LAUSR

Pollutant concentrations in influents into constructed wetlands (CWs) are highly fluctuating and may vary over several orders of magnitude, leading to large uncertainties in removal performance assessment when using pollutant concentrations in the influent and effluent directly. Incorporating a probabilistic approach into removal performance assessment and needed area estimation of CWs could advantage decision making regarding wastewater treatment and engineering applications. A series of three-stage surface-flow CWs (SFCWs) were constructed for treating ammonium-rich swine wastewater. The surface removal rate and removal efficiency of ammonium nitrogen in the SFCWs using the probabilistic approach were 0.27-3.23 g m-2 d-1 and 43.0-99.9% (95% confidence interval (CI)), which were consistent with the deterministic approach (95% CI: 0.24-3.18 g m-2 d-1 and 70.4-99.9%). The needed SFCW area was estimated as 6.6 (95% CI: 1.4-17.8) to 29.7 (95% CI: 6.4-80.1) m2 for required removal efficiency from 40% to 90% for 0.18 m3 d-1 swine wastewater with different strengthens. For specific removal efficiency of 90%, the needed CW areas was 13.9 (95%CI: 4.9-42.7), 25.1 (95%CI: 5.9-66.0), 33.5 (95%CI: 13.5-87.1), and 40.8 (95%CI: 16.2-89.4) m2 for influent ammonium loading rate of 0.18-2.7, 2.7-14.4, 14.4-36, and 36-60 g d-1, respectively. The first-order removal constant of ammonium nitrogen decreased logarithmically with increasing influent and effluent concentration/loading rate in the SFCW units (p < 0.001), which was responsible for the needed SFCW areas covering a wide range. The reliability analysis confirmed the results from the probabilistic approach were appropriate. The present study shed new lights on the performance evaluation and design of CWs for treating wastewater with highly-fluctuating concentrations using a probabilistic approach.