Hydrologic Modeling of Distributed Stormwater Control Measure Retrofit and Examination of Impact of Subcatchment Discretization in PCSWMM

Abstract

Modifications made to the landscape during urbanization can lead to increased stormwater runoff volumes and peak flows. Low-impact development (LID) techniques and green stormwater infrastructure are implemented to minimize the effects of urbanization on downstream environments. The United States Environmental Protection Agency’s Storm Water Management Model (SWMM) can be used to simulate rainfall and runoff and evaluate the performance of existing or proposed urban water infrastructure. This study examined the performance of SWMM models developed using the Personal Computer Stormwater Management Model (PCSWMM) for a residential catchment before and after being retrofit with LID practices. Three preretrofit models at varying resolutions (low, middle, and high) and one postretrofit model were created and calibrated with observed stormwater outflow and rainfall data from a previous study in Wilmington, North Carolina. The LID retrofit included a bioretention cell bump-out and two permeable-pavement parking stalls. The uncalibrated low-resolution model was unsatisfactory for both runoff volume [Nash-Sutcliffe Efficiency (NSE) =0.49] and peak flow (NSE=−0.29) and overpredicted both with a percent bias of 65% and 70%, respectively. After calibration, the three preretrofit models all had very good agreement between observed and predicted results for runoff volume (NSE>0.95) and good agreement for peak flow (NSE>0.85). Runoff volume was still overpredicted by 23% in the middle-resolution model, but the difference in peak flows was minimal at −4%. The level of subcatchment discretization for the preretrofit models had little impact on overall model performance. Following the inclusion of LID retrofits, the postretrofit model was acceptable for event volumes (NSE=0.66) and peak flows (NSE=0.78), but overpredicted volumes by 56% and underpredicted peak flows by 10%. Using observed postretrofit data, the postretrofit model was recalibrated and greatly improved agreement between the observed and predicted data for runoff volume (NSE=0.88) while remaining acceptable for peak flow (NSE>0.78). More specifically, the calibrated postretrofit model reduced the overprediction of event volumes to 21%. These results highlight the need for observed rainfall and runoff data for both preretrofit and postretrofit calibration when modeling in SWMM.