ABSTRACT
Horizontal subsurface flow constructed wetlands have a variety of complex and interrelated physical, chemical, and biological processes, so the mathematical representation for these processes is difficult. It was decided to use artificial neural networks (ANNs) and statistical analysis (SPSS) in this study. Artificial neural networks were used for modeling the input variables and forecast the output concentrations. Statistical analysis for the two stages was modeled using stochastic package for social science. Several regression equations were tested and the best ones were chosen. The best equation which gave the good convergence of the target about the line of perfect agreement was applied. A comparison between the experimental measured data and both ANNs and SPSS results are presented and the best modeling is chosen to represent these data. It is important to model the pollutants in set up stage as the porosity of all cells decreases with time from start of operation and the wetland plants and attached biofilm are growing.
