2025/04/06 by María Alejandra Pimiento, José Anta, Andrés Torres · 1 voice
Environmental Science · #Hydrological Forecasting Using AI #Hydrology and Watershed Management Studies #Fish Ecology and Management Studies
paper · doi:10.1016/j.hazadv.2025.100708
openalex publication_date 2025/04/06 · openalex created_date 2025/10/10 · openalex updated_date 2026/06/11
• Method for classifying variables closely associated with heavy metal risk indices. • Methodology for modeling risk indices of heavy metals in urban runoff sediments. • A SVM model was tested for the Pb enrichment factor for an urban river in Bogotá. Despite the growing application of machine learning (ML) in water quality assessment and pollution source identification, its potential for predicting environmental risk indices in urban stormwater sediments remains largely unexplored. Conventional models struggle to capture complex interactions among hydrological variables, sediments and pollution parameters. This study uses ML techniques to enhance sediment quality assessment to address this gap. The case study focuses on sediments from the Molinos River in Bogotá, Colombia, characterized by particle size distribution (PSD), heavy metal (HM) concentrations, and environmental risk indices. Cohen's Kappa coefficient was used to evaluate the relationship between the enrichment factor (EF) of Ni and Pb, PSD, and hydrological variables as rainfall data. A support vector machine model using an ANOVA kernel, validated through multiple calibration and validation datasets, demonstrated the feasibility of predicting sediment-related risks in urban drainage systems. The best model successfully predicted Pb EF levels for 7 of 8 samples, achieving a Cohen's Kappa coefficient of 0.71 ( p = 0.037), indicating substantial agreement. These findings highlight the potential of ML models to predict sediment EF using rainfall data, providing a practical tool for environmental risk assessment. By enabling predictions of contamination levels, this methodology enhances decision-making and promotes more sustainable urban water management strategies.