Modelling the Hydrological Performance of Stormwater Ponds Using Observed Water Level Data

Abstract

Stormwater modelling software operates on fundamental hydraulic and hydrological principles to predict and simulate the flow and movement of stormwater in large catchments. There are significant gaps in studies that use both modelling and field measurements to evaluate the water quantity control performance of stormwater management (SWM) ponds many years after construction. This study aims to provide new insights into the ability of design hydrologic models to predict the actual performance of aging SWM ponds. Models created during the design and planning stages of two SWM ponds in Vaughan, Canada were re-created in Visual OTTHYMO (VO), and modelled water levels were compared against observed pond depths for eight rainfall events at each pond. The results showed that 50% of the 16 modelled runs reasonably reflected the observed conditions, with 88% (14 of 16 events) achieving normalized root mean square error (RMSE) values below 0.5 and 50% attaining R2 values greater than 0.7. A one-way sensitivity analysis showed that total impervious area (TIMP), modified curve number (CN*), and final infiltration rate (Fc) had a significant impact on the modelled peak runoff rates. To improve model performance, measuring the onsite discharge and re-creating the stage- storage-discharge relationship for as-built conditions is recommended.