Empirical regression models for estimating nitrogen removal in a stormwater wetland during dry and wet days

Due to the highly variable hydrologic quantity and quality of stormwater runoff, which requires more complex models for proper prediction of treatment, a relatively few and site-specific models for stormwater wetlands have been developed. In this study, regression models based on extensive operational data and wastewater wetlands were adapted to a stormwater wetland receiving both base flow and storm flow from an agricultural area. The models were calibrated in Excel Solver using 15 sets of operational data gathered from random sampling during dry days. The calibrated models were then applied to 20 sets of event mean concentration data from composite sampling during 20 independent rainfall events. For dry days, the models estimated effluent concentrations of nitrogen species that were close to the measured values. However, overestimations during wet days were made for NH3-N and total Kjeldahl nitrogen, which resulted from higher hydraulic loading rates and influent nitrogen concentrations during storm flows. The results showed that biological nitrification and denitrification was the major nitrogen removal mechanism during dry days. Meanwhile, during wet days, the prevailing aerobic conditions decreased the denitrification capacity of the wetland, and sedimentation of particulate organic nitrogen and particle-associated forms of nitrogen was increased.