Conference Agenda

Rivers provide many ecosystem services. However, in densely populated urban areas, they are polluted by multiple sources including combined sewer overflows (CSOs). By studying the behavior of the combined sewer system, whose capacity can be saturated during heavy rainfall episodes, it is possible to anticipate these contamination periods. In addition, by using an inverse method (quantifying causes from effects), it would be possible to determine the quantity of contaminants released by the CSO. In this paper, we present the design of the method on the study site of the Seine River, in Paris. The spatial extension of the combined sewer network is described by a directed graph. Temporal consistency is achieved through event definition. The use of these data for CSO assessment would make it possible to anticipate contamination and guide decisions to contain it.

Characterization of sources of microbiological contamination in the Seine River during the Paris Olympics 2024 using inverse modeling

Urban runoff is a pathway of particles and particle-bound contaminants (PBCs) towards urban water bodies. These contaminants are emitted from urban wet weather discharges (i.e., combined and separate sewer networks) during storm events, creating health risks for humans, ecosystems and degrading water quality downstream. Due to the spatial and temporal variability of PBCs associated with different urban sources, accurately quantifying, characterising and identifying their transport pathways remains a major challenge. Modelling is a useful tool to study the dynamic behaviour of urban drainage systems. Hence, to understand the PBCs pathways at the interface between sewer systems and streams, we have established an integrated monitoring network in a small stream located Dresden, Germany. We further developed an integrated model that incorporates both combined and separate sewer network discharges to evaluate their impact on the stream water quality. By integrating long-term, high-resolution monitoring with catchment-scale modelling, this study offers new insights into the export and dynamics of TSS and PBCs transport in urbanised catchments. Our findings highlight challenges and opportunities related to modelling PBCs with an integrated approach, showing that embracing the complexity of real-world settings is necessary to support the development of management alternatives to reduce the export of pollutants.

Modelling pollutant dynamics in Combined Sewer Overflows (CSOs) remains challenging due to complex transport processes and limited high-resolution data. Total Suspended Solids (TSS), key water quality indicator, are influenced by localized hydraulic conditions and sediment characteristics. This study presents the LSR model as a solution to the limitations of existing models, which often assume uniform spatial heterogeneity and particle size distributions. By integrating spatial heterogeneity, localized dynamics, and inter-event heterogeneity, the LSR model aims to enhance the accuracy of TSS predictions. Events with resuspension potential were analyzed using three structures of increasing model heterogeneity. Increasing model heterogeneity improves accuracy, but its benefits depend on data quality. The best performance was achieved using hotspot-specific resuspension with well-calibrated hydraulics and uniform sediment characteristics due to less accurate TSS calibration. Additionally, three hotspot contribution approaches were tested. The results showed that uniform activation imposes artificial uniformity and oversimplifies real spatial dynamics, limiting the model’s ability to capture true resuspension behavior. In contrast, selective hotspot activation led to notable improvements. These findings underscore the importance of accounting for spatial and inter-event heterogeneity in improving TSS predictions and highlight the potential of the LSR model to capture these dynamics and support urban drainage management.

Biocides are commonly used in facade coating materials; however, their release into stormwater runoff endangers both aquatic and terrestrial ecosystems. Despite the rising need to tackle this problem, urban-scale modeling of biocide emissions remains underexplored due to the complexity of biocide behavior, variability in emissions, and limited knowledge on biocide stocks. The main objective of this research is to couple a biocide emission model with the Town Energy Balance(TEB) model, a distributed hydroclimatic model, to simulate urban-scale emissions and their contributions to stormwater systems. This paper develops and evaluates a methodological framework for distributed modeling of biocide emissions from buildings’ façades at an urban scale. The developed model calculates the quantity of biocides released from building facades by wind-driven rain(WDR) at an hourly timestep. The application of the developed model is done on 2 residential areas. A single-mesh analysis using Monte Carlo simulations is performed to examine the parameters impact on total emissions. Then, a multi-mesh analysis is used, where meshes are incrementally added until the average emission per mesh stabilizes. This work determines the applicability of the model and the appropriate spatial scale for urban modeling of biocide runoff, hence providing reliable emission estimates at the urban scale.

Surface water quality is increasingly at risk due to anthropogenic activities and climate change, leading to issues such as eutrophication that threaten aquatic ecosystems and human well-being. This study harnesses the power of Artificial Intelligence (AI), specifically deep learning and domain adaptation techniques, to predict nitrate concentrations using readily measurable parameters such as electrical conductivity (EC), pH, and temperature. We propose the Multi-Domain Adaptation for Regression under Conditional Shift (DARC) framework, designed to tackle data scarcity and marginal shifts between catchments. By incorporating a Modified Pairwise Similarity Preserver (MPSP) loss function, our model achieved an NSE value of 0.44 using only seven data points from the target dataset, outperforming traditional linear regression, which failed to reach comparable performance even with more than 20 data points. This study highlights the potential of AI-based domain adaptation methods as cost-effective, scalable solutions for water quality monitoring, addressing global environmental challenges through improved prediction and management of surface water resources

The removal of trace-level organic chemicals (TrOCs) from stormwater is critical due to their persistent, toxic, and mobile nature. Stormwater biofilters have shown promise in mitigating TrOCs through physical, chemical, and biological mechanisms. However, detecting TrOCs is time- and resource-intensive, underscoring the need to identify effective surrogates for real-time monitoring. This study evaluates the predictive potential of 11 surrogates, including nine water quality parameters (e.g., Total Organic Carbon (TOC), Dissolved Oxygen (DO), UVA₂₅₄) and two operational parameters (e.g., oxidation-reduction potential (ORP) and soil moisture), in stormwater biofilters under varying rainfall events. Using machine learning (ML) models such as Random Forest and XGBoost, acceptable prediction accuracy was achieved (R² > 0.5) for TrOCs, e.g., Caffeine, DEET, and Diuron. Additionally, a LSTM deep learning model was introduced, leveraging soil moisture and operational time (e.g., water release schedules) to predict ORP under different Real Time Control (RTC) biofilters (with averaged testing NSE above 0.7). Integrating parameters such as soil moisture and ORP, alongside operational time, provided new insights into biofilter behaviour, enabling accurate and efficient TrOCs removal predictions. This approach significantly advances automating biofilter performance monitoring, reducing the reliance on labour-intensive sampling, and enhancing the practical applicability of biofilters in stormwater management.