National Sanitation Foundation (NSF)-WQI model and modified rating curves for specific parameters with statistical dispersion. The calculated WQI values for 209 samples ranged from 36 (Bad) to 82 (Good) on a scale of 100. More than 70 % of the samples were in the medium or bad, and the rest were in the good category. The trend in WQI across the rivers indicated higher values during the wet season, attributed to the dilation from local rainfall. By incorporating a well-distributed dataset spanning several years, this statistical approach effectively minimizes the subjectivity and bias in developing WQI models for rivers in Dhaka, contributing to more robust future model development. Moreover, this study introduces a modern approach for assessing the river water quality of Dhaka city that can be incorporated into the river pollution control strategies.