Bayesian Analysis Of Stormwater Quality Treatment: Application To A Surface Sand Filter And A Subsurface Gravel Wetland

In a previous study of the same research project, Avellaneda et al. (2010) reported the application of a transport model for simulating contaminant removal by a sand filter. The model was based on the one-dimensional advection-dispersion equation. The unknown parameters of the model were the contaminant deposition rate and the hydrodynamic dispersion. The model was calibrated on a storm by storm basis and optimized parameter values were provided for 15 rainfall-runoff events. Although a statistical summary of optimized parameter values was provided, they recommended these results as input information for the implementation of a more rigorous statistical approach towards the description of parameter uncertainty. In this study, a Bayesian statistical approach is used to determine parameter uncertainty for the stormwater treatment model and monitoring data reported by Avellaneda et al. (2010). The objectives of this study are: (1) to develop a Bayesian stormwater quality approach for two stormwater treatment systems: a subsurface gravel wetland and a sand filter; (2) to determine the posterior probability distribution (PPD) of the deposition rate and the hydrodynamic dispersion; and (3) to perform Monte Carlo simulations to estimate effluent pollutant concentrations from the stormwater systems using the calculated PPDs. Two stormwater treatment systems were selected for this study: a sand filter and a subsurface gravel wetland. Both systems are located at the University of New Hampshire Stormwater Center in Durham, New Hampshire (USA). Influent and effluent for these two stormwater treatment systems were monitored between August 2004 and September 2006. A total of 15 storms were collected for the subsurface gravel wetland and 16 storms for the sand filter. Runoff constituent analysis included: TSS, TPH-D, and Zn. Results indicate that the mean particle deposition rate of the sand filter is higher than that of the subsurface gravel wetland. The deposition rate parameter quantifies the removal characteristics of the system. However, the overall performance of the studied systems is also a function of design characteristics such as cross-sectional area and filter length. Results also indicate higher dispersion for the subsurface gravel wetland than for the sand filter. Higher hydrodynamic dispersion values for the subsurface gravel wetland could be explained by the higher particle size of its filter media. Avellaneda, P., Ballestero, T., Roseen, R., and Houle, J. (2010). Modeling urban stormwater quality treatment: Model development and application to a surface sand filter. Journal of Environmental Engineering., 136(1), 68-77.