2D hydraulic modelling of a flood vulnerable neighborhood

A two-dimensional (2D) hydraulic model was created to simulate flood conditions in a flood vulnerable community in Toronto, Ontario. The model illustrates how the Regional Storm flood wave propagates through a neighborhood, identifying the sequence of urban and riverine flooding. This information could be used to develop emergency evacuation routes for the neighborhood in the event of a flood. The model could also be integrated within a flood forecasting model to provide real-time flood threat estimation.

The flood vulnerable neighborhood abuts the West Don River, which meanders through the neighborhood via sections of concrete channel and confined valley sections. The contributing upstream drainage area is approximately 87 km2. The neighborhood was developed prior to laws that prevent development in the Regional Storm floodplain. Riverine flooding falls under the jurisdiction of the Toronto Region Conservation Authority (TRCA). The TRCA has floodplain mapping which illustrates the extent of flooding within the neighborhood. The floodplain maps were developed using a one dimensional (1D) steady state hydraulic HEC-RAS model. While the HEC-RAS model defined the extent of flooding, it does not explain how flooding progresses within the neighborhood. A 2D model does not have this limitation as flows are dynamically routed in discrete elements representing the underlying topography and drainage network.

The 2D hydraulic model was created using PCSWMM Professional 2D. Topographic data was generated with 1 m spacing, with an estimated vertical accuracy of 0.5 m. Information within the channel banks including section elevations and crossing details were imported from the HEC-RAS model. The Regional Storm hydrograph was generated using Visual OTTHYMO 2.0 run under the Hurricane Hazel hyetograph. The PCSWMM 2D mesh was created using a grid spacing of 10 m. Existing building footprints are identified within PCSWMM and the grid was generated to form around the buildings.

Preliminary 2D modelling results are generally commensurate with the 1D HEC-RAS model. The extent of flooding between the two models are comparable and with sufficient agreement to confirm the reasonableness of the results. The HEC-RAS model generally had higher flooding depths, which in one area resulted in a wider floodplain. However, it is recognized that 1D steady state modelling is more conservative than unsteady state 2D modelling as the flood wave is not routed and storage behind crossings are not considered.

The 2D modelling yields valuable information on the chronology of flooding within the vulnerable area. This information is of key importance in developing flood response and emergency evacuation plans. Information such as depth and velocity of flooding on roads determine which roads are safe for people and vehicles to traverse. Future works will examine integrating the 2D model into an existing TRCA PCSWMM food forecasting model. This would be a useful tool that could model near-term predicted flood conditions, providing emergency management staff advance information for warning, preparedness and response to flood threats.