We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Integrated assessment of urban pond water quality using multivariate statistics and explainable machine learning in Chattogram, Bangladesh
Summary
Researchers tested nearly 500 urban pond samples across a fast-growing Bangladeshi city and found that most ponds, used by communities for daily water needs, had poor water quality, with widespread heavy metal contamination, murky water, and microplastics found in every single neighborhood tested. Even though arsenic levels technically met legal limits, long-term exposure calculations suggest people drinking this water face real health risks, including increased cancer risk, highlighting why cities need better monitoring of these overlooked local water sources as urban populations keep growing.
In fast-growing cities, urban ponds are significant sources of potable and/or community water, but their water quality is seldom evaluated at the extensive spatial level. To the best of our knowledge, this study is among the first integrated assessments of urban pond water quality at the ward scale across all 41 administrative wards of Chittagong City Corporation (CCC) in Bangladesh. During post monsoon (November-December 2025), 492 water samples were collected from the ponds and tested for physicochemical parameters and heavy metals, while 410 samples were separately processed for microplastic analysis. A Water Quality Index (WQI), Heavy Metal Pollution Index (HPI), human health risk assessment, Principal Component Analysis (PCA), Hierarchical Cluster Analysis (HCA), and ten supervised machine learning models were combined to characterize water quality and determine dominant water quality drivers. The ward average WQI ranged from 78.33 to 166.11 (mean: 115.63), with 85.4% of wards being classified as Poor, signifying that the overall quality of water is deteriorating with the major problem being turbidity, nutrient enrichment and fecal contamination. All the wards had HPI value above the critical threshold (HPI>100), the highest heavy metal contamination was near the industrial and port adjacent area. Although the measured arsenic concentrations remained below the ECR 2023 permissible limit, the conservative lifetime oral-exposure assessment indicated elevated non-carcinogenic and carcinogenic risks, particularly due to arsenic. Microplastics were found in all wards and the most common polymers found were PET, PP, HDPE and LDPE. A significant amount of spatial heterogeneity in the city was identified by PCA and HCA which showed distinct environmental gradients and four clusters of pollution. ElasticNet performed best in predicting the WQI (R² = 0.8425, RMSE = 31.87, MAPE = 18.87%), and among the evaluated machine learning models, turbidity was the most significant factor influencing the variability of WQI was identified by the SHAP analysis. The integrated framework developed in this study could serve as a solid foundation for the monitoring, cluster-based remediation and risk-based implementation of the Environment Conservation Rules (ECR 2023), and can be transferred to other rapidly urbanizing areas for the assessment of urban freshwater systems.