Trade-offs Between Discretization Approaches in Urban Stormwater Modeling: Accuracy, Interpretability, and Practical Implications

Zhaokai Dong, Sabrina Jivani, Pradeep Goel, Clare E. Robinson

Stormwater models are important tools for urban drainage design, planning, and analysis, but their performance and interpretation depend heavily on how spatial discretization is handled. This study evaluates the influence of two common discretization strategies – topography- and sewer geometry-based – on hydrological representation and simulation accuracy in the Storm Water Management Model (SWMM), using a mixed urban and peri-urban watershed in London, ON, Canada. Leveraging long-term flow data from multiple monitoring locations across the watershed, we systematically evaluated the effects of discretization strategy across different rainfall conditions and land use settings (e.g., urban vs. peri-urban) using continuous and event-based simulations, as well as a fixed-effects regression model. The two models with different discretization approaches showed no significant differences in simulating outlet flows, indicating that discretization choice had limited impact on outlet flow simulations. However, the topography-based model yielded parameter values with greater hydrological interpretability and, accordingly, performed better at simulating flows at locations within the watershed. In addition, model performance was strongly influenced by rainfall depth and land use characteristics, with significantly improved results observed during larger storm events and in the urban watershed. The strengths and limitations of the two discretization approaches are laid out based on the study findings. Ultimately, the study demonstrates that discretization choice can significantly influence model structure, parameter interpretation, and spatial simulation accuracy, particularly in watersheds with heterogeneous topography and mixed drainage systems, and should therefore be carefully considered in stormwater modeling and scenario planning.