Steamboat Creek is contaminated by agricultural and stormwater runoff. Exploratory, descriptive, and statistical modeling methods using the SAS software (version 9.13) were performed for quantifying and predicting the performance of a small-scale constructed wetlands. The performance of ten constructed wetland trains with different combinations of sediment and source water are discussed. Water sampling was conducted twice a month and analyzed for total nitrogen (TN), total phosphorus (TP), total suspended solids (TSS), and total organic carbon (TOC) in the wetlands.; Data exploration using box plots was performed to check for trends, outliers, and distribution patterns. Efficiency of nutrient removal associated with each train, wetland design, seasonal changes, and interaction between wetland designs and seasons was analyzed using Mulitvariate Analysis of Variance (MANOVA). Canonical correlations between operational conditions such as flow, pH, and temperature and the removal of TN, TP, TSS, and TOC were examined. Scatter plots were used to explore the linear trends, significance, and the range of the majority values in a correlation. Multivariate Wilk's Lambda test statistics were used to check for the overall significance in canonical correlations. Following canonical correlations, multiple regression models were developed for TN, TP, TSS, and TOC with respect to predictors flow, pH, and temperature.; The results from the statistical analyses performed on the data from the small-scale constructed wetlands demonstrated that the construction of a large-scale wetlands system would be an effective method to improve water quality in Steamboat Creek by reducing the loadings of TN, TP, TSS, and TOC into the lower Truckee River.
展开▼
