Predicting Performance for Constructed Storm-Water Wetlands

Abstract

Water quality treatment via constructed storm-water wetlands (CSWs) is intimately linked to system hydraulics. Previous works have attempted to define the relationship between performance and wetland design variables (e.g., length, width, area). However, these works suffer from two major flaws: small sample size and/or nonrandom samples. The authors provide a framework herein to overcome these flaws. The goals of this research were to develop a methodology for creating randomly generated wetland designs and to use these designs to develop a set of equations for predicting peak flow reduction. Two thousand randomly generated wetland designs were generated using a five-tiered approach. Channel length and roughness were highly correlated with peak flow reduction and explained 83% of the total variability within the data set. Because of the large number of randomly generated designs, the regression equations presented herein prevent bias toward nonrandom designs. These equations represent the most general predictive performance equations developed to date and can be used to aid in CSW design.