An evaluation of methodologies for the generation of stochastic hydraulic conductivity fields in highly heterogeneous aquifers

Stochastic techniques, such as Monte Carlo experiments, are more and more f requently used for the study of flow and transport in heterogeneous aquifer s. When the aquifer is composed of distinct hydrofacies, a common way to mo del heterogeneity is to first generate equally-possible hydrofacies fields, and then convert these hydrofacies fields into hydraulic conductivity (Kj fields by assigning a single K value to each facies. This technique assumes relative homogeneity of K within each facies but may not be appropriate fo r the most conductive facies that often exhibits substantial variability. I n this paper, we assessed the impacts of assigning multiple random K, rathe r than a uniform K value, to the highly conductive facies on the results of a flow and transport model. A set of fifty stochastic hydrofacies maps dep icting an environment similar to the Snake River Plain aquifer (SRPA) in so uth-east Idaho were generated. Simulations demonstrated that a uniform K va lue, if carefully chosen, can reasonably reproduce the specific discharges and early particle arrival times produced by multiple K values. Yet, the re sults obtained with a uniform K value are dramatically less variable than t hose obtained with multiple K values. It is therefore concluded that stocha stic simulations with uniform K assigned to the most conductive and variabl e facies do not necessarily portray the entire uncertainty in the analyses.