Abstract
Multiple regression models, principal component analysis, redundancy analysis, and the self-organizing map (SOM) model were applied to assess the effects of physico-chemical parameters on the treatment performance of contaminated wetland sediments in replicate integrated constructed wetland (ICW) mesocosms treating either farmyard runoff or domestic wastewater. Dissolved oxygen concentration and conductivity were correlated with the reduction of ammonia-nitrogen (NH4-N) and/or molybdate reactive phosphorus (MRP), and the redox potential and conductivity were related to chemical oxygen demand (COD) removal. SOM was selected as the prediction model to provide numerical estimations for the performance of ICW mesocosms. The model was validated, indicating that NH4-N, MRP, and COD removal can be predicted by input variables that are quick and cost-effective to measure. The SOM model is an appropriate method for monitoring the performance of mature ICWs as a source of contaminants such as nitrogen and phosphorus. Although no reduction in the overall performance was observed for ICW Sites 7 and 11, this laboratory-scale study provides valuable warning signs regarding the loss of contaminant removal for ICW systems that have been in operation for a relatively long time.
Original language | English |
---|---|
Pages (from-to) | 1009-1017 |
Number of pages | 9 |
Journal | Journal of Environmental Engineering, ASCE |
Volume | 138 |
Issue number | 10 |
DOIs | |
Publication status | Published - 1 Oct 2012 |
Externally published | Yes |
Keywords
- Mathematical models
- Neural networks
- Sediment
- Wastewater management
- Wetland
ASJC Scopus subject areas
- Environmental Engineering
- Environmental Chemistry
- Civil and Structural Engineering
- General Environmental Science