Modeling bioretention hydrology: Quantifying the performance of DRAINMOD-Urban and the SWMM LID module

Abstract

• Bioretention volume and hydrograph performance of two calibrated models was compared. • SWMM produced good predicted volumes and outflow hydrographs even when uncalibrated. • DRAINMOD-Urban better represented measured drainage hydrograph shape than SWMM. • Hydrograph-calibration was closest to measured hydrographs and volumes in DRAINMOD-Urban. • In SWMM, the calibration method optimized measured hydrographs or volumes but not both. Bioretention systems have become a leading infiltration-based Low Impact Development (LID) practice to reduce urban stormwater runoff volumes and peak flows. Although these systems have performed well in many site-scale field studies, modeling of bioretention systems has received less attention. Additional studies are needed which calibrate various models to field measurements to investigate and optimize the performance of individual LID practices and effectively scale local interventions to the watershed. DRAINMOD-Urban has been successfully applied to bioretention at the site-scale due to its advanced soil–water accounting using the soil–water characteristic curve and its ability to explicitly model underdrains and internal water storage (IWS) zones. At the same time, the U.S. EPA Stormwater Management Model (SWMM) has become one of the most widely used urban drainage models. The latest version, SWMM5, included dedicated LID modules including a routine for bioretention modeling. In this study, DRAINMOD-Urban and the SWMM LID module were compared through detailed analysis of the internal processes of each model as well as through model calibration and output investigation. The objective was to identify the strengths and weaknesses of each model and compare the performance of both models to a single bioretention cell. Both SWMM and DRAINMOD-Urban were evaluated in calibrated and uncalibrated scenarios since urban drainage models often remain uncalibrated for planning scenario analysis. Following calibration, DRAINMOD-Urban was superior for replicating drainage hydrographs (NSE = 0.60) while SWMM produced better overflow hydrographs (NSE = 0.57). Specifically, SWMM often output a maximum drainage rate that caused rectangular drainage hydrographs, but DRAINMOD-Urban was better able to match the shape of measured drainage hydrographs. While the DRAINMOD-Urban model output was in good agreement with measured drainage and overflow event volumes when calibrated (drainage NSE = 0.83, overflow NSE = 0.57–0.66), SWMM was closer to measured volumes even when uncalibrated (drainage NSE = 0.70–0.93, overflow NSE = 0.59–0.81). This study improved existing knowledge of the SWMM LID module by calibrating to field-collected data from a single bioretention cell for the first time in literature. Furthermore, the results of this study indicate an opportunity for model coupling that could combine the strengths and weaknesses of each model and improve bioretention cell modeling.