Development of a Hybrid Machine Learning-Based Framework for Monitoring and Predicting Odor Pollution in Urban Surface Water Canals: A Case Study of the Ghiyasvand Canal, Tehran

Articles in Press

Document Type : Article

Authors

1 Department of Civil Engineering, Sharif University of Technology, Tehran, Iran.

2 Water Resources Management, Group, Department of Civil Engineering, Sharif University of Technology.

3 Environmental Engineering Group, Department of Civil Engineering, Sharif University of Technology, Tehran, Iran.

10.24200/j30.2025.66941.3433

Abstract

Unpleasant odors emitted from urban surface‐runoff canals pose a significant environmental and public health challenge in large cities such as Tehran. These odors, primarily caused by anaerobic decomposition of organic matter and the inflow of untreated wastewater, degrade water quality, generate public dissatisfaction, and reduce the environmental health and livability of urban areas. Despite their widespread impact, the spatial distribution and intensity of odor emissions in such canals are rarely monitored systematically. To fill this gap, this study introduces a practical framework for monitoring, quantifying, and predicting odor intensity in urban surface water canals. The proposed framework was implemented and evaluated in a pilot project along the Ghiasvand Canal in Tehran. Weekly field sampling was carried out at ten critical locations over a ten‐week period, during which key water‐quality parameters (including pH, electrical conductivity (EC), total dissolved solids (TDS, and dissolved oxygen (DO), and water temperature) and meteorological variables (such as air temperature, wind speed, and relative humidity) were collected. In addition, odor intensity was measured with a portable Odor meter. Pearson, Spearman, and Kendall correlation analyses, along with a random forest regression model, were employed to examine and predict the relationships between physicochemical and atmospheric variables and the odor intensity. Correlation analyses indicated that water temperature, electrical conductivity, and air temperature were positively correlated with odor intensity, whereas DO showed a negative correlation, indicating its critical role in odor suppression. The developed model performed well in predicting odor intensity, achieving an accuracy of 83% for both training and testing data. This study demonstrates the potential of integrated field monitoring and machine learning approaches to support practical odor management in urban water systems, leading to improved environmental quality and public well-being. While the framework was applied to a specific case in Tehran, the results and approach are broadly applicable to similar urban settings facing odor-related challenges.

Keywords

Main Subjects

Sharif Journal of Civil Engineering

Articles in Press, Accepted Manuscript
Available Online from 22 October 2025

Files

Share

How to cite

Statistics