Determining the entrapment architecture in the DNAPL source zone using down- gradient mass flux measurements: A combined numerical and stochastic study Satawat Saenton and Tissa H. Illangasekare Colorado School of Mines, ssaenton@mines.edu, tillanga@mines.edu, Golden, CO, USA ABSTRACT The presence of dense non-aqueous phase liquids (DNAPL) in source zones in the subsurface generates continuous mass flux long after the initial spill. Observed dissolved concentration in a screened monitor- ing well downstream of a DNAPL source zone will provide very little information on the entrapment archi- tecture that is needed to design effective remediation schemes. Non-intrusive techniques based on tradi- tional partitioning tracer methods will only provide volume estimates that are in error under complex en- trapment architecture that may include pools. This paper presents an alternate non-intrusive technique where measured spatial distribution of mass flux emanating from the source zone could be used to get knowledge of entrapment architecture that contributes to the generation of the plume. Determination of locations where mass flux is generated within the source zone will assist in targeting the hotspots for DNAPL mass removal through remediation. It is our hypothesis that mass flux emanating from the en- trapment zone is controlled largely by the vertical distribution of the DNAPL. The proposed method analy- ses measured vertical distribution of mass flux using a modified mass transfer model based on MOD- FLOW and RT3D and inverse modeling using UCODE to determine DNAPL entrapment morphology as well as the hydrodynamically accessible mass in the source zone. This theoretical study uses subsurface stochastic methods to address issues related to heterogeneity within the source zone. INTRODUCTION Non-aqueous phase liquids (NAPLs) such as gasoline and chlorinated solvents are common organic compounds found in contaminated soils and aquifers [Mercer and Cohen, 1990]. These organic liquids in the entrapment zone slowly dissolve into the flowing ground water generating a downstream contaminant plume in which concentrations are usually found to exceed regulatory standards. Mass transfer limitations due to rate-limited dissolution result in a long-term persistence of the source even in cases of small en- trapped DNAPL mass. The identification of DNAPL sources located beneath the water table is critical to achieve goals of site remediation and aquifer restoration. Traditional soil coring is not widely used in DNAPL contaminated sites since it is considered to be intrusive and results in cross contamination and remobilization of entrapped DNAPLs. In addition, the estimation of entrapped mass using point data ob- tained from cores is strongly sensitive to the choice of interpolation techniques and the estimates tends to associated with high uncertainty [Pankow and Cherry, 1996]. The emergence of partitioning interwell tracer tests offers an in situ, non-intrusive site characterization method that minimizes the risk of remobilization of DNAPL [Jin et al., 1995]. This technique utilizes con- servative and suite of partitioning tracers to detect the presence and to estimate the hydrodynamically ac- cessible DNAPL mass in the entrapment zone [Rao et al., 2000]. Results obtained from this technique provide only the average saturation and total DNAPL volume in the tracer-swept zone. In addition, analy- ses of the tracer data are limited to the use of method of temporal moment where equilibrium partitioning is assumed. This may not be the case when higher saturation zones of DNAPLs are encountered be- cause the limited contact time between flowing aqueous phase containing tracers and DNAPL is not suffi- cient for equilibrium partitioning to take place. As a result, DNAPL volume is always underestimated [Dai et al., 2001]. Current research by Moreno-Barbero and Illangasekare [2003] is attempting to improve the accuracy of this technique in detecting and distinguishing DNAPL in pools and intermediate saturation zones as well as estimating the volume. The methods they have developed treat a pool to have two zones: a zone of high saturation at the bottom bounded at the top by a transition zone where NAPL satu- ration decreases gradually to residual. Preliminary results indicate that multiple tracers test detect por- tions of DNAPL which is mostly in the transition zone whereas DNAPL in the high saturation zone is by-