[1] The estimation of mass discharges from contaminated sites is valuable when evaluating the potential risk to down-gradient receptors, when assessing the efficiency of a site remediation, or when determining the degree of natural attenuation. Given the many applications of mass discharge estimation, it is important to quantify the associated uncertainties. Here a rigorous approach for quantifying the uncertainty in the mass discharge across a multilevel control plane is presented. The method accounts for (1) conceptual model uncertainty using multiple conceptual models and Bayesian model averaging (BMA), (2) heterogeneity through Bayesian geostatistics with an uncertain geostatistical model, and (3) measurement uncertainty. Through unconditional and conditional Monte Carlo simulation, ensembles of steady state plume realizations are generated. The conditional ensembles honor all measured data at the control plane for each of the conceptual models considered. The probability distribution of mass discharge is obtained by combining all ensembles via BMA. The method was applied to a trichloroethylene-contaminated site located in northern Copenhagen. Four essentially different conceptual models based on two source zone models and two geological models were set up for this site, each providing substantially different prior mass discharge distributions. After conditioning to data, the predicted mass discharge distributions from each of the four conceptual models all approach each other. This indicates that the data set available at the site is strong and that the estimated mass discharge is robust to the assumed conceptual models and their boundary conditions. On the basis of the results, we discuss which of the conceptual models is most likely to reflect the true site conditions and analyze the relative proportions and importance of uncertainties as well as the impact of different data types on mass discharge uncertainty.

No abstract

Data obtained from a field study of an aquifer contaminated by landfill leachate and related laboratory experiments suggest that natural attenuation of phenoxy acid herbicides such as mecoprop (MCPP) occurs in the transition zone between the anaerobic plume core and the overlying aerobic water body. The location of this transition zone is assumed to be primarily controlled by vertical transverse dispersion processes occurring downstream of the pollution source. A reactive transport modeling study was carried out to evaluate this conceptual model. The transport was simulated for a two-dimensional vertical cross section to quantify the combined physical, geochemical, and microbial processes that affect the fate of the phenoxy acid herbicides. The simulations, showing removal of phenoxy acids, an increase of phenoxy acid degraders in the fringe zone, and a dependency of the results on vertical transverse dispersivity, are compatible with the hypothesis of fringe-controlled aerobic biodegradation of the phenoxy acids.

Abstract. A continuous, natural gradient, field injection experiment, involving six herbicides and a tracer, was performed in a shallow aerobic aquifer near Vejen, Denmark. Bentazone, (_+)-2-(4-chloro-2-methylphenoxy) propanoic acid (MCPP), dichlorprop, isoproturon, and the dichlobenil metabolite 2,6-dichlor-benzamide (BAM) were injected along with 2-methyl-4,6-dinitrophenol (not discussed in this paper) and the tracer bromide. The injection lasted for 216 days and created a continuous plume in the aquifer. The plume was monitored in three dimensions in 96 multilevel samplers of 6-9 points each for 230 days, with selected individual points for a longer period. The bromide plume followed a complex path through the monitoring network downgradient of the injection wells. The plume movement was controlled by spatially varied hydraulic conductivities of the sand deposit and influenced by asynchronous seasonal variation in groundwater potentials. An average flow velocity of 0.5 m/d was observed, as depicted by bromide. Bentazone, BAM, MCPP, and dichlorprop retardation was negligible, and only slight retardation of isoproturon was observed in the continuous injection experiment and a preceding pulse experiment. No degradation of bentazone was observed in the aerobic aquifer during the monitoring period. BAM and isoproturon were not degraded within 5 rn downgradient of the injection. The two phenoxy acids MCPP and dichlorprop were both degraded in the aerobic aquifer. Near the source a lag phase was observed followed by fast degradation of the phenoxy acids, indicating growth kinetics. The phenoxy acids were completely degraded within 1 rn downgradient of the injection wells, resulting in the plumes being divided into small plumes at the injection wells and pulses farther downgradient. During the lag phase, phenoxy acids had spread beyond the 25 rn long monitoring network. However, the mass of the phenoxy acids passing the 10-25 rn fences never matched the corresponding bentazone or bromide masses, and the pulse was observed to shrink in size. This indicates that this pulse of phenoxy acids was being partially degraded at a low rate as it traveled through the aquifer.