Determination of regression model parameter for constructed wetland using operating data

Seven years of performance data from a free surface flow constructed wetland system receiving agricultural runoff were used to determine treatment performance and to develop regression and wetland design models. Removal rates by the wetland were 21–43.6% for 5-day biochemical oxygen demand (BOD₅), 49.0–58.1% for total phosphorous (TP), 24.1–46.0% for total nitrogen (TN), and 57.6–77.8% for total suspended solids (TSS). First-order area-based rate constant (k ₂₀) values for BOD₅ were 15.48 m/year in the early stage of observation and decreased to 12.00 m/year for the stable period. Similar results were found for TP, for which k ₂₀ values were 18.72 m/year in the early stage and 14.92 m/year for the stable period. For TN, k ₂₀ values in the early stage (21.32 m/year) were slightly lower than those for the stable period (38.02 m/year). Finally, TSS had values of 132.4 and 172.6 m/year in the early and stable periods, respectively. The low k ₂₀ for BOD₅ was not important for nonpoint source pollution control in the constructed wetland because these kinds of wetlands mainly focus on nitrogen and phosphorus retention. The wetland area and outlet concentration could be approximately predicted using the first-order kinetic model, but the maturity and hydraulic loading rate should be considered for more accurate prediction.