Imaging reactive minerals in the subsurface using inverse reactive transport modeling: an example relevant for arsenic mobilization

Wednesday, 17 December 2014
Sarah Fakhreddine1, Jonghyun Harry Lee1, Peter K Kitanidis1, Scott E Fendorf1 and Massimo Rolle1,2, (1)Stanford University, Stanford, CA, United States, (2)University of Tuebingen, Tuebingen, Germany
The spatial distribution of naturally occurring, arsenic-bearing minerals in the subsurface is a key factor for determining the fate and transport of arsenic in groundwater systems. However, the direct measurement and estimation of these heterogeneously distributed minerals are often costly and difficult to obtain. While previous studies have shown the utility of using indirect measurements combined with inverse modeling techniques for tomography of physical properties including hydraulic conductivity, these methods have seldom been used to image geochemical properties. In this study, we use synthetic applications to demonstrate the ability of inverse modeling techniques to image reactive mineral lenses in the subsurface and quantify estimation error using indirect, commonly measured groundwater parameters. Specifically, we simulate the mobilization of arsenic via kinetic oxidative dissolution of arsenopyrite due to dissolved oxygen in the ambient groundwater. The modeling investigation is carried out at various scales and considers different flow-through domains including (i) a 1D lab-scale column (80 cm), (ii) a 2D lab-scale setup (30 cm x 60 cm) and (iii) a 2D field-scale domain (4 m x 20 m). Next, the simulated dissolved oxygen data and forward reactive transport model are used to image the spatial distribution of arsenopyrite using the Principal Component Geostatistical Approach (PCGA) for inverse modeling. The PCGA is a matrix-free geostatistical inversion approach that uses the leading principal components of the prior information to save computational costs and can be easily linked with any simulation software. Our results show that the PCGA can be used to image randomly distributed arsenopyrite lenses at various scales and has the potential to be employed at the field-scale to map complex distributions of reactive minerals in the subsurface based on the measurement of dissolved constituents in groundwater.