Bayesian Chemical Mass Balance Method for Surface Water Contaminant Source Apportionment

A Bayesian chemical mass balance (CMB) source apportionment method is developed using the Markov Chain Monte Carlo (MCMC) approach. Compared with deterministic approaches, the Bayesian method is capable of accounting for the measurement errors and the impact of variability of the source elemental compositions resulting from the heterogeneities and estimate the uncertainties associated with the estimated source contributions. The method estimates the joint probability densities and consequently, the credible intervals and correlation matrices of source contributions of various sources into a receiving water using observed elemental profiles of samples from both potential sources and the receiving surface waters. The model is applied to samples collected from possible sources and runoff and stream flow from two stream crossing sites along Highway 89 in the Lake Tahoe Basin. The contributing sources of total dissolved nitrogen, total dissolved phosphorus concentrations, and microparticles (<20 mu m) from traffic and non-traffic-related sources have been evaluated using the method. The results showed that the model is capable of predicting the source contributions for the dissolved and particulate samples. During both rain events and snowmelt, deicing salt is a major source of dissolved solids whereas vegetation was to be the major source of dissolved nitrogen and phosphorus. Soil was found to be the primary source of microparticles and particulate phosphorus. The method does not definitively show contribution from traffic-related sources; however, in some cases it cannot conclusively rule out a significant contribution from them. The observed credible intervals for dissolved constituents were narrower compared with microparticle samples. DOI: 10.1061/(ASCE)EE.1943-7870.0000645. (C) 2013 American Society of Civil Engineers.