Monitoring and evaluation of constructed wetlands for wastewater treatment

To better understand how constructed wetland treatment systems work, the treatment performance of four constructed wetlands (two overland flow and two subsurface flow) was evaluated from December, 1994 through December, 1995. The wetlands provide tertiary treatment of a Southeast Tennessee food processing plant's wastewater. All four wetlands were established with a variety of wetland plants. Wetland influent and effluent were sampled weekly. Samples were tested for nitrate, ammonia, TKN, BOD₅ and other pollutants. Multiple samples were collected to maintain quality control.

Bench-scale test cells were established on-site. Each (1.5 m x 0.5 m) (5 ft by 1.7 ft) bench-scale cell corresponded to a full-scale wetland, and was set up with similar plant populations, plant types and horizontal velocities. The bench-scale cells were sampled in a similar manner to the full-scale wetlands.

The results show that a high degree of treatment is being realized for the measured pollutants. The system appeared to be lightly loaded for the pollutants measured. During the course of the study the average loading for nitrate, ammonia, TKN and BOD₅ was 25,17,17, and 34% (respectively) of the maximum observed loading. Even during the periods of maximum loading, the system achieved high removals for the monitored compounds.

Air, substrate and water temperatures, photosynthetically active radiation (PAR), and precipitation were monitored in order to correlate treatment performance with environmental conditions. However, hydraulic loading to the wetlands varied considerably due to disturbances in the wastewater treatment plant that provides influent to the constructed wetlands. Because of these variations, correlations between treatment performance and environmental conditions could not be made. However, the environmental data was used in calculating evapotranspiration (ET), which was needed in order to compute areal mass removal rates (AMRR).

No significant differences in treatment performance between plant types or wetland types were seen. The bench-scale cell performance appeared to be slightly lower (per unit area) than the full-scale marshes, but generally followed the same removal trends. The removal kinetics of nitrate, TKN, ammonia, and BOD₅ fit a first order model; i.e., the removal rates went up in proportion to inlet concentration.

In order to determine flow characteristics through the two wetland types, a tracer test was performed. Results of the tracer test show that the soil marsh tested exhibited plug-flow. The gravel marsh tested also exhibited plug-flow characteristics, but with some short circuiting.