Lessons learned from the modeling of nature-based solutions for urban flood mitigation in Ottawa, Canada

• Modeled 64 storm-NBS scenarios using 1D-2D Topographic and Lot-level PCSWMM models. • Full-scale BRCs outperformed PRIV and ROW under all storm intensities and models • Compared PRIV vs. ROW BRC placements using total and area-normalized flood metrics. • Showed Topographic models better captured rear-yard flow and NBS placement effects. • Found storm hyetograph timing strongly shaped bioretention performance efficiency. Urbanization has expanded impervious surfaces globally, increasing surface flood risk—a challenge intensified by climate change. Nature-Based Solutions are increasingly adopted to manage urban stormwater and restore hydrologic function. This study evaluates the flood mitigation potential of bioretention cells (BRCs) in two Ottawa catchments called Byron & Golden, and Crystal Beach, using two modeling approaches: (1) Lot-level delineation, assigning subcatchments to individual properties, and (2) Topographic delineation, based on natural drainage patterns. City of Ottawa provided PCSWMM models were refined and calibrated to simulate BRC performance under three scenarios: placement only in the road right-of-way (ROW), only on private property (PRIV), and Full-scale implementation (both). Simulations were conducted for 10-, 50-, and 100-year 4-hour Chicago design storms, as well as for the observed August 2023 flash flood event. Results show that BRCs significantly reduced flood volume, peak discharge, and inundation extent, with Full-scale implementation achieving the greatest reductions. Across all storms and catchments, flood volume reductions ranged from 12 to 56% and peak flow reductions ranged from 12.5 to 63%, while flooded extent reductions ranged from 10 to 75% (Full-scale: 39–75%). PRIV configurations often produced larger absolute reductions due to more distributed runoff capture and infiltration. However, per-hectare analysis revealed that ROW BRCs sometimes had higher efficiency, particularly in Crystal Beach under topographic routing. Topographic models, which routed surface flow over pervious areas, projected greater reductions (in total flood volume, peak flow, and flooded extent) than lot-level models, which routed flow towards impervious roadways. The August 2023 storm, despite similar rainfall depth to the 50-year event (74.6 vs. 70 mm), yielded stronger reductions due to its front-loaded hyetograph—e.g., Full-scale flood-volume reductions were 29–48% in Crystal Beach and 35–56% in Byron & Golden. The findings of this study underscore the importance of distributed BRCs, performance normalization, and testing under both synthetic and observed storms to inform urban flood mitigation strategies.