Considering the effect of groundwater on bioretention using the Storm Water Management Model

Abstract

The Storm Water Management Model (SWMM), with its recently released low impact development (LID) module, is among several models used for the performance evaluation of LID facilities in reducing runoff and pollutants. Modeling is often difficult because of the variety of factors affecting the LID system. Among these factors, the effect of groundwater can be important in the LID modeling results due to the possibility of its interaction with LID. In this study, the performance of the SWMM-LID controls in predicting runoff from bioretention cells was evaluated for a site under groundwater influence. In addition, for considering the groundwater effect in the model, this study explores the utility of the SWMM groundwater model in predicting runoff under groundwater influence. Runoff from the considered watershed draining into the bioretention cells was well-simulated with very favorable performance statistic values (r2 = 0.96, NSE = 0.94, % difference = 2.76). However, comparison of simulated with observed runoff from bioretention cells produced weaker statistical values (r2 = 0.69, NSE = 0.65, % difference = 18.22), which is thought to be due to the presence of events affected by groundwater interference. Removal of these events and recalibration were able to improve the overall results, suggesting that the influence of groundwater should be taken into account for better LID modeling of the study site. In order to consider the groundwater influence, the SWMM groundwater model was used in tandem with LID controls to provide an additional influent source to bioretention cells. This resulted in a good fit for two events which were thought to be impacted by groundwater (events in which outflow exceeded inflow) and overall better performance (r2 = 0.95, NSE = 0.95, % difference = 3.49) compared to the results obtained by using only LID controls. In conclusion, the SWMM groundwater model can help deal with groundwater-impacted events. However, for better representation of the phenomenon, the LID module itself needs to be improved to account for direct interaction with groundwater.