Main content area

Data-driven modeling of phosphorus (P) dynamics in low-P stormwater wetlands

Juston, John M., Kadlec, Robert H.
Environmental modelling & software 2019 v.118 pp. 226-240
algorithms, data collection, equations, models, phosphorus, pollution load, stormwater, stormwater management, time series analysis, water treatment, wetlands
The ability to simulate wetland phosphorus (P) concentration dynamics in the low-P domain (<20 μg/L of total P) has been poor. This paper develops a new model algorithm to describe P removal in Stormwater Treatment Areas (STAs) for Everglades restoration. It includes a performance review of previous low-P models. Analysis of inflow-outflow pulse responses at daily and monthly time steps in the STAs revealed surprisingly counterintuitive dynamics (lowest outflow concentrations occurring during highest flow events). Analysis of internal P concentration and flux data (from within STAs) informed new flow-dependent P removal and zero-order return flux formulations. The resulting parsimonious model is based directly on these two core equations. Calibration and testing are presented, using four independent datasets (6–36 month duration) including outflow P time series and internal water treatment profiles. This data-driven approach significantly improved dynamic simulations (R2 = 0.4–0.7) and identified the determinant role of internal P loading in low-P wetlands.