Uncertainty analysis in quantification of near source zone mass discharge.

Mass discharge, defined as the contaminant mass per unit time ([M/T]) migrating across a hypothetical control plane orthogonal to the mean groundwater flow, is an important metric for the assessment of potential benefits of dense nonaqueous phase liquid (DNAPL) source zone depletion, design of source - plume remediation systems, and estimation of natural attenuation rates and the effectiveness of biodegradation for many contaminants. Mass discharge estimated in the field is always subject to uncertainty; however, the uncertainty is typically not accounted for in analyses. Incorporating accurately quantified uncertainty of mass discharge is critical for a proper application of this metric in decision making processes. This research systematically investigates the uncertainty of mass discharge estimated in the field using measurements obtained with the Multilevel Sampling method (MLS) and the Passive Flux Meter method (PFM). For each measurement technology, a geostatistical approach is developed to quantify mass discharge and the associated uncertainty using existing field measurements. A multi-stage multi-criteria spatial sampling algorithm is also developed to collect field measurements more efficiently to improve the accuracy and precision of the mass discharge uncertainty model through the delineation of hot spots. The developed methodologies are applied on numerically simulated plume data that represent both mildly heterogeneous and highly heterogeneous aquifer systems. Results for one-stage sampling using a regular sampling design show that, even for mildly heterogeneous systems, as much as 6-7% of the transect needs to be sampled to provide an accurate model of uncertainty. This requirement is encountered for transects in which the concentration distribution is characterized by a small number of 'hot spots'. The required sampling effort is even higher for highly heterogeneous systems. It is unrealistic to assume that this level of sampling can be implemented in the field. Thus, this finding suggests that most field applications to date may not have been based upon a sample size sufficient to accurately quantify the uncertainty of mass discharge, and the estimated mass discharge may have large errors. The developed sampling algorithm can help to reduce the required sampling density by as much as 50%, depending upon the number and magnitude of hot spots.Ph.D.Applied SciencesEarth SciencesEnvironmental engineeringHydrologic sciencesUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/126945/2/3287566.pd